Spectrophotometry

and Spectrofluorimetry
Practical Approach Series

Related Practical Approach Series Titles

Flow Cytometry 3/e *
High Resolution Chromatography
Affinity Separations
Enzyme Assays

* indicates a forthcoming title

Please see the Practical Approach series website at

http://www.oup.co.uk/pas
for full contents lists of all Practical Approach titles.
 Spectrophotometry
and Spectrofluorimetry
A Practical Approach

Edited by
Michael. G. Gore
Division of Biochemistry and Molecular Biology,
School of Biological Sciences, University of
Southampton.

OXPORD
UNIVERSITY PRESS
OXFORD
UNIVERSITY PRESS

Great Clarendon Street, Oxford 0X2 6DP

Oxford University Press is a department of the University of Oxford.
It furthers the University's objective of excellence in research,
scholarship, and education by publishing worldwide in
Oxford New York
Athens Auckland Bangkok Bogota Buenos Aires Calcutta Cape Town
Chennai Dar es Salaam Delhi Florence Hong Kong Istanbul Karachi
Kuala Lumpur Madrid Melbourne Mexico City Mumbai Nairobi Paris
Sao Paulo Singapore Taipei Tokyo Toronto Warsaw
with associated companies in Berlin Ibadan
Oxford is a registered trade mark of Oxford University Press in the UK
and in certain other countries
Published in the United States by Oxford University Press Inc., New York
© Oxford University Press, 2000
The moral rights of the author have been asserted
Database right Oxford University Press (maker)
First published 2000
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any
means, without the prior permission in writing of Oxford University
Press, or as expressly permitted by law, or under terms agreed with the
appropriate reprographics rights organization. Enquiries concerning
reproduction outside the scope of the above should be sent to the Rights
Department, Oxford University Press, at the address above
You must not circulate this book in any other binding or cover and you
must impose this same condition on any acquirer
A catalogue record for this book is available from the British Library
Library of Congress Cataloguing in Publication Data
Spectrophotometry and spectrofluorimetry : a practical approach /
edited by Michael G. Gore.
(The practical approach series ; 225)
1. Spectrophotometry. 2. Fluorimetry. 3. Biochemistry-Technique.
I. Gore, Michael G. II. Series.
QP519.9.S58 S643 2000 572'.36-dc21 99-057576
ISBN 0 19 963813 6 (Hbk)
ISBN 0 19 963812 8 (Pbk.)
1 3 5 7 9 1 08 6 4 2
Typeset in Swift by Footnote Graphics, Warminster, Wilts
Printed in Great Britain on acid-free paper
by The Bath Press, Avon
 Preface

Optical spectroscopy underpins the day to day operations of most laboratories in

the chemical, biological and medical sciences, and this book is the second
edition of a text, first published in 1987, intended to help the reader understand
the background concepts to spectrophotometry and spectrofluorimetry. The
intervening years have seen a steady development in the design and quality of
laboratory instrumentation in contrast to the dramatic increase in the use of
microprocessors and computers to assist data handling and presentation. It is
therefore only too easy for an inexperienced researcher to obtain processed data
without perhaps appreciating the limitations of the instrumentation or tech-
nique, or the assumptions made in the final data output. An understanding of the
underlying physical principles greatly assists its successful use and helps avoid
experiments which are too demanding for the technique or instrumentation
available.
This edition is intended to build upon and extend concepts established in the
first edition of the series. It therefore contains chapters addressing the principles
of spectrophotometry, spectrophotometric assays, spectrofluorimetry, time
resolved fluorescence and phosphorescence studies, circular dichroism and pre-
equilibrium spectroscopic techniques. In all of the chapters the emphasis is
placed upon practical aspects, with protocols to guide readers through test
experiments, with sufficient theory and references to provide interested readers
with a starting point for more in-depth studies. Other chapters are included to
introduce subjects that have traditionally depended upon spectroscopy, such as
basic enzyme kinetics, ligand binding, data handling and the more recently
established interest in the study of protein and DNA stability. Finally, the
concept of 'global analysis' is introduced to provide the reader with an insight to
this method of utilizing the vast arrays of experimental data provided by current
methodologies.
It is assumed throughout, that readers will have available to them com-
mercially available instrumentation. In these chapters, authors have sometimes
described named instruments used in their studies and laboratories. Readers
should note that these are only given as examples, not as recommendations,
comparable instruments are available from other manufacturers.
PREFACE

Finally, I would like thank the authors for their participation and contributions
to the text.
It was with deep regret that I learned of the demise of Professor Jorge E.
Churchich shortly before the completion of this book. He was a dedicated
scientist who inspired all who worked with him.

Southampton M.G.G.
1999

VI
 Contents

List of protocols page xv

Abbreviations xvii

1 Introduction to light absorption: visible and ultraviolet spectra 1

Robert K. Poole and Uldis Kalnenieks
1 Introduction 1
Radiation and light 1
UV and visible spectra 2
2 Spectrophotometry 2
The Beer-Lambert law 2
Deviations from the Beer-Lambert law 3
Absorbance or light scattering? 4
3 Spectra of some important naturally occurring chromophores 6
Amino acids and proteins 6
Nucleic acids 6
NAD(P)H 7
Carotenoids 7
Haem proteins 7
4 Spectrophotometer configurations 9
Single beam spectrophotometers 9
Split beam or 'double beam' spectrophotometers 9
Dual-wavelength spectrophotometers 10
Multi-wavelength spectrophotometers 13
Diode array spectrophotometers 13
Microwell plate-reading spectrophotometers 16
Reflectance methods 16
Novel double monochromator methods 16
Computing and Spectrophotometry 17
5 Choice of Spectrophotometer operating conditions 17
Wavelength range and light source 17
Spectral versus natural bandwidth 17
Spectral resolution 19
Scan speed and instrument response time 20
Temperature 20

vii
CONTENTS

6 Use of the spectrophotometer 22

The choices 22
Baselines 22
Isosbestic points 24
Wavelength and absorbance calibrations 24
Choice and use of cuvettes (cells) 25
A detailed example: recording of a cytochrome difference spectrum
(reduced minus oxidized) 27
Post-scan options 30
Acknowledgements 31
References 31

2 Fluorescence principles and measurement 33

Arthur G. Szabo
1 Introduction 33
2 Physical principles 33
The absorption process 33
Excited singlet state deactivation processes 35
3 Fluorescence parameters 39
Fluorescence spectrum 39
Fluorescence quantum yield, FF 39
Singlet and radiative lifetime 40
4 Fluorescence spectrometers 40
The light source 42
Wavelength selectors 42
Sample excitation components 44
Sample compartment 44
Fluorescence path optical components 45
Fluorescence instrumentation electronics 46
5 Fluorescence spectra 47
Inner filter effect 47
Light scattering 48
Instrumental settings 50
Fluorescence spectral corrections 51
Excitation spectra 54
Quantum yield measurement 55
6 Fluorescence applications 58
Fluorescence resonance energy transfer 60
Fluorescence anisotropy 60
Protein fluorescence 61
Fluorescence quenching 62
7 Conclusion 66
References 66

3 Time-resolved fluorescence and phosphorescence

spectroscopy 69
Thomas D. Bradrick and Jorge E. Churchich
1 Introduction 69

viii
 CONTENTS

2 Background 69
Basic photophysics and time dependence of fluorescence and phosphorescence
decays 69
Fluorescence and phosphorescence energy transfer and sensitized luminescence 71
Observed time dependence of fluorescence 72
Decay associated spectra (DAS) and discrete lifetimes versus lifetime
distributions 75
Polarized excitation and emission anisotropy decay 75
Data analysis 78
3 Equipment for time-resolved fluorescence measurements 84
Excitation sources 85
Detectors 86
Recording electronics 86
4 Phosphorescence 89
Phosphorescence of proteins 89
Time-dependent phosphorescence anisotropy 93
Acknowledgements 95
References 95

4 Introduction to circular dichroism 99

Alison Rodger and Matthew A. Ismail
1 Introduction 99
Circular dichroism 99
Optical rotataiy dispersion 101
Chapter outline 101
2 Measuring a CD spectrum 103
The instrumentation 103
The sample 103
The cuvette 303
The baseline and zeroing 105
The parameters 105
Noise reduction 107
3 Equations of CD spectroscopy 110
Degenerate coupled-oscillator CD 111
Non-degenerate coupled-oscillator CD 114
Carbonyl n-ir* CD 115
d-d transitions of tris-chelate transition metal complexes 117
Optical activity (optical rotation, OR) 119
Dissymmetry factor 119
4 Units of CD spectroscopy 139
5 Circular dichroism of biomolecules 321
Introduction 121
CD of polypeptides and proteins 321
Protein UV spectroscopy 121
Protein structure determination from CD 323
Determining the percentage of different structural units in a protein from peptide
region CD spectra 125
Other applications of protein CD 328
Membrane proteins 328
DNA geometry and CD spectra 329
UV spectroscopy of the DNA bases 133

ix
CONTENTS

Nucleic acid CD 132

DNA/ligand interactions 134
References 138
General CD references 339

5 Quantitative determination of equilibrium binding isotherms for

multiple ligand-macromolecule interactions using spectroscopic
methods 141
Wlodzimierz Bujalowskt and Maria J. Jezewska
1 Introduction 141
2 Thermodynamic basis of quantitative spectroscopic titrations 143
The signal used to monitor ligand-macromolecule interactions originates from
the macromolecule 344
Signal used to monitor the interactions originates from the ligand 354
3 Summary 363
Acknowledgement 364
References 164

6 Steady-state kinetics 367

Athel Cornish-Bowden
1 Introduction to rate equations, first-order, second-order reactions etc. 167
2 Units 368
3 Basic assumptions in steady-state kinetics 369
4 Measurement of specific activity 3 70
5 Graphical determination of Km and V 372
6 Inhibition of enzyme activity 3 74
7 Specificity 375
8 Activators 376
9 Environmental effects on enzyme activity 3 77
pH 377
Temperature 178
10 Cooperativity 179
11 Experimental conditions for kinetic studies 383
12 Concluding remarks 182
References 382

7 Spectrophotometric assays 383

T. J. Mantle and D. A. Harris
1 Introduction 383
Spectrophotometers 383
Beer-Lambert Law 184
The nature of the sample 384
2 Some general comments on, and practical aspects of, assay design 385
Accuracy and precision 386
 CONTENTS

3 End point and rate assays 187

End point assays 187
Rate assays 188
Rate assays involving amplification 189
4. Spectrophotometric assays for proteins 189
A280 190
The Biuret method 191
The Lowry method 191
The bicinchoninic assay 192
Dye-binding assay 193
Fluorimetric assay 193
5. Spectrophotometric assays for nucleic acids 194
6. Enzyme-based Spectrophotometric assays 195
Some general points on assay design 195
Amount of enzyme required 195
Determination of glucose—a comparison of two methods 197
7. Luminescence-based assays 199
8. Spectrophotometric assays of enzymes 200
Some elementary enzyme kinetics 200
Continuous assays 201
Stopped assays 201
Coupled assays 203
Plate readers 203
Centrifugal analysers 203
9. Spectrophotometric assays for protein amino acid side chains 204
Cysteine 204
Lysine 205
Tyrosine 205
Histidine 206
Tryptophan 206
10. Concluding remarks 206
References 207

8 Stopped-flow spectroscopy 209

M. T. Wilson and J. Torres
1. Introduction 209
2. Features of the basic instrument 210
Instruments available 211
3. Measurement at a single wavelength 212
Setting up 212
Selecting the wavelength for a real experiment 212
The form of a simple progress curve: Making sure the apparatus is mixing and
transferring reactants to the observation chamber rapidly 213
Measurement of the dead time 214
Amplitude of the signal 217
Assigning a signal 218
4. Determining rate constants 220
First-order processes 221
Second-order processes 223

xi
CONTENTS

5. Multiwavelength detection: diode array 'rapid scan methods' 227

SVD (singular value decomposition) 228
Fitting to a mechanism 233
Acknowledgements 239
References 239

9 Stopped-flow fluorescence spectroscopy 241

Michael G. Gore and Stephen P. Bottomley
1. Introduction 241
2. Instrumentation 242
Data collection 242
Instrument calibration, stability and dead time 243
Measuring mixing efficiency 243
Sample preparation 244
Artefacts 245
Temperature effects 245
Density differences between the two solutions 246
3. Factors affecting the sensitivity of the optical system 246
Slit width 246
Selection of wavelength of emitted light 247
Voltage applied to PMT 247
4. Selection of reporter group 249
Intrinsic reporter groups 249
Examples of the use of protein fluorescence to follow binding reactions 250
Use of ligand fluorescence to monitor binding reactions 257
Extrinsic probes 260
Examples of reactions monitored by changes in fluorescence of covalently
attached fluorophores 262
References 263

10 Stopped-flow circular dichrolsm 265

Alison Rodger and Michael J. Carey
1. Introduction 265
2. Instrumentation considerations 266
Available budget 266
Sensitivity required 267
Minimum dead time required 267
Corrosiveness and adsorbance of the samples to be studied 267
Type of flow system: stepper motor or compressed air driven 268
Size and design of the optical cell 268
The number of syringes and mixing stages required 269
Mixing ratios required and whether these need to be variable from experiment to
experiment 269
Sample viscosity 269
Wavelength range for detection, wavelength scanning and bandwidth 270
Software 270
3. Currently available instrumentation 271
Stopped-flow attachment for CD spectropolarimeters 271
Integrated stopped-flow CD systems 273

xii
 CONTENTS

4. Additional experimental considerations 274

Parameters 274
Zero-time and dead-time calibration 275
Baseline 275
System tests 275
5. Examples 278
Stopped-flow CD and lysozyme folding 279
DNA as a catalytic template 280
References 283

11 Spectrophotometry and fluorlmetry of cellular compartments and

Intracellular processes 283
C. Lindsay Bashford
1. Introduction 283
2. Experimental design 284
Apparatus 284
Light sources 287
Wavelength selection 288
Chromophore selection 288
Characterization and calibration of optical signals 292
3. Examples 293
Oxidation-reduction state of tissue mitochondria 293
Membrane potential of cells and organelles 295
pH of cellular compartments 299
Membrane cycling and recycling 303
4. Future prospects 304
Acknowledgements 304
References 304

12 Use of optical spectroscopic methods to study the thermodynamic

stability of proteins 307
Maurice R Eftink and Haripada Matty
1. Introduction 307
2. Basic thermodynamic principles 308
The two-state model 308
Thermal unfolding 309
Denaturant induced unfolding 309
Acid induced unfolding 310
Pressure induced unfolding 330
Simulations 311
Unfolding of oligomeric proteins 311
3. Practical considerations and deviations from the two-state model 316
Existence of equilibrium intermediates 336
Kinetic considerations 316
Irreversibility 317
Baseline considerations 318
Interfering substances 319
Global analysis 320

xiii
CONTENTS

4. Advantages of different spectroscopic signals 320

Absorbance 320
Circular dichroism 322
Fluorescence 323
5. Concluding remarks 325
Acknowledgements 325
References 326

13 The use of spectroscopic techniques In the study of DNA stability 329

John SantaLutia, Jr
1. Introduction 329
2. Overview of UV melting 330
Strengths and weaknesses of UV melting and calorimetry 333
3. Sample 334
Sequence design 334
Redundant design of motifs 335
Sample preparation 336
Choice of buffer 339
4. Instrumentation 343
Microvolume cuvettes and aluminium cuvette adapters 343
Spectrophotometer 344
5. Data analysis 348
Curve fitting to calculate thermodynamic parameters 348
Presentation of normalized absorbance curves 352
Error analysis 353
References 354
A1 List of suppliers 357
Index 363

xiv
 Protocol list

Use of the spectrophotometer

Recording the absolute UV-vis spectrum of a soluble protein in solution 29
Recording the visible difference spectrum of a membrane-bound chromophore in a turbid
sample (cells or membranes) 29
Fluorescence spectrometers
Preparation of reference solution 45
Fluorescence spectra
Generation of spectral correction factors (250-600 nm) 53
Fluorescence applications
Measurement of IF of a sample 59
Acrylamide quenching of protein fluorescence 65
Phosphorescence
Phosphorescence sample deoxygenation 90
Eosine labelling of proteins for time-resolved phosphorescence measurements 94
Measuring a CD spectrum
Measuring a routine CD spectrum 108
Circular dichroism of biomolecules
Application of CDsstr to compute protein secondary structure from CD spectra 126
Thermodynamic basis of quantitative spectroscopic titrations
Fluorescence titration of the DnaB helicase with TNP-ADP 146
Fluorescence titration of eADP with the DnaB helicase 156
Spectrophotometrlc assays for proteins
A Biuret assay to determine protein concentration 190
A Lowry assay to determine protein concentration 191
A bicinchoninic assay to determine protein concentration 192
A dye-binding assay for the estimation of protein concentrations 193
A fluorescence assay for the estimation of protein concentrations 194
Enzyme-based Spectrophotometric assays
An assay for glucose using glucose oxidase 198
An assay for glucose using glucose 6-phosphate dehydrogenase 198

XV
PROTOCOL LIST

Spectrophotometric assays of enzymes

A phosphomolybdate assay for inorganic phosphate released by ATPase activity 202
Measurement at a single wavelength
To test the stability of the detection system 212
To determine the time between initiation of the signal recording and the stop of the
flow of solution 214
Measurement of the 'dead time' of the instrument 216
Determination of a kinetic difference spectrum 219
Determining rate constants
An experiment to determine the effect of a 'pH jump' on the absorption of
carboxymethylated horse heart cytochrome c 222
Multi-wavelength detection: diode array 'rapid scan methods'
Reduction of the metmyoglobin-cyanide complex by sodium dithionite 234
The SVD procedure 236
Instrumentation
To check the mixing efficiency of the stopped-flow instrument 243
Selection of reporter groups
Determination of kobs for the binding of warfarin to human serum albumin 259
Spectrophotometry and fluorimetry of cellular compartments and Intracellular
processes - Examples
Reflectance spectra of freeze-trapped rat kidney 293
Measurement of plasma membrane potential using oxonol-V and cells in suspension 297
Labelling of endosomes with fluorescein conjugates 300
Measurement of cytoplasmic Ca2+ concentration or pH using acetoxymethylester-linked
indicators 302
Basic thermodynamic principles
Observation of the unfolding of a protein 313
Observation of the unfolding of a protein, e.g. lambda Cro 314
The use of spectroscopic techniques in the study of DNA stability - Sample
SEP-PAK desalting 337
Measurement of optical density units 338
Sample degassing 340
Mixing equal concentrations of non-self-complementary strands 341
Sample dilution scheme 342
Instrumentation
Typical melting protocol 345
Sample recovery 346
Cleaning of quartz cuvettes 347

xvi
 Abbreviations

A absorbance, absorption
c, C concentration
D attenuance (A)
DSC differential scanning calorimetry
e molar absorbance coefficient
eATP 1, N6-ethenoadenosine diphosphate
FAD(H) Flavin adenine dinucleotide (reduced form)
Gdn-HCl guanidine hydrochloride
HEPES N-2-hydroxyethylpiperazine-N'-2-ethanesulphonic acid
I intensity of transmitted light
I0 intensity of incident light
ITC isothermal titration calorimetry
l pathlength
MES 2-(N-morpholino)ethanesulphonic acid
MOPS 3-(N-morpholino)propanesulphonic acid
NAD(H) Nicotinamide adenine dinucleotide (reduced form)
OD optical density
PIPES piperazine-N,N'-bis(2-ethanesulphonic acid)
PMT Photomultiplier tube
oF Quantum yield of fluorescence
s second
T transmittance, transmission or temperature
TNP-ATP 2'(3')-o-(2,4,6-trinitrophenyl)adenosine 5'-triphosphate
Tris tris(hydroxymethyl)aminomethane
TRP Tryptophan
TYR Tyrosine
UV ultraviolet

xvii
This page intentionally left blank
 Chapter 1
Introduction to light
absorption: visible and
ultraviolet spectra
Robert K. Poole* and Uldis Kalnenieks1
* Krebs Institute for Biomolecular Research, Dept. of Molecular Biology
and Biotechnology, The University of Sheffield, Firth Court,
Western Bank, Sheffield S10 2TN
f
Institute of Microbiology and Biotechnology, University of Latvia,
Kronvalda Boulevard 4, Riga, LV-1586, Latvia

1 Introduction
1.1 Radiation and light
light is a form of electromagnetic radiation, usually a mixture of waves having
different wavelengths. The wavelength of light, expressed by the symbol \, is
defined as the distance between two crests (or troughs) of a wave, measured in
the direction of its progression. The unit used is the nanometre (nm, 10-9 m).
Light that the human eye can sense is called visible light. Each colour that we
perceive corresponds to a certain wavelength band in the 400-700 nm region
(Figure 1). Spectrophotometry in its biochemical applications is generally con-
cerned with the ultraviolet (UV, 185-400 nm), visible (400-700 nm) and infrared
(700-15 000 nm) regions of the electromagnetic radiation spectrum, the former
two being most common in laboratory practice.

550 600 650

Wavelength (nm)

Figure 1 The visible spectrum. Colours shown are of the light beam at the wavelengths on
the scale so the colour the eye perceives of a reflectant object in 'white' light is that due to
wavelengths that are not absorbed. Thus, the 'green oxidase' (cytochrome d) looks green but
absorbs light in the blue and red regions of the spectrum.
ROBERT K. POOLE AND ULDIS KALNENIEKS

1.2 UV and visible spectra

The wavelength of light is inversely related to its energy (E), according to the
equation:
E= ch/A 1
where c denotes the speed of light, and h is Planck's constant. UV radiation,
therefore, has greater energy than the visible, and visible radiation has greater
energy than the infrared. Light of certain wavelengths can be selectively
absorbed by a substance according to its molecular structure. Absorption of light
energy occurs when the incident photon carries energy equal to the difference
in energy between two allowed states of the valency electrons, the photon
promoting the transition of an electron from the lower to the higher energy
state. Thus biochemical spectrophotometry may be referred to as electronic
absorption spectroscopy. The excited electrons afterwards lose energy by the
process of heat radiation, and return to the initial ground state. An absorption
spectrum is obtained by successively changing the wavelength of mono-
chromatic light falling on the substance, and recording the change of light
absorption. Spectra are presented by plotting the wavelengths (generally nm or
um) on the abscissa and the degree of absorption (transmittance or absorbance)
on the ordinate. For more information on the theory of light absorption, see
Brown (1) and Chapters 2, 3 and 4.

2 Spectrophotometry
2,1 The Beer-Lambert law
The most widespread use of UV and visible spectroscopy in biochemistry is in
the quantitative determination of absorbing species (chromophores), known as
spectrophotometry. All spectrophotometric methods that measure absorption,
including various enzyme assays, detection of proteins, nucleic acids and dif-
ferent metabolites, reside upon two basic rules, which combined are known as
the Beer-Lambert law. Lambert's law states that the fraction of light absorbed by a
transparent medium is independent of the incident light intensity, and each successive layer
of the medium absorbs an equal fraction of the light passing through it. This leads to an
exponential decay of the light intensity along the light path in the sample,
which can be expressed mathematically, as follows:
loglo(I0/I) = kl 2
where I0 is the intensity of the incident light, I is the intensity of transmitted
light, I is the length of the light-path in the spectrophotometer cuvette, and k is a
constant for the medium, which is deciphered by Beer's law. Beer's law claims
that the amount of light absorbed is proportional to the number of molecules of the chromo-
phore through which the light passes. In other words, the constant k is proportional to
the concentration (c) of the chromophore: k = ec, where e is the molar
absorption coefficient, a property of the chromophore itself. e is numerically
 INTRODUCTION TO LIGHT ABSORPTION: VISIBLE AND ULTRAVIOLET SPECTRA

equal to the absorption of a molar (1 mol litre-1) solution in a 1 cm light-path.

The units are litre mol-1 cm-1 or M"1 cm-1, and not cm2 mol-1.
Hence, the expression of the combined Beer-Lambert law is:
Iog10(lo/I) = ed 3
The term Iog10 (Io/I) is called the absorbance (A), synonymous with attenuance
(D). Extinction (E)has been used in the past but is now discouraged. Pathlength
(usually in cm) and wavelength (in nm, without units) are sometimes given as
subscripts, e.g. A1 cm 550. Note that, in difference spectroscopy, a difference
molar absorption coefficient (e) may be defined (2) by the equation:
AA = Aecl 4
where M is the difference in absorption for the chromophore in the two
environments.
Sometimes the passage of light through the cuvette is described in terms
of transmittance, or transmission (T): T = I/I0 and is generally expressed as a
percentage. It is important to note, however, that only absorbance, not
transmittance, is linearly proportional to the chromophore concentration. In
quantitative analysis, where it is required to obtain the concentration of
substance, therefore, absorbance is more commonly used. The relation between
these two parameters is given by the following:
A = log10(1/T) 5
Thus, when A = 2 only 1% of the incident light is transmitted, while at A = 3,
only 0.1% is transmitted.

2.2 Deviations from the Beer-Lambert law

According to the Beer-Lambert law, absorbance should be linearly proportional
to the concentration of the chromophore. However, there are systems that
apparently do not obey this rule. The examples that follow are amplified in (2).
2.2.1 High absorbance
At high absorbances, deviation from linearity is caused by stray light. This is
defined as light received at the detector that is not anticipated in the spectral
band isolated at the monochromator. Stray light is usually 'white', i.e. having
the same composition as the source or ambient illumination, and arises from (i)
the monochromator and/or (ii) light leaks in the sample and detector region.
The first ought to be eliminated by good monochromator design and filters, the
second by good design and construction or masking. When the selected wave-
length is absorbed by the sample, the proportional contribution of the stray
light to the transmitted intensity increases because it is of a wavelength that is
not absorbed. For practical purposes, it is necessary to keep light of the chosen
wavelength (Ic) about 10 times greater than the stray light (Is). It is prudent to
check the stray light characteristics of any spectrophotometer by measuring A
for a range of standards. Severe deviations from linearity may occur at A values
as low as 2 or as high as 6. A more detailed treatment is given in (3).
ROBERT K. POOLE AND ULDIS KALNENIEKS

2.2.2 Concentration variations

Apparent deviations from the Beer-Lambert law arise when concentration
variation causes changes in the distribution of several chromophore species in
the solution. In these cases, if the absorption contribution from each species is
considered, the Beer-Lambert law is obeyed by each species, but it is the vari-
ation in species concentrations which produces the apparent deviation. For
example, dimerization of the chromophore molecules at higher concentrations
may take place, providing a dimer with an e value other than that of the mono-
meric form, causing a non-linear dependence of absorbance upon the solute
concentration.

2.2.3 Coupled reactions

In colorimetric assays, where the compound under study has first to react with a
colour reagent in order to be determined, non-linearity at higher concentrations
can arise when insufficient colour reagent is present. Likewise, non-linearity at
low concentrations may be caused if insufficient time is allowed for completion
of the colour reaction, since the rate of reaction at low concentrations is lowest.

2.3 Absorbance or light scattering?

Spectrophotometers are designed for measurement of light absorption, but in
laboratory practice it is common for the sample under investigation to scatter
light as well as absorb it. As we shall see (Sections 4.2 to 4.4), instruments have
been designed to allow maximum sensitivity of absorbance measurements even
in the face of a very high degree of light scattering. It is quite possible to
measure absorbance changes of 0.005A in a suspension of cells that has an
apparent absorbance of 3 (see later, Figure 15). However, in some applications, it
is scattering that we wish to measure, not absorbance. One of the most com-
mon of such applications occurs in microbiology where turbidity is routinely
used as a measure of biomass concentration.
It has been argued before (4) that measurements on bacterial cultures, for ex-
ample, where scattering is considerable should not be referred to as 'absorbance'
or 'attenuance' but as 'apparent absorbance' or 'optical density'. The spectro-
photometer output (chart recorder, analogue or digital readout, or computer
monitor) may say 'absorbance' but in fact most of the signal will be due to light
scattering. When describing measurements of 'apparent absorbance' or 'optical
density', the pathlength (or type of tube etc.) and wavelength (in nm) of the
incident light should be given (e.g. OD600) and, very importantly, the type of
spectrophotometer (maker and model) should be cited (because optical design
greatly influences readings; see below) and the extent of any dilution used
should be cited. It is prudent to prepare a calibration curve using a range of cell
concentrations and determine the 'apparent absorbance' or 'optical density' at
which the response of the spectrophotometer becomes non-linear, which may
be as low as 0.6-0.7. The use of instruments specifically designed to measure
turbid samples, such as nephelometers or Klett meters, should be reported in
 INTRODUCTION TO LIGHT ABSORPTION: VISIBLE AND ULTRAVIOLET SPECTRA

appropriate units. It is sometimes said that Klett units are meaningless (because
they cannot be defined as A can) but, in fact, they are more useful: the variation
between Klett meters is likely to be much smaller than the variation obtained
when measuring a turbid sample in different 'biochemical' spectrophotometers.
Arthur Koch has written eloquently (5, 6) of the factors that influence light
scattering by particles in suspension and of our ability to detect sensitively such
light scattering. In the most practically oriented of these papers (6), Koch com-
pares experimentally various laboratory spectrophotometers and wavelengths
for their suitability as 'turbidimeters'; variation is due largely to what portion of
the scattered light is viewed by the detector. Most of the light scattered by a
microbial suspension deviates by only a few degrees (7°) from the incident beam
(Figure 2). Thus, using an instrument in which the detector is placed as far as
possible from the cuvette will greatly enhance sensitivity, since a low concen-
tration of scattering particles will be sufficient to scatter some of the light from
the incident beam. It can be very instructive to measure optical density of a
single sample (perhaps fixed with formaldehyde to prevent further growth and

Figure 2 Properties of ideal turbidimeter and spectrophotometers. For a clear sample, the
position of the detector (PMT, photomultiplier tube) is unimportant (A). If a turbid sample is
examined in a spectrophotometer where the PMT is far removed from the cuvette (B), the
instrument will behave as a sensitive turbidimeter because it will detect little of the
scattered light. If a turbid sample is examined in a spectrophotometer where the PMT is
close to the cuvette (C), the instrument will detect absorbance even though much of the light
is scattered.
ROBERT K. POOLE AND ULDIS KALNENIEKS

changes in cell size) in as many spectrophotometers as possible, as Koch has

done for some now-obsolete instruments. As a rule, choosing a lower wave-
length will offer greater sensitivity of the optical density measurement, but a
longer wavelength will give greater linearity over the same range of cell sus-
pensions. Although an ideal turbidimeter would receive none of the scattered
light (Figure 2), many biochemical spectrophotometers are designed to view as
much of this as possible generally by locating the detector as close to the
sample as possible. Beware of using spectrophotometers that offer a choice of
cuvette positions in terms of distance from the detector; frighteningly large
discrepancies between OD readings at such positions are likely.

3 Spectra of some important naturally occurring

chromophores
A very few examples of biochemical applications follow, betraying our personal
interests in respiratory metabolism.

3.1 Amino acids and proteins

The spectra of amino acids are determined by the nature of their side chains.
None of the amino acids has an absorption extending into the visible region,
and, in the absence of additional chromophores (haem groups, Cu ions, flavin,
etc.), proteins are therefore colourless. Amino acids with aromatic side chains,
in particular tryptophan and tyrosine, account for most of the absorption seen
in proteins below 300 nm. The spectral properties of amino acids absorbing in
the UV region are given in Section 6.3 of Chapter 2. Due to the contribution of
tyrosine and tryptophan side chains, protein absorption is maximal at about
280 nm, and can be used for approximate estimation of protein concentration
in purified samples or those not containing other species which absorb in the
same wavelength region. Absorption at 280 nm is a convenient protein assay in
following the elution of a sample from a chromatographic column, and the
A260/A280 ratio is a common method of assessing samples of nucleic acid for
freedom from protein (7). Note that the tyrosine absorption spectrum is pH-
dependent, due to its hydroxyl group, which dissociates at high-medium pH.

3.2 Nucleic acids

The spectra of a free base and the corresponding nucleoside are quite similar,
since only the purine or pyrimidine base component is a chromophore, giving
rise to absorption in the region between about 250 and 270 nm at neutral pH.
Because of protonation of the base nitrogen atoms, spectra of these compounds
are also highly pH-dependent. Nucleic acids have an absorption maximum close
to 260 nm. The variation in absorption coefficient of DNA and RNA per nucleo-
tide residue is small; hence the absorption at 260 nm can be used as a measure
of total nucleic acid concentration (7). However, the absorption of a native
nucleic acid molecule cannot be expressed as a sum of absorbances of all the
 INTRODUCTION TO LIGHT ABSORPTION: VISIBLE AND ULTRAVIOLET SPECTRA

component nucleotides. This is known as the hypochromism of nucleic acids,

and arises from the weak interactions of the neighbouring residues with one
another. Nucleic acids can be denatured by exposure to heat or extreme pH.
During the denaturation process, the double helix of the native DNA unwinds,
the two strands separate as random coils, and weak interactions between neigh-
bouring nucleotides are diminished. As a result, denaturation is accompanied
by an absorbance increase at 260 nm, the so-called hyperchromic effect.
Measurement of the extent of the hyperchromic effect allows the progress of
denaturation to be followed, for example, as a function of temperature, and the
melting point of a DNA sample to be determined (see Chapter 13 for more
details).

3.3 NAD(P)H
Upon reduction of both nicotinamide adenine dinucleotide (NAD) and nicotina-
mide adenine dinucleotide phosphate (NADP) to NADH and NADPH, respect-
ively, a strong absorption peak appears at 340 nm. The millimolar absorption
coefficient at this wavelength for both reduced coenzymes is 6.22 mM-1 cm-1. As
these compounds function as coenzymes of many dehydrogenases, a large
number of enzyme reactions can be followed (either directly, or using coupled
assays) by monitoring the change of absorbance at 340 nm (8).

3.4 Carotenoids
Carotenoids are members of a large class of red, yellow and orange plant pig-
ments, also found in the membranes of some bacteria. Carotenoids contain long
carbon chains with many conjugated double bonds. Delocalization of electrons
is responsible for their strong absorption in the visible region. In general, the
more conjugated double bonds a molecule contains, the stronger and more red-
shifted its light absorption is. Carotenoids are used as intrinsic membrane probes
of transmembrane electric potential in photosynthetic energy-transducing
membranes (9) due to their very rapid electrochromic shift of absorption
maximum in response to transmembrane electric field.

3.5 Haem proteins

The haem group consists of a porphyrin ring with a ferrous or ferric iron co-
ordinated centrally. It serves as a prosthetic group of haemoglobins, myoglobin,
hydroperoxidases, and a vast number of cytochromes. The conjugated double
bond system of the porphyrin ring causes a strong absorption in haemo-
proteins, termed the a, B and y bands. Typically, a bands occur at longest
wavelengths (550 to 650 nm), y bands at the shortest wavelengths (also called
Soret bands, after the Swiss scientist who first examined the near UV region of
cytochromes), and (3 bands lie between. The characteristic absorbances for re-
duced (ferrous) haemoproteins are generally studied in reduced minus oxidized
difference spectra (see Figure 3 for absolute and difference spectra of cytochrome
c). The positions of these absorbances are determined by the nature of the
ROBERT K. POOLE AND ULDIS KALNENIEKS

550.5

AA = 0.5

480 500 520 540 560 580 600

Wavelength (nm)
Figure 3 Absolute and difference spectra of cytochrome c. A is the absolute spectrum of the
oxidized (Fem) form of purified horse heart cytochrome c, recorded with reference to a sample
of buffer, with 500 nm as the reference wavelength in a dual wavelength scanning
spectrophotometer. B is the absolute spectrum of the reduced (Fe") form of the same
sample after treatment with a few grains of sodium dithionite. Isosbestic points occur
wherever the spectra of the oxidized and reference forms cross. C (shifted up) is the
difference spectrum obtained by subtraction of A from B. Spectra were recorded in a
Johnson Foundation/Current Designs Inc. SDB4 Dual-Wavelength Spectrophotometer and
manipulated using Soft SDB software. Conditions were: room temperature (about 20 °C); 10
mm pathlength; scan speed, 4.25 nm s-1; spectral bandwidth, 2 nm. The positions of the B
(520.5 nm) and a (550.5 nm) bands are marked.

substituent groups attached to the porphyrin ring and thus the particular type
of haem, its redox state, the spin state of the iron and the presence of bound
ligands.
The haem iron of cytochromes, globins, oxygenases and hydroperoxidases
frequently bind ligands resulting in characteristic absorbance changes. For
example, carbon monoxide (CO) binds to the reduced form of certain haem
proteins, acting as a competitive inhibitor with respect to oxygen in respiration.
The most common laboratory application is in CO difference spectroscopy, i.e.
the identification of CO-binding haem proteins by recording the difference
between a reduced sample and another that has been treated with CO. For
example, a terminal oxidase common in bacteria, cytochrome bo', binds CO at
the oxygen-reactive haem O to give an adduct with an absorption maximum at
about 415 nm. Since the reduced but unligated haem O has an absorption maxi-
mum at about 430 nm, a difference spectrum (reduced + CO minus reduced) will
show a peak near 415 nm and a trough near 430 nm. Also, the CO-Fe bond is
photodissociable so that for certain haem proteins a second type of difference
spectrum, the photodissociation spectrum, can be recorded. This is the differ-
ence between the CO-ligated sample after photolysis (i.e. reduced, unligated)
minus the same sample before photolysis (reduced, CO-ligated). For an excellent
 INTRODUCTION TO LIGHT ABSORPTION: VISIBLE AND ULTRAVIOLET SPECTRA

account of the use of cytochrome spectroscopy as applied to bacterial cyto-

chromes, see Wood (10).

4 Spectrophotometer configurations
4.1 Single beam spectrophotometers
Single beam instruments are widely used for routine laboratory measurements
at a single wavelength. As shown in Figure 4, only a single light beam from the
monochromator passes through the sample compartment. The absorbance zero
or 100 per cent transmittance is adjusted with a cuvette containing buffer or
solvent. Then the absorbance (or transmittance) value of a sample cuvette con-
taining sample solution is measured. It is necessary to put the sample and the
reference in the optical beam alternately, by manual operation, typically with
an interval of a few seconds required to change the cuvettes. However, if full
spectra of the sample and reference are required, the interval between the two
measurements is likely to be several minutes, with the consequent risk that
lamp drift or other sources of instability can lead to errors.
It is impossible to give more detailed information or protocols for using the
wide range of single-beam instruments on the market. The advice given here is
that which must apply throughout this chapter: consult the manufacturer's
manual.

4.2 Split beam or 'double beam' spectrophotometers

Confusion of nomenclature is rife here. Certain manufacturers and the authors
prefer 'split beam' since it emphasizes that a single beam of monochromatic light
is split spatially. 'Double beam' is better reserved for two beams of different
wavelengths (see Section 4.3).
In the split beam mode, corrections for variations in light source intensity
are made automatically because the incident beam is divided between reference
and sample materials, as shown in Figure 5. The monochromatic light from the
monochromator is split (chopped) into reference and sample beams by a
chopper mirror. The two transmitted light beams of identical wavelength are
now separated in time and space, one beam passes through the sample and the
other through a standard or reference. The beams are sequentially detected as

Light source Monochromator Sample compartment Detector Amplifier Hardware and software
for display

Figure 4 Components and their arrangement in a single beam Spectrophotometer. For

details, see text.
ROBERT K. POOLE AND ULDIS KALNENIEKS

Reference cell

Sample cell

Light source Monochromator Sample compartment Detector Amplifier Hardware and software
for display

Figure 5 Components and their arrangement in a split beam ('double beam')

spectrophotometer. For details, see text.

the reference signal 10 and the sample signal I, and the sample absorbance is
obtained as the logarithm of the ratio of the two signals, which, according to
the Beer-Lambert law, is independent of the incident light intensity. In com-
paring a sample and reference at each wavelength, the split beam spectrophoto-
meter produces a 'difference' spectrum (instead of two separately acquired
'absolute' spectra, one for the reference and one for the sample, in a single-
beam device). In the difference spectrum, those features common to sample and
reference, like buffer absorption, moderate turbidity, or optical imperfections,
are cancelled out.
The split beam mode is used to measure either absolute spectra (i.e. the
difference between a sample in one beam and the solvent, or buffer, or even air
in the other beam) or a difference spectrum. For the latter, the intensities of the
two beams after passing through the cuvettes are again compared but, instead
of comparing a sample to a non-absorbing reference, two nearly identical
samples are compared. The difference between the two may be amplified be-
cause any interfering or background absorbance or light scattering is common
to both samples. It should be noted that in difference spectra both peaks
(positive differences) and troughs (negative differences) are anticipated accord-
ing to whether the 'sample' absorbs more or less light, respectively, than the
reference.
The use of a chopper, usually vibrating or rotating with a frequency of about
50 Hz, limits the kinetic resolution of the split-beam instruments. Spectral
changes occurring on a faster timescale cannot be monitored. Recent advances
in component design with improved stability have to some extent limited the
advantages of split-beam methods and there has been a resurgence of interest in
and availability of simpler single-beam designs.

4.3 Dual-wavelength spectrophotometers

The most distinctive feature of the dual-wavelength spectrophotometer is the
presence of two monochromators, the outputs from which are directed to a
single sample, again traditionally via a chopper mirror. Consider first the
operation of a dual-wavelength non-scanning spectrophotometer. Rather than
comparing the absorbance of a sample and reference at the same wavelength (as

10
 INTRODUCTION TO LIGHT ABSORPTION: VISIBLE AND ULTRAVIOLET SPECTRA

Lamp and supply

Amplifier and (1) YT recorder or computer.

demodulator (2) XY recorder or computer.

Figure 6 Components and their arrangement in a dual-wavelength spectrophotometer. In its

basic mode of operation, both A1 and A2 are fixed and kinetic data may be obtained (A4 A 1 - A 2
versus time). In the dual-wavelength scanning mode, A2 is set at a reference wavelength and
A1 is scanned to generate a spectrum referenced to A2. HV, high voltage; PMT,
photomultiplier tube. For details, see text.

in the split beam mode; Section 4.2), the absorbance of a single sample is
compared at two wavelengths—'sample' (A1 in Figure 6) and 'reference' (A2 in
Figure 6). The sample A1 is selected to correspond to the absorbance maximum of
the signal under study (e.g. 550 nm for cytochrome c) and the reference A2 is
chosen to be, ideally, at an isosbestic point (see Section 6.3) or other wavelength
close to A1 where significant absorbance changes are not anticipated. Wave-
length pairs in the fixed dual X mode should be chosen to optimize:
(1) sensitivity, i.e. high signal:noise (choose A for high absorbance coefficient);
(2) specificity (select for freedom from other components); and
(3) freedom from light scattering changes (select pairs as close as possible,
ideally 15-40 nm apart).
In this way, A2 serves as a reference measurement for the observed absorbance
change to be studied at A1. Thus, if there is an undesired change in the absolute
spectrum of the sample during measurement, e.g. due to particles settling out,
the effects on X1 - X2 will be minimized since the undesired change will be
recorded at both wavelengths. Results are presented on a YT recorder, computer
etc.
The dual-wavelength scanning spectrophotometer was developed by Britton
Chance in the Johnson Foundation in the early 1950s and remains immensely
useful and in production (Figure 7). A single light source is used for two
independent monochromators; one is set to a reference wavelength while the

11
ROBERT K. POOLE AND ULDIS KALNENIEKS

Figure 7 Photograph of a custom-built dual-wavelength spectrophotometer designed and

constructed by the University of Pennsylvania School of Medicine Research Instrumentation
Shop and Current Designs Inc. The instrument has the optical components in a classical
dual-wavelength configuration, except that use of monochromators with entrance and exit
slits 'in-line' (not at 90°) to each other requires 'elbow'-type mirror lens adapters to bring the
two beams to the chopper. Light from a 45 W tungsten halogen source (L) is focused on the
entrance slits of two Jobin-Yvon H20 monochromators {M}, having aberration-corrected
holographic gratings with focal length 200 mm, f/4,2. One of the monochromators is driven
by a J-Y TTL Stepper Interface (S) over the desired spectral range while the second is set at
the reference wavelength, selected to lie at an isosbestic point or a region of the spectrum
adjacent to regions of interest. The output of the monochromators is modulated by a
resonant tuning fork vibrating mirror (V) and focused on the cuvette (C). Transmitted light is
measured by a Hamamatsu R928 side window photomuitiplier tube (P) positioned about 50
mm from the nearest edge of the cuvette. The log of the photocurrcnt is taken prior to
digitization of the data. A Scanner Interface Unit (I) incorporates the high voltage supply and
servo circuit, log amp, chopper driver and demodulator 18 bit ADC. and interface to a
Macintosh computer. Spectral data arc analysed using SoftSDB (Current Designs Inc.) and
subsequently by CA-Cricket Graph III, Microsoft Excei or other desired software.

other scans the desired spectral range. The two monochromator exit beams
intercept at right angles at a vibrating chopper mirror which transmits only one
of the beams at a time through the cell (the other beam being reflected). The
two beams arrive at the detector, generally a photomultiplier tube, separated, in
time; the signals are separated electronically, compared and recorded. Results
are presented on an XY recorder, computer etc.
In the non-scanning mode, dual-wavelength instruments are generally used
to study kinetic changes at a pair of wavelengths (sample-reference) allowing
the observation of small absorbance changes despite a very high initial absorb-
ance or in the face of light scattering changes arising from particles settling, for
example. In the dual-wavelength scanning mode, spectra are obtained with
reference to an isosbestic point or other suitable reference wavelength. Note
that since only one cuvette is scanned at a time, difference spectra are obtained
only by computing the difference between two scans. In either mode, the dual-
wavelength instrument is valuable for optically isolating components in mix-
tures using wavelength(s) in which the desired component changes absorption
and other components do not.

12
 INTRODUCTION TO LIGHT ABSORPTION: VISIBLE AND ULTRAVIOLET SPECTRA

4.4 Multi-wavelength spectrophotometers

A novel design was introduced by Chance and colleagues in 1975 (11). It allows
four- or eight-channel measurements to be made and can serve as fluorometer,
reflectometer or spectrophotometer. In essence, four or eight interference
filters are mounted in a disc, rotating about an axis parallel to the light beam
from a halogen source. The speed of the disc (driven as a turbine from a filtered,
compressed air supply or by an electric motor) provides a variable chopping
frequency, typically of 100 Hz. Full details are given in reference (11).

4.5 Diode array spectrophotometers

Diode array spectrophotometers are particularly suited to the single-beam mode
since spectra are acquired very quickly (see Section 4.1). In conventional spectro-
photometers, it has been customary to use as the detector a photomultiplier or
photodiode which is sensitive to a wide range of wavelengths (wavelength
specificity being provided by a monochromator between the lamp emitting
polychromatic light and the sample). The development of diode array spectro-
photometry and the introduction in 1979 of commercially available instruments
has led to a wide range of applications, most of which exploit the ability of the
diode array to detect light at a large number of discrete but closely spaced
wavelengths simultaneously.
A diode array consist of a series of photodiode detectors arranged side-by-side
on a silicon crystal. The diode array typically has a photosensitive area of a few
square millimetres. Each diode is dedicated to measuring a finite but narrow
band of the spectrum and all diodes are connected to a common output line. By
measuring the variation in light intensity over the entire wavelength range, the
absorption spectrum is measured.
Figure 8 shows the arrangement of the basic components. Polychromatic light
from an appropriate source is passed through the sample area and focused on
the entrance slit of a grating or polychromator, which disperses light onto the
array. Note that the relative positions of the sample and dispersive element are
reversed relative to a conventional spectrophotometer ('reverse optics'). The
bandwidth of light detected by each diode is related to the size of the poly-
chromator entrance slit and the physical size of the diode. A shutter is used to
cut off light from the source until a measurement is to be made. This prevents
continuous illumination of the sample with high intensity radiation, minim-
izing undesirable photochemical effects (see later).
In conventional scanning spectrophotometers, the instrumental spectral
bandwidth (see later, Section 5.2) is primarily a function of the entrance and
exit slit widths of the monochromator and the dispersion generated by the
grating; resolutions of 0.5 to 2 nm are common. In a diode array instrument, the
spectral resolution is dependent on the number of photodiodes dedicated to
measurement in the specified wavelength range (see Figure 9). The spectrum can
be considered to be digitized, with the distance between successive diodes the
sampling interval. Higher resolution requires more (and smaller) diodes.

13
ROBERT K. POOLE AND ULDIS KALNENIEKS

Figure 9 The discontinuous spectrum produced by a diode array spectrophotometer.

14
 INTRODUCTION TO LIGHT ABSORPTION: VISIBLE AND ULTRAVIOLET SPECTRA

The main advantages of diode array spectrophotometry are:

• it is very fast because of parallel data acquisition and electronic scanning
• excellent wavelength reproducibility
• reliability, due to the absence of moving parts in a monochromator, and
solid-state technology
The most common applications currently are in rapid reaction kinetic studies
where a diode array detector can be mounted to a stopped-flow spectra-
photometer (several commercial instruments of this type are available, for
example, from Hi-Tech Scientific Instruments and Applied Photophysics) and in
monitoring fractions during chromatography. Figure 10 shows an application in
the study of a flavohaemoglobin (12) and illustrates the ability to record spectra
over a wide wavelength range, revealing simultaneous changes in three chromo-
phores over an observation period in which temporal resolution from milli-
seconds to hundreds of seconds was achieved. Again exploiting the ability to
obtain a spectrum virtually instantaneously, another recent application involves
obtaining absorbance spectra during transient steady states in a growing
bacterial culture (13).
Potential problems include the use of very high light intensities with possible

Wavelength/rim

Figure 10 Art example of a diode array stopped-flow spectrophotometric experiment showing

the ability to display complete spectral changes over time. The reduction of 5 um cytochrome
c (550 and 520 nm signals) by Escherichia coli flavohaemoglobin (Hmp, 4 um) in the
presence of oxygen and NADH is shown. The FAD of Hmp is largely responsible for the
trough centred near 480 nm and the oxygenated haem absorbs at 540 and 580 nm. The
spectra were recorded by Dr Yutaha Orii (Kyoto University) on a rapid-scan stopped-flow
spectro photo meter constructed in his laboratory. For further details see (12). This figure is
reproduced from Poole, R. K., Rogers, N. J., D'Mello, R. A, M,, Hughes, M. N. and Orii, Y.
(1997). Microbiology 143, 1557-1565 with permission from the Society for Microbiology.

15
ROBERT K. POOLE AND ULDIS KALNENIEKS

implications for photosensitive materials (like flavins), cost (although direct

comparisons with conventional instruments are difficult), and difficulties in
obtaining high resolution spectra with extremely turbid light-scattering samples.

4.6 Microwell plate-reading spectrophotometers

Microplate readers have been available for over 10 years, but those available
now offer a range of useful operating modes that can find numerous applica-
tions in screening multiple samples in 96-well microwells of 'microtitre' plates.
Applications include analysis of microbial growth turbidimetrically, colori-
metric assays (e.g. NAD(P)H-linked or dye reduction), immunoassays, platelet
aggregation assays and measurements of nucleic acids and proteins (see Chapter
7). In one current instrument, Molecular Devices Spectramax 340PC, wave-
lengths between 340 and 85 nm can be scanned in 1 nm increments. The depth
of the liquid in each well is determined and values normalized to 1 cm. Sample
volumes between 30 and 100 jjJ can be measured. The dual-scanning mono-
chromator versions of such readers now available for fluorescence work are
outside the scope of this chapter.

4.7 Reflectance methods

For opaque samples, such as microbial cell cultures, animal tissues and cell
films, it is possible to obtain a spectrum by measuring not the (very small)
amount of transmitted light but rather the reflectance of light from the sample
surface. An example of the use of fibre optic light guides to record diffuse
reflectance spectra from brain in vivo was given by Bashford (14). Essentially, a
bifurcated light guide was used to transmit light from the source to the brain
surface and the reflected light was transmitted to the photomultiplier detector
by the second limb of the light guide. A further example using a custom-built
apparatus can be found in (15). For reflectance measurements from skin, a
Dermaspectrometer (Cortex Technology, Hadsund, Denmark) has diodes for
emission centred at 568 nm (for haemoglobin) and 655 nm (for melanin) (16).
Note that the method devised by Brown et al. (17) for screening yeast colonies on
agar plates is actually a transmittance method but a fibre optic light guide is
used to direct light from the monochromator to the colony in situ on the Petri
dish; a photodiode is placed beneath the colony.

4.8 Novel double monochromator methods

A unique design has been implemented by Olis Inc. in the Olis RSM-1000
incorporating a patented subtractive double-grating monochromator, capable of
1000 scans/s. In the middle plane of the monochromator is a spinning disc
containing a number of slits which acquires millisecond 'snap-shots' of a rapid
reaction. The RSM 1000 Spectrophotometer offers an alternative to diode array
systems for millisecond spectral scanning and avoids the high light intensities
required for diode array measurements. The instrument can be used for
absorbance, fluorescence and circular dichroism.

16
 INTRODUCTION TO LIGHT ABSORPTION: VISIBLE AND ULTRAVIOLET SPECTRA

4.9 Computing and spectrophotometry

It is now so common for spectrophotometers of any of the above designs to be
equipped with computer control, that it is easily forgotten that optically excel-
lent instruments abound that predate the Macintosh and PC era. For instance,
dual wavelength instruments designed and constructed in the Johnson Research
Foundation laboratory of Britton Chance in the 1950s were fitted with on-board
micro-processing for spectrum acquisition, subtraction and other manipulations
a quarter of century ago. These were fitted with superb optics but, with the
ability to 'store' only four spectra, now appear rather primitive. Fortunately a
few designers and manufacturers offer packages for computerization of an
existing spectrophotometer. One such example, for Macintosh-based spectro-
scopy, is Current Designs, Inc. The digital signal processor (DSP)-based 'Scanner
Interface Unit' is powerful and flexible, so that it can be readily used as the
hardware basis for this task. The DSP system can perform signal averaging,
wavelength scan, control, chopped beam demodulation, and almost any other
sort of processing, with the specifics handled in software. Second, since it is easy
to add other channels of analogue or digital inputs, oxygen electrode readings,
for instance, can be acquired and displayed in real time along with absorbance.
Furthermore, the Current Designs Macintosh software, for example, has been
put in the public domain, allowing researchers and other developers to build an
ever-enlarging library of enhancements, analysis functions and hardware
control packages.

5 Choice of spectrophotometer operating

conditions
An increasing number of commercial spectrophotometers offer programs
designed for certain tasks—kinetics, wavelength scan, etc. However, unless the
reason why such a program may specify a particular scan rate or slit width, for
example, is understood, serious mistakes can be made.

5.1 Wavelength range and light source

The wavelength settings or wavelength range may be set manually by turning
the monochromator(s) to the desired X. or inserting filters, or by instructing a
microprocessor or computer which controls the apparatus to select the wave-
length. Decisions will generally be based on published optical properties or
prior experience. References cited in (18) provide useful lists of absorbance
maxima.

5.2 Spectral versus natural bandwidth

This is perhaps the least understood of the commonly set variables. The ability
of a spectrophotometer to (a) resolve peaks, that is to distinguish between two
absorption bands close together, and (b) to measure their intensity accurately, is

17
ROBERT K. POOLE AND ULDIS KALNENIEKS

dependent on spectral bandwidth and natural bandwidth. The former, but not
the latter, can be set by the operator.
To understand spectral bandwidth, it must be appreciated that 'mono-
chromatic' light from a monochromator never consists only of radiation of
precisely one wavelength (as does the light from a laser). Instead, it has a certain
bandwidth, including smaller amounts of light with shorter and with longer
wavelengths than the particular chosen wavelength. The spectral bandwidth is
defined as the band of wavelengths contained in the central half of the entire
band passed by the exit slit of the monochromator (Figure 11). It is the exit slit
(generally a few mm across) whose physical width can be set by the investigator.
To calculate spectral bandwidth from this slit width, we need to know the
reciprocal dispersion of the monochromator, which can be found in manu-
facturers' literature. Typical values of dX/dx, where x is the slit width, are 2 or 4
nm mm"1. Thus, in the latter case, when the exit slit is opened to 0.5 mm, say,
the spectral bandwidth is 2 nm.
In contrast, the natural bandwidth is an intrinsic property of the sample,
independent of the instrument bandwidth, and is defined as the width (in nm)
at half the height of the sample absorption peak, as shown in Figure 11. For
example, the value for the natural bandwidth of the 340 nm peak of NADH is
58 nm, whereas for most cytochromes at room temperature the natural
bandwidths in the a-region are of the order of 10 nm. It is easy to conceive that
having too broad a spectral bandwidth would result in an apparent decrease of
sample absorption. This is because the incident light would contain a large
fraction of radiation with wavelengths poorly absorbed by the sample.
A rarely seen graph, allowing errors in peak height measurements to be
predicted, is shown in Figure 12. This shows the dependence of the absorbance
peak magnitude on the ratio of the spectral bandwidth to the natural band-

270 350 390

Figure 11 The relationship between natural and spectral bandwidths. The natural bandwidth
shown is for NADH (58 nm, curve A). Curve B is a schematic representation of the spectral
bandwidth of the monochromator exit beam. Taken from (18) where the original source is
cited.

18
 INTRODUCTION TO LIGHT ABSORPTION: VISIBLE AND ULTRAVIOLET SPECTRA

6 8 10

Figure 12 Graph allowing calculation of the error in measured peak height for a given
monochromator spectral bandwidth and natural bandwidth of an absorption band. Spectral
bandwidth can be obtained from manufacturers' information or by knowing the physical slit
width of the monochromator and the reciprocal dispersion (nm/mm). Since natural
bandwidths of, for example, cytochrome absorption bands are about 10 nm at room
temperature a spectral bandwidth of 2.5 nm (ratio on the abscissa of 0.25) will introduce no
more than a 3% error in measured peak height. However, for low temperature spectra,
spectral bandwidths of about 0.5 nm are required.

width. Obviously, the best spectral bandwidth to be selected on a spectro-

photometer is 1/8 to 1/10 as wide as the natural bandwidth of the sample
absorption peak. However, if too narrow a spectral bandwidth is selected, too
small an amount of light energy will reach the sample and, as a result, electronic
noise (defined as any undesired recorder pen or-display excursion) will be
higher, decreasing the accuracy of the measurement. On the other hand, for
example, use of a spectral bandwidth of 8 nm would give rise to an error of
about 22% if attempting to measure the intensity of a cytochrome absorption
band with a natural bandwidth of 10 nm, but in a measurement of NADH the
error would be less than 0.5%. For quantitative work with compounds having
narrow natural bandwidths, a compromise must be reached. Probably the best
solution is to run the spectrum trying different spectral bandwidths remember-
ing that because of sample light scattering and variations in detector response,
one slit width may not be ideal for all wavelengths. Examples of such a test were
shown by Poole and Bashford (18) in their Figure 7.

5.3 Spectral resolution

Spectral resolution is a measure of the capability of a spectrophotometer to
distinguish between two closely adjacent wavelengths. The wavelengths are
considered to be resolved if the trough between the two peaks in the spectrum

19
ROBERT K. POOLE AND ULDIS KALNENIEKS

Figure 13 Resolution of two adjacent bands. In A, the separation between the bands (AX) is
sufficient for them to be resolved using the criterion that the trough between the adjacent
peaks is no higher than 0.8 of the peak height. In B, the separation between the bands
(AX') is just sufficient for the presence of two bands to be seen. In C, the separation
between the bands (AX") is insufficient for resolution and the bands cannot be
distinguished.

obtained is lower than 80% of the peak height. This is sometimes called the
Rayleigh criterion (see Figure 13). Although spectral resolution is a parameter
that greatly concerns the practical spectroscopist, it is inextricably linked to
spectral bandwidth (see above).

5.4 Scan speed and instrument response time

These variables are closely inter-related and again their selection is a matter of
compromise. Scanning a spectrum too fast for the instrument's response time
can lead to serious skewing of the observed peak, with loss of peak intensity
and blurring of fine detail in the signal (see the experiment shown in Figure 8 of
(18)). On the other hand, scanning too slowly can allow the sample to change
chemically or physically (e.g. by cell or membrane particles settling out of
suspension).

5.5 Temperature
The temperature-controlled sample holders available in most commercial
spectrophotometers, when coupled to an external circulating heater or cooler,
can control temperature in the approximate range 0 to 40°C. Specialized
cuvettes are not necessary and the device is useful for kinetic measurements or
observation of a labile sample. However, there are many advantages in kinetics
and wavelength scanning of being able to operate at much lower temperatures.

5.5.1 77 K
Spectroscopy at 77 K (liquid nitrogen temperatures) has been widely used to
detect differences in the absorption spectra of closely related haemoproteins, to
trap unstable intermediates and steady-states of oxidation and reduction, to
slow down rapid rates of reaction, and to detect and measure very low concen-
trations of haemoproteins. The main effects are a sharpening and enhancement

20
 INTRODUCTION TO LIGHT ABSORPTION: VISIBLE AND ULTRAVIOLET SPECTRA

of the absorption bands and a blue shift of 1-4 nm. The absorbance changes are
due to:
(1) a true temperature dependence of the absorbance of the chromophore,
resulting in narrowing, sharpening and shifting to shorter wavelengths of
the bands;
(2) light-scattering changes in the medium, which result in an effective in-
crease in pathlength by multiple internal reflections from ice crystals. [It is
worth noting that enhancement effects of related origin may be observed in
highly scattering suspensions (e.g. of intact cells) at room temperature.]
Reference (19) is a classic account of the practical uses of low temperature
spectroscopy. The enhancement effects are strongly dependent on the suspend-
ing medium and the method used to freeze it (20). For example, a devitrified (i.e.
polycrystalline) 1.4 M sucrose solution can give a 25-fold intensification at
— 190°C, compared with 20°C. The presence of an organic solvent also makes
the enhancement factors more reproducible than in dilute buffers. Glycerol has
the added advantage of suppressing the pH changes that occur on freezing, as
does a careful choice of buffer. The pH of Tris buffer changes dramatically
(increasing) on lowering the temperature, while phosphate and acetate buffers,
for example, are relatively temperature-independent. For a detailed account of
the choice of buffers in low temperature work, Douzou (21) should be
consulted.
Accessories for recording spectra at 77 K are available for many but not all
spectrophotometers. The attachments are generally expensive and many workers
have designed and constructed simple devices for their own instruments.
References to examples are given by Jones and Poole (22).

5.5.2 Other sub-zero temperatures

In media, such as those containing mannitol or sucrose, which do not exhibit
phase transitions in the experimental range of temperatures, the absorption
peak heights and peak areas are nearly linear functions of temperature between
-40°C and liquid nitrogen temperatures. Furthermore, if the sample is warmed
within this temperature range, and cooled again to the original temperature,
the resultant spectrum is indistinguishable from that recorded originally (19).
At temperatures higher than -40°C, the absorbance intensity decreases abruptly
and re-freezing does not fully restore the enhancement.
For kinetic studies at sub-zero temperatures, or where it is required to study
the effect of temperatures on an absorption spectrum (e.g. in investigating
haemoprotein spin-states), it is sometimes necessary to use temperatures other
than 77 K. Temperatures between about 0 and —40°C can be maintained simply
by circulating cooling water with ethylene glycol (as antifreeze) around a
cuvette containing the ethylene glycol-supplemented sample. For temperatures
between about -40°C and -140°C, as required for kinetic studies of ligand
binding to terminal oxidases and of subsequent electron transfer, the procedure
developed by Chance and co-workers (23) for their 'triple trapping' studies of

21
ROBERT K. POOLE AND ULDIS KALNENIEKS

mitochondrial cytochrome c oxidases is appropriate. Here, the temperature is

maintained by a steady flow of nitrogen gas, cooled by passing through a copper
coil immersed in liquid nitrogen and re-heated by a thermostatically controlled
resistor. Measurements of the sample temperature with a thermocouple next to
the cuvette show that temperatures are maintained to better than 0.5°C by the
thermostat. This system is simple, reliable and relatively inexpensive, but not
available commercially. The techniques of cryoenzymology and spectroscopy
below 77 K are outside the scope of this chapter but further information can be
found in reference (21).

6 Use of the spectrophotometer

6.1 The choices
To some extent, the choice of instrument to be used for obtaining the spectrum
of a biochemical sample will be dictated by what is available. For recording an
absolute spectrum of a routine sample, such as a dye or chromophore in clear
solution, almost any optical configuration could be used, namely single beam,
split-beam, dual-wavelength or diode array. If, on the other hand, the investi-
gator wishes to record, say, a difference spectrum of chromophores in a highly
turbid membrane or cell suspension, or perhaps measure time-resolved changes,
special techniques and instrumentation will be required. Fast kinetic methods
are covered in Chapters 8, 9 and 10. As an illustration of the choices that have to
be made in 'everyday' routine spectral scanning of clear solutions and in more
challenging situations like turbid samples, a guide to selection of operating
conditions (Figure 14) and two illustrative protocols will be described. However,
it cannot be overemphasized that these are not experimental recipes: every
possible permutation of sample, spectrophotometer and information sought
will require careful consideration of experimental design. Before presenting
these protocols, some fundamental practical aspects must be considered.

6.2 Baselines
A baseline is the wavelength-dependent difference in absorbance either (i)
between two cuvettes in a split beam apparatus when their contents are thought
to be identical or (ii) between two scans of one cuvette in a dual-wavelength
spectrophotometer over a period when the contents are thought not to have
changed. Any irregularities in a baseline will be included in subsequent dif-
ference spectra and may be superimposed on the spectral regions of interest. In
extreme cases, baseline irregularities may be mistaken for spectral peaks and
troughs, and a steeply sloping baseline can alter the apparent position of an
absorption peak. Therefore, baseline flatness should always be checked before
recording a difference spectrum and either presented with the latter or, at least,
reported. The following additional measures can be taken to improve baseline
flatness.

22
 INTRODUCTION TO LIGHT ABSORPTION: VISIBLE AND ULTRAVIOLET SPECTRA

wide

Figure 14 Scheme to show choices necessary for obtaining a routine absorbance spectrum.

6.2.1 Electronic baseline correction facilities

In most modern instruments, any irregularities are 'memorized' by the instru-
ment in an initial scan (normally not plotted) of the cuvette(s) and then sub-
tracted from subsequent scans. Remember, however, that in this case, any
dissimilarities in the cuvettes are essentially hidden.

6.2.2 Sample preparation

Give extra care to cleanliness and matching of the two samples, the proper
alignment of the cuvette(s) and the identicality of the cuvettes' contents. Where
both test and reference cuvettes contain samples treated in some way, such as is
the case in the recording of the baseline (reduced minus reduced) for a CO
difference spectrum (Section 3.5), it is generally preferable to split the sample
between the cuvettes after (in this example) reduction, rather than try to
prepare two samples identically.

23
ROBERT K. POOLE AND ULDIS KALNENIEKS

6.2.3 Manual correction

In the absence of the above facilities, the uncorrected baseline can be recorded
and subtracted from the difference spectrum manually at, say, 2 nm intervals.
The difference spectrum can be assumed to cross the baseline at a known
isosbestic point (see below). The method is of course tedious.

6,3 Isosbestic points

An isosbestic point is a wavelength where the absorptions (or e) of two light-
absorbing forms are equal (Greek iso, equal; sbestos, extinguished), and may be
seen, for example, where a difference spectrum crosses an appropriate baseline
one or more times. The isosbestic point is useful in both quantitative and
qualitative work. Where a clear isosbestic point occurs during the course of a
reaction, it is often taken as evidence that only two species are involved, since it
is considered unlikely that a third absorbing species would also have an identical
e at these wavelengths. It is more likely that a third species may simply have no
absorption at this wavelength. However, the lack of an isosbestic point does
indicate the presence of a third component. The isosbestic point is more
generally used as a reference wavelength to which the absorbance at a nearby
wavelength may be referred in (i) setting up a dual wavelength spectro-
photometer or (ii) measuring from an absorption spectrum the concentration of
a component using Ac. Less frequently, it is a wavelength suitable for quanti-
fication of the total amount of the two species present. Finally, in the pres-
entation of many stacked spectra, quoting the isosbestic point(s) eliminates the
need to draw baselines for each spectrum.

6.4 Wavelength and absorbance calibrations

Factors that can affect the apparent position of an absorption maximum are
described later. Even when operating conditions are properly chosen, the poss-
ibility remains that the spectrophotometer generates peaks that are a nano-
metre or more from the expected position. Wavelength accuracy can be
checked using the bright line spectrum (Amax = 656.1 nm) of the deuterium
light source; alternatively standards for checking the wavelength calibration are
available. Most commonly, these are rare earth oxides or a mixture of oxides
such as holmium and/or didymium; such 'filters' should be scanned with air as
the reference or blank. Holmium has nine useful maxima between 241.5 and
637.5 nm and didymium five between 573 and 803 nm (see manufacturers'
charts). A quick and easy check of both wavelength calibration and accuracy of
absorbance measurement is to use a 0.2 mM solution of potassium chromate in
0.05 M KOH. Two broad absorption bands at 275 nm and 375 nm should be seen
with molar absorption coefficients, respectively, of 3680 M-1 cm-1 and 4820 M-1
cm-1 in a 1 cm cell (18). It is useful to run calibration spectra periodically, dating
the scans, to check for degradation of accuracy and calibration.

24
 INTRODUCTION TO LIGHT ABSORPTION: VISIBLE AND ULTRAVIOLET SPECTRA

6.5 Choice and use of cuvettes (cells)

6.5.1 Pattern
The most common type of absorption cell or cuvette is the square cell, open at
the top, with a pathlength of 10 mm and working volume of 2-3 ml. Only
fluorescence cuvettes are polished on all faces. Manufacturers' catalogues illus-
trate designs for special purposes, with pathlengths ranging from 1 to 100 mm.
The essential feature is parallelism of the two end plates which limit the
pathlengths. The magnitude of the other horizontal axis (i.e. perpendicular to
the light beam) is dictated by the volume of the sample and the desirability of
keeping the light beam clear of the side walls, where intense reflection can
occur. 'Semi-micro' cuvettes (i.e. pathlength 10 mm, capacity about 0.65 ml)
appear to be convenient in reducing the volume of sample needed, but light
may pass around the edge of the sample as well as through it. These should be
either blackened (these are commonly available from manufacturers such as
Starna) or 'masked', or else a narrow spectral bandwidth must be selected (but
see Section 5.2). (In this laboratory highly turbid samples are always recorded in
conventional 1 cm square cells.) It is a useful exercise to determine the mini-
mum volume that can be used with each set of cuvettes; this is done by
gradually adding a coloured or turbid sample until the absorption reading is
constant. The working volume can sometimes be reduced further by including a
spacer below the cuvettes.

6.5.2 Mixing
Cuvette lids or, better, ground stoppers prevent evaporation of solvent and/or
contamination of the cuvette contents. Tapered Teflon stoppers can be drilled to
accommodate the fine needle of a syringe such as those supplied by Hamilton.
Mixing of cuvette contents can be problematic, especially when the cuvette is
almost full and a lid is fitted; a few glass beads (small enough to lie below the
light path) can provide sufficient agitation when a cuvette is inverted. An
alternative is to use disposable mixers or stirring paddles (Kartell). Those with
small horizontal platforms ('plumpers') can be loaded with up to about 100 ul of
an addition, then rotated in the cuvette between thumb and forefinger to
provide simultaneous and fairly rapid addition and mixing. For semi-micro
cuvettes, the mixers and paddles can be trimmed with a blade.
Where the contents of a cuvette must be continuously stirred during the
scanning of a spectrum, a small magnetic bar can be rotated magnetically
within the cuvette. Some (e.g. Spinbars from Bel-Art Products) are circular with
four small vanes and rotate smoothly inside a 10 X 10 mm cuvette. The small
stirrer motor is usually located under the cuvette and the power supply is
outside the spectrophotometer. A mu metal shield protects the photomultiplier
from 'noise'. High quality dc brush motors driven in a feedback loop to maintain
constant speed, not voltage, minimize stirring artefacts, especially important if,
for example, oxygen electrode measurements are to be made in parallel with

25
ROBERT K. POOLE AND ULDIS KALNENIEKS

absorbance. Such stirrers are available from The University of Pennsylvania

Research Instrumentation Shops.
Although suitable for modest stirring of small volumes, such stirrers are
sometimes inadequate for mixing the contents of a larger cell such as that
required in potentiometric redox titrations (24). Here, a longer magnetic stir-
ring bar rotates around a horizontal axis just above the light beam in the
main body of the liquid sample (typically 5-7 ml). The bar can be driven by a
button magnet mounted on the shaft of a small 12-V variable-speed electric
motor.

6.5.3 Optical material

The choice of glass, silica (quartz) or plastic cuvettes will be primarily dictated
by whether they will be used in the UV or visible spectral regions. Only silica is
suitable below about 360 nm. Plastic disposable cuvettes are cheap, popular and
can give perfectly acceptable results. Their less-than-perfect optical faces may be
of little consequence when working with highly light-scattering samples.

6.5.4 Special applications

Examples of cuvettes for special purposes and which must usually be con-
structed by the user are those for retaining standard e.p.r. tubes in a single
beam, for low-temperature work and photodissociation spectra, and for
potentiometric titrations.

6.5.5 Cleaning
The importance of clean cuvettes is self-evident. Routinely, all non-disposable
cuvettes should be emptied immediately after use, rinsed repeatedly in the
solvent (e.g. water), then with clean ethanol or acetone and dried with low
pressure air or nitrogen from a cylinder. It is prudent to install a filter (such as
those with pore sizes of 0.45 um used in filter sterilization) in the gas line.
Cuvette washers (e.g. Aldrich) wash, rinse, and dry cuvettes. Cotton wool 'buds'
can also be useful for dislodging interior, stubborn marks and for drying. The
outside optical surfaces should be polished with clean lens tissue. Note that
plastic 'squeezy' bottles generally used for solvents contain plasticizers such as
butyl phthalate, which can interfere with critical UV spectra.
The Perspex windows of low-temperature cuvettes easily become scratched
and will eventually crack. However, provided they are still liquid-tight (at least
for as long as it takes to plunge them and their contents into liquid nitrogen)
they are quite serviceable in this condition. The opacity of the Perspex is
negligible compared with that of the frozen sample.
Neglected cuvettes that cannot be dismantled for cleaning may require soak-
ing overnight in concentrated sulphuric acid containing a few crystals of
dichromate or permanganate, or boiling in distilled water containing a
laboratory detergent.

26
 INTRODUCTION TO LIGHT ABSORPTION: VISIBLE AND ULTRAVIOLET SPECTRA

6.6 A detailed example: recording of a cytochrome

difference spectrum (reduced minus oxidized)
The most common method of cytochrome spectral analysis is the recording, with
a split-beam spectrophotometer, of a spectrum that represents the difference
between an oxidized and a reduced sample. Such spectra may provide data on the
identity and the amount of the cytochrome types present, and may be obtained
with suspensions of intact cells, subcellular fractions derived therefrom or puri-
fied preparations. An example of such a spectrum obtained with pure cytochrome
c was shown as Figure 3. Note, however, that attempts to obtain a spectrum of a
cytochrome in situ, i.e. in intact cells or mitochondria, will be frustrated by the
intense light scattering of the sample. Figure 15 shows three attempts using dif-
ferent approaches. The bacterium Zymomonas mobilts has a low content of
cytochromes and is a challenging material for spectrophotometry (25). The abso-
lute spectrum of the suspension used (i.e. recorded using a blank) reveals a high
signal, particularly at lower wavelengths, due almost entirely to turbidity. When
the spectrum is taken in the dual-wavelength scanning mode with 500 run as a
reference wavelength, the absorbance difference across the spectrum is reduced
by more than 10-fold (compare scales in Figures 15A and B). The true contribution
of cytochromes is seen in the difference spectrum, obtained by taking oxidized
and reduced spectra in the dual-wavelength scanning mode and subtracting the
former from the latter in the instrument software. The cytochrome absorbance
peaks represent less than 1% of the turbidity measured in A.
A compromise must be reached in the choice of scan speed, slit width and
response time, these factors being inter-related. Typical' or starting values for
the parameters in such an experiment might be 1-4 nm s-1, 2 nm spectral
bandwidth and 0.1 s, respectively.
Full reduction of a sample can generally be achieved by adding a few grains
of sodium dithionite (Na2S2O4) to a cell or membrane suspension, but may take
several minutes for completion. It is important that the dithionite should be
fresh. It is useful to split a newly purchased sample into small aliquots in, say, 5
ml screw-capped bottles and to use each for only a month before discarding.
Alternatively, the preparation can be allowed to become anoxic by respiration
of a suitable oxidizable substrate. Sodium borohydride is also a useful reductant.
Oxidation may be achieved by using an exogenous oxidant or by vigorous
aeration of the sample just prior to scanning the spectrum. Difficulty can be
experienced if significant levels of endogenous reductants are present or if a
respiratory chain is terminated by an oxidase with a particularly high O2 affinity
or is capable of particularly rapid rates of electron flux. In such cases, O2 may be
generated in the sample by adding H2O2 and catalase, although preparations of
cells or sub-cellular fractions may exhibit sufficient endogenous catalase activity.
Alternatively, potassium ferricyanide (ferro/ferricyanide +0.43 V), hexachloro-
iridate (iridium (IV) chloride), or ammonium persulphate (S2082-/2SO4-2, +2.0 V)
may be added as oxidants (E°' values in parentheses). The first two are highly
coloured however; aqueous solutions of ferricyanide absorb strongly between

27
ROBERT K. POOLE AND ULDIS KALNENIEKS

A = 0.25

A = 0.025

400 500 600 700

Wavelength (nm)

Figure 15 Attempts to record absorbance spectra in a very turbid cell suspension. A shows
the 'absolute' spectrum (a scan of a cuvette with the buffer taken as the baseline) of
reduced membranes of the bacterium Zymomonas mobilis, acquired with a single-beam
instrument (Beckman DU 650). B shows A4 acquired with a dual-wavelength instrument
(SDB-4) with 500 nm as the reference wavelength. C is the reduced minus 'as prepared'
difference spectrum obtained with the same dual-wavelength instrument. The concentration
of membrane protein was 10 mg ml-1; sodium dithionite was used as the reductant.

370 and 460 nm (Xmax420 nm), whilst hexachloroiridate solutions have complex
spectra in this region with Amax at 488 nm. Ammonium persulphate is
colourless. Addition of these oxidants to final concentrations of about 2 mm is
usually satisfactory, although it is common practice to add 'a few grains' to 1-4
ml samples. Further details of these procedures and references are given by
Jones and Poole (22).

28
 INTRODUCTION TO LIGHT ABSORPTION: VISIBLE AND ULTRAVIOLET SPECTRA

Protocol 1
Recording the absolute UV-vis spectrum of a soluble
protein in solution
Equipment and reagents
• Protein in buffer, same buffer for • Split-beam scanning spectrophotometer
dilution. Impossible to be prescriptive (or almost any other kind!)
• Reagents for treatment of protein (e.g.
reductant for haem protein)

Method
1 Estimate absorption anticipated at A max , either from knowledge of protein con-
centration and anticipated absorbance coefficient, or from preliminary spectral
scan,
2 Decide on dilution required or adjustment of absorbance signals to be recorded by
choice of pathlength.
3 Establish and set wavelength limits. Select light source (e.g. UV and/or vis).
4 Choose cuvette material (e.g. silica/quartz for UV).
5 Select response time, slit width, scan speed. Use guidelines given in this chapter for
'normal' conditions, follow published protocols, or assess result of present scan
and adjust conditions accordingly (see figure 14).
6 Add buffer to two cuvettes, install in instrument, scan baseline, plot or store,
7 Keep buffer in reference cuvette, add protein to sample cuvette or replace contents
of sample cuvette with protein in buffer, install in instrument, scan, plot or store.
8 Compare baseline (step 6) and absolute spectrum of spectrum (step 7) visually or
subtract first from second.

Protocol 2
Recording the visible difference spectrum of a membrane-
bound chromophore in a turbid sample (cells or
membranes)
Equipment and reagents
• Cells or membranes in buffer, same buffer Split-beam or dual-wavelength scanning
for dilution. Impossible to be prescriptive spectrophotometer
• Reagents for treatment of protein (e.g.
reductant and oxidant for haem protein)

29
ROBERT K. POOLE AND ULDIS KALNENIEKS

Protocol 2 continued

Method
1 Estimate absorption anticipated at Xmax, either from knowledge of protein
concentration and anticipated absorbance coefficient, or from preliminary spectral
scan.
2 Decide on dilution required or adjustment of absorbance signals to be recorded by
choice of pathlength."
3 Establish and set wavelength limits. Select light source (i.e. vis).
4 Choose cuvette material (generally glass but cheap even scratched cuvettes may not
degrade signals).
5 Select response time, slit width, scan speed. Use guidelines given in this chapter for
'normal' conditions, follow published protocols, or assess result of present scan
and adjust conditions accordingly (see Figure 14).
6 Add identical, well mixed samples of cells or membranes to two cuvettes,b install in
instrument, scan baseline, plot or store.
7 Keep sample in reference cuvette, add reagent that will generate the required
difference (e.g. reductant) to sample cuvette, install in instrument, scan, plot or
store.
8 Compare baseline (step 6) and difference spectrum of spectrum (step 7) by
subtracting first from second.
" It is strongly recommended that the entire wavelength range is scanned at various sample
concentrations to avoid signal loss by light scattering at lower wavelengths (see Section 2.3).
b
In haem protein work, for example, this baseline may be that of the untreated or oxidized
sample.

6,7 Post-scan options

6.7.1 Derivative spectra
In its simplest form, a derivative spectrum is a plot of dA/dX versus \ rather than
A versus A. The derivative spectrum of a single, symmetrical peak crosses the
baseline at or near the peak in the starting spectrum. More complex forms arise
if two peaks are close together. If this first derivative spectrum is subjected to
the same derivatization, a second derivative spectrum is obtained in which the
original peak is replaced by a trough at the peak position and 'wings' appear
above the baseline. Examples were shown in (18) and the technique is discussed
in detail by Butler (26). The main application of such derivatives is in resolution
of partly overlapping absorption bands, made possible by the progressive band
narrowing that occurs as derivatives are taken.
Some spectrophotometers allow derivative spectra to be obtained by scan-
ning two monochromators offset by a small, fixed wavelength interval (as in the
old Aminco DW-2a spectrophotometer). This results in a difference signal pro-
 INTRODUCTION TO LIGHT ABSORPTION: VISIBLE AND ULTRAVIOLET SPECTRA

portional to AA/AA. However, more often, derivative spectra are now obtained
by computing data sets obtained in a more conventional scanning mode. Plotting
derivatives up to the second is frequently a standard option in modern scanning
spectrophotometers.

6.7.2 Data handling

An increasing number of commercially available spectrophotometers are con-
trolled by a computer and allow the user to extract spectral data in two-column
format (A and X) for subsequent manipulation and plotting in publication-ready
form. Commercial graphing packages, such as Cricket Graph III, Deltagraph and
many more can be used to smooth spectra, perform subtractions and additions
of whole spectra, normalizations and calculations of derivatives.

Acknowledgements
RKP is grateful to the large family of research students, postdoctoral researchers
and visiting scientists to his laboratory who have put these techniques through
their paces in obtaining spectra of microbial haem proteins. Professor Britton
Chance has been a source of inspiration for 25 years and introduced us to the
Research Instrumentation Shop (RIS) at the University of Pennsylvania School of
Medicine. RIS staff Norman Graham and Bill Pennie and Current Designs
President Ben Dugan have provided us with machines and limitless help, in-
cluding the reading of a draft of this chapter. We thank Hewlett-Packard,
Molecular Devices and Shimadzu for information on their products. We are
very grateful to Mark Johnson for expertly producing many of the figures and
Ron Adams for Figure 7. Work in this laboratory was supported by BBSRC and
The Royal Society/NATO.

References
1. Brown, S. B. (1980). In An introduction to spectroscopy for biochemists (ed. S. B. Brown),
pp. 1-13. Academic Press, London.
2. Brown, S. B. (1980). In An introduction to spectroscopy for biochemists (ed. S. B. Brown),
pp. 14-69. Academic Press, London.
3. Bashford, C. L. (1987). In Spectrophotometry and spectrofluorimetry—A practical approach
(ed. D. A. Harris and C. L. Bashford), pp. 1-22. IRL Press, Oxford.
4. Fewson, C. A., Poole, R. K. and Thurston, C. F. (1984). Society for General Microbiology
Quarterly, 11, 87.
5. Koch, A. L and Ehrenfeld, E. (1968). Biochimico et Biophysica Acta, 165, 262-273.
6. Koch, A. L. (1970). Analytical Biochemistry, 38, 252.
7. Sambrook, J., Fritsch, E. F. and Maniatis, T. (1989). Molecular cloning, a laboratory
manual, 2nd edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor
Laboratory, NY.
8. Harris, D. A. (1987). In Spectrophotometry and spectrofluorimetry—A practical approach
(ed. D. A. Harris and C. L. Bashford), pp. 49-90. IRL Press, Oxford.
9. Jackson, J. B. and Crofts, A. R. (1969). FEBS letters, 4, 185
10. Wood, P. M. (1984). Biochimica et Biophysica Acta, 768, 293.
11. Chance, B., Legallais, V., Sorge, J. and Graham, N. (1975). Analytical Biochemistry, 66,
498.

31
ROBERT K. POOLE AND ULDIS KALNENIEKS

12. Poole, R. K., Rogers, N. J., D'Mello, R. A. M., Hughes, M. N. and Orii, Y. (1997).
Microbiology, 143,1557.
13. Kavanagh, E. P., Callis, J. B., Edwards, S. E., Poole, R. K. & Hill, S. (1998). Microbiology,
144, 2271.
14. Bashford, C. L. (1987). In Spectrophotometry and spectrofluorimetry—A practical approach
(ed. D. A. Harris and C. L. Bashford), pp. 115-135. IRL Press, Oxford.
15. Chen, S. S., Yoshihara, H., Harada, N., Seiyama, A., Watanabe, M., Kosaka, H.,
Kawano, S., Fusamoto, H., Kamada, T. and Shiga, T. (1993) American Journal of
Physiology, 264, G375.
16. Thibodeau, E. A. and D'Ambrosio, J. A. (1997) European Journal of Oral Science, 105, 373.
17. Brown, S., Colson, A. M., Meunier, B. and Rich, P. R. (1993). European Journal of
Biochemistry, 213, 137.
18. Poole, R. K. and Bashford, C. L. (1987). In Spectrophotometry and spectrofluorimetry—A
practical approach (ed. D. A. Harris and C. L Bashford), pp. 23-48. IRL Press, Oxford.
19. Wilson, D. F. (1967). Archives of Biochemistry and Biophysics, 121, 757.
20. Vincent, J.-C, Kumar, C. and Chance, B. (1982). Analytical Biochemistry, 126, 86.
21. Douzou, P. (1977) Cryobiochemistry. Academic Press, London.
22. Jones, C. W. and Poole, R. K. (1985). In Methods in Microbiology (ed. G. Gottschalk), 18,
285. Academic Press, London.
23. Chance, B. (1978). In Methods in Enzymology (ed. S. Fleischer and L. Packer), 54, 102.
Academic Press, New York.
24. Dutton, P. L. (1978). In Methods in Enzymology (ed. S. Fleischer and L. Packer), 54, 411.
Academic Press, New York.
25. Kalnenieks, U., Galinina, N., Bringer-Meyer, S. and Poole, R. K. (1998) FBMS
Microbiology Letters, 168, 91.
26. Butler, W. L. (1979). In Methods in Enzymology (ed. S. Fleischer and L. Packer), 56, 501.
Academic Press, New York.

32
 Chapter 2
Fluorescence principles and
measurement
Arthur G. Szabo
School of Physical Sciences, University of Windsor, 401 Sunset Avenue,
Windsor, Canada N9B 3P4

1 Introduction
Fluorescence spectrometry is the most extensively used optical spectroscopic
method in analytical measurement and scientific investigation. During the past
five years more than 60000 scientific articles have been published in which
fluorescence spectroscopy has been used. The large number of applications
ranges from the analytical determination of trace metals in the environment to
pH measurements in whole cells under physiological conditions. In the scientific
research laboratory, fluorescence spectroscopy is being used or applied to study
the fundamental physical processes of molecules; structure-function relation-
ships and interactions of biomolecules such as proteins and nucleic acids; struc-
tures and activity within whole cells using such instrumentation as confocal
microscopy; and DNA sequencing in genomic characterization. In analytical
applications the use of fluorescence is dominant in clinical laboratories where
fluorescence immunoassays have largely replaced radioimmunoassay techniques.
There are two main reasons for this extensive use of fluorescence spectro-
scopy. Foremost is the high level of sensitivity and wide dynamic range that can
be achieved. There are a large number of laboratories that have reported single
molecule detection. Secondly, the instrumentation required is convenient and
for most purposes can be purchased at a modest cost. While improvements and
advances continue to be reported fluorescence instrumentation has reached a
high level of maturity.

2 Physical principles
2.1 The absorption process
A review of the physical principles of the fluorescence phenomenon permits
one to understand the origins of the information content that fluorescence
measurements can provide. A molecule absorbs electromagnetic radiation
through a quantum mechanical process where the molecule is transformed
from a 'ground' state to an 'excited' state. The energy of the absorbed photon of

33
ARTHUR G. SZABO

light corresponds to the energy difference between these two states. In the case
of light in the ultraviolet and visible spectral range of 200 nm to 800 nm that
corresponds to energies of 143 to 35.8 kcal mol-1. The absorption of light results
in an electronic transition in the atom or molecule. In atoms this involves the
promotion of an electron from an outer shell orbital to an empty orbital of
higher energy. In molecules, an electron is promoted from the highest occupied
molecular orbital (HOMO) to the lowest unoccupied molecular orbital (LUMO),
This process is governed by a number of selection rules, and the transition
probability for the absorption of a photon is reflected in the molecule's molar
absorptivity, e (M-1 cm -1 ), that is determined from the Beer-Lambert law.
Figure 1(a) provides an example of the absorption process in the case of a
carbon-carbon double bond. On interaction with a photon of appropriate

U) (b)

Figure 1 (a) A TT-TT* transition of a C-C double bond, with the lower energy TT state showing
the IT bonding orbital filled with two electrons with antiparallel spin and the unfilled TT
antibonding orbital (shaded), and the higher energy IT* state with the electron spins still
antiparallel but with one electron in the IT bonding orbital and one in the IT antibonding
orbital, {b) A TT-O-* transition of a C-C double bond with the same lower energy electron
configuration as in figure l(a), and the higher energy a* state with one electron in the IT
bonding orbital and one electron in the IT antibonding orbital (shaded).
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

energy an electron from the IT bonding orbital of the ground electronic state is
promoted to a IT* antibonding orbital (LUMO). The energy difference between
these two states corresponds to the absorbed photon energy, AE = hc/A, where h
is Planck's constant, c is the velocity of light in a vacuum and A is the
wavelength of the light. This excitation process is termed aIT-IT*transition and
the excited state is referred to as a IT-IT* state. There are other antibonding
orbitals of higher energy, and in this example such an orbital is a O*
antibonding orbital. The excitation of an electron from the IT bonding orbital to
the o* antibonding orbital would occur when the photon energy corresponds to
the energy difference between the ground state and the TT-o* excited state
(Figure 1(b)).
One of the selection rules that governs the transition probability is that the
spin of the electrons must be retained, i.e. the electron spins in the ground state
IT bonding orbital are paired and antiparallel and they retain their antiparallel
spin in the transition to the excited IT-IT* state. The multiplicity, M, of these
states describes the spin states of the electrons. Electrons are designated to have
a spin quantum number, S, of either +1/2 or -1/2 and so the multiplicity of the
state can be calculated according to equation V.
M=2|SS| + 1 1
In the ground state ES = 0, and so M = 1 and this state is designated as a ground
singlet state, S0. In the IT-IT* excited state the original electron spins are con-
served and M = 1. This excited state is then designated as a singlet state. In the
example in Figure 1 the TT-TT* state is the first excited singlet state, S1, and the
TT-O* state is the second excited singlet state, S2.

2.2 Excited singlet state deactivation processes

A modified Jablonski diagram (Figure 2) is a convenient description of the
photon absorption processes and the deactivation processes of excited states.
Associated with each electronic state are a set of vibrational states that corres-
pond to different vibrational energies of the molecule. In turn associated with
each vibrational state are sets of rotational states. The energy difference between
the vibrational states corresponds to the infrared frequencies, and the rotational
energy level differences correspond to microwave frequencies. At 20°C most
molecules exist in the lowest vibrational level of the ground state. The total
energy, E, of a particular state of a molecule can be represented as the sum of
the electronic energy, Eel, the vibrational energy, Evib, and the rotational energy,
Erot-
2
The overall change of energy as a result of the absorption of a photon of light is:
3
The absorption of a photon by a molecule is usually to a vibrational level above
the 'zeroth' vibrational level of the excited singlet state. The molecular
absorption spectrum traces out the energy difference between the ground state

35
ARTHUR G.

Figure 2 A modified Jablonski diagram showing the different transitions and processes
between the excited states. S0(0), S±(0) and S2(0) being the 0 level vibrational energy level
of ground state, first singlet state and second singlet state, respectively. 71(0) is the lowest
vibrational level of the first triplet state and P represents the energy level of a stable
photoproduct which may be formed. The transitions with the arrows pointing towards the top
of the figure represent absorption transitions and they are described in the text. The dashed
arrows pointing towards the bottom of the figure represent competing pathways by which the
excited molecule deactivates. I represents internal conversion processes; ISC, intersystem
crossing; NR, non-radiative deactivation; PR, photoproduct formation; hv the radiative or
fluorescence process.

Wavelength (nm)
Figure 3 An absorption spectrum representing the three absorption transitions designated
1, 2 and 3, in Figure 2.

36
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

and the several vibrational levels of the excited singlet states. Figure 3 is a
representation of a typical absorption spectrum, where absorption at A1 corres-
ponds to the absorption transition marked 1 (figure 2). Another wavelength, X2,
would correspond to transition 2 (Figure 2) to an upper vibrational level of S1,
and A.3 corresponds to absorption transition 3 (Figure 2) to S2.
The light absorption process occurs very rapidly (10-15 s) to form the excited
state, so fast that there is no change in the nuclear coordinates of the molecule
during the absorption process (Frank Condon transition). If the absorption is
to S2, the molecule undergoes a very rapid process (10-14 s) known as internal
conversion from the lowest vibrational level of S2 to an upper vibrational level
of Sj. In solution the molecule rapidly loses the excess vibrational energy
through collision with solvent molecules. The net result is that within 10-13 to
10-14 s of excitation the molecule is in the lowest (0th) vibrational energy level of
s s
1 ( 1(0))- The absorption process has very rapidly created a metastable state that
can undergo a number of competing processes that were not available to the
ground state.

2.2.1 Intersystem crossing

The process known as intersystem crossing occurs when one of the electrons
undergoes a spin inversion from either + 1/2 to-1/2or vice versa. This results in an
electronic state of lower energy (single electrons in different orbitals prefer to
have parallel spin) with a multiplicity M = 3, which is thus designated as a triplet
state, T1. The rate constant that is associated with the intersystem crossing
process is kisc.

2.2.2 Photoproduct formation

Since Sl is an electronic state that has a different electronic distribution from S0,
the ground state, the molecule may undergo changes in molecular orbital over-
lap. A new bonding configuration may result in the formation of a new molecule
or photoproduct. This is a photochemical process that has an associated rate
constant, kPR, that competes with other S: deactivation processes.

2.2.3 Non-radiative deactivation

Another process of internal conversion can occur when the 0th level of Sl
overlaps with a very high energy vibrational level of S0. The electron moves back
into the unfilled bonding orbital, and the molecule then relaxes very rapidly
through a series of vibrational steps to the lowest level of S0. Non-radiative
relaxation to an upper vibrational level of S0 can also occur through interactions
with solvent. The excess energy of the molecule is converted into heat energy.
These non-radiative deactivation pathways of S1 to S0 are grouped into a single
rate process described by the non-radiative rate constant, kNR.

2.2.4 Radiative deactivation (fluorescence)

The molecule can also return to one of several vibrational/rotational energy
levels of S0 by the emission of a photon of light that corresponds to the energy

37
ARTHUR G. SZABO

difference between S1(o) and a particular vibrational level of S1 (transition 4 in

Figure 2). The emission of light from an upper singlet state to the ground state is
known as fluorescence. In this radiative process the electron in the IT* anti-
bonding orbital makes a quantum jump back into the half filled IT bonding
orbital emitting a photon of light whose energy corresponds to the energy
difference between these two states. The rate constant that is associated with
the fluorescence radiative process is defined as kR. Just as the absorption of light
occurs between the lowest vibrational state of S0 (S0(0)) and one of several
vibrational levels of S1, the fluorescence radiative process takes place from SJO)
to any of the vibrational levels of S0.

2.2.5 Competitive deactivation processes

All of these different processes of deactivation of S1 compete with one another.
Later other rate processes of deactivation of S1 will be described that will also be
in competition with the processes discussed above. These different kinetic
processes are summarized in Table \.

Table 1 Summary of deactivation processes of the excited singlet state of a molecule, M

Absorption of a photon: M(So) + hv - M(S1)

after vibrational relaxation
to S1(0)
Intersystem crossing: M(S1) - M(T1 kisc
Photochemistry: M(S1) — Product kpR

Non-radiative relaxation: M(S1) - M(S0) kNR

Radiative (fluorescence): M(Si) - M(S0) + hv ka
Quenching: M(S1) + [Q] - M(So) +[Q] kQ[Q]
Resonance energy transfer M(S1) + Acceptor - M(So) + Acceptor kET
Where M(S1) corresponds to the electronically excited molecule, M.

It is this competition between the different processes that provides an im-

portant aspect of the information content of fluorescence spectroscopy and
makes it so useful in the determination of the properties of molecules. If the
molecule that is excited undergoes changes in its structure or environment,
then usually one of the rate constants, kisc, kNR, or kPR is affected with a
concomitant change in the relative probability of the fluorescent process.

2.2.6 Phosphorescence
Phosphorescence is also a radiative process. It describes the radiative process
originating from T1 to S0. The required spin inversion of one of the electrons
when it moves from the LUMO to the half filled HOMO is a forbidden process.
When observed it occurs on a timescale much longer than the fluorescence
process. In most cases it can only be observed in frozen glass solutions at low
temperature (77 K).

38
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

3 Fluorescence parameters
3.1 Fluorescence spectrum
The light emitted by a molecule from Sj(0) corresponds to the energy difference
between St(0) and different vibrational levels of S0, i.e. S0(n). There will be a
spectrum of energies of the fluorescence photons and this energy spectrum
(Figure 4) is what is measured on a fluorescence spectrometer. The wavelength,
X0 corresponds to the radiative transition from S1(O) to S0(0). This is the highest
energy radiative transition and is the energy difference between these two
states. AtXMAX,the wavelength corresponding to the highest fluorescence in-
tensity, the transition from S1(O) to S0(n) represents the most probable radiative
transition between these two electronic states.

3.2 Fluorescence quantum yield, oF

The quantum yield of fluorescence, oF is defined according to:
oF = number of photons emitted/total number of photons absorbed
Obviously the maximum value of oF is 1, that is all molecules that have been
excited to S1 return to S0 by the emission of a light photon. It can be shown that
of is related to the several rate constants discussed above:

KISC

Other rate processes to be described later will be included in this relationship.

The denominator of this expression is related to A, the absorbance of the

Wavelength (nm)
Figure 4 A typical fluorescence spectrum representing the energy distribution of the photons
from S1(0) to different vibrational levels of the ground state, S0.

39
ARTHUR G. SZABO

molecule. If I0 is the intensity of exciting light in photons s-1 then the integrated
intensity of fluorescence, IF (photons s1) is:
IF (total number of photons absorbed / s, IA). ( o F ) 5
From the Beer-Lambert law:
6
where I is the transmitted light intensity in photons s-1.
7
and
8
Thus the integrated fluorescence signal depends on the incident light intensity,
the absorbance of the sample at the wavelength of excitation, EX, and oF.

3.3 Singlet and radiative lifetime

If one excites a sample with an infinitely narrow pulse of light creating a
population of excited singlet state molecules M(S1), one can write a differential
equation that relates the concentration of M(Sl) at any time, t, to the various
deactivation processes:

and on integration this leads to:

where M(SJ(t) is the concentration M(Sl) at any time t; M(S1)(0) is the con-
centration of MfS at time t = 0; and Sk = kR + kNR + kisc + kpR- Since IF x M(Sl)
then one can write the expression:

By definition the lifetime of the excited singlet state, M(S1), is:

showing that the singlet excited state has a finite lifetime that is dependent on
the efficiency of the several competing deactivating processes. The radiative
lifetime of S1 is defined according to:

Thus if the only deactivation process of S1 was fluorescence then TS = TR, and
oF = 1. An important relationship is:

When one measures TS and 4F then the intrinsic molecular rate constant,
kR = TR, can be calculated.

40
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

4 Fluorescence spectrometers
The above section has outlined the physical parameters that describe the fluor-
escence process. One can measure the fluorescence spectrum, F(h),the singlet
excited state lifetime, and determine the .These parameters can be inter-
preted in terms of the structure, environment and dynamics of the molecule of
interest. In this section, the different optical and electronic components com-
prising an instrument that can measure F(h) and QF will be described. This
instrument is generally known as a steady state fluorescence spectrometer, since
it integrates the fluorescence intensity over a given time period. Time-resolved
fluorescence instrumentation that is used to measure the excited singlet state
decay times is described in Chapter 3.
Figure 5 is a block diagram of the essential components of a fluorescence
spectrometer. All fluorescence spectrometers comprise a light source, an excita-
tion wavelength selector (the wavelength of excitation is determined from the
molecule’sabsorption spectrum), a sample holder, a second wavelength selector
to isolate the fluorescence wavelength, and a photodetector that converts any
incident fluorescence photons into an electronic signal. Another important

Flgure 5 A block diagram of a typical fluorescence spectrophotometer. L represents the light

source; ExM and EmM, the excitation and emission monochromators, respectively; BS is a
quartz glass beam splitter; F/P is a filter or polarizer; Sa is the sample holder; Sle and SIX
are the entrance and exit slits of the monochromators; RB is the rhodamine B sample used
in the reference channel; F; is the red filter in that channel; PMT, represents the
photomultiplier tubes which detect the fluorescence and reference fluorescence signals, S
and R, respectively; the shaded arrows represent the excitation light beam; the solid arrow
pointing towards RB represents c. 10%of the light reflected from BS; and the arrows from
Sa represent a portion of the fluorescence light intensity that is emitted in a sphere
surrounding the sample.

41
ARTHUR G. SZABO

component, the reference channel, is a system to create an electronic signal

that is proportional to the incident light intensity on the sample.

4.1 The light source

When a solution sample is excited to Sj the fluorescence that emanates from
the sample propagates in all directions. The fluorescence photons that pass
through the emission monochromator (EmM, Figure 5) are only a small fraction
of the total fluorescence intensity. It is necessary therefore to use a light source
that has a relatively high intensity light output. In very simple fluorescence
spectrometers used for routine dedicated analyses, a Mercury arc lamp may be
satisfactory, but it only produces light at a limited number of wavelengths in the
ultraviolet and visible spectral range (e.g. 253.7, 302.2, 313.2, 435.8, 546.1 nm).
The continuum lamps used in absorption spectrometry, deuterium and quartz/
iodine sources have too low an intensity and hence the fluorescence intensity is
also very low. Most fluorescence spectrometers use a high pressure xenon arc
lamp as an excitation source since it produces a useful high intensity continuum
light spectrum from 200 nm to 1000 nm (Figure 6). These lamps are designated
according to the electrical power consumed, i.e. 150 W, 450 W, 1000 W. In
instruments recently made available, a 75 W xenon arc is used, but it is pulsed,
so that the peak intensity of each pulse is high.

4.2 Wavelength selectors

In simple spectrometers that have a dedicated use, optical glass filters are often
used. There are a very large number of bandpass filters available commercially

Wavelength (nm)

Figure 6 A scan of the spectral output of a 450 W xenon arc lamp measured by placing an
optically dense sample of rhodamine B in the sample holder. The fluorescence passed
through a 610 nm cut-off filter, and EmM was set at 620 nm. Excitation bandpass, 2 nm;
emission bandpass 2 nm.

42
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

Wavelength (nm)
Figure 7 A trace of the monochromator spectral bandpass, with an emission slit width of
2 mm, and a reciprocal dispersion, Cof 4 nm mm-1.

that can be used to isolate a particular line of a light source, such as a mercury
arc, a portion of the continuum spectrum of a xenon arc, or a portion of the
sample's fluorescence spectrum. Besides selecting for the maximum trans-
mission wavelength of a filter, one should also consider the half bandwidth that
defines the range of wavelengths that the filter will transmit. In addition one
must be cognisant that the intensity of transmitted light through a filter will be
reduced as the bandwidth is reduced.
The excitation wavelength, AEX, or fluorescence wavelength, AEM, of light is
often selected through a monochromator. The dispersion element in a mono-
chromator may be either a grating or a quartz prism. A grating monochromator
is preferred since the reciprocal dispersion, D-1, is constant over all A, while for
a prism this parameter varies with X and is cumbersome to account for. Hence a
quartz monochromator is rarely used in most standard laboratory instruments.
D-1 is the parameter that relates the effective bandwidth of A passing through a
monochromator to the slit widths (mm) of the monochromator and has units of
nm mm-1 (Figure 7). The value of D-1 depends on the quality of the mono-
chromator, but typically it ranges from 2 nm mm-1 to 8 nm mm-1.
One further point to note is that a grating monochromator disperses light in
a series of orders. Thus a monochromator setting at 500 nm will also pass light
whose wavelength is 250 nm since the second order of dispersion of light at A1
will occur at 2A1.Placing an appropriate cut-off filter after a monochromator,
i.e. a Pyrex filter in this example, can efficiently remove this second-order
dispersion.

43
ARTHUR G. SZABO

4.3 Sample excitation components

Light from the xenon arc is collected with appropriate mirrors in the lamp
housing and then focused through a lens into a light beam. The appropriate AEx
is determined from the absorption spectrum of the sample and is selected using
either a filter or monochromator (ExM, Figure 5). In most laboratory instruments
a monochromator is used to select \EX and entrance and exit slit widths set the
effective bandwidth of \. Choice of AEX and slit widths are discussed below. An
excitation double monochromator is an advantage when one is working with
samples that are highly light scattering, such as cellular suspensions, micelles,
or membrane preparations. This dramatically reduces the stray light that passes
through the monochromator and falls on the sample. The use of a double
monochromator will result in a lower light throughput and may necessitate
using a higher power light source.
The light beam passing through ExM is collected by a lens that collimates the
light beam. The next element that is usually encountered is a thin piece of
quartz glass, placed at an angle of 45° to the light beam. It can be appreciated
that any fluctuation or 'jitter' in the light intensity of the xenon arc will cause a
corresponding fluctuation in the sample fluorescence signal, since the fluor-
escence intensity is proportional to the intensity of the excitation. In addition
the light output of a xenon arc decreases as the lamp ages. Further, if one wishes
to compare the fluorescence intensity at two or more AEX then the intensity
spectrum of the light source at the two wavelengths needs to be accounted for.
The beam splitter reflects a small fraction (c. 10%) of light into a 'reference'
channel having a cell containing a concentrated ethylene glycol solution of one
of the rhodamine dyes, usually rhodamine B (RB). The high concentration of RB
(Protocol 1) results in the solution having an absorbance >2.5 from 250 to 600 nm.
Effectively all light at X < 600 nm incident on this cell is absorbed, and most
within the first mm of the solution. The dye fluorescence occurs at A > 600 nm
and a red cut-off filter placed after the dye cell permits only the dye fluorescence
to pass. Since the RB absorbance is so high any variation in lamp intensity will be
directly reflected in dye fluorescence intensity. The fluorescent photons from
the dye fall onto a photodiode or photomultiplier that subsequently generates
an electronic signal, R, that is proportional to the reference fluorescence
intensity.

4.4 Sample compartment

The sample cells used in fluorescence spectroscopy have all four sides clear,
since in most instruments the fluorescence photons are detected at an angle of
90° to the propagation direction of the excitation beam. The 90° geometry is
used in order to minimize any interference in the fluorescence light detection
by the excitation light beam. In most cases the cuvettes are made from fused
silica (quartz) but if the excitation and fluorescence wavelengths are above 300
nm plastic disposable cuvettes may be used.
The deactivation processes of S1(O) are often highly temperature dependent

44
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

Protocol 1
Preparation of reference solution
Equipment and supplies
• High purity rhodamine B Pasteur pipette
• Ethanol 1-cm square quartz stoppered cuvette or 1-
• Ethylene glycol, spectral grade cm triangular quartz stoppered cuvette
• Methanol, spectral grade Sonication bath
• 25-ml scintillation vial with a screw cap Optical tissues
• Polyfilm Latex gloves

A. Solution preparation
1 The rhodamine dyes are considered to be toxic. It is important to wear gloves while
handling the dyes and to avoid all spills. Weigh 15 mg of rhodamine B into the
scintillation vial.
2 Add 0.5 ml of ethanol to aid the solubility of the rhodamine. Add 4.5 ml of ethylene
glycol and cap the vial. Shake the solution and then hold it in a warm sonication
bath until there is no evidence of any particulate matter in the sample. The
sonicator assists in the solution of the rhodamine.
3 Totally fill the cuvette with the rhodamine solution using a Pasteur pipette.
Stopper the cuvette, and wrap the stopper and top of the cuvette with a small piece
of polyfilm. Wipe the outside of the cuvette with a tissue wet with methanol and
then a dry tissue,
4 Place the cuvette into the reference cell holder of the instrument.
5 The solution should be replaced approximately every six months of steady use.
B. Cleaning the optical elements of the reference channel
1 It is a good idea to clean the optical elements of the reference channel. The beam
splitter and red cut-off filter should be carefully wiped with a tissue wet with
methanol followed by wiping with a dry tissue.

and so the cuvette should be contained in a cell holder that can be thermo-
stattcd. The fluorescence photons are then collected with a lens and focused
onto the fluorescence wavelength selector.
A sample holder that can hold four cuvettes in a rotating turret is an advantage.
In this way other samples and blank solutions can be maintained at the same
temperature and one can simply rotate between the four cuvettes.

4.5 Fluorescence path optical components

In simple fluorimeters where a limited number of analytes are measured a
bandpass or cut-off filter may be used to select the fluorescence photons. If a

45
ARTHUR G. SZABO

cut-off filter is used, then light of all X above the cut-off X will be detected. This
will permit the measurement of very low fluorescence light intensities and
correspondingly low concentrations of the analyte. In standard laboratory
fluorimeters where a large variety of different fluorescence measurements will
be performed, a grating monochromator (EmM, Figure 5) is used to select AEM.
The fluorescence from the sample is focused onto the entrance slit of EmM and
after passing through the exit slit falls onto the photocathode of a photo-
multiplier tube (PMT).

4.6 Fluorescence instrumentation electronics

The PMT converts the photon intensity into an amplified electronic signal that
can be processed with electronic circuitry in either an analogue or digital mode.
The photon energy falling on a photocathode of a PMT causes the ejection of
'photoelectrons' from the bi-alkali metal surface. These electrons in turn cause
the ejection of electrons from a dynode, and this process is repeated at several
subsequent dynodes until a very large number of electrons reach the anode of
the PMT. The electron current is proportional to the number of photons falling
on the photocathode surface and hence the fluorescence intensity. The anode
current then passes into electronic circuitry where it is translated into a voltage
signal. This analogue signal, S, is proportional to the average photon density
falling on the PMT. By appropriate electronic circuitry the signal S can be
divided by the internal reference signal, R, to provide the ratio, IF = S/R and thus
IF will be independent of any variation of excitation light intensity. The signal,
IF, can then be relayed to a readout device such as a chart recorder. More
commonly, today's instruments are interfaced to a personal computer where
the electronic signals can be stored, processed or displayed.
4.6.1 Photon counting
Many of the current spectrofluorophotometers operate in a digital or photon
counting mode. In these instruments, the electronic circuitry permits recording
of individual fluorescence photons as an anode pulse. This has several positive
advantages. Most importantly thermionic emission of electrons from any of the
dynodes of the PMT that also results in an anode signal will be rejected by the
electronics of the instrument. These pulses will have a lower voltage than an
anode pulse originating from a fluorescence photon that falls on the photo-
cathode. This latter anode pulse will have a narrow band of a characteristic
voltage, and pulses of lower voltage can be electronically discriminated and not
recorded. By contrast, in an analogue instrument all thermionic emission from
the dynodes is recorded and hence the background or 'dark' signal will be much
higher than in a photon counting instrument. Since one is counting photons, it
can be readily appreciated that the instrumental sensitivity will be higher in
these instruments, as single photon events will be detected. In photon counting,
the detection system is much more stable and less affected by small variations
in the PMT voltage. A disadvantage is that stray light sources or light leaky
monochromators may cause a background signal from such parasitic light. One

46
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

additional disadvantage is that high light intensity levels result in a significant

non-linearity, as two photons arriving simultaneously at the photocathode will
only be counted as a single photon, or depending on the electronics will be
rejected entirely. However, this situation can be easily avoided in a number of
convenient and different ways. The enhanced sensitivity and lower background
of a photon counting instrument outweighs any disadvantages.

4.6.2 Computer interfacing

Most instruments sold today that function in either an analogue or photon
counting mode are interfaced to a personal computer. The computer software
from the supplier should control the entire operation of the spectrofluoro-
photometer, save the setting of the slits. The software will control the stepping
motors for changing the monochromator A, the sample PMT and reference PMT
voltages, and in some cases the rotation of the sample turret and any light
polarizers (vide infra) in the instrument. The computer interfacing will be used
to set all instrument parameters required for the recording of a fluorescence
spectrum, including the spectral wavelength range, the scan rate (nm s-1) and
the related integration time at each A position (photon counting instrument).
Because the fluorescence spectrum is recorded and stored in the computer
memory, it subsequently can be conveniently manipulated. Blank or reference
spectra can be subtracted; spectra can be normalized, smoothed according to
some spline function and integrated, and multiplied or divided by different
factors for spectral comparison; and differences between spectra can be re-
viewed. Finally, the output of such instruments can be conveniently formatted
to whatever representation one requires.

5 Fluorescence spectra
5.1 Inner filter effect
After the sample's absorption spectrum has been measured, one can select an ex-
citation wavelength, AEX, that falls within the absorption envelope. There will be
other factors that may influence the selection of AEX. If the sample is homogene-
ous and contains a single molecular chromophore, then any wavelength may be
appropriate. In the case of samples that contain more than a single molecular
chromophore, such as in a protein (vide infra), then the choice of XEX will depend
on whether preferential excitation of one of those chromophores is desired.
An important parameter of the sample that must be considered is the
absorbance of the sample at AEX. Equation 8 shows that the rate of fluorescence
emission, IF, is determined by the rate of light absorbance. In dilute solution
equation 8 can be expanded in terms of a series and where ecl is small then it can
be approximated that:
IF = I0(ecl) oF 15
This approximation can lead to an error known as the inner filter effect. The
solution at the front of the cell is exposed to a higher excitation light intensity

47
ARTHUR G. SZABO

Table 2 Percentage error due to inner filter effect

Absorbance Percentage error In IF

0.01 1.1
0.05 5.5
0.10 10.6
0.20

than the sample near the opposite side, because of the absorbance of the inter-
vening solution. This will introduce an error and non-linearity in the relation
between IF and A. The terms in the series expansion of equation 8 that have been
neglected in equation 15 can be evaluated and the percentage error in the
fluorescence signal owing to the inner filter effect can be calculated (Table 2).
One usually adjusts the concentration of the sample so that A at AEX is <0.1 in
order to minimize inner filter errors. If it is necessary to perform an experiment
at a certain concentration of the sample and this results in a very high absorb-
ance at the absorption maximum, XAMAX. then a XEX can be selected on the low
energy side of the absorption spectrum where A < 0.1.

5.2 Light scattering

All samples will scatter the exciting light to some extent. There are two scatter-
ing phenomena that the operator must be aware of. One is elastic or Rayleigh
scattering and this depends on the particle size of the solute or any suspended
material. The intensity of Rayleigh scattering, IRS, is proportional to r6/A4, where
r is the particle radius. Scattering from the sample results in a large 'scatter'
signal at emission wavelengths, XEM, close to XEX. Figure 8 demonstrates the
origin of this effect. In this example XEX and XEM are 5 nm apart. If the bandpass
of each monochromator is 4 nm (typical of many measurements), there is an
overlap in the bandpass curves and light wavelengths in the shaded area will
contribute to the signal in the emission detector owing to sample scattering.
Obviously, one way to reduce such a scatter signal would be to reduce the
bandpass of both ExM and EmM. However, this significantly reduces the signal
that one is measuring. Thus a compromise needs to be reached between scatter
contribution and fluorescence signal intensity. As stated above most fluor-
escence spectral measurements use bandwidths of 4 nm. Therefore, one usually
begins the recording of the fluorescence spectrum from a XEM that is 10 nm
higher than XEX. In a solution of a single chromophore this scattering artefact
can be easily avoided by choosing anAEXclose to XAMAX. In such a case, one should
choose a starting wavelength for the fluorescence measurement, XI,that is on
the low energy (red) edge of the absorption spectrum. Otherwise another inner
filter error is encountered where there is appreciable self-absorption of the
fluorescence in the absorption envelope. Measuring the signal from a solvent
blank cannot correctly account for elastic scattering as there is no scattering
solute in the blank. In the case of fluorescence measurements of cell suspen-

48
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

Wavelength (nm)
Figure 8 A demonstration of the overlap of excitation and emission monochromator
bandpass whenAEXand AEM are close to one another thus explaining the observation of
Rayleigh scattering of AEX.

sions, dispersed particulate samples, and micelles or vesicle samples, Rayleigh

scattering is a significant problem. A double monochromator that has lower
stray light and a narrower bandpass can reduce this effect. A method to correct
for the scattering of such samples has been proposed and applied with some
success (1, 2).
The second type of scattering that cannot be avoided is inelastic Raman
scattering of the solvent. In this phenomenon some of the excitation light
energy is abstracted by vibrational modes of the solvent molecules. In the case
of water or hydroxylic solvents, the most dominant vibration that absorbs this
energy is the 0-H stretch. The energy of the O-H vibration is observed near
3300 cm-1. The Raman signal from the solvent will be observed at a wavelength
that is 3300 cm-1 lower in energy than AEX. The A of Raman scattering,ARA,can
be calculated in the following way:

For example with AEX = 290 nm a Raman scatter peak will be observed at ARA =
321 nm. Figure 9, includes a spectrum of an aqueous blank that shows this
Raman band and the effect of the Raman signal is seen on the sample fluor-
escence spectrum. In the latter case it appears as a weak shoulder on the high
energy side of the fluorescence wavelength maximum, XFMAX- When the sample
fluorescence is intense the contribution of the Raman band will be negligible. A
simple test to determine if a spurious peak is due to a Raman scattering signal is

49
ARTHUR G. SZABO

Wavelength (nm)
Figure 9 The dashed curve is the fluorescence spectrum of a protein sample withAEX= 290
nm. The solid curve with circles is the spectrum of the buffer blank, showing a maximum
signal at 321 nm, corresponding to the Raman scatter of the water. The solid curve is the
signal resulting from subtracting the buffer blank from the raw spectral data.

to shift XEX by 10 nm to a lower wavelength. The Raman band will shift in the
same direction.

5.3 Instrumental settings

The first step is to set ExM at XEX considering the factors discussed above. In
most cases an excitation bandpass of 4 nm is satisfactory. If the sample fluor-
escence is weak and Rayleigh scattering can be avoided then the fluorescence
signal can be increased by increasing the slit width of ExM. By varying XEM of
EmM in a regular manner the fluorescence signal, IF, plotted against XEM gives
the sample fluorescence spectrum FM(A) (Figure 9). A 4 nm bandpass is typical of
most measurements unless there are very narrow vibrational band features in
the fluorescence spectrum that one wishes to resolve. The fluorescence scan is
initiated at AI that is usually 10 nm > AEX in order to avoid a scattered light com-
ponent in the spectrum. A wavelength range of between 100-200 nm is scanned
depending on the sample fluorescence spectrum. It is good practice to record
the fluorescence spectrum to AF where the fluorescence intensity is reduced to a
value close to zero. Recall, the recorded signal has been ratioed by the reference
signal, R. Since most instruments now use stepping motors to scan the mono-
chromators the step size becomes important. In the interest of good spectral
resolution a step size of no greater than 1 nm should be selected. The integra-
tion time chosen for each step should be sufficient to obtain a good signal to
noise, S/N, ratio. An integration time of 0.5 s or 1 s is usually sufficient. If a very
wide wavelength range is being scanned then the total time to acquire the total

50
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

spectrum becomes a consideration. This becomes a highly important factor if

the sample undergoes any appreciable photodecomposition (kPR) as this will
reduce or modify the fluorescence signal.
It is always necessary to record a solvent/buffer blank under identical instru-
mental conditions. This can be subsequently subtracted from the sample
fluorescence spectrum for quantitative purposes.

5.4 Fluorescence spectral corrections

When fluorescence spectra having different shapes and X distributions are to be
compared quantitatively it is necessary to correct for the wavelength-dependent
sensitivity of the EmM/PMT combination. If the spectra of a solute under differ-
ent sample conditions have identical shape and only vary in intensity, then no
instrumental correction is necessary. PMTs do not have constant photon
sensitivity across the useful UV-VIS spectral range. A typical photon sensitivity
curve is shown in Figure 10 for a Hamamatsu R955 PMT, a PMT that is used in
many instruments. It would be desirable is to have a flat wavelength response.
Obviously, this sensitivity versus A variation will distort the true fluorescence
spectrum of the solute and lead to errors in quantitation, especially where
spectral integration is performed. Superimposed on the PMT A sensitivity
variation is a smaller effect of the light throughput dependency on \ of the
monochromator.
There are three methods of generating spectral correction curves to account
for these sensitivity variations. One is to measure the fluorescence spectra of
standard compounds whose true fluorescence spectra have been determined
and reported (3). Then the sensitivity factors could be calculated by comparing
the experimentally measured spectra with the reference spectrum.

600 too no
Wavelength (nm)
Figure 10 The spectral sensitivity curve of a Hamamatsu R955 photomultiplier tube.

51
ARTHUR G. SZABO

A second method that is often used is to illuminate a solid scatterer, such as

MgO, which is held in the sample holder, using a calibrated standard tungsten
lamp available from the National Institute of Standard Technology (US). Such
lamps act as a black body radiator that has a smooth light output intensity
variation with wavelength. Light from the scatterer passes through EmM and
is detected by the PMT. The measured spectral intensity curve is compared to
the lamp calibration values and a reliable spectral correction curve can be
generated and stored in the instrument computer for use in correcting spectra.
A limitation of this procedure is that such tungsten lamps have negligible
output at X. < 360 nm. Thus a correction curve can be determined from
360-800 nm with this lamp system. A further consideration is the necessity of
purchasing a calibrated lamp, housing and power supply at an appreciable
cost, in order to determine a correction factor curve that needs to be measured
only rarely.
A method that has shown to be quite satisfactory and relatively straight-
forward is described in Protocol 2. In this method a diffuser plate or an MgO
scatterer plate is placed in the sample holder and then both ExM and EmM are
scanned synchronously from 250-600 nm. This will produce a spectrum of the
xenon arc source that has been distorted by the sensitivity variation of the
emission detection system. The reflector is replaced by a concentrated sample of
rhodamine B in a triangular cuvette. The cuvette is oriented so that the
hypotenuse is on the face opposite to EmM. This is a similar configuration to
the cell in the reference channel of the instrument. A red cut-off filter is placed
in front of the entrance slit to EmM and XEM is set to 620 nm. Then ExM is
scanned from 250-600 nm and the spectrum is recorded. This spectrum is a
close representation of the true xenon lamp intensity profile. Often neutral
density filters will have to be placed in the emission beam in order to prevent
signal saturation of the PMT. The two curves thus generated can be compared
and the correction factor curve, C(\), calculated:

where FRB(XEX. 620 nm) is the recorded spectrum of rhodamine B with EmM
constant at 620 nm and AEX is varied, and FMM (AEX, AEM) is the recorded spectrum
using a diffuser plate of MgO and both ExM and EmM are varied synchronously.
If FM(X) represents the measured uncorrected sample fluorescence spectrum
then a corrected sample fluorescence spectrum can be obtained by a multi-
plication of C(X) and FM(X) according to

Thus the fluorescence signal at each measured X of the sample is multiplied by

the correction factor at that X. Again this process can be easily achieved in the
data acquisition computer.

52
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

Protocol 2
Generation of spectral correction factors (250-600 nm)
Equipment and supplies
• A quartz diffuser plate mounted in a • A triangular quartz cuvette filled with a
holder so that the plate is oriented at an concentrated ethylene glycol solution of
angle of 45° to the excitation light rhodamine B (see Protocol 1)
propagation direction or a metal plate • Quartz neutral density filters
coated with an MgO film. This plate
• Fine wire mesh screens
should be mounted in a holder so that it
makes an angle of 35° to the excitation • 610-njn glass cut-off filter
light propagation direction.

A. Diffuser set-up and signal levels

1 Place the diffuser plate or MgO plate into the sample holder. In the case of the
diffuser plate, the incident light face should be on the side opposite to the EmM
entrance slit. In this way light that passes through the plate and is scattered by it
will fall on the entrance slits, but reflected light will be avoided. For the MgO plate
the incident face should be on the same side as the entrance slit of EmM.
2 Adjust the slits of ExM to a 1-nm bandpass. The bandpass of EmM should be 4 nm
or whatever baadpass is commonly used for fluorescence spectral determination.
Place a Neutral Density (ND) filter with ND = 1 in front of EmM. Set both
monochromators to 400 nm and open the shutters of the instrument. Note the
signal level of the emission detector. The level should be well below that where the
PMT is being saturated with light. If saturation is observed then place additional
ND filters or fine wire mesh screens in front of the EmM entrance slits until a
modest signal level is achieved.
B. Instrument settings for F L (A)
1 Set the instrument so that both monochromators will run synchronously.
2 Set both monochromators to 250 nm.
3 Set the wavelength scan range to 250-600 nm.
4 Set the instrument so that the signal detected is not ratioed by the reference signal, R.
5 Set a scan rate of 2 nm s-1 if the instrument monochromators are not controlled by
stepping motors. In the case where the monochromators are controlled by stepping
motors set the step size to 0.5 nm and the integration time to 0.2 s. This step size
will be such that most sample fluorescence spectra can be corrected.
C. Record spectrum FL(A)
1 Record the spectral light signal between 250-600 nm with both monochromators
scanning synchronously. This spectrum designated FL(A) = FMM (A EX , A tM ) should be
similar in shape to the spectrum of the xenon lamp. It is distorted by the A
sensitivity of the emission detection system.
ARTHUR G. SZABO

Protocol 2 continued

D. Rhodamlne sample
1 Replace the diffuser plate by the triangular cuvette of rhodamine. The incident face
should be opposite to the entrance slit of EmM. In this way the only light incident
on EmM will be rhodamine B fluorescence. Reflected light is directed in the
opposite direction.
E. Instrument settings for FT(A)
1 Remove the ND filters and any wire mesh screens and replace them with the 610-
nm cut-off filter.
2 Set ExM to 250 nm and EmM to 620 nm.
3 The wavelength scan range and scan rate should be the same as for the previous
spectrum.
4 Set the instrument so that only ExM will scan,
5 Record the spectral light signal between AEX 250-600 nm. This spectrum is
considered to be the true lamp spectrum, FT(A) = FRB(AEX. 620 nm). Since EmM and
the PMT only 'saw' a single wavelength (620 nm) any A bias was removed from the
emission detector.
F. Data handling and C(A)
1 Using the computer software divide ST(A) by SL(A).
2 Normalize the correction factor curve that is obtained so that the value at 400 nm
is 1. Most software packages have a normalization command. The curve thus
obtained (equation 17) can be used to correct all fluorescence spectra between 250
and 600 nm.
3 A fluorescence spectrum is corrected according to equation 18.

5.5 Excitation spectra

A useful fluorescence spectral measurement especially when studying mixtures
of substances is an excitation spectrum. A EM . a wavelength representing a point
on the fluorescence spectrum, is held constant and then AEX is scanned by
changing the A of ExM. This scan is usually carried out across the wavelength
range of the absorption spectrum of the sample. The spectrum that is traced out
should correspond closely with the absorption spectrum of the molecule that is
responsible for the fluorescence (Figure 11). This spectrum reveals whether the
sample is homogeneous, and whether all fluorescence features result from a
single molecule. For example in Figure 12(a) the fluorescence spectrum shows a
pronounced shoulder at long wavelengths. By measuring the excitation spec-
trum at two different fluorescence wavelengths, one at the high energy (blue)
(AB) side and the other at the low energy (red) (AR) side of the spectrum, the
shape of the absorption spectrum of the molecule or molecules from which this
fluorescence originates can be obtained. Figure 12(b) shows the result where the

54
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

Figure 11 The excitation spectrum of a protein sample with the AEM = 340 nm.

fluorescence spectrum in Figure 12a results from a mixture of two molecules.

Each molecule has a different absorption and fluorescence spectrum. If the
fluorescence originated from the excitation of a single chromophore, then the
excitation spectra will have identical shape. The intensity will be proportional
to the fluorescence intensity at each AEM. Excitation spectra can also provide
information as to whether FRET is occurring between two chromophores in a
sample, since the excitation spectrum will include signals from the donor
absorbance.

5.6 Quantum yield measurement

In many studies using fluorescence for analytical purposes, knowledge of the
absolute quantum yield oF is not necessary. Usually one is only comparing
fluorescence intensities of different samples and so the relative fluorescence
intensity offers a satisfactory measurement. In many other studies one wants to
compare the fluorescence efficiencies of the samples in order to relate the
fluorescence to the structural properties and interactions of the molecule. This
is the situation when biomolecules such as proteins are being studied. It is
necessary then to determine oF of the fluorescent chromophore. This is also the
case if one is interested in the photophysical parameters of a chromophore.
When combined with the measured singlet lifetime, TS, according to equation 14,
then the radiative rate constant, kR can be calculated. This permits the evalua-
tion of the rate constant of the non-radiative deactivation processes, (kNR, of the
excited singlet state of the chromophoric molecule being studied. Since kR is an
intrinsic rate constant of a chromophore, direct comparison of oF of the

55
ARTHUR G. SZABO

Wavelength (nm)
(a)

Wavelength (nm)
(b)

Figure 12 (a) A fluorescence spectrum showing two spectral maxima. Xr is the fluorescence
wavelength setting selected to determine the excitation spectrum responsible for the low
energy fluorescence maximum; Xb is the fluorescence wavelength setting selected to
determine the excitation spectrum responsible for the high energy fluorescence maximum,
(b) The solid line is the excitation spectrum when EmM was set at Xr; the dashed line is the
excitation spectrum when EmM was set at \b for the sample whose fluorescence is shown
in (a).

56
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

chromophore under different conditions permits the correlation of oF with

molecular structural properties and inter or intramolecular interactions.
The measurement of the oF of a sample is generally carried out by com-
paring the sample fluorescence to that of a standard molecule whose oF has
been determined using an absolute method. The basis of this 'secondary'
method of comparison of oF of the sample to that of a standard is the following

where SM(X) and SS(X) are the integrated areas of the corrected fluorescence
spectra of the sample, M, and the standard, S, respectively; and oF (M) and oF (S)
are their respective quantum yields; and AM and As are their absorbancies at AEX-
Thus

which on rearranging gives

f--oF(S) 22
AM
The determination of oF of a molecule by a primary method is a very difficult
task and has been the subject of several reviews (4, 5). The quantum yields of
several molecules under rigorous experimental conditions have been deter-
mined (6, 7). A selection that is generally useful is listed in Table 3. When such a
standard molecule is chosen in order to measure oF of the molecule of interest,
the exact sample conditions for the standard must be followed. Temperature
control is also critical owing to the temperature dependency of kNR. One of the
most often used standards is quinine sulphate in 1 N H2S04 at concentrations
<10-4 M. At 25°C its oF is 0.546 (AEX = 365 nm). For protein fluorescence, where
tryptophan is the fluorescent chromophore, N-acetyltryptophanamide (NATA) at
pH 7, 20 °C is used as a secondary standard. The accepted value of oF of NATA is
0.14 (8).
When the oF of the sample is determined, F(A) of the sample and the stand-
ard should cover a similar X range. Obviously, both the sample and standard

Table 3 Fluorescence quantum yield standards

Standard/conditions Spectral range (nm) Excitation / (nm) oF

N-Acetyltryptophanamide/ 300-450 280-300 0.14

aqueous buffer, pH 6-7, 20°C
Quinine sulphate/ 380-580 260-390 0.546
1 N H2SO4, 25°C
2-Aminopyridine/ 315-480 250-300 0.66
0.1 N H2S04, 20°C
9,10-Diphenylanthracene/ 390-530 320-380 0.95
ethanol, 20°C, degas

57
ARTHUR G. SZABO

should absorb at the same XEX- Since it is important to minimize inner filter
effects the absorbance of both the sample and the standard should be <0.05 at
XEX. This becomes one of the largest sources of error in measuring oF since
there is a large percentage error in measuring such low absorbancies. A means
of reducing this error is to measure the absorbance of the sample and standard
in a 5 cm cylindrical cuvette whose outside diameter is 7 mm. The absorbances
of these solutions (c. 2 ml volume) are measured in this cuvette, and then the
solutions are transferred to a 1-cm fluorescence cuvette for the fluorescence
measurements. In this way an absorbance that is five times that in a 1 cm
cuvette is measured and the error in the absorbance is. significantly reduced.
The fluorescence spectra of both the sample and standard should be scanned
over an identical scan range, and it is important that FS(X), the fluorescence
spectrum of the standard, should have reached a value close to zero on its red
edge. The solvent blank is subtracted from these spectra. Since it is unlikely that
the sample and the standard will have identical F(A), the spectra should be cor-
rected for the X sensitivity of the PMT. The resulting corrected spectra, F(A)coRR
are then integrated across the same spectral range, i.e. from Xr to AF. This
integration can usually be conducted in the instrument computer. Alternately,
the integration can be accomplished using Simpson's rule or a trapezoidal
approximation over narrow X ranges (1-5 nm). Prior to the availability of com-
puter interfaced instruments, the integration was accomplished by cutting out
the spectral recording on a chart paper and weighing the paper cut-out.
When one is measuring oF of a fluorophore in an aqueous solution and the
standard is in a non-aqueous solvent then a refractive index correction is
recommended. This correction is applied according to:

23
n2(standard)
where oF' is the measured quantum yield without correction for refractive
index difference, oF is the corrected quantum yield value, n2(sample) is the
square of the refractive index of the solvent of the sample solution, and
n2(standard) is the square of the refractive index of the solvent of the standard
solution. If the sample had been dissolved in benzene and the standard was
quinine sulphate the correction would have been 27%.

6 Fluorescence applications
The application of fluorescence measurements to a large variety of chemical,
biochemical, biological and physical problems is extensive. The majority of
fluorescence applications involve the use of extrinsic fluorescence probes. These
are chromophoric molecules that are attached to or adsorbed onto another
molecule and their fluorescence is measured. These molecules probe the prop-
erties of the substances to which they are attached. The catalogue and manual
published by Molecular Probes (9) is an excellent collection of a large number of
different examples of the use of fluorescence, especially as they relate to bio-

58
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

Protocol 3
Measurement of OF of a sample
Equipment and reagents
• Purified sample • Four 1-cm quartz cuvettes
• Purified standard • A 5-cm quartz cylindrical cuvette
• Purified solvents (outside diameter = 0.7 cm)

A. Selection of standard
1 A standard material is selected from a list of quantum yield standards (6, 7 and
Table 3). The standard should have an absorption spectrum in the same range as
that of the sample. The fluorescence spectra of the standard and sample should
span a similar A range.

B. Excitation wavelength selection and absorbance measurement

1 Select AEX for the sample and if there is a single chromophoric species in the
sample then this wavelength should be one where the absorbance does not change
rapidly. The standard should also absorb at this A.
2 Measure and record the absorbance of the sample and standard at this AEX,
subtracting any solvent blank contribution, using the 5-cm cylindrical cuvette.
Adjust the concentration of the sample and the standard so that A(5 cm) - 0.25.
The absorption spectrophotometer bandpass in these measurements should be
nearly the same as the bandpass of ExM, that is typically 4 nm.
3 Transfer the solutions to separate 1-cm rectangular quartz fluorescence cuvettes.
Two cuvettes should also contain the respective blanks for the sample and standard.
C. Spectral measurement
1 Set the bandpass of ExM and EmM to an appropriate value, typically 4 nm.
2 Measure F(A) of the sample and standard over the identical A range. It is important
that the fluorescence of both are close to zero at the long A of the scan. The initial
or starting wavelength, A1. should be selected so that the fluorescence of both
solutions is as close to zero as possible. This may be difficult to achieve in highly
scattering solutions.
3 The scan rate or step size/integration time should be set to minimize the noise on
the spectrum,
4 Measure the background signal of the solvent blanks for the two solutions.
D. Data handling and calculations
1 Subtract the solvent blank spectra from the fluorescence spectra of the sample and
standard.
2 Multiply these blank subtracted spectra by the correction factor curve, C(A)
(Protocol 2).

59
ARTHUR G. SZABO

Protocol continue

3 Integrate f(A)CORR over an identical A range to obtain SM(A) and SS(A).

4 Calculate oF(M) of the sample by substituting the appropriate values into equation 22.
5 If the solvents used are different then the value of oF(M) should be corrected for
the refractive index (equation 23).

chemical, chemical and biological investigations. Intrinsic fluorescence probes

are those chromophoric species that occur as natural constituents of the sub-
stances under investigation. An area where intrinsic fluorescent chromophoric
components have been extensively utilized is in the field of protein fluorescence
(see Chapter 9).

6.1 Fluorescence resonance energy transfer

A highly useful application of fluorescence spectroscopy is the technique
known as fluorescence resonance energy transfer (FRET). This is treated in more
detail in Chapter 3. The basic principle is to have a pair of chromophores attached
to a molecule. One chromophore, the donor, absorbs the excitation energy and
transfers this excitation energy to the second chromophore, the acceptor,
through a dipole-dipole interaction mechanism. The Forster equation describes
the physical terms that determine the efficiency of this process. FRET is used as
a 'spectroscopic ruler' to estimate the distance between attachment points of
the donor and acceptor on a large macro molecule such as a protein. It can also
be useful in providing information on distance distributions owing to the
backbone dynamics of the molecule.

6.2 Fluorescence anisotropy

Fluorescence anisotropy measurements and their applications are covered in
Chapter 3, Plane polarized exciting light can be obtained by placing a Polaroid
sheet or a Clan Taylor prism in front of the sample holder. The polarization or
anisotropy of the fluorescence of the sample is then determined by placing
another polarizer in front of'EmM, The fluorescence that is polarized parallel, I],
and perpendicular, J±, to the excitation polarization direction is then measured.
The fluorescence anisotropy, r, of the sample is defined according to

The denominator is proportional to the total fluorescence intensity, r is in the

main a measure of the rotational motion or dynamics of the molecule, the latter
being very useful in obtaining flexibility parameters of local segments of a
macromolecule.
6.2.1 Polarization effects on fluorescence intensity
If one is measuring the oF of a macromolecule such as a protein, or a large
molecular complex, it is advised that two additional optical components be

60
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

included in the spectrophotofluorimeter. The light exiting ExM is not randomly

polarized, i.e. it is unpolarized but it will have a polarization bias in either the
vertical or horizontal plane. As a result the macromolecular fluorescence will be
anisotropic owing to slow rotational tumbling and the fluorescence will be
biased in a particular polarization direction. This can be corrected by placing a
depolarizer (a wedge plate) optic in the excitation light beam after the beam
splitter, which results in the excitation light being unpolarized. A polarizer is
placed in front of the slits of EmM and oriented at an angle of 35° to the vertical
plane. The fluorescence passing this polarizer will be proportional to the total
intensity and will avoid any polarization bias.
Another point to be aware of is that holographic grating monochromators
have a polarization anomaly known as the Wood's anomaly. At certain wave-
lengths the light is polarized entirely in the horizontal plane. This can cause
important spectral artefacts such as false maxima or false shoulders. The
anomaly can be avoided by using an excitation depolarizer and emission
polarizer as described above.

6.3 Protein fluorescence

One research area that has benefited from extensive fluorescence investigations
is the field of protein structure-function studies. There are three aromatic amino
acids that absorb light in the ultraviolet spectral range, phenylalanine (Phe),
tyrosine (Tyr) and tryptophan (Trp). Because of their molar absorptivity the
fluorescence of Tyr (s276 = 1405 M-1 cm-1) and Trp (E280 = 5579 M-1 cm-1) have
been extensively used to study proteins.
Trp, an indole amino acid, has received by far the greatest attention. There
are several reasons for this. Usually the number of Trp residues in a protein are
limited and there are several proteins containing only a single Trp residue. The
absorption spectrum of Trp with its higher E compared to Tyr has an extended
absorbance beyond the normal absorbance of Tyr. This permits the preferential
excitation of the Trp residues in a protein. Because of the electronic properties
of the Trp excited state, its fluorescence spectrum is highly dependent on the
nature of its environment. Selective intramolecular interactions can lead to
fluorescence enhancement or quenching. The origin of these effects are now
reasonably well understood permitting a discussion of the local molecular
structure in the vicinity of the Trp residue. Obviously if there are several Trp
residues in a protein it is difficult to make specific interpretations of their
individual behaviour. However, the fluorescence of such proteins can still be
useful if all one wishes to determine is an association constant. The literature
on Trp fluorescence and its application to the study of protein structures, inter-
actions and dynamics is vast and the information content of Trp fluorescence is
considerable.
Time-resolved fluorescence studies have shown that ground state conform-
ational heterogeneity of the Trp residue can be revealed. This heterogeneity has
been attributed to the fluorescence of different rotamers of the indole ring

61
ARTHUR G. SZABO

around the Ca-CB bond (10,11). Recently it has been shown how Tip analogues
when substituted into proteins in place of the normal Trp residue can provide
useful information on molecular details of interacting segments of protein-
protein and protein-nucleic acid complexes (12). These analogues have extended
ultraviolet absorption beyond the normal Trp absorbance and hence can be
selectively excited in the presence of a large number of normal Trp residues.
It is beyond the scope of this chapter to review this literature. The reader is
referred to several review articles that provide a useful summary (13-15).
It is useful, however, to discuss a few practical aspects of measuring Trp
fluorescence and some of the information that can be obtained. One of the first
factors that it is necessary to consider is the number of Tyr residues in a protein
that contains a single Trp residue. In most cases where the number of Tyr
residues is small, XEX = 295 nm will predominantly excite the single Trp residue
and only Trp fluorescence will be observed. This is because Tyr absorbance at
295 nm is very small (E295 = 18 M-1 cm-1) but not zero. When there are a large
number of Tyr residues in a protein, the sum of the absorbance of these residues
makes a modest contribution to the total absorbance and in these cases the
fluorescence generated with AEX = 295 nm will contain Tyr spectral components
(16). Also FRET from Tyr to Trp may occur and affect the integrated fluorescence
intensity. In order to virtually eliminate the Tyr contribution to the fluores-
cence, A EX can be changed to 300 nm. However, this generates another problem,
especially when one measures oF, since at typical protein sample concen-
trations of 1-5 x 10-6 M the absorbance at 300 nm is very low.
There is also another cautionary note to consider when the oF of proteins is
being measured. In order to selectively excite Trp fluorescence AEX is usually
chosen to be 295 nm or greater in order to minimize the Tyr absorbance. An
assumption that is usually made is that the molar absorptivity at 295 nm of the
particular protein under the different conditions of the experiments is the
same. For example if the thermal unfolding or chemical denaturation of a
protein is being studied using fluorescence hardly any attention is given to the
fact that E295 is in fact changing. This A. is on the very steep falling slope of the
protein absorption spectrum. Changes in E295 are not readily apparent on casual
inspection of the absorption spectrum. As a Trp residue that may originally
have been buried in the interior of a protein becomes solvent exposed, E295 will
change. When performing chemical denaturation experiments this can be
accounted for. However, corrections for these changes in thermal unfolding are
difficult to estimate.

6.4 Fluorescence quenching

6.4.1 Intramolecular quenching
In proteins the Trp fluorescence is often quenched by intramolecular inter-
actions with different amino acids. These quenching processes contribute to the
magnitude of the non-radiative relaxation rate constant, kNR, of the excited
state. Cysteine and especially disulfide bridges will quench Trp fluorescence

62
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

provided they are within 6-10 A of the Trp residue. This quenching process is
thought to be due to an electronic exchange process.
Histidine may quench Trp fluorescence through a proton transfer mech-
anism. There is evidence (17) that position 4 of the indole ring acquires more
electron density in the excited singlet state and can accept a proton from a
proton donor such as histidine.
The neutral amide function of glutamine or asparagine, as well as the amide
backbone of the protein, can quench Trp fluorescence. This is another short-
range electron transfer process. The short lifetime of one of the Trp fluor-
escence decay components may be due to the close proximity of the indole ring
to the amide backbone in that particular rotainer state.
When the Trp residue is located in a hydrophobic environment of the
protein its oF is enhanced and importantly the fluorescence spectral maximum
is shifted to a much higher energy than that normally seen for a solvent
exposed residue.

6.4.2 Intermolecular quenching

A very informative experiment that can be performed with a protein is the
quenching caused by the addition of a quenching molecule or ion to the
solution. Three species are often used as quenchers, acrylamide, iodide ion, I-,
and cesium ion, Cs+. The former is a neutral molecule while the latter two are
negatively and positively charged, respectively. The concentrations of quencher
that are used range from 0.01 to 0.5 M. In the case of proteins, acrylamide
quenching has been extensively used to determine the degree of exposure of
Trp residues. The experiment is conducted by adding aliquots of a concentrated
acrylamide solution to the protein and measuring the fluorescence of the
solution after each addition. This quenching process is described according to

and the second-order rate constant, kq, is the kinetic term that is a measure of
the process. In the absence of external quencher, Q,

where

and is the solute quantum yield in the absence of added quencher. In the
presence of quencher

which gives the ratio

63
ARTHUR G. SZABO

0.3

[Quencher] M
Figure 13 A set of typical Stern Volmer fluorescence quenching plots. •, a linear Stern
Volmer plot; A, a plot showing upward curvature owing to a high concentration of quencher
surrounding the fluorophore; •. a plot showing downward curvature owing to the presence of
two fluorescent chromophores being quenched with different efficiencies; *, a linear Stern
Volmer plot representing the case where the chromophore is less accessible to the quencher.

and since the unquenched singlet lifetime, TS = l/(kR + ko), equation 29 can be
simplified to

•= 1+ 30

This is the well known Stern Volmer relationship that relates the fluorescence
yield of the sample to the concentration of the added quencher. A plot of o 0 F /o F
or, since oF F, F°/F versus [Q] should yield a straight line with an intercept of 1,
and the Stern Volmer slope, KSV = kQTs (Figure 13).
Quenching by the addition of an external quencher is a diffusion controlled
process. As a first approximation the value of Ksv reveals the degree of exposure
to the solvent of a Trp residue. A high value suggests a solvent exposed residue,
while a low value indicates that the residue is buried inside the protein. How-
ever, this is not strictly correct since this assumes that TS is the same for both
cases. A more accurate estimation of the degree of exposure of Trp is to calculate
kQfrom Ksv = kQTS. In this case the value of TS should be the intensity weighted
decay tune when multi-exponential decay kinetics are observed (that is often the
case even for a single Trp protein). The intensity weighted decay time is

where 04 and TJ are the pre-exponential term and decay time of the ith fluor-
escence decay component, respectively. It is important to use the intensity

64
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

weighted lifetime to calculate values of kq since in a steady-state measurement

the fluorescent component that makes the largest contribution to the total fluor-
escence, the largest value of err, will be the one that dominates in the fluorescence
quenching experiment.
In a protein containing more than one Trp residue the Stern Volmer
quenching curve often shows downward curvature (Figure 13). This would result
because the less accessible Trp residues will be quenched after the more ex-
posed residue, and will have a lower value of kQ. However, the direction of
curvature will depend on the relative values of kQ and TS, especially in the case
of a single chromophore when upward curvature is observed. This is due to the
fact that at a high quencher concentration diffusion controlled kinetics are not
observed, and there will be an appreciable fraction of the chromophore that is
surrounded by the quencher at the instant of excitation. That molecule will be
immediately quenched and F will be less than that expected for a diffusion
controlled process.
In acrylamide quenching experiments it is important that XEX - 290 nm.
Acrylamide has an appreciable absorption at 285 nm. At the high concentra-
tions of acrylamide that are normally used (0.01-0.5 M) a large proportion of
excitation light at 285 nm will be absorbed by the acrylamide. This is an inner
filter effect that will render the data and analysis invalid.
When quenching by I~ or Cs+ is the experimental procedure, the ionic
strength of the solution must be held constant. In quenching experiments mM
concentrations of quencher are used. Constant ionic strength of the solutions
are maintained by the appropriate addition of salt solutions.

Protocol 4
Acrylamide quenching of protein fluorescence
Equipment and supplies
• High purity spectral grade actylamide • Volumetric pipettes
• High purity buffer for solution of the • 1-cm quartzfluorescencecuvettes
protein

A. Solution preparation
1 A stock solution of 5 M acrylamide is prepared in the sample buffer.
2 2.00 ml of a protein sample is prepared to give an absorbance at AEX- 0,05.
B. Fluorescence and tftratlon measurement
1 Record the fluorescence spectrum of the protein and of the buffer blank.
2 Add aliquots of acrylamide to the protein sample and measure the fluorescence
spectrum of the sample after each addition of acrylamide. The first addition should
be 5 fil of stock acrylamide, to give [acrylamide] = 0.0125 M. A second 5-ul aliquot is

65
ARTHUR G. SZABO

Protocol 4

added followed by a third aliquot of 10 fil. Subsequently five separate aliquots of

20 jil are added, followed by two separate aliquots of 40 ul each. This will provide a
final [acrylamide] - 0.5 M.
3 The same aliquots of acrylamide should be added to 2.00 ml of the buffer blank and
the background signal measured,
C. Data handling and analysis
1 The buffer blank signal is subtracted from the corresponding sample fluorescence
spectrum.
2 The integrated areas of each blank subtracted spectrum, SM(A) are determined. The
integrated area of the initial sample, [acrylamide] = 0M, gives F°. Each of the other
integrations are the corresponding values of F at the different concentrations of
acrylamide.
3 Calculate the ratio F °/F for each acrylamide concentration.
4 Correct the acrylamide concentration for the dilution factor of each sample
according to
[acrylamide] = [acrylamide stock]VA/(2000 + VA) M 32
where VA is the cumulative volume of acrylamide stock added for each sample
measured.
5 Plot the ratio F °/F versus [acrylamide] for each sample.
6 After inspection of the plot, determine the slope and intercept of those points that
appear to obey a linear relationship. For this a linear regression analysis is usually
performed.
7 The slope gives KSV, the parameter related to the accessibility of the Trp residue.
The intercept should be close to 1.
8 From time resolved fluorescence experiments determine the intensity-weighted
singlet decay tune, <t s >, and calculate kQ from Ksv =

7 Conclusion
The description above is a summary of the important practical aspects of steady
state fluorescence spectroscopic measurements. Further details and applications
can be obtained from a number of reviews and publications, a selection of
which are referenced below (18-21).

References
1. Teale,J. (1969) Photochem. Photobiol. 10: 363-374.
2. Lenz, B. R., Moore, B, M., and Barrow, D. A. (1979) Biophys. J. 25: 489-494.
3. Melhuish, W. H. (1972) J. Res. Nat. Bur. Stand. 76A: 547.
4. Demas.J. N, and Crosby. C. A. (1971) J. Phys. Chem. 75: 991

66
 FLUORESCENCE PRINCIPLES AND MEASUREMENT

5. Birks, J. B. (1977) Standardization in spectrophotometry and luminescence measurements (ed.

K. D. Mielenz, R. A. Velapoldi and R. Mavrodineanu). US Department of Commerce.
6. Miller, J. N. (1981) Standards influorescencespectroscopy. Chapman and Hall, London.
7. Parker, C. A. (1968) Photoluminescence of solutions. Elsevier, London.
8. Chen, R. F. (1967) Anal. Lett. 1: 35
9. Haugland, R. P. (1996) Handbook of fluorescent probes and research chemicals (6th Edn).
Molecular Probes, Eugene.
10. Dahms, T. E. S., Willis, K. J., and Szabo, A. G. (1995) J. Am. Chem. Soc. 117: 2321-2326.
11. Willis, K. J., Neugebauer, W., Sikorska, M., and Szabo, A. G. (1994) Biophys.J. 66:
1623-1630.
12. Ross, J. B. A., Szabo, A. G., and Hogue, C. W. V. (1997) Methods of Enzymology 278:
151-189 (ed. L. Brand and M. L. Johnson). Academic Press, New York.
13. Eftink, M. R. (1991) Protein structure determination: Methods of biochemical analysis 35:
127-207 (ed. Suelter, C. H.). John Wiley, New York.
14. Szabo A. G. (1989) The enzyme catalysis process (ed. A. Cooper, J. L. Houben, and L. C.
Chien), pp. 1213-1239. Plenum, New York.
15. Millar, D. P. (1996) Curr. Opin. Struct. Biol. 6: 637-642.
16. Willis, K. J. and Szabo, A. G. (1989) Biochemistry 28: 4902-4908.
17. Saito, L, Sugiyama, H., Yamamoto, A., Muramatsu, S. and Matsuura, T. (1984) J. Am.
Chem. Soc. 106: 4286-4287.
18. Millar, D. P. (1996) Curr. Opin. Struct. Biol. 6: 322-326.
19. Eftink, M. R. (1998) Biochemistry (Mosc.) 63: 276-284
20. Lee, Y. C. (1997);. Biochem. 121:818-825.
21. Soper, S. A., Warner, I. M., and McGown, L. B. (1998) Anal. Chem. 70: 477R-494R.

67
This page intentionally left blank
 Chapter 3
Time-resolved fluorescence
and phosphorescence
spectroscopy
Thomas D. Bradrick* and Jorge E. Churchicht
* Optical Spectroscopy Section/LBC/NHLBI, Bldg. 10, Rm 5D14,
MSC1412, National Institutes of Health, Bethesda, MD 20892-1412, USA
T
Dept. of Applied Biology and Chemical Technology, Hong Kong
Polytechnic University, Hong Kong, PRC

1 Introduction
In this chapter, we outline the basic theory and methodology for making time-
resolved fluorescence and phosphorescence measurements. We begin with a
brief discussion of the intrinsic time dependence of fluorescence and phosphor-
escence decays, and also introduce several important photophysical concepts.
Energy transfer measurements, which are important for determining molecular
distances, are then addressed, followed by the convolution integral (which
describes the luminescence decay that is actually observed in pulsed excitation
experiments) and the relationships that are used to determine fluorescence
lifetimes from phase/modulation data. Polarized fluorescence measurements,
which are an important tool for following molecular motions, are then
discussed, and the fluorescence portion of the chapter concludes with an
overview of data analysis and brief descriptions of the instrumentation that is
used in making time-resolved fluorescence measurements. The remainder of
the chapter is then devoted to a discussion of phosphorescence spectroscopy.

2 Background
2.1 Basic photophysics and time dependence of
fluorescence and phosphorescence decays
The intrinsic time dependence of fluorescence decays is derived in Section 3.3 of
Chapter 2. There it is shown that if a population of excited singlet-state
molecules is generated instantaneously, its size decreases exponentially with
time, as does the intensity of the emitted photons. The fluorescence lifetime TS
is typically used to describe the rate of decay, where Ts-1 = kR + kNR + kISC + kPR
(equation 12, Chapter 2) and kR, kNR, kISC and kpR are the rate constants for the

69
THOMAS D. BRADRICK AND JORGE E. CHURCHICH

various parallel unimolecular de-excitation processes (Section 2.2 and Figure 2,

Chapter 2).1 If there are several non-interacting fluorophore species in solution,
their excited states will decay independently and a sum of exponentials will be
needed to describe how the sample's fluorescence intensity decreases with time.2
The general expression for the intrinsic fluorescence decay F(t) is therefore

where the sum is over all the different fluorophore species 'i' in solution, and
we now express as ai their individual pre-exponential factors. The af depend on
the species' respective kR values, concentrations and extinction coefficients, as
well as on the exciting light intensity and geometric factors (among other
things). In carrying out fluorescence lifetime measurements, some of these
variables are typically not quantified; as their product is thus not known before
hand, it is simply expressed as a constant that is determined when the data are
analysed (see Section 2.6).
The steady-state fluorescence intensity I(A) at emission wavelength X is
related to the time-resolved decay parameters through

where we now write xj as a function of the emission wavelength. The fractional

intensity fi of the ith decay component is given by

Analogous to the fluorescence quantum yield (equation 4, Chapter 2), we can

define a quantum yield for intersystem crossing (Section 2.2.1, Chapter 2),
which is the fraction of initially excited molecules that makes it from S1 to T1
and is given by

Phosphorescence is a longer-lived emission than fluorescence, with the

former having lifetimes varying from microseconds (us = 10-6 s) to seconds, as
compared with picoseconds (ps = 10-12 s) to nanoseconds (ns = 10-9 s) for the
latter. Even though phosphorescence emission is a two-step process (inter-
1
Excited singlet-state molecules may also participate in a number of bimolecular reactions such as
energy transfer, excited-state complex (excimer) formation, and excited-state proton transfer, all of
which result in non-exponential fluorescence decays. A discussion of the kinetics of these process-
es is beyond the scope of this book, however. The interested reader may wish to consult (1).
2
'Other' fluorescent species need not be different molecules. Depending on its environment,
the fluorescence decay of a single tryptophan residue at some position in a protein can be com-
plex, as several alternate processes, such as electron and proton transfer, and the existence of dif-
ferent microenvironments for different rotational orientations of the indole side chain, may
result in the residue having several different decay times.

70
 TIME-RESOLVED FLUORESCENCE AND PHOSPHORESCENCE SPECTROSCOPY

system crossing followed by a radiative return to S0 from T1), generally kKC is

very much greater than the rate of triplet state de-excitation, and creation of
the triplet state population is essentially complete before emission from it
proceeds. For the unimolecular decay of the triplet state this means that its
population and the phosphorescence intensity also decay exponentially with
time. The rate of decay may be described using the phosphorescence lifetime or
decay time, which is given by TP = (kp + k^)-1, where kp and kqp are the re-
spective rate constants for the radiative and non-radiative relaxation of T1.
The phosphorescence quantum yield is the fraction of absorbed photons that
is emitted as phosphorescence and so is the product of the fraction that makes
it from S1 to T1 (i.e. $isc). and the fraction of the triplet state molecules that
returns to S0 by emitting a photon:

The ratio of the quantum yields for phosphorescence and fluorescence is then

In the event that kQp kp, equation 6 becomes

(i.e. op = TSkISC = oISC). The intersystem crossing rate constant (kISC) is sensitive
to spin orbit coupling by heavy atoms (2).

2.2 Ruorescence and phosphorescence energy transfer

and sensitized luminescence
The theory of energy transfer via dipole-dipole interactions was originally
developed by Forster (3, 4) to explain singlet-singlet transfer: D(S1) + A(S0) -
D(S0) + A(S1). It is also applicable to triplet-singlet transfer (5, 6): D(T1) + A(S0) -
D(S0) + A(S1). (Here, D is the donor, A is the acceptor and their electronic states
are indicated in parentheses.) The strength of the interaction is usually expressed
in terms of a 'critical radius,' Ro. At this distance of separation between donor
and acceptor, the probability that energy transfer will take place is 50%.
According to Forster' s theory

where/ is the overlap integral

K2 is a dimensionless orientational factor, o0 is the quantum yield of the donor

in the absence of energy transfer, and n is the refractive index of the medium.
In equation 9 the emission spectrum of the donor (fD) is normalized to unity (i.e.
ffD(v)dv = 1) and EA is the molar extinction coefficient of the acceptor. Both fD

71
THOMAS D. BRADRICK AND JORGE E. CHURCHICH

and eA are plotted versus wavenumber (v), which is the reciprocal of the wave-
length (A., in cm).Rovalues for a number of donor-acceptor pairs free in solution
(for which K2 is 2/3) may be found in (7); otherwise, they must be calculated
using equations 8 and 9.
The efficiency of energy transfer is given by

where oD and oDA are the donor's quantum yields and TD and TDA are its decay
times in the absence and presence of the acceptor, respectively. (The inter-
changeability of quantum yield and lifetime follows from equation 14 of Chapter
2). Because of the transfer efficiency's sensitivity to the donor-acceptor distance R

energy transfer measurements can be used to determine molecular distances. If

the equipment is available, it is generally easier to do this accurately by
measuring emission lifetimes than it is by measuring quantum yields. The
reader may wish to consult (8) for additional details.
Molecules in their triplet states may also participate in collisional transfer
processes known as triplet-triplet energy transfer: D(T1) + A(S0) - D(S0) + A(T1).
This, however, takes place through an electron exchange mechanism that
involves overlap of the electronic clouds of the donor and acceptor molecules.
Consequently, the interaction energy cannot be determined from optical
measurements of the donor. (Triplet-triplet energy transfer by dipole-dipole
interaction would not be expected to occur over long distances since the
electronic transitions in both the acceptor and donor are spin-forbidden.)
Radiationless energy transfer between triplet levels becomes appreciable when
the acceptor concentration reaches the level of 10 mM or higher. The sensitized
phosphorescence yield of the acceptor resulting from D(T1) - A(T1) transfer is
given by the equation

where oPD and oPD are the phosphorescence quantum yields of the donor in the
absence and presence of the acceptor, respectively, and oPA is the phos-
phorescence quantum yield of the acceptor. When the forbidden nature of a
transition in the donor results in a corresponding increase in the lifetime of its
excited state (as is the case for the triplet state of many molecules), energy ex-
change over long distances can be high (provided there is a good overlap of the
spectra concerned) as the distance over which the triplet-state molecule can
diffuse is relatively great.

2.3 Observed time dependence of fluorescence

2.3.1 Introduction to convolution integral
In Section 2.1 we stated that the fluorescence intensity due to a population
of excited-state molecules decreases exponentially with time following their

72
 TIME-RESOLVED FLUORESCENCE AND PHOSPHORESCENCE SPECTROSCOPY

instantaneous excitation. In practice, however, excitation is never instantaneous

but is achieved either by using pulses that have a significant duration compared
to the fluorescence lifetimes one is measuring (pulsed excitation, Section 2.3.2),
or else by continuously illuminating the sample with light whose intensity
varies sinusoidally with time (traditional phase/modulation, Section 2.3.3). Both
types of excitation distort the observed fluorescence decay away from being
exponential, as the decaying excited-state populations created by earlier parts of
the excitation pulse are subsequently replenished by its later parts. This distor-
tion complicates obtaining the fluorescence decay times (T) and relative ampli-
tudes (a) from the collected data. In the case of pulsed excitation, the observed
fluorescence is described by an expression called the convolution integral.

2.3.2 Pulsed excitation and the convolution integral

An excitation pulse L that has finite duration is treated as consisting of a
succession of instantaneous (so-called '8-function') pulses of varying 'heights' or
intensities, each of which generates a subpopulation of excited-state molecules
that immediately begins to decay. The relative sizes of these different sub-
populations and thus their relative fluorescence intensities are proportional to
the intensities of the 8-function pulse components that generated them. The
observed fluorescence intensity at a particular time t is then the sum of the
contributions each of these sub-populations makes:

Here, L(t1) is the height and At' the width of the excitation pulse's 8-function
component at time t1. F(t) (the intrinsic fluorescence decay) is given by equation 1
and its argument t - t1 accounts for the fact that the elapsed time varies
between when the different subpopulations are created (at times t1) and when
the fluorescence intensity is measured. The (relative) 'height' of L at t1( appro-
priately 'scales' the magnitude of the resulting fluorescence response. The effect
of convolving an exponential decay with an excitation pulse of finite width will
be seen in Section 2.6.4, where we describe the analysis of pulsed-excitation
data. A fuller discussion of the convolution integral may be found in (9).3,4
3
Photomultiplier tubes, which are commonly used to convert light into an electronic signal for
further processing (see Section 3.2), typically distort both the measured excitation pulse profile
and observed fluorescence decay by broadening them. For that reason, the measured L(t') is often
called the instrument response function as it is actually a convolution of the excitation pulse
and the photomultiplier response. It is, in fact, the 'spread' in the signal caused by the photo-
multiplier, rather than that due to a laser pulse, that most distorts the observed fluorescence
decay away from being exponential and requires we use the convolution integral description.
4
On the much longer timescale over which phosphorescence is emitted, nanosecond pulses that
populate S1 'look' like 8-function pulses. In that case, the triplet state population is still formed
virtually instantaneously and the phosphorescence decay remains undistorted and decays expo-
nentially. Usually the phosphorescence decay profile does not have to be described by a convo-
lution integral unless one is using a xenon flashlamp for excitation, for which the pulses are a
few tens of microseconds wide, with a weak 100 p,s tail.

73
THOMAS D. BRADRICK AND JORGE E. CHURCHICH

2.3.3 Phase/modulation relationships

Traditional phase/modulation fluorometry involves measuring fluorescence life-
times by continuously illuminating the sample with light whose intensity varies
sinusoidally. The resulting fluorescence intensity also varies sinusoidally at the
same frequency as that of the exciting light, but is delayed or 'phase shifted' by
an angle 9 with respect to the latter (Figure 1). Also, the 'modulation' of the fluor-
escence (mF = b/a, the ratio of the amplitude of oscillation to the average intens-
ity; see Figure 1) is reduced compared to that of the exciting light (mL = B/A; see
Figure 1).
The phase shift and demodulation both depend on the fluorescence lifetimes
and amplitudes and on the modulation frequency to according to (10)

where

and the sums are over the number of exponential terms 'i' that are being used
to model the fluorescence decay. fi is the fractional intensity of the ith decay
component (equation 3).
Thus, by measuring the phase shift 0 and the demodulation m, one can deter-
mine the fluorescence decay times (T) and amplitudes (a). Such measurements
are typically made sequentially over as broad a frequency range as possible in
order to minimize the error in the extracted parameters, and to enable one
to distinguish between different multi-exponential fluorescence decays (see

Figure 1 Illustration of exciting light intensity L(f) and fluorescence response Fobs(t) for a
phase/modulation experiment. B and b are the respective amplitudes of oscillation of the
two waves, while A and a are their respective average intensities. 6 is the angle by which
Fobs(t) is delayed or phase shifted with respect to L(t').

74
 TIME-RESOLVED FLUORESCENCE AND PHOSPHORESCENCE SPECTROSCOPY

Section 2.6.5). (Note: MHz modulation frequencies are required for measuring
nanosecond fluorescence lifetimes.) The reader who desires additional details
about phase/modulation fluorometry may wish to consult (10).

2.4 Decay associated spectra (DAS) and discrete lifetimes

versus lifetime distributions
Due to space limitations, we can only make brief mention of DAS and lifetime
distributions. DAS (decay associated spectra) are the emission spectra associ-
ated with the different decay species and are described in (11). Examples of
applications may be found in (12-14). The treatment of fluorescence lifetimes
as distributions (rather than as the discrete values implied in Section 2.1) is dis-
cussed in (15-18). Commercially available software packages (see Section 2.6.1)
are typically able to carry out the necessary analysis to obtain DAS from decay
measurements made at multiple emission wavelengths, or to obtain lifetime
distributions.

2.5 Polarized excitation and emission anisotropy decay

2.5.1 Introduction to emission anisotropy
Implicit in the previous discussion of fluorescence decay was that the fluorescing
sample was being excited using unpolarized light. In fluorescence anisotropy
decay measurements, the sample is excited using vertically polarized light and
the horizontal and vertical polarization components of the fluorescence decay
are separately recorded. The observed decrease in the intensity of either com-
ponent is due both to the decrease in the size of the population of excited-state
molecules, and the tumbling motion of the molecule, which reorients the
transition dipole moment between absorption and emission and so changes the
intensity of the light observed using a particular emission polarizer orientation.
Thus, the polarized fluorescence decay components contain information about
the motional dynamics of the molecule.
The time-dependent emission anisotropy r(t) is defined as

where I||(t) is the emission polarization component with orientation parallel to

that of the exciting light (i.e. vertical), I±(t) is the emission polarization com-
ponent with orientation perpendicular to that of the exciting light (i.e. hori-
zontal) and g is a factor that corrects for any bias the emission monochromator
and detector may have towards one orientation over another. (Typically, mono-
chromators preferentially transmit vertically polarized light. Systems for making
time-resolved measurements may come equipped with a 'depolarizer' placed in
front of the entrance slits of the emission monochromator that negates the
effects of the bias and causes g to equal unity. Otherwise, g is determined by
exciting the fluorophore with horizontally-polarized light and taking the ratio

75
THOMAS D. BRADRICK AND JORGE E. CHURCHICH

of the resulting vertical and horizontal emission intensities.) The denominator

of equation 16 (IB(t) + 2gI-(t)) is the total fluorescence intensity of the sample.
The expected time dependence of the emission anisotropy has been worked
out for several important models, including those detailed in the following
sections.
2.5.2 Freely-rotating spherical molecule
In this case, the time-dependent fluorescence anisotropy is given by

(A globular protein with a fluorophore (either intrinsic or extrinsic) rigidly

attached so that the latter's motion is dominated by the tumbling of the
protein, would be an example of this as the protein is completely free to rotate.)
Here, r0 is the initial anisotropy (i.e. r at t = 0) and o is the rotational correlation
time. The latter quantity is a measure of how fast the molecule tumbles, which,
for a spherical molecule, is given by

where n is the micro-viscosity of the sphere's environment, V is the sphere's

hydrodynamic volume, k is Boltzmann's constant and T is the absolute tempera-
ture. A good 'rule of thumb' is that a protein's rotational correlation
time in nanoseconds should be about half its molecular weight in kDa.
(e.g. we expect a 16-kDa globular protein to have a o of about 8 ns.) Protein
unfolding and the concentration-dependent association of protein subunits are
examples of processes that would be expected to cause o to change and so can
be examined using time-resolved emission anisotropy.

2.5.3 Rod-shaped molecule with restricted motion

For a rod-shaped molecule with its transition dipole moment oriented along the
rod's long axis and whose motion is confined to a cone, with one end of the rod
fixed at the cone's apex, the emission anisotropy is given by

Because the molecule's motion is restricted, the anisotropy does not decay to
zero but reaches some final constant value, r00. Equation 19 is typically used to
model the anisotropy decay of a rod-shaped fluorophore such as l-(4-
trimethylammoniumphenyl)-6-phenyl-l,3,5-hexatriene (TMA-DPH) embedded in
a phospholipid membrane, where it reports on the order and dynamics of the
latter. The so-called 'order parameter' S of the membrane is defined as

where 9 is the half angle of the cone. A more general model for restricted
motion is

76
 TIME-RESOLVED FLUORESCENCE AND PHOSPHORESCENCE SPECTROSCOPY

A more complete description of the theory of emission anisotropy may be

found in (19) (which also contains specific examples of the technique's use as
well as the appropriate equations for phase/modulation measurements).

2.5.4 Examples
A study of the effects of melittin on phospholipid membranes that included
time-resolved fluorescence anisotropy measurements of l,6-diphenyl-l,3,5-
hexatriene (DPH, the parent compound of TMA-DPH) incorporated into the
membranes is described in (20). (Note: TMA-DPH is now the preferred of the two
probes as its cationic tertiary amine helps to anchor one end of the probe
molecule to the bilayer-water interface.)
References (21-23) describe time-resolved fluorescence anisotropy studies
(using TMA-DPH and DPH as fluorescent probes), of the effect of immune com-
plexes on macrophage membrane fluidity, the effect of succinate and phenyl
succinate on the fluidity of the inner mitochondrial membrane, and the role of
cholesterol in modulating the fluidity of renal cortical brush border and
basolateral membranes, respectively.
A time-resolved fluorescence anisotropy study of the base motions in DNA,
made using the intrinsic fluorescence of the bases themselves is described in
(24). Large amplitude motions were found on the picosecond timescale that are
sensitive to solvent viscosity. This provides support for the idea that base
motions contribute substantially to the observed fluorescence depolarization in
DNA and is consistent with a mechanism that involves concerted motions in the
interior of the polymer.
Reference (25) details a study in which time-resolved fluorescence intensity
and anisotropy decay measurements were combined with stopped-flow mixing
to monitor the local and global dynamics of E. coli dihydrofolate reductase
during refolding. A general review of such (so-called) 'double kinetic' experi-
ments is given in (26).
Recent contributions that both time-resolved fluorescence intensity and
anisotropy decay studies have made to our understanding of the structure and
dynamics of proteins and nucleic acids are reviewed in (27).

2.5.5 Caveats
It is very important that the lifetime of the fluorophore (either intrinsic or
extrinsic) being used to measure the anisotropy decay be appropriate to the
timescale of the motions under examination. Too short a lifetime will mean
that the fluorescence intensity will 'die away' before much rotational motion
takes place; the anisotropy will remain 'high' over the course of the fluorescence
intensity decay and the latter's components will contain limited information
about the motional dynamics. On the other hand, too long a lifetime means that
the fluorescence will become too quickly depolarized on the timescale of the
intensity decay, so that again the signal will contain limited information about
the dynamics. Generally, tryptophan fluorescence (with a lifetime of a couple of
nanoseconds) can be reliably used to measure o values of up to about 25 ns.

77
THOMAS D. BRADRICK AND JORGE E. CHURCHICH

(The required accuracy for o depends, of course, on the purpose of the measure-
ment.) Protein motions with o values greater than this will probably require
that a longer-lived extrinsic fluorophore (e.g. pyrene) be attached to the
molecule. Motions on the microsecond to second timescales can be monitored
using phosphorescent probes (see Section 4.2).
In the event that one wishes to measure only the fluorescence decay times of
a sample, but is forced to make the measurement with polarizers present, it is
imperative that the emission polarizer be oriented at an angle of 54.7° — 55°
with respect to the vertical excitation orientation (the so-called 'magic angle') in
order to correct for the presence of the polarizers. Otherwise, as the 'apparent'
decay rates of the horizontal and vertical fluorescence components are due to
both the depletion of the excited state population and to fluorophore motion (as
we mentioned in Section 2.5.1), the measured 'lifetimes' will be incorrect.
Proteins are generally excited at 295 nm in order to avoid (i) energy transfer
from tyrosine to tryptophan residues, which can occur at shorter wavelengths
(e.g. at 280 nm where tryptophan has its maximum absorbance) and which
depolarizes the fluorescence, and (ii) depolarization of tryptophan fluorescence
due to relaxation from its 1Lb to 1La electronic states. (At shorter wavelengths
electrons are excited to both levels, after which electrons in the higher level
relax radiationlessly to the lower, which also depolarizes the fluorescence.) One
would like to maximize the initial anisotropy in order to obtain the most
sensitivity for polarized fluorescence measurements.
Due to a process called photoselection, r0 can have a maximum value of 0.4.
A measured value for r0 that is significantly less than this strongly suggests that
some fast depolarization process is taking place that is beyond the equipment's
temporal resolution. Energy transfer, excimer formation, fast local libration and
radiationless relaxation between two different electronic states (such as that
from 1Lb to 1La in tryptophan) are some examples of processes that result in a
significant decrease in r0.

2.6 Data analysis

2.6.1 Introduction to least-squares analysis of data
After the data from a time-resolved fluorescence measurement have been col-
lected, they must be analyzed to obtain the lifetimes (T) and amplitudes (a) and,
in the case of polarized fluorescence, the anisotropies (r) and rotational correla-
tion times (o). For results from pulsed-excitation experiments, this involves per-
forming a non-linear least-squares fit of the convolution integral (equation 13) to
the experimental data. (The analysis of phase/modulation data is described in
Section 2.6.5.) The goal is the same as that in linear least-squares fitting (or
linear regression analysis) in that the parameters of the model function (the
slope andy-axis intercept of a straight line in the case of linear regression, the a
and T values in the case of fluorescence intensity decay) are adjusted so that the
fitted line evenly 'cuts' through the distribution of noise in the data and balances
the (squared) vertical distances from the fitted line to the data points above and

78
 TIME-RESOLVED FLUORESCENCE AND PHOSPHORESCENCE SPECTROSCOPY

below it. These vertical distances or differences are called residuals; mathemati-
cally, the best fit is defined as being that which minimizes the sum of their
squared values. (The residuals are squared as the maximum likelihood of a
normal distribution is then at their centres.) The individual residuals are often
weighted according to their uncertainties (if known), so that the 'fit' is not
unfairly influenced by values that possess a relatively high degree of
uncertainty. If the sum of the squares of the weighted residuals is 'rescaled'
according to the 'number of degrees of freedom,' we have the definition of x2,
which has an optimal value of unity.
A general discussion of the use of least-squares fitting in fluorescence
measurements may be found in (28). The global analysis of fluorescence data is
discussed in (29). Commercially available time-resolved fluorimeters are typically
sold with data analysis software included. Available stand-alone packages in-
clude the Globals Unlimited suite, which is capable of analysing both time- and
frequency-domain data, stopped-flow kinetics, etc. The Center for Fluorescence
Spectroscopy at the University of Maryland (USA) also offers software for
frequency- and time-domain fluorescence lifetime analysis.

2.6.2 Goodness-of-fit criteria

Strictly speaking, one does not know before hand how many exponentials will
be required to adequately describe a particular fluorescence decay, so that one
has to perform several separate fits, each using a different number of such
terms. We therefore need criteria to use in choosing which model best
describes the data. The standard ones are the forms of:
(1) the residuals;
(2) the autocorrelation function values; and
(3) the x2 value.
Note: the autocorrelation function looks' for correlations between residuals
separated by various amounts of time. Specifically, its ith value expresses the
amount of correlation between pairs of residuals separated by i - 1 time points
(or recorder channels). If there is no correlation amongst the residuals, then the
autocorrelation function values should be completely random. There are
typically only half as many (plus one) autocorrelation function values as there
are residuals, owing to the way the former are calculated. Also, by definition,
the first value is equal to unity, as each residual is completely correlated with
itself. Typically, autocorrelation values have not been calculated in the case of
non-linear least-squares fitting of phase/modulation data.

2.6.3 Initial guesses.

In the case of linear regression analysis, the problem has an 'analytic' solution:
finding the slope and y-intercept of the best-fitting straight line is reduced to
'plugging' the x- and y-coordinates of the data points into two formulae. There is
no such analytic solution for the problem of non-linear regression; the best-
fitting values for the parameters have to be found by an iterative process that

79
THOMAS D. BRADRICK AND JORGE E. CHURCHICH

proceeds from some starting point. One begins the fitting procedure using
initial 'guesses' for the parameter values, and the fitting program employs one
of several techniques to try to reach the best-fitting values by 'stepping' towards
them from the initial guesses. The values for the initial guesses need to be
chosen with care as the fitting program can have difficulty reaching the best-
fitting values if the former are too far away from the latter. For a single
exponential fit, one can estimate the lifetime from the rate at which the
fluorescence decays at long times (i.e. far away from the excitation pulse and
the distortion it causes). For tryptophan fluorescence in proteins, a clustering of
values around 1, 3 and 7 ns occurs, which can serve as initial guesses. For other
fluorophores, published values can be used. The fitting procedure is less
sensitive to the initial guesses for the amplitudes, so arbitrary values for them
often work.

2.6.4 Analysis of pulsed excitation data

1. Non-polarized (or total fluorescence intensity) decay data. We illustrate this with an
example.

Time-

Figure 2 Results of a single-exponential fit to synthetic data F generated by convolving a

double exponential decay with the laser pulse profile L. In panel A, the single-exponential fit
can also be seen as it falls slightly below F at the fluorescence peak. Panel B shows a plot
of the resulting weighted residuals versus time while panel C shows the plot of the
autocorrelation (A.C.) function values for the residuals.

80
 TIME-RESOLVED FLUORESCENCE AND PHOSPHORESCENCE SPECTROSCOPY

Rgures 2 and 3 show how the appearance of the residuals and autocorrelation
function for a pulsed excitation experiment typically depend on the appropri-
ateness of the fitting function. In panel A of both figures, L shows an actual
excitation pulse profile (more properly, the instrument response function) that
was generated by an argon ion laser that pumped a dye laser circulating
rhodamine 6G, the tuned output of which was frequency-doubled to 295 nm
(nanometer = 1CT9 m = 10 A) by passage through a B-barium borate crystal (cf.
Section 3.1.1). The pulse was recorded using a photomultiplier after being
scattered by a dilute silica suspension in distilled water. Synthetic data (F in the
figures) were generated by convolving a double exponential decay (x1 = 2.5, T1
= 2.5 ns, a 2 = 0.5, T 2 = 5.0 ns) with the laser pulse profile. Random Gaussian
noise, appropriately scaled for single photon counting (cf. Section 3.3.1), was
then added to this 'signal'. The resulting peak intensity was about 45500
counts.
Figure 2 shows the results of fitting the convolution of a single exponential to
this decay, which yielded best-fitting values of a = 2.8 and T = 3.0 ns. This
lifetime is close to the intensity-weighted average lifetime of 3.2 ns (given by
E/iTi, where the fi are the fractional intensities (equation 3)), which is usually to

Time-

Figure 3 Results of a double-exponential fit to synthetic data F generated by convolving a

double exponential decay with the laser pulse profile L. Panel B shows a plot of the resulting
weighted residuals versus time while panel C shows the plot of the autocorrelation (A.C.)
function values for the residuals.

81
THOMAS D. BRADRICK AND JORGE E. CHURCHICH

be expected. The fit can also be seen in panel A as it falls somewhat below the
'data' at the peak of the fluorescence curve. Otherwise, the two seem to agree
fairly well, especially at long times. The weighted residuals (Figure 2B) indicate,
however, that the single exponential fit deviates from the data essentially every-
where along the decay curve, and the autocorrelation of the residuals (Figure 2C)
is also far from being random, which confirms that a definite trend exists in the
residuals. The x2 value of 14.88 also indicates a poor fit. (Recall that the aim of
the fitting procedure is to 'adjust' the model function so that it evenly 'cuts'
through the noise in the data. If the model is fundamentally inadequate, no
amount of adjustment will enable one to obtain a good fit with a random dis-
tribution in the residuals.)
Figure 3 shows the results of the double exponential fit that was attempted
next. In this case the fit appears to completely overlay the data in panel A and
this impression is borne out by the completely random distribution of the
residuals around the zero line (panel B). The random distribution of the
autocorrelation function values (Figure 3C) confirms that there is indeed no
trend in the residuals, while a x2 value of 0.84 (close to unity) further attests to
the goodness of the fit. A fit of a triple exponential decay was also attempted
(results not shown). A x2 value of 0.85 indicated that the addition of another
term was not warranted, and this judgement was reinforced by the fact that
two of the three exponentials had nearly identical lifetimes. (The general rule is
that, for -300 degrees of freedom (i.e. —305 data points), the addition of
another exponential term is not justified if it results in a less than 10%
improvement in x2-) Thus, the decay is taken to be double exponential, the
results for which were x1 = 2.55, T1 = 2.52 ns, x2 = 0.45 and T2 = 4.61 ns. These
values are in excellent agreement with the generating function concerning the
short lifetime component. The amplitude and lifetime values of the long life-
time component are each about 10% in error, presumably because that 'species'
only contributed about 28% of the total fluorescence intensity at the chosen
emission wavelength, and the two lifetimes only differ by a factor of two. Much
greater accuracy can be obtained by carrying out a global analysis that includes
fluorescence decays measured at other emission wavelengths (i.e. DAS, Section
2.4) at which the longer lifetime component would make a different con-
tribution.

2. Polarized fluorescence data. As the emission anisotropy (equation 16) is not a

directly measured quantity, polarized fluorescence data cannot be analysed by
fitting calculated experimental values for r(t) with a simple convolution of the
lamp profile and a model function (e.g. equation 17 for a rotating sphere).
Instead, the polarized data are often analysed in two steps. First, the total
fluorescence decay (I||(t) + 2gI1(t) in the denominator of equation 16, and typically
referred to as the 'sum' data, S(t)) is fitted as described above. Once the appro-
priate expression F(t) for the intrinsic fluorescence decay has been obtained, the
product F(t)r(t)Lij is then convolved with the lamp profile and fitted to the
'difference' data D(t) = I||(t) -gI1(t) (the numerator of equation 16). Here, r(t) is the

82
 TIME-RESOLVED FLUORESCENCE AND PHOSPHORESCENCE SPECTROSCOPY

model function for the anisotropy decay and Ij, is an 'association matrix' (made
up of ones and zeros) that combines the appropriate intensity and anisotropy
decay components, depending on whether the decay is associative or non-
associative (30). The expression for r(t) can be modified as required until a good
fit is obtained, where x2 and the form of the residuals and autocorrelation again
serve as the fitting criteria as described previously. Alternatively, I||(t) = F(t)[l +
2r(t)Lij]/3 and J±(t) = F(t)[1 - r(t)Lij]/3g convolved with the lamp pulse profile may
be fitted simultaneously to their respective polarized fluorescence decay
components in a global analysis.

2.6.5 Analysis of phase/modulation fluorescence decay data

In figure 4 we show as an example the results from an analysis of synthetic
phase/modulation data. In this case, equations 14 and 15 were used to generate
phase angle and modulation values for a double exponential fluorescence decay

100

Figure 4 Results of single- and double-exponential fits to synthetic phase/modulation data

for a double-exponential decay. In panel A, the solid circles (•) are the phase angle values
(in degrees) and the open circles (o) are the percentage modulation values (modulation x
100%), both plotted as a function of the modulation frequency of the exciting light. The solid
and dashed lines show the respective results of single- and double-exponential analyses of
these data. Panels B and C show the residuals from those fits to the phase (B) and %
modulation (C) data. For both types of residuals, the solid circles (•) indicate the residual
values from the single-exponential analysis, while the solid squares (•) indicate the values
from the double-exponential analysis.

83
THOMAS D. BRADRICK AND JORGE E. CHURCHICH

for which a, = 0.33, Ta = 2.0 ns, a2 = 0.67 and T2 = 5.0 ns. Gaussian random
noise with standard deviations of 0.2° and 0.005 (10, 18) for the phase angle and
modulation, respectively, were added to the calculated values. These standard
deviations were also used as the inverse weighting factors in the non-linear
least-squares analysis of the 'data.' (Note: for phase/modulation data the weight-
ing factors, which are the reciprocals of the standard deviations, are estimated
from repeated measurements so that, depending on their accuracy, the optimal
X2 value may not be unity. The appropriate number of decay components is
simply that which minimizes x2.)
The 'data' are shown in Figure 4A, where the closed circles are the phase angle
values and the open circles are the per cent modulation (modulation x 100%)
values, both plotted as functions of the modulation frequency (in MHz). The
solid lines (which clearly do not everywhere overlay the data) show the results
of simultaneous fits of the single-exponential expressions (9 = arctan(2TTfr) and
m = (1 + 4TT2f2T2)-1/2) to the data (x2 = 30.1), which yielded a value of 4.1 ns for
the single lifetime. This value is close to the intensity-weighted average lifetime
(T) of 4.5 ns, which is usually to be expected. The residuals for this fit (closed
circles in panels B and C for the phase and per cent modulation residuals,
respectively), clearly show trends, which reinforces our impression from panel
A that this decay is not well modelled by a single exponential. The dashed lines
in Figure 4A (which do seem to fit the data well) show the results of a double-
exponential fit to the data (x2 = 0.57), which gave values of x1 = 0.345, T1= 2.04
ns, x2 = 0.655 and T2 = 5.02 ns. These are in excellent agreement with the values
used to generate the data, and the goodness of the fit is indicated by the random
distribution in the residuals (solid squares for the phase and per cent modu-
lation residuals in panels B and C, respectively). Usually, the autocorrelation
function values are not calculated for fits to phase/modulation data. The reader
may consult (10) for the equations used in analysing polarized phase/modulation
data.

3 Equipment for time-resolved fluorescence

measurements
Experimental set-ups for time-resolved fluorescence measurements can be
broken down into the following component blocks:
(1) an excitation source;
(2) a sample chamber with optics for focusing the exciting light and fluor-
escence;
(3) a detector for converting fluorescence into an electronic signal; and
(4) the recording electronics for processing the detector's output.
Below, we briefly describe the options available for (1), (3) and (4). We also
mention several commercially available systems as examples.

84
 TIME-RESOLVED FLUORESCENCE AND PHOSPHORESCENCE SPECTROSCOPY

3.1 Excitation sources

3.1.1 Lasers
Lasers are a source of coherent, monochromatic, highly-collimated radiation. UV
tunability requirements for protein and DNA fluorescence measurements mean
that the typical set-up usually involves either a mode-locked argon-ion or Nd:YAG
laser synchronously pumping a dye laser.5 Different dyes give laser output at dif-
ferent wavelengths. Rhodamine 6G, for example, can provide pulses at 590 nm,
which can then be frequency-doubled to 295 nm for exciting tryptophan fluor-
escence. Frequency doubling is a non-linear optical effect (called second
harmonic generation) that results from the interaction of the laser beam's
intense electric field with the doubling crystal. B-Barium borate (BBO) crystals can
provide a doubling efficiency of greater than 15%. (Note: third harmonic
generation using the tunable, modulated output of titanium-sapphire lasers is
beginning to be used to provide short UV pulses.) The repetition rate of the
pump-dye laser combination is typically high enough that the time between
successive pulses is much shorter than the time needed to accommodate a
complete fluorescence decay. A technique called cavity dumping is used to select
out, for instance, one in twenty pulses for subsequent doubling. This is
accomplished by placing an acousto-optic deflector inside the dye cavity that is
periodically triggered with a radio frequency burst. The reader who desires
additional information about lasers may consult (31-34).

3.1.2 Flash lamps

A flash lamp consists of two electrodes (separated by a narrow gap) contained in
a housing and surrounded by a gas under pressure. A device called a thyratron
rapidly switches voltage pulses to the electrodes so that an electrical discharge
takes place across the gap -50 000-100 000 times per second (Hz). Each dis-
charge excites the atoms/molecules of gas in the region so that they give off a
pulse of light, the profile of which (in terms of either time or wavelength)
depends on the pressure and composition of the gas. These pulses typically have
a breadth of several nanoseconds, with a tail.

3.1.3 Synchrotron radiation

Synchrotron radiation can also be used to provide excitation pulses for perform-
ing time-resolved fluorescence measurements. We will limit our comments,
however, as the necessary support facilities required for a synchrotron radiation
source restrict them to being located at major universities and national
laboratories. Such facilities include: the Aladdin Biofluorescence Center (ABC) at
5
In mode-locking, the combination of different standing wave 'modes' that the laser cavity sup-
ports is fixed so that superposition of the different modes gives short pulses. Although argon-ion
and Nd:YAG lasers do not themselves produce the shortest mode-locked pulses, their periodical-
ly-modulated output can synchronously pump or drive a dye laser (whose mode-locking has been
matched to that of the pump laser), which then produces much shorter pulses than the pump
laser.

85
THOMAS D. BRADRICK AND JORGE E. CHURCHICH

the Synchrotron Radiation Center of the University of Wisconsin, Madison

and the National Synchrotron Light Source (NSLS) at Brookhaven National
Laboratory (both in North America); the Synchrotron Radiation Source (SRS) at
Daresbury (UK); and the Laboratoire pour I'Utilisation du Rayonnement
Electromagnetique (LURE) at the Universite de Paris-Sud at Orsay (France). (Links
to the websites of other facilities may be found at the LURE website.) The
interested reader may find (35) provides additional useful information.

3.2 Detectors
Commercially available time-resolved fluorescence spectrometers typically use
photomultiplier tubes (PMTs) to convert light into an electronic signal for
further processing. As was pointed out in Footnote 3, however, PMTs distort the
fluorescence decay profile by broadening it; this places a limit on how short a
fluorescence decay time a photomultiplier-based time-resolved system can
measure. PMTs are described in Section 4.6 of Chapter 2 ; the reader can also
find additional information in (36).
Other, 'faster' detectors that can be used in place of a conventional PMT
include microchannel plate photomultiplier tubes (MCP-PMTs) (31) and streak
cameras (37). Because of their expense, the use of these devices is usually con-
fined to 'home built' fluorimeters found in dedicated fluorescence laboratories,
and is therefore not discussed here.

3.3 Recording electronics

The type of recording electronics used depends on whether pulsed excitation
(either pulse sampling or time-correlated single photon counting) or phase/
modulation measurement are being made. Therefore, we discuss the electronics
in the context of the method.
3.3.1 Pulse methods
For the case of pulsed excitation (as described in the discussion of the con-
volution integral, Section 2.3.2), a full fluorescence decay profile cannot be
captured following excitation by just a single pulse; the emission intensity is too
dim, and recording devices are generally too slow to allow that. Instead, a com-
posite decay profile is assembled from recorded 'portions' of many (typically
hundreds of thousands of) consecutively generated fluorescence decays. Because
the final result is a composite 'picture,' it is very important that the excitation
source provides stable pulses. If, for example, the excitation pulse profile
changes between when the fluorescence decay and excitation pulse profiles are
recorded, then the convolution of the latter with the proper expression for the
intrinsic fluorescence decay F(t) (equation 1) will not fit the recorded decay and
reliable lifetimes etc. will not be extracted.
In both of the pulse excitation techniques we describe below, the excitation
pulse is also used to provide a triggering signal that the recording electronics
uses to synchronize events. In the case of flash lamp excitation, this is com-
monly accomplished using an 'antenna pick-up'. For laser pulse excitation,

86
 TIME-RESOLVED FLUORESCENCE AND PHOSPHORESCENCE SPECTROSCOPY

a portion of the beam may be reflected into a photodiode whose output pro-
vides the triggering pulse. Often there has to be a delay between when the
triggering signal arrives at the recording device and when the latter can first
begin to process the incoming fluorescence signal from the detector. The arrival
of the fluorescence signal typically has to be delayed to allow for this, which can
be accomplished by simply increasing the length of cable connecting the
detector and the recorder. (It takes an electronic signal about 1 ns to travel one
foot (—0.3 m).) What happens next depends on the recording technique being
used.

1. Pulse sampling. In this technique, a device called a boxcar averager may be

used to record the signal coming from the detector. One can think of a boxcar
averager as having an electronic 'shutter' that is opened for a brief period tg at
some time td following the reception of the triggering signal at tt (see Figure 5J.
(This process is called gating, as the effect is like that of opening and closing a
gate to control something's entry. Note: the 'width' of tg in Figure 5 has been
greatly exaggerated in order to make it clearly visible. Typically, the time course
of a fluorescence decay will be divided up into -256 or 512 such segments or
'channels,' rather than the two dozen or so suggested by the figure.)
For a pre-set number of consecutive pulses, the shutter will always open at
the same time td after triggering and for the same duration tg to allow a small
'slice' (denoted by the hatched area under the decay curve in Figure 5) of the
detector's output signal to enter the recorder. There it is used to charge a
capacitor, whose charge increases with each repetition of the pulse. If enough
pulses are sampled in this way, random noise in the measured decay segment

Luminescence decay

Figure 5 Events occurring during excitation of a sample in the case of pulse sampling
(fluorescence) and phosphorescence measurements, te is the width at half height of the
excitation pulse, td is the delay from the arrival of the triggering signal at tt to the beginning
of observation and tg is the gate width of the detector, during which time the system
measures the hatched portion of the decay curve.

87
THOMAS D. BRADRICK AND JORGE E. CHURCHICH

will tend to average out. Once the pre-set number of samples has been reached,
the capacitor is discharged through a resistor, whose voltage is digitized and
stored. The whole process then begins again, with the shutter now opening
after a new delay time of td' = td + tg. In this way, the shutter's 'window' is
scanned across the decay profile, which allows the latter to be recorded in
discrete, consecutive segments.
Alternatively, the voltage being applied to the detector can be sequentially
gated and the signal from the latter stored in a multi-channel analyser (MCA) as
described in (38). (Note: the main disadvantage that pulse-sampling methods
have is that the PMT is operated in the linear or low-gain analogue mode, so
that its output changes if its high-voltage supply 'drifts.')
Boxcar averagers and their use in signal recovery are described in (39). PTI
(Photon Technology International) manufactures a series of fluorescence life-
time spectrometers (models C-70 to C-73) that use a stroboscopic detection
technique (i.e. a gated PMT, as described in the previous paragraph) and either
nanosecond flashlamp or nitrogen/dye laser excitation.

2. Time-correlated single photon counting (TCSPC). In this technique, the photon

flux is kept low enough and the PMT voltage high enough that the arrival of
individual photons can be detected (see Section 4.6.1 of Chapter 2 ). As in pulse
sampling, the lamp pulse provides a signal that initiates a timing or 'clock'
sequence. The electronics then wait for the pulse from the arrival of the first
photon, which stops the 'clock.' (The 'clock' or chronometer is actually a
capacitor that is charged at a constant rate during the period between the start-
and stop-pulses, so that the total charge is proportional to the elapsed time.
After the arrival of the stop-pulse, the charge is measured and digitized, and one
'count' is added to the appropriate 'bin' or channel of a multi-channel analyser
(MCA) as determined by the just-measured arrival time of the first photon. The
module that performs all this is called a TAC, or time-to-amplitude converter.)
After the arrival of the first photon, the electronics are oblivious to pulses
caused by any other photons until after the next excitation pulse, at which
point the process is repeated.
Fluorescence decay in a population of excited-state molecules is a random
process, so that the arrival time of the first photon is also random, although it is
more likely to occur at the peak of the fluorescence decay than it is along the
latter's 'tail.' Because of this inherent statistical nature, over many excitation
cycles one gradually builds us a statistical profile of the likelihood of emission
versus time that corresponds to the decay profile. For a large enough number of
counts, the uncertainty in the number of counts in a particular channel is equal
to the count number's square root, which gives us the weighting factors to be
used in the least-squares analysis of the data.
Another important component of a TCSPC system is a device called a
constant fraction discriminator (CFD), whose primary function is to compensate
for amplitude variations in the triggering pulses and yield reliable timing.
A comprehensive treatment of TCSPC can be found in (40), which contains

88
 TIME-RESOLVED FLUORESCENCE AND PHOSPHORESCENCE SPECTROSCOPY

introductory chapters on fluorescence and the basic principles of the technique

as well as chapters dealing with light sources, photomultipliers, electronics,
data analysis, time-resolved emission spectra (including DAS) and time-resolved
fluorescence anisotropy. The reader may also wish to consult (41). Commercially
available TCSPC systems include Quantum Northwest's QNW 1000 time-
resolved fluorimeter and Edinburgh Instruments' FL900 fluorescence lifetime
spectrometer.

3.3.2 Phase/modulation equipment

In traditional phase/modulation fluorometry, the output of a xenon arc lamp or
laser is modulated at the desired frequency F. The gain of the fluorescence de-
tector (e.g. a PMT or MCP-PMT) is then modulated at a slightly off-set frequency
F + 8F (where 8F is typically 25, 40 or 800 Hz), so that when the excitation and
emission signals are recombined they mix and one observes the phenomenon
called 'beats', in which the product of the signals is modulated at the frequency
off-set, 8F. (Beating is the same phenomenon one perceives by listening to two
vibrating tuning forks of slightly different frequency.) The phase and modula-
tion are related to 8F which can easily be measured as the value of the latter is
low.
ISA's (Spex's) Fluorolog Tau-3 lifetime system and Spectronic Instruments'
SLM-AMINCO 48000 DSCF spectrofluorimeter both use xenon flash lamp
excitation (typically 150-450 W) and have modulation frequencies of up to 310
MHz. (Most systems can also be operated as steady-state fluorimeters). SLM also
manufactures a multi-harmonic system based on a pulse- modulated continuous
wave light source.

4 Phosphorescence
4.1 Phosphorescence of proteins
The triplet lifetimes of aromatic amino acid residues in proteins show a strong
dependence on the viscosity of the matrix. In rigid glasses consisting of mix-
tures containing glycerol/water 50:50 v/v at 77 K, the phosphorescence lifetime
of free tryptophan is typically around 6 s (42). Smaller values of the phosphor-
escence lifetime of proteins, excited at 295 nm, imply specific interactions:
e.g. electron transfer with a group in the immediate vicinity (43) or inter-
tryptophanyl energy transfer. The phosphorescence spectra of proteins are
dominated by tryptophanyl groups that show vibronic structures with a 0-0
vibrational band centred at around 412 nm. The well-resolved vibrational struc-
ture of the phosphorescence emission allows one in some cases to distinguish
between the emission spectra of tryptophanyl groups positioned in environ-
ments of different polarity (44).
For phosphorescence measurements, the samples are typically excited using
a pulsed xenon flash lamp. A commercially available instrument (the Perkin-
Elmer LS-50B luminescence spectrophotometer) is suitable for phosphorescence

89
THOMAS D, BRADRICK AND JORGE E. CHURCHICH

measurements designed to record emission spectra and determine phosphores-

cence decay times. The events occurring during the excitation of the sample
with a pulsed xenon source in the phosphorescence mode are the same as those
illustrated in Figure 5 for pulse sampling of fluorescence. In this case, the xenon
lamp gives an excitation pulse with a width tf at half-peak intensity of around 10
H.s. During excitation, the phosphorescence increases to a maximum value of I0
and then decays exponentially to zero. The gating of the sample photo multipli-
er is delayed by a time (t d ), so that it no longer coincides with the lamp flash. The
gate width of the photomultiplier detector is controlled by the gate time (tg) as
indicated in the figure. If a delay greater than 0.1 ms is selected for the phos-
phorescence measurements, no fluorescence signal is detected.
In typical phosphorescence experiments, the delay time is varied between 0
and 900 ms, whilst the gate time changes from 0.01 to 500 ms. Both td and tg can
be varied in multiples of 0.01 ms. The phosphorescence of proteins can be
recorded in rigid matrices at 77 K (liquid nitrogen temperature) or in aqueous
solutions at room temperature (45). The central requirement for the observation
of protein phosphorescence in solution is that dissolved oxygen be reduced to a
sufficiently low level, since oxygen can efficiently quench the tryptophan triplet
state. Removal of oxygen from the protein solution can be achieved by repeated
application of moderate vacuum followed by the inlet of argon. Anoxia can also
be attained and maintained by the use of an enzyme system consisting of
glucose oxidase, glucose and catalase (46) as outlined in Protocol 1.
A check on the thoroughness of deoxygenation is provided by the depend-
ence of the phosphorescence lifetime on the amount of oxygen present in the
system. Complete removal of 02 from a solution of alkaline phosphatase (10 (J.M)

Protocol 1
Phosphorescence sample deoxygenation
Reagents
• Glucose oxidase and catalase (Sigma) and • Protein buffer, pH 7
glucose (Fisher)

Method
1 Prepare stock solutions of glucose oxidase (20 mg/ml) and catalase (2 mg/ml) in
protein buffer.a
2 To the desired amount of lyophilized protein, add 3 mg of glucose for every ml of
final sample volume. Add the desired amount of protein buffer and place the
solution in a cuvette.
3 For every ml of protein/glucose solution, add 1 (ul each of the glucose oxidase and
catalase stock solutions, stopper the cuvette and mix.
a These solutions should keep for a couple of weeks when stored at 4 °C.

90
 TIME-RESOLVED FLUORESCENCE AND PHOSPHORESCENCE SPECTROSCOPY

at 25°C yields a phosphorescence lifetime of 1.5 s (47) when excited in the

spectral region coinciding with the absorption of the protein. The phosphor-
escence lifetime is decreased to 1.2 s in the presence of nanomolar concentra-
tions of oxygen.
It is well established that oxygen quenches both excited singlet and triplet
states of aromatic compounds. Thus, indole derivatives in solution at room
temperature display a ratio for the rate constant kNR/kQP of approximately two
(48, 49). However, the quenching of alkaline phosphatase phosphorescence by
oxygen proceeds at a rate kQP = 1.2 x 106 M-1 s-1 (47). The magnitude of this rate
constant can only be explained in terms of hindered diffusion of the quencher
through the protein matrix. Apparently, one of the tryptophanyl residues of
alkaline phosphatase is buried within a shell formed by a-helical rods and (3-
pleated sheets. This particular agglomerate (the so-called 'knot' structure (50))
prevents the penetration of small molecules like oxygen.
The technique of phosphorescence spectroscopy in either rigid glasses or
aqueous solutions can be used to study the spectroscopic properties of trypto-
phanyl residues in wild-type and mutant forms of proteins. As an example,
samples of bovine brain inositol monophosphatase containing three, two and
one tryptophanyl groups/monomers were examined by phosphorescence
spectroscopy at 77 K in rigid glasses. The locations of the three tryptophanyl
residues in the wild-type protein have been provided by X-ray crystallography
(51). Trp5 is located at the N-terminus, Trp87 is separated by 12 A from the
metal chelating centre of Mg2+, and Trp219 is near the catalytic centre. Mutated
forms of the enzyme containing one (double) and two (single) mutant trypto-
phanyl residues/monomer were constructed (52). The wild-type enzyme and
variant W219F display similar phosphorescence spectra (Figure 6A) and the
emission decays monoexponentially with a phosphorescence lifetime of 4.0 s.
By contrast, the phosphorescence emission of the double mutant is blue shifted
and the 0-0 vibronic transition is centred at around 409 nm. On the other hand,
the double mutant (W219F, W5F) exhibits a shorter decay component (2.75 s)
which also fits monoexponential kinetics. The substantial decrease in the
phosphorescence lifetime is interpreted to mean that the degree of rigidity of
the environment surrounding Trp87 has been perturbed as a result of replace-
ment of Trp5 and Trp219 by the amino acid phenylalanine. The conformational
dynamics of the wild-type and mutant forms of inositol monophosphatase were
also examined by measuring resonance energy transfer. Because of the lumin-
escence properties of the ion Tb3+ (53, 54), the lanthanide bound to the catalytic
site of inositol monophosphatase shows a strong sensitized luminescence upon
excitation of the tryptophanyl groups of the protein. Two major bands centred
at 490 and 545 nm, and two minor bands at 580 and 620 nm are detected upon
excitation at 295 nm (Figure 6B).
The efficiency of energy transfer for the wild-type enzyme is greater than
that observed for either the single mutant (W219F) or the double mutant
(W219F, W5F) (Table 1). This behaviour was expected since Trp219 is near the
metal binding centre (7 A); it contributes significantly to the sensitized lumin-

91
THOMAS D. BRADRICK AND JORGE E. CHURCHICH

320 340 360 380 400 420 440 460 480 500
WAVELENGTH (nm)

WAVE LENGTH (nm)

Figure 6 A: phosphorescence spectra of inositol monophosphatase. The phosphorescence

spectra were recorded at 77 K with samples dissolved in the solvent system,
glycerol-Tris-HCI (pH 7.5) (1:1 v/v). The protein concentrations are 16 uM. Results were
obtained for wild-type protein (—), single mutant (W219F) (•) and double mutant containing
Trp87 (...). The delay time was 0.1 ms and the gate time was 20 ms. B: Luminescence

92
 TIME-RESOLVED FLUORESCENCE AND PHOSPHORESCENCE SPECTROSCOPY

escence because the critical transfer distance is 3 A. On the other hand, Trp87
and Trp5, separated by 12 and 22 A, respectively, from the metal binding site,
show weaker sensitized luminescence. However, the significant decrease in the
sensitized luminescence of the double mutant as compared to the single mutant
can be attributed to two independent factors: i.e. change in the distance of the
donor-acceptor pair (Trp87-Tb3+) and change in coordination efficiency as
demonstrated by the binding studies (Table 1).
The reader may find (55) and (56) provide additional useful information about
phosphorescence measurements.

Table 1. Binding of To3+ to myoinositol monophosphatase (1.2 uM) at pH 7.5 in 10 mM

Tris-HCI

Sample JtduO Number of Efficiency* of

binding sites energy transfer
Wild-type 1.6 1 0.016
W219F 1.7 1 0.004
W291F, W5F 50 1.3 0.0005
-19
aCalculated using equation 12. For the wild-type enzyme, J(v) = 0.73 x 10 M-1 cm3, R0 = 3.1 A and
R = 6 A.

4.2 Time-dependent phosphorescence anisotropy

In Section 2.5 we described the use of time-resolved fluorescence anisotropy for
monitoring protein motion on the nanosecond timescale. For motion on much
longer timescales, time-resolved phosphorescence anisotropy can be used
instead. The latter technique has been employed, for example, to examine the
rotational motion of membrane-bound proteins labelled with the triplet probe
eosin (57, 58). Prior to making the measurements, the protein is labelled with
eosine-maleimide as described in Protocol 2.
An oxygen-consuming reaction, involving glucose oxidase, glucose and
catalase (Protocol 1) is then employed to create anaerobic conditions and ln(t) and
Ix(t) are measured. The anisotropy data are generally analysed using as a model
either a cylinder or ellipsoid undergoing restricted rotation and wobbling
(equation 21). In the case of eosin-labelled band 3 in erythrocyte membranes,
three rotational correlation times (oi) were determined, ranging from 0.03 to
values greater than 1 ms (59). The phosphorescence lifetime of eosin-tagged
protein is 2.5 ms.
The presence of several rotational correlation times was interpreted to mean
that several classes of band 3 proteins differing in rotational motion exist in
ghosts and intact cells. A typical anisotropy decay from (59) is shown in Figure 7.

spectra of inositol monophosphatase (2 uM) in the presence of 10 uM TbCI3 before (—) and
after addition of 10 mM MgCI2 (...). The samples were dissolved in 10 mM Tris-HCI buffer
(pH 7.5) at room temperature. The excitation wavelength was 295 nm. The delay time was
0.1 ms, gate time was 1 ms and the slit widths for excitation and emission were 5 nm.

93
 0.05
2.0 4.0 6.0 8.0 10.0

Milliseconds
Figure 7 Phosphorescence anisotropy decay in ghosts. The anisotropy decay of eosin-
labelled proteins was monitored as a function of time. The data were fitted according to
equation 21, Reprinted in part with permission from Matayoshi, E. D. and Jovin, T. M. (1991)
Biochemistry, 30, 3527-3538, Copyright 1991 American Chemical Society.

94
 TIME-RESOLVED FLUORESCENCE AND PHOSPHORESCENCE SPECTROSCOPY

Acknowledgements
T. D. B. would like to express his gratitude to the National Institute of Neuro-
logical Disorders and Stroke, the National Institutes of Health (NIH), for its
continuing support in the form of a postdoctoral fellowship, and to Dr Jay R.
Knutson, Chief, Optical Spectroscopy Section, National Heart, Lung and Blood
Institute, NIH, for welcoming him into the latter's laboratory. In addition, he
would like to thank Dr Knutson and Dr Solon Georghiou for their critical
reading of and helpful comments concerning the fluorescence portion of this
chapter. J. E. C. wishes to express his appreciation for the support given by the
Hong Kong Technical University during his stay there as a visiting professor.

References
1. Lakowicz, J. R. (1983). Principles of fluorescence spectroscopy, p. 383. Plenum Press, NY.
2. McGlynn, S. P., Azumi, T. and Kinoshita, M. (1969). Molecular spectroscopy of the triplet
state, p. 261. Prentice Hall, NJ.
3. Forster, T. B. (1959). Discuss. Faraday Soc., 27, 7.
4. Forster, T. B. (1965). In Modern quantum chemistry, Part III (ed. O. Sinanoglu), p. 93.
Academic Press, NY.
5. Kellog, R. E. (1967). J. Chem. Phys., 47, 3403.
6. Ermolaev, V. L and Sveshnikova, E. B. (1964). Opt. Spectroscopy (English Translation), 14,
320.
7. Berlman, I. B. (1973). Energy transfer parameters of aromatic molecules. Academic Press,
NY.
8. Cheung, H. C. (1991). In Topics in fluorescence spectroscopy, vol. 2: principles (ed. J. R.
Lakowicz), p. 127. Plenum Press, NY.
9. Fernandez, S. M. and Sha'afi, R. I. (1983). In Fast methods in physical biochemistry and cell
biology (ed. S. M. Fernandez and R. I. Sha'afi), p. 1. Elsevier, NY.
10. Lakowicz, J. R. and Gryczynski, I. (1991). In Topics in fluorescence spectroscopy, vol. 1:
techniques (ed. J. R. Lakowicz), p. 293. Plenum Press, NY.
11. Knutson, J. R., Beechem, J. M. and Brand, L. (1983). Chem. Phys. Lett., 102, 501.
12. Royer, C. A., Mann, C. J. and Matthews, C. R. (1993). Protein Sci., 2,1844.
13. Shen, F., Triezenberg, S. J., Hensley, P., Porter, D. and Knutson, J. R. (1996). J. Biol.
Chem., 271, 4819.
14. Green, S. M., Knutson, J. R. and Hensley, P. (1990). Biochemistry, 29, 9159.
15. Alcala, J. R., Gratton, E. and Prendergast, F. G. (1987). Biophys. J., 51, 587.
16. Alcala, J. R., Gratton, E. and Prendergast, F. G. (1987) Biophys./., 51, 597.
17. Alcala, J. R., Gratton, E. and Prendergast, F. G. (1987). Biophys. J., 51, 925.
18. Knutson, J. R. (1992). In Methods in enzymology (ed. L. Brand and M. L Johnson). Vol.
210, p. 357. Academic Press, NY.
19. Steiner, R. F. (1991). In Topics in fluorescence spectroscopy, volume 2: principles (ed. J. R.
Lakowicz), p. 1. Plenum Press, NY.
20. Bradrick, T. D., Dasseux, J.-L, Abdalla, M., Aminzadeh, A. and Georghiou, S. (1987).
Biocm'm. Biophys. Ada, 900, 17.
21. Petty, H. R., Niebylski, C. D. and Francis, J. W. (1987). Biochemistry, 26, 6340.
22. Mutet, C., Duportail, G., Cremel, G. and Waksman, A. (1984). Biochem. Biophys. Res.
Commun., 119, 854.
23. Molitoris, B. A. and Hoilien, C. (1987). J. Membrane Biol, 99, 165.
24. Georghiou, S., Bradrick, T. D., Philippetis, A. and Beechem, J. M. (1996). Biophys. J., 70,
1909.

95
THOMAS D. BRADRICK AND JORGE E. CHURCHICH

25. Jones, B. E., Beechem, J. M. and Matthews, C. R. (1995). Biochemistry, 34, 1867.
26. Beechem, J. M. (1997). In Methods in enzymology (ed. L. Brand and M. L. Johnson). Vol.
278, p. 24. Academic Press, NY.
27. Millar, D. P. (1996). Curr. opin. struct, biol, 6, 637.
28. Straume, M., Frasier-Cadoret, S. G. and Johnson, M. L. (1991). In Topics in fluorescence
spectroscopy, vol. 2: principles (ed. J. R. Lakowicz), p. 177. Plenum Press, NY.
29. Beechem, J. M., Gratton, E., Ameloot, M., Knutson, J. R. and Brand, L. (1991). In
Topics influorescencespectroscopy, vol. 2: principles (ed. J. R. Lakowicz), p. 241. Plenum
Press, NY.
30. Brand, L., Knutson, J. R., Davenport, L, Beechem, J. M., Dale, R. E., Walbridge, D. G.
and Kowalczyk, A. A. (1985). In Spectroscopy and. the dynamics of molecular biological
systems (ed. P. M. Bayley and R. E. Dale), p. 259. Academic Press, London.
31. Small, E. W. (1991). In Topics in fluorescence spectroscopy, vol. 1: techniques (ed. J. R.
Lakowicz), p. 97. Plenum Press, NY.
32. Rabek.J. F. (1982). Experimental methods in photochemistry and photophysics, p. 593. Wiley,
NY.
33. Silfvast, W. T. (1996). Laser fundamentals. Cambridge University Press, NY.
34. Davis, C. C. (1996). Losers and electro-optics: fundamentals and engineering. Cambridge
University Press, Cambridge, UK.
35. Munro, I. H. and Martin, M. M. (1991). In Topics in fluorescence spectroscopy, vol. 1:
techniques (ed. J. R. Lakowicz), p. 261. Plenum Press, NY.
36. Rabek, J. F. (1982). Experimental methods in photochemistry and photophysics, p. 459. Wiley,
NY.
37. Nordlund, T. M. (1991). In Topics in fluorescence spectroscopy, vol. 1: techniques (ed. J. R.
Lakowicz), p. 183. Plenum Press, NY.
38. Lakowicz, J. R. (1983). Principles of fluorescence spectroscopy, p. 62. Plenum Press, NY.
39. Rabek, J. F. (1982). Experimental methods in photochemistry and photophysics, p. 554. Wiley,
NY.
40. O'Connor, D. V. and Phillips, D. (1984). Time-correlated single photon counting. Academic
Press, London.
41. Birch, D. J. S. and Imhof, R. E. (1991). In Topics in fluorescence spectroscopy, vol. V.
techniques (ed. J. R. Lakowicz), p. 1. Plenum Press, NY.
42. Longworth, J. W. (1971). In Excited states of proteins and nucleic adds (ed. R. F. Steiner
and I. Weinryb), p. 319. Plenum Press, NY.
43. Churchich, J. E. (1966). Biochim. Biophys. Acta, 120, 406.
44. Strambini, G. B. and Gonnelli, M. (1990). Biochemistry, 29, 196.
45. Saviotti, M. L. and Galley, W. C. (1974). Proc. Natl. Acad. Sci. USA, 71, 4154.
46. Englander, S. W., Calhoun, D. B. and Englander, J. J. (1987). Anal. Biochem., 161,
300.
47. Strambini, G. B. (1987). Biophys.]., 52, 23.
48. Lakowicz, J. R. and Weber, G. (1973). Biochemistry, 12, 4161.
49. Guzeman, O. L, Kaufman, J. F. and Porter, G. (1973). J. Chem. Soc., Faraday Trans. II, 69,
708.
50. Gregory, R. B. and Lumry, R. (1985). Biopolymers, 24, 301.
51. Bone, R., Spinger, J. P. and Atack, J. R. (1992). Proc. Natl. Acad. Sci. USA, 89, 10031.
52. Gore, M. G., Greasley, P., McAllister, G. and Ragan, C. I. (1993). Biochem. J., 296, 811.
53. Evans, C. H. (1990). Biochemistry of the lanthanides. Plenum Press, NY.
54. Moreno, F., Corrales, S., Garcia Blanco, F., Gore, M. G., Rees-Milton, K. and
Churchich, J. E. (1996). Eur.J. Biochem., 240, 435.
55. Vanderkooi, J. M. (1992). In Topics in fluorescence spectroscopy, vol. 3: biochemical
applications (ed. J. R. Lakowicz), p. 113. Plenum Press, NY.

96
 TIME-RESOLVED FLUORESCENCE AND PHOSPHORESCENCE SPECTROSCOPY

56. Schauerte, J. A., Steel, D. G. and Gafhi, A. (1997). In Methods in enzymology (ed. L. Brand
and M. L. Johnson), Vol. 278, p. 49. Academic Press, NY.
57. Brochon, J.-C., Wahl, P. and Auchet, J.-C. (1972). Eur. J. Biochem., 25, 20.
58. Cherry, R. J. and Schneider, G. (1976). Biochemistry, 15, 3657.
59. Matayoshi, E. D. andjovin, T. M. (1991). Biochemistry, 30, 3527.

97
This page intentionally left blank
 Chapter 4
Introduction to circular
dichroism
Alison Rodger and Matthew A. Ismail
Department of Chemistry, University of Warwick, Coventry CV4 7AL

1 Introduction
1.1 Circular dichroism
Circular dichroism (CD) is the ideal technique for studying chiral molecules in
solution. It is uniquely sensitive to the asymmetry of the system. These features
make it particularly attractive for biological systems. CD is by definition the
difference in absorption, A, of left and right circularly polarized light (CPL):

CPL has the electric field vector of the electromagnetic radiation retaining con-
stant magnitude in time but tracing out a helix about the propagation direction
(Figure 1). Following the optics convention we take the tip of the electric field
vector of right CPL to trace out a right-handed helix in space at any instant of
time (Figure 2) (1, 2).
CD spectra can in principle be measured with any frequency of electro-
magnetic radiation. In practice, most CD spectroscopy involves the ultraviolet-
visible (UV-visible) regions of the spectrum and electronic transitions, though
increasing progress is being made with measuring the CD spectra of vibrational
transitions using infrared radiation. We shall limit our consideration to elec-
tronic CD spectroscopy since the practical considerations for vibrational CD
differ from those for electronic CD.
For randomly oriented samples, such as solutions, a net CD signal will only

left riant unoriented

polarized polarized
K chiral
light light molecule CD spectrum

Figure 1 Schematic illustration of CD. (Source: Circular dichroism and linear dichroism, A.
Rodger and B. Norden, 1997, by permission of Oxford University Press.)

99
ALISON RODGER AND MATTHEW A. ISMAIL

(a) Linearly polarized (b) Right circularly polarized

B

Figure 2 (a) Linearly and (b) circularly polarized electromagnetic radiation. The arrows denote
direction of the electric field £. At each point in space or time, the magnetic field, B, is
perpendicular to the electric field such that k (the propagation direction), E and B form a
right-handed axis system. (Source Circular dichroism and linear dichroism, A. Rodger and B.
Nordgn, 1997, by permission of Oxford University Press.)

be observed for chiral molecules (ones that cannot be superposed on their

mirror images (3)). Oriented samples of achiral molecules, such as crystals, will
also give a CD spectrum unless the optical axis of the sample aligns with the
propagation direction of the radiation. However, such spectra are seldom useful.
CD is now a routine tool in many laboratories. The most common applica-
tions include proving that a chiral molecule has indeed been synthesized or
resolved into pure enantiomers and probing the structure of biological macro-
molecules, in particular determining the a-helical content of proteins. Figure 3
gives an example of a CD spectrum. The key points to remember are that a CD
signal is observed only at wavelengths where the sample absorbs radiation, i.e.
under absorption bands, and the signal may be positive or negative depending
on the handedness of the molecules in the sample and the transition being
studied. In this chapter we shall limit our consideration to wavelength scans.
Kinetics will be covered in Chapter 10.

Wavelength (nm)

Figure 3 CD spectrum of trans-bis-(4-W-methylpyridiniumyl)diphenylporphyrin (8 uM) and

poly[d(A-T)]2 (40 uM) in phosphate buffer (1 mM; pH 7) and NaCI (20 mM).

100
 INTRODUCTION TO CIRCULAR DICHROISM

1.2 Optical rotatary dispersion

Solutions of chiral molecules, in addition to absorbing left and right CPL
differently, rotate the plane of the polarization of an incident linearly polarized
light beam (1, 2). The amount of rotation varies with wavelength, resulting in
an optical rotatory dispersion (ORD) spectrum. As Fresnel realized, the phenom-
enon of ORD is understood when one realizes that a linearly polarized light
beam may be expressed as the sum of two CPL beams. The left and right CPL
beams travel through a chiral solution at different speeds due to their different
refractive indices. Thus when the beams recombine upon emerging from the
sample, their phase difference has changed and the resulting linearly polarized
beam has a different orientation from the incident beam. Due to the relation-
ship between refraction and absorption (2) a complete CD spectrum may be
converted to an ORD spectrum by using the Kramers-Kronig relations. ORD is
now little used except as a means of quantifying the chirality of a solution using
a single wavelength of light, usually the sodium D line at 589 nm. As discussed
below it also has an effect on the units used in CD spectroscopy.

1.3 Chapter outline

CD spectroscopy has become relatively straightforward with the modern instru-
mentation available. In the next section, how to measure a CD spectrum is thus
preceded by a description of the instrumentation. A full understanding of CD
requires some knowledge of CD theory; this is the subject matter of Section 3,
Section 4 summarizes the units used for CD. The chapter concludes with a
discussion of the CD of biomolecules, with particular reference to spectra/
structure correlations. More detail on most aspects may be found in the alpha-
betical General references list.

2 Measuring a CD spectrum
2.1 The instrumentation
Measuring a CD spectrum is a routine procedure assuming one has access to a
CD instrument. CD instruments are usually referred to as circular dichrometers
or spectropolarimeters; however, dichrograph and spectropolarograph better
describe the output of modern instruments. If your sample gives a good UV-
visible absorbance spectrum then it is highly likely that (if it is chiral) you will
get a good CD spectrum. The essential features of a CD spectrometer are a source
of (more-or-less) monochromatic left and right circularly polarized light and a
means of detecting the difference in absorbance of the two polarizations of
light. As the absorbance signal of a typical CD spectrum is a fraction of a per
cent of that observed in the normal absorbance spectrum, the normal method
of achieving these requirements is to implement a polarization phase-
modulation technique.
In most commercial CD spectrometers, a photoelastic modulator (in older

101
ALISON RODGER AND MATTHEW A. ISMAIL

instruments a Pockels cell) produces alternatively right and left circularly polar-
ized light with a switching frequency of typically 50 kHz. The light intensity is
constant, but upon passage through a sample exhibiting CD an intensity fluctu-
ation (corresponding to the different absorptions of left and right circularly
polarized light) that is in phase with the modulator frequency appears. The un-
absorbed photons hit a PMT which produces a current whose magnitude
depends on the number of incident photons. This current is detected by a lock-
in amplifier. Thus, the DC (direct current) component of the PMT current de-
pends on the total absorption of light by the sample (and on lamp intensity and
monochromator characteristics), and the AC (alternating current) component
relates to the CD (a lock-in amplifier is needed to determine the phase of the AC
component which contains information about the sign of the CD). A larger
voltage is applied to the PMT when the number of incident (i.e. non-absorbed)
photons is smaller, thus the high tension (HT) voltage on a CD machine is a rough guide
to the absorbance of a sample. The HT voltage can usually be monitored if one
wishes to do so. It is advisable to take this option even with routine samples, as
it is an easy way to detect sample abnormalities.
The CD spectrometers described above are single beam instruments. A new
double beam instrument from On-Line Instrument Systems (OLIS) has recently
come on the market, which uses sample and reference beams to enable direct
calculation of the CD signal. The instrument utilizes two modulated CPL beams
out of phase by 180° which are passed through the sample and detected simul-
taneously. The absorption of left and right CPL is therefore measured directly.
The technique has the advantage that factors such as lamp fluctuations are
negated since they affect both beams equally, whereas the CD behaviour is
characteristic to each beam. The use of a sample/reference system results in
zero drift over time and nearly flat baselines. However, the two beams pass
through different parts of the sample and cuvette. This may or may not be
important.
The radiation source in UV-visible CD spectrometers is a high energy
(150-450 W) xenon arc lamp. In some instruments the lamp is water cooled.
This can prove the most problematic part of the whole instrument since if the
local water supply is used for a flow through system one is at the mercy of its
purity, hardness etc. Alternatively, if a self-contained system is used, then the
fairly narrow bore of the tubing that the water flows through requires a
reasonably efficient pump, and the water must be kept free from algae etc. and
must be cooled. A narrow bore central heating system pump may be the
cheapest solution to the pumping problem and using car radiator cooling fluid
inhibits the growth of most things if regularly changed. Avoid using clear
tubing if possible. Despite the water consumption required, a water flow heat
exchanger may be the only cooling option that is sufficiently efficient to cool
the water (even for a self contained system) for a 450 W lamp. If at all possible,
an instrument with an air cooled lamp is to be preferred. This no longer seems
to involve accepting a loss in sensitivity.
Another requirement of the high energy lamps used in CD machines is that

102
 INTRODUCTION TO CIRCULAR DICHROISM

the optics of the instrument need to be purged with nitrogen gas to avoid ozone
being created and reacting with surfaces of mirrors etc. The nitrogen purging is
also required for the sample compartment when running below 200 nm to
avoid having O2 in the sample compartment absorbing the incident radiation.
In practice this means a moderate nitrogen flow rate (3-5 cm3/min) at all times,
with an increase to 20 or more cm3/min when collecting data below 200 nm.

2.2 The sample

Most CD spectrometers are single beamed instruments and can tolerate a maxi-
mum absorbance of about 2 absorbance units before the number of photons
passing through the sample become too small to be measured accurately. It is
preferable to have the maximum absorbance less than 1.5 (the optimum value
is usually approximately 1 unless the dissymmetry factor (see below) is large). It
is good practice to check that the CD signal is proportional to the sample con-
centration by running a spectrum of a diluted sample and checking that the
signal scales with concentration. If it does not, then you have evidence of
solute-solute interactions in your sample.
As the wavelength and absorbance ranges are the same for both normal
absorption and CD it is advisable to run a normal absorption spectrum of the
sample first (see Protocol 1). Always leave the reference beam of the absorbance
spectrometer empty for this experiment. In most cases the normal absorption
spectrum will indicate absorbance maxima. If these do not align (assuming the
instruments are correctly calibrated) with the CD maxima this provides
valuable information when it comes to interpreting the spectrum (see below).

2.3 The cuvette

2.3.1 Cuvette width and sample volume
AH of the light beam incident upon the cell must pass through the sample and
not be clipped or reflected by the walls or base of the cell or the meniscus of the
solution, otherwise the measured spectrum is affected by scattered light. Thus
the narrow cells often used to minimize sample volume in a normal absorption
spectrophotometer cannot be used for CD unless the light beam is chopped or
focused prior to hitting the cell. While focusing of the light beam is possible,
one must ensure that the lenses used for the focusing are not themselves signifi-
cantly birefringent (CD active) and also that the whole light beam incident on
the sample is collected by the PMT. The light beam must not be focused too
tightly on the PMT itself otherwise the PMT may be damaged. Business cards are
ideal for inserting in the light beam at -550 nm (the green light most easily
detected by our eyes) to see the beam width and height, though note that the
beam width is dependent on the instrument slit width, which in turn depends
on the lamp energy so may be larger in the UV region than at 550 nm. A typical
unmasked, unfocused beam size is about 8 mm by 10 mm.

103
ALISON RODGER AND MATTHEW A. ISMAIL

2.3.2 Cuvette pathlength and shape

For UV-visible CD, high quality quartz cuvettes are required that transmit the
full wavelength range of UV-visible light. In the visible region, glass may be
used but it is generally advisable to use quartz even here. Plastic cuvettes
typically have high intrinsic birefringence so should be avoided. In any case, the
need to run a baseline of each cuvette used (see below) removes the usual
attraction of disposable plastic cuvettes.
As with normal absorbance spectroscopy, the default cuvette pathlength is
usually 1 cm. Most CD spectrometers have cuvette holders of better design than
absorption spectrometers and will properly hold a range of cuvette sizes (and
shapes). So, if the sample has a large absorbance signal, say greater than 2
absorbance units, try using a shorter pathlength cell. If much of the absorbance
is due to solvent or buffer, this is a particularly attractive option if the analyte
concentration can be increased.
Either cylindrical or rectangular cuvettes may be used for CD. Cylindrical
cells are usually deemed to have lower birefringence (baseline CD) than
rectangular cuvettes; however, if UV and CD 'matching' is requested when the
cuvettes are purchased, rectangular cuvettes seem to be equally good. Water
jacketed cylindrical cells enable the sample to be thermostatted most simply
and also take the least sample volume for a given pathlength. With these
cuvettes, you must check that the configuration of your light beam and cuvette
holder ensure that the light beam passes through the sample and not the quartz
walls and cooling water parts of the cuvette. The light beam will almost
certainly need masking for this to be achieved.
Rectangular cells have a number of advantages over cylindrical cuvettes for
the 1 mm and longer pathlength experiments. They are cheaper, may be used
in standard absorption spectrophotometers (so CD and normal absorption data
may be collected on exactly the same sample), and may be used for serial
titration experiments as -60% of a rectangular cell can be empty for the first
spectrum and then gradually filled. Titration experiments where spectra are
collected as a function of concentration, ionic strength, pH etc. often involve
adding solution to the cuvette. A simple way to avoid dilution effects is as
follows. Consider a starting sample that has concentration x M of species X. Each
time y (ul of Y is added, also add y ul of a 2x M solution of X. The concentration of
X remains constant at x M. An infinite number of variations on this theme are
possible.
If pathlengths of 0.1 mm or less are required, it is probably best to use
demountable cuvettes where the sample is dropped onto a quartz disc that is
etched to a predefined depth and then another quartz disk is carefully placed on
top. In this case sample recovery is very difficult, so the smaller volume of cylin-
drical cells makes them more attractive than rectangular ones. A simple cuvette
holder for demountable cylindrical UV cells may be created from an infrared
cell holder and pieces of rubber (a mouse mat proves ideal) with holes drilled at
the appropriate places to allow the holder assembly screws to pass through
(Figure 4). Any such cell holder must be located perpendicular to the light beam

104
 INTRODUCTION TO CIRCULAR DICHROISM

Cylindrical quartz short pathlength

UV-visible cuvette plates
Spacer Spacer

Nuts
Metal plates
Figure 4 Cell holder for cylindrical quartz UV-visibfe short pathlength [less than 0.1 mm)
cuvettes.

with the whole incident light beam passing through the cell windows when
data is being collected.

2.4 The baseline and zeroing

A machine baseline (i.e. CD spectrum of air) measured on a standard CD
spectrometer will not be flat. However, instruments seldom have a facility for
inputting a baseline for a series of experiments. This is in part because the
cuvette used for an experiment will also have its own CD spectrum; even
matched cuvettes have slightly different intrinsic spectra. So always collect a
baseline spectrum of the solvent/buffer under the same conditions as the
sample spectrum using the same cuvette in the same orientation with respect
to the light beam. Subtract the baseline spectrum from the sample spectrum to
produce the final CD plot. It is usually the case that with small CD signals, even
when the baseline is subtracted, the CD is not exactly zero outside the absorp-
tion envelope. However, if it is flat outside the absorption envelope, the spec-
trum may be zeroed by adding or subtracting a constant (either within the
spectrometer software or using your chosen data-plotting software).

2.5 The parameters

2.5.1 Wavelength range
CD spectrometers usually scan from longer wavelengths to shorter ones. Select
a starting wavelength so that there is at least 20 nm of zero absorbance beyond
the normal absorption envelope(s) of interest. When the baseline spectrum is
subtracted from the sample spectrum (see Section 2.4), the region outside the
absorption envelope should be flat. If it is not, then this probably means either
there is a very weak absorbance band that has a large dissymmetry factor (see
Section 3.6) and hence a large CD signal compared with its normal absorbance
intensity, or, more probably, there is some light scattering of the sample.
Sources of light scattering include slightly dirty cuvettes (inside or outside),

105
ALISON RODGER AND MATTHEW A. ISMAIL

Wavelength (nm)
Figure 5 Normal absorption spectra of 100 uM calf thymus DNA in (i) 10 mM NaCI and 1 mM
sodium cacodylate buffer (solid line) and (ii) 10 mw NaCI, 1 mw sodium cacodylate and
50 uM spermine (broken line), which is sufficient to condense the DNA and provide light
scattering that may appear as an absorbance signal outside the absorption envelope.

undissolved sample, condensation of samples (a particular problem with DNA

when highly charged cations are being added; see Sections 5.10 and 5.11), and
particulate samples (of particular relevance for membrane proteins; see Section
5.7). Given that the wavelengths of light being scattered are less than 1 um, one
does not necessarily expect to be able to see by eye the presence of such par-
ticles. Scattering effects can usually be distinguished from absorption effects, as
the scattering more-or-less linearly increases with decreasing wavelength,
whereas an absorbance band should increase and then decrease. In addition, if
light scattering is the source of a sloping baseline, the normal absorption will
show a similar effect, as illustrated in Figure 5.

2.5.2 Signal to noise ratio

The signal to noise ratio in a CD spectrum increases with: V". where n is the
number of times the spectrum is accumulated (and the data averaged); VT,
where T is the time over which the machine averages each data point; and VI,
where I is the intensity of the light beam (which is in turn influenced by the
bandwidth, see below). Most CD spectrometers have both short timescale (milli-
second to minutes) and long timescale (minutes to hours) baseline variations. If
the CD signal is large, both can be ignored. To avoid any problem from short
timescale variations, collect data averaged over a number of faster scans rather
than one slow one. Longer timescale fluctuations can be more problematic and
are usually dealt with by alternating collection of sample and baseline spectra
or by assuming it is simply a linear drift that can be removed by subtracting or
adding a constant value to each spectrum (Section 2.4). While perhaps appear-
ing to be wishful thinking, the assumption of only linear baseline drift seems to
be valid especially with more recent instrumentation.

106
 INTRODUCTION TO CIRCULAR DICHROISM

2.5.3 Parameter sets

CD spectrometers give the operator considerable control over response time (T),
scan speed (s), bandwidth (b) (the wavelength range (error) of the incident light),
and data interval (d}. To optimize signal to noise effects T should be selected to
be as large as possible subject to T x s - b/2. If T is too long for the chosen s and
b, then maxima of peaks (both positive and negative) will be cut off and their
wavelengths shifted. A control scan using T' = -T/2 (or s' = 2s) should be used to
check that spectra are not being distorted by the chosen parameters. The data
interval determines how often a data point is collected. There is no point in
having, for example, b = 5 nm and d = 0.1 nm as all that will be achieved is very
long run times and many measurements being performed where essentially the
same light intensity and wavelength distribution pass through the sample. It
would be much more sensible to measure far fewer points more accurately.
Scan speeds of 100 nm/min, T = 1 s, b = 1 nm and a data step of 0.5 nm
(though see Section 5.2 for protein structure determination applications) seem
to be a good starting point as a parameter set for most experiments where the
samples have the broad band shapes usually found for solution samples. If the
bandwidth is fixed, the instrument will be programmed to control the slit width
directly. There are occasions where one may wish to adjust the slit width
manually. However, such applications are specialized and require care to avoid
damaging the instrument.
In general, it is usually advisable to perform a fast preliminary scan to
determine whether there is any point in collecting an accurate spectrum and
whether the sensitivity scale has been chosen correctly to appropriately display
the CD at all wavelengths of the spectrum. Having the display scale exactly
right is not important on modem instruments as data can be re-scaled to a great
extent (though not usually from the highest to the lowest ranges); however, it is
nice to see the spectrum as it is being collected.

2.6 Noise reduction

If the parameters for a run have been chosen correctly then there is no point in
smoothing data sets by averaging the signal from neighbouring points. All that
is achieved, with an intensity versus wavelength scan, is what would have been
more efficiently achieved by having a different parameter set. When a noisy CD
spectrum is plotted, however, it is often possible to sketch the spectrum that
one 'knows' is really there as illustrated in Figure 6. Sometimes what one should
do is repeat the experiment by averaging over more scans or increasing T and
reducing the scan speed. However, this is not always practicable if the runs are
already long and/or a series of spectra needs to be collected within the lifetime
of a sample. In this case, one wishes to have a computational option that will do
numerically what your eye and pencil can do. The 'Noise Reduction' option on
the Jasco software, for example, is designed for this. It is a Fourier transform
smoothing routine that proceeds by transforming the spectrum from the wave-
length domain into a frequency domain where the high and low frequency

107
ALISON RODGER AND MATTHEW A. ISMAIL

4 -

2 -

0 -

Wavelength (nm)

Figure 6 CD spectrum of Hoechst 33258 (20 mM). calf thymus DNA (40 mM), NaCI (20 mM)
and phosphate buffer (1 mM; pH 7) (broken line) where it is clear what the noise reduced
spectrum (solid line) should look like.

signals (hopefully corresponding to the noise) can be cut out. Transforming

back to the wavelength domain ideally gives you the spectrum you 'knew' was
hiding under the noise envelope. To ensure this is indeed the case, always over-
lay the noise-reduced result on the original data set and use your eye to decide
the validity of the transformed spectrum. If it does not 'look right', reject the
result and alter the cut-off parameters. The dangers of this type of noise
reduction are either the introduction of 'vibronic' structure or the truncation of
peaks.

Protocol 1
Measuring a routine CD spectrum
Equipment
• UV-visible absorbance spectrophotometer • Quartz cuvette of an appropriate type (see
• Computer-controlled CD spectrometer Section 2.3)
• CD spectrum measurement and data • cm3 volumes of sample and reference
analysis software solutions of uM concentrations

A. Sample preparation
1 Determine an appropriate cuvette type for the sample. Use a 1-mm pathlength
quartz cuvette for strongly absorbing samples or when only a small volume of the
sample is available. Otherwise, use a 10-mm pathlength rectangular quartz cuvette."
2 Run a UV-visible spectrum of the buffer/solvent over the wavelength range of
interest. If using a double beam UV-visible spectrophotometer, ensure that the
spectrum is collected with the reference beam empty.b

108
 INTRODUCTION TO CIRCULAR DICHROISM

Protocol 1 continued

3 Prepare a solution of the sample such that the sample absorption at the wave-
length of interest is approximately 1 but not more than 1.5 absorbance units. Run a
UV/visible absorbance spectrum, again with an empty reference beam.
4 Compare the sample and buffer solvent UV-visible spectra and check that the
wavelength maximum of interest is due to sample absorption and not some other
component of the system.

B. CD parameter selection
1 Set the wavelength range for the CD spectrum to begin at least 20 nm beyond the
absorption envelope of interest and ensure that the scan begins in a region of the
spectrum where there is no absorption.
2 Place the quartz cuvette containing the sample solution into the sample compart-
ment and ensure that the light beam is not clipping the meniscus of the solution
(increase the volume of liquid in the cuvette if necessary or perhaps raise the
height of the cuvette) or the walls or base of the cuvette (in which case lower the
cuvette or use a different cuvette type). You may be able to remove the cuvette
holder assembly to view down the light beam path. Alternatively use a mirror.
3 Run a fast, approximate scan of the sample to determine an appropriate CD
sensitivity level. Use T = 0.25 s, b - 0.5 nm and s = 500 nm min-1 accumulated over
one scan only,
4 For the accurate spectrum, set T,.Sand b such that T is as large as possible subject to
T x s - b/2. Use s - 100 nm min-1, T - 1 s, b = 1 nm and d = 0.5 nm as a starting
point for routine work. Specify up to eight accumulated scans depending on the
quality of the spectrum required/ In some instances more scans may be required. If
you are uncertain of what to choose, select a large number and stop the run when
the spectrum 'looks OK'. To ensure proper averaging over the number of scans, it is
advisable to stop the instrument while it is rewinding between scans.
5 If collecting CD data for input into the structure fitting program CDsstr (see
Protocol 2), set d = 1 nm.

C. Measuring the CD spectrum

1 Collect a sample CD spectrum using the above parameters.
2 Run the reference spectrum in the same cuvette as was used for the sample
spectrum measurement with the same parameters.d As with the sample spectrum,
ensure that no clipping of the light beam is occurring.

D. Processing the CD data

1 Obtain the baseline corrected CD spectrum by using your CD data analysis software
to subtract the reference spectrum from the sample spectrum.
2 If the spectrum is excessively noisy and adjusting the instrumental parameters to
acquire a better spectrum is not practical, apply a noise reduction algorithm (a
Fourier transform routine, for example) to improve the spectrum.

109
ALISON RODGER AND MATTHEW A. ISMAIL

Protocol 1 continued

3 Overlay the final spectrum on the original spectrum to ensure no peak shape
distortion or data mishandling has occurred.
" Other cuvette types may be appropriate in certain circumstances. See main text for details.
b
If you put the solvent or buffer in the reference beam of a double beam normal absorption
spectrometer you may get a good spectrum even if the solvent/buffer has a significant
absorbance as the instrument will subtract the absorbance of the reference cuvette; however,
there is no such compensation mechanisms in a single beam CD spectrometer.
'The signal to noise ratio is dependent on a number of factors as detailed in the text.
However, increasing the number of accumulated scans is a convenient practical method of
improving the quality of the CD spectrum.
d
By using the same cuvette for both sample and reference measurements, absorption of CPL
by the cuvette is eliminated from the final spectrum.

3 Equations of CD spectroscopy
An electronic transition occurs because either the electric field or the magnetic
field (or both) of the radiation 'pushes' the electrons to a new stationary state.
The effect of the electric field is to cause a linear rearrangement of the electrons;
the net linear displacement of charge during any transition is therefore called
the electric dipole transition moment (edtm) of the transition and is denoted by the
vector u. The magnetic field induces a circular rearrangement of electron
density. The net circulation of charge is the magnetic dipule transition moment
(mdtm), m. In an achiral molecule the net electron redistribution is always
planar. It is usually linear (having u = 0, m - 0 so being electric dipole allowed
(eda) and magnetic dipole forbidden (mdf)) or circular (having u - 0, m 1= 0 so being
electric dipole forbidden (edf) and magnetic dipole allowed (mda)), but it may also be a
planar spiral in molecules of low symmetry.
In a chiral molecule the electron rearrangement during a transition is always
helical which requires that u and m have a parallel component. The Rosenfeld
equation for the CD intensity (or CD strength or rotational strength, R) of a
transition in a collection of randomly oriented chiral molecules is (4):

where 'Im' denotes 'imaginary part of (the magnetic dipole moment operator
contains a factor of \/-1), u is the edtm for the transition from the final to the
initial state, and m is the mdtm for the reverse transition.
CD may sometimes be used to deduce the structure of a system. This is only
really viable when the system can be considered as a collection of chromo-
phores (spectroscopically well-defined subunits of a molecule) each of which is
only slightly perturbed by the rest of the system. In the rest of this section we
shall consider the coupling of two intrinsically achiral chromophorcs that are
chirally oriented with respect to one another. A range of applications of this
theory is given in (4). We have to treat eda/mdf transitions separately from

110
 INTRODUCTION TO CIRCULAR DICHROISM

mda/edf transitions since eda transitions require the induction of a magnetic

component, whereas mda transitions require an induced electric component.
The dependence of the different kinds of induced moments on the geometry of
the system is very different, thus when we wish to extract geometric inform-
ation from CD we must be aware of which situation we have.
The CD spectrum induced into eda transitions is also different if it arises
from the coupling of identical chromophores (Section 3.1) rather than from the
coupling of non-identical chromophores (Section 3.2). We shall quote the results
for each case with little justification. Full derivations using degenerate and non-
degenerate perturbation theory may be found in (4). The two cases could be
treated as one using a full secular-determinant approach. This would have the
advantage of naturally including the near-degenerate case, but has the dis-
advantage of not leading to transparent geometry/CD correlations.

3.1 Degenerate coupled-oscillator CD

Consider a molecule composed of two identical chromophores A and C. The
largest coupling takes place between eda transitions in the two chromophores
occurring at the same energy (e) so they are degenerate (5). The transition
moments of two degenerate transitions have the same length but different
origins and orientations. The two degenerate edtms couple together to make
two equal magnitude opposite handedness transition helices, one from in phase
and one from out of phase coupling. Thus we expect to see a CD spectrum with
two equal magnitude opposite-signed peaks at energies very close to e (Figure 7).
We choose the axis system and vectors illustrated in Figure 8 with A in front
and C behind. The x-axis lies along RAC, the vector from the A origin to that of C,
and z lies along the y-z projection of ua, the A chromophore edtm. We also
define three angles: a (0 < a < 180°) the angle between RAC and ua, y (0 - y <

Figure 7 CD resulting from the coupling of one transition in each of two chromophores
where both transitions are eda and degenerate (occurring at energy E). The characteristic
form of an excitonic spectrum results from cancellation of overlapping positive and negative
bands. (Source: Circular dichroism and linear dichroism, A. Rodger and B. Norden, 1997, by
permission of Oxford University Press.)

111
ALISON RODGER AND MATTHEW A. ISMAIL

180°) the angle between -RAC and uc. and T the angle passed through in going
from the y-z projection of ua to that of uC in an anticlockwise direction as illustrated
in Figure 8. Thus

We shall refer to the system as being right handed if the three vectors uc, ua
and RAC (CAR) form a right-handed coordinate system. The angles relate to the
vectors as follows (4):

where A denotes a unit vector.

The energies at which one finds the CD signals resulting from the coupling of
two degenerate chromophores are:

where e± means that e+ takes the upper of the signs of any ± or + in an

equation and (4)

CAR right-handed

Figure 8 Diagram illustrating the geometry and coordinates for the A/C system described in
the text. Note that T is the angle taken in the anticlockwise direction between the projections
of the edtms onto the y-z plane when the observer is looking down the x-axis. 0° - T <
180° if uc x ua.RAC > 0 (i.e. if the three vectors form a right-handed parallelepiped) and
180° - T < 360° if uc x (ua'RAc - 0. (Source: Circular dichroism and linear dichroism, A.
Rodger and B. Norden, 1997, by permission of Oxford University Press.)

112
 INTRODUCTION TO CIRCULAR DICHROISM

where u is the length of u3 and uc.

The two CD bands have sign and magnitude given by the following vectorial
equation (4)

The CD spectrum we expect to see for the coupling of two degenerate

transitions on two chromophores is as follows.
1. There are two bands of equal magnitude and opposite sign, R+ centred at e+
and R- centred at e-.
2. The CD strength R depends only on the strength of the transitions in the
isolated chromophores and on their orientations relative to one another and
to the vector connecting the two chromophores.
3. The sign of the CD of the '+' state is positive if uc, ua and RAC form a right-
handed axis system.
4. Vis small, so the two bands are centred close to one another and a significant
part of the total intensity of each band is cancelled. Hence the spectral form
illustrated in Figure 7.
5. Although the CD strength is a maximum when T, a, and y are all 90° (and the
system is achiral), under these circumstances V = 0, so the positive and
negative CD signals are centred at e and exactly cancel, resulting in no CD
signal as expected for an achiral molecule.
6. According to equation 10 the CD strength should get larger as the distance
between the chromophores increases. However, V concomitantly decreases
so the two component bands come closer together and more and more can-
cellation occurs, thus avoiding the nonsense situation of two chromophores
too far apart to interact having an infinitely large CD signal. The extent of
cancellation depends upon the shapes of the bands but goes approximately
linearly in Vthus the CD signal effectively decreases as RAC-2.
7. The CD of two enantiomers may be seen to be equal and opposite by
reflecting the system about the plane containing ua and RAC.
Equation 10 describes the magnitude of the area under each CD band resulting
from the coupling of ua and uc; the actual appearance of the spectrum is also
dependent on the band shape, which is usually Gaussian or Lorentzian, so we
get spectra as illustrated throughout this chapter rather than single sharp lines
at a precise energies.

113
ALISON RODGER AND MATTHEW A. ISMAIL

3.2 Non-degenerate coupled-oscillator CD

When eda transitions occur in two non-identical achiral chromophores, as with
the degenerate case, a linear motion of charge at a distance (in another chromo-
phore) has a circular (magnetic) component locally. Thus if the two transitions
couple, an induced CD signal (4) is found. The main differences between the
non-degenerate and degenerate cases are that for the non-degenerate case:
(1) the two CD bands resulting from the coupling occur at (strictly very close
to) the energies of the two non-degenerate transitions, rather than as an
exciton couplet where most of the CD intensity is cancelled, and
(2) the coupling between non-degenerate transitions is not as strong as that
between degenerate transitions.
The CD induced at ea into the transition on A by its coupling with an edtm of
different energy in C is (4)
-eaecV

c,ua
-eaec - 3RAC1
X ua.RAC} 11
KAC

{sin a sin 7 cos T + 2cos a cos -y}sin a sin -y sin T

A number of general points about non-degenerate coupled-oscillator CD

spectra (Figure 9) follow from equations 10 and 11.
1. The CD strength decreases with increasing A to C distance according to RAc-2.
This is the smallest distance-dependence for any CD mechanism; coupled-
oscillator CDs are therefore generally larger than ones arising from other
chromophore coupling mechanisms (4). Hence if a transition is both eda and
mda, then the coupled-oscillator mechanism generally dominates the
observed CD.

(a)

Wavelength - Wavelength •

Figure 9 Schematic illustration of non-degenerate coupled-oscillator CD spectra for a = -y =

90°: (a) 0 - 2T < 180° and (b) 180° - 2T < 360° (see Figure 8 for definition of T). (Source:
Circular dichroism and linear dichroism, A. Rodger and B. Norden, 1997, by permission of
Oxford University Press.)

114
 INTRODUCTION TO CIRCULAR DICHROISM

2. It is often the case that a = y - 90°. When this situation arises, the geometry
factor in equation 11 is cosTsinT = 1/2-isinf^T) so the CD is a maximum when ua
and u'' are oriented at 45° to one another.
3. If we wish to know the combined effect of many transitions in C on one
transition in A, we simply introduce a sum over c into equation I1.
4. The CD induced at t-c is determined by simply permuting all the 'a' and 'c'
and 'A' and 'C' labels in equation 11, and we see that K(ec) = -R(eJ).

33 Carbonyl n-TT* CD
Long before any theoretical understanding of the carbonyl n—TT* transition CD
had been achieved, it was found empirically that each atom (or group of atoms)
of the molecule induces a contribution to the observed CD whose sign is
determined by the octant in which the atom lies (6). Any atom that has an
equivalent 'partner' in a neighbouring octant has its contribution to the CD
cancelled so it is not a net perturber and may be ignored. Thus the CD of an
n—TT* transition has its sign determined by the factor (Figure 10)

where (x,y,z) are the coordinates of the unit vector along the line from the
carbonyl to the net perturbcr in the coordinate system of Figure 10. It was later
realized (7) that the relative contributions of two or more net perturbers scale
with their polarizability and their distance from the carbonyl group as indicated
below.
The equation describing the octant rule dependence of the dynamic coupling
induced CD of an n-rr* transition is derived as follows (4, 7). One first divides
the molecule into an achiral chromophorc A composed of the C=0 bond and
the two carbon atoms to which it is joined (Figure 11). The rest of the molecule is
divided into chromophores, C, which are subumits of the molecule that do not
exchange electrons with the rest of the molecule. For a molecule where (apart

Figure 10 The octant rule. (Source: Circular dichroism and linear dichroism, A. Rodger and
B. Norden, 1997, by permission of Oxford University Press.)

115
ALISON RODGER AND MATTHEW A. ISMAIL

Figure 11 Geometry of a carbonyl A/C system (cf. Figure 8). (Source: Circular dichroism and
linear dichroism, A. Rodger and B. Norden, 1997, by permission of Oxford University Press.)

from the carbonyl group) all the valence electrons are in sigma bonds, taking
the C to be bonds is a good approximation (7-10). Alternatively, the C may be
defined to be atoms such as carbons and nitrogens. Although the bond approxi-
mation is the better one, in most cases the difference in xyz for the middle of a
bond and one of the terminal atoms of the bond is not great. If there are TT-bonds
or non-bonding electrons in the molecule, then the unit where the electrons are
localized is often more than a single bond, e.g. if the molecule has a -COOH,
— NH2 or — C6H5 group, these will need to be treated as units.
We may write the CD of the n—*tt* transition in terms of an A optical factor,
f(A), and a C optical/geometry factor, g(C), (which includes the energy terms):

The first non-zero term of the CD strength for the carbonyl n-rr* transition has
an A optical factor

where ml is the mdtm of the n—TT*transition, Q_ay is the x-y component of

quadrupole transition moments of other transitions on A and a sum over them
is implied. The optical/geometry factor is (Figure 11) (4, 8):

where a(ea) is the isotropic polarizability of A at ea.

Thus, to apply the octant rule in the revised form indicated by equation 16
first identify the chromophores, C, into which the non-carbonyl part of the
molecule is to be divided (usually sigma bonds and functional groups as
discussed above). Second, determine for each C,

116
 INTRODUCTION TO CIRCULAR DICHROISM

• its isotropic polarizability (8)

• its distance from the carbonyl chromophore origin (which we take to be the
centre of the C=0 bond)
• the unit vector from the carbonyl origin to the Ci origin
Equation 16 may then be evaluated for each C in terms of the parameter f(A)
which we assume is negative and constant for all ketone carbonyls. In some
instances other terms in the full expression for the CD must be included (10).

3.4 d-d transitions of tris-chelate transition metal

complexes
Another class of mda/edf transitions that has been extensively studied are the
d-d transitions of transition metal complexes (Figure 12). A particularly success-
ful empirical sector rule for this system is as follows: the CD of the E band of
tris-chelate complexes such as [Co(propylenediamine)3]3+ is larger than that of
the A2 band, and the sign of the E band of the A enantiomer is negative (Figure
13). As the A2 and E bands of d-d transitions lie very close in energy (these D3

Figure 12 A-Enantiomer tris-chelate transition metal complexes illustrating the geometry of

the fee and mer isomers adopted by [Co(propylenediamine)3]3+ and the lel and ob
conformers of [Co(ethylenediamine)3]3+. (Source: Circular dichroism and linear dichroism, A.
Rodger and B. Norden, 1997, by permission of Oxford University Press.)

117
ALISON RODGER AND MATTHEW A. ISMAIL

Figure 13 d-d CD spectrum of A-[Co((R)-propylenediamine}3]3+. (Source: Circular dichroism

and linear dichroism, A. Rodger and B. Norden, 1997, by permission of Oxford University
Press.)

symmetry tris-chelate molecules are close in geometry to octahedral, in which

case the transitions would be degenerate) so they overlap. The isotropic polar-
izability equations analogous to equation 16 for this case are (11)

where Q.ax2 etc. andHax3zetc. are, respectively, quadrupole and hexadecapole

transition moments of A (11). The distance dependence of these equations
would lead us to expect the E band CD to be larger than the A2 band—i.e. the
'dominant E band rule'.
There has been some controversy about the dominant E band rule: when the
CD spectrum of crystalline [Co(ethylenediamine)3]3+ was measured it was found
that the crystal CD for the A2 component of the Tlg d-d band had almost the
same magnitude as that of the E band and both were an order of magnitude
larger than those observed in the solution spectrum—so the solution spectrum
appears to result from the cancellation of two large (A2 and E) bands (4). This
raises the question of why the E band in solution spectra is always the larger if

118
 INTRODUCTION TO CIRCULAR DICHROISM

the observed solution CD is only a small percentage of the total crystal signal.
The obvious solution, namely crystal packing effects, has been eliminated as a
possible explanation. The dilemma is probably resolved by realizing that the
above equations give the CD intensity for a D3d environment about the metal. In
practice, the symmetry is lower and other terms contribute to the CD of each
transition; however, these other contributions are equal and opposite for the A2
and E bands so cancel due to the small energy difference of the two bands (4).

3.5 Optical activity (optical rotation, OR)

Optical rotation (Section 1.2) is the difference in refractive indices of left and
right circularly polarized light upon passing through the medium:

where (n1 — nr) is called the circular Inrefringence and A and d have the same units
(usually dm). OR is usually measured by determining the rotation of linearly
polarized light upon passing through the solution. If the linearly polarized light
is rotated clockwise (viewed looking into the light source), the OR is called a
positive or right (dextro) OR. Optical rotation as a function of \ is called ORD
(see above). ORD is an S-shaped curve centred at the CD maximum (so-called
anomalous ORD). For a positive CD band the long-wavelength side of the ORD
curve is a large positive ORD contribution. The short-wavelength side is less
positive or negative. ORD is also non-zero away from absorption bands. Hence
xD values (the ORD at the sodium D line) may be used to characterize the
enantiomeric excess of a solution.

3.6 Dissymmetry factor

A measure of the size of a CD signal is given by the dissymmetry factor g for the
transition:

where R is the rotational strength of the transition, D is the dipole strength of

the transition and c is the speed of light.

4 Units of CD spectroscopy
The relationship between CD signal (in absorbance units) and sample concentra-
tion and pathlength is analogous to the Beer-Lambert law for absorbance,
namely:

where C is the sample concentration in mol dm -3 , 1 is pathlength in cm and Ae

is in units of mol-1 dm3 cm -1 . The units used for extinction coefficients are
almost always absorbance units mol-1 dm3 cm-1, which gives values that are ten
times greater than those in SI units. The moles to which one refers may be

119
ALISON RODGER AND MATTHEW A. ISMAIL

either in terms of molecules or, for macromolecules, in terms of residues such

as DNA bases or protein amino acids.
A general summary of possible CD units and their interrelationship follows.
In absorbance units we write

Alternatively in molar units it is

where er is the molar extinction coefficient for the absorption of right circularly
polarized light.
CD is often given in terms of the ellipticity, 0. 0 is obtained from the ratio of
the minor and major axes of the ellipse traced out by the electric field vector of
the elliptically polarized light that emerges from a circularly dichroic sample
onto which linearly polarized light was incident.

where n1 is the absorption index for left circularly polarized light. The wave-
length, A, must be in the same units as J. For small ellipticities

Upon converting to millidegrees, as commonly used for CD:

0 / millidegrees = 32 980 Cl(e1 - er)
= 32 980 (A - Ar)
Most CD spectrometers, although they actually measure differential absorb-
ance, produce a CD spectrum in ellipticity, with units 9, in millidegrees, versus
A,rather than AA versus X. Most CD machines will perform the conversion from
ellipticity to absorbance units on request.
Another old measure of CD is the specific ellipticity

where Cg is sample concentration in g cm-3 and d is pathlength in dm.

The related molar ellipticity is

where M is molar mass in g, C is in mol dm-3, and I is in cm. If dm-3 is sub-

stituted by 1000 cm-3 we see that molar ellipticity can have the commonly used
units of deg cm2 dmol-1. Thus

120
 INTRODUCTION TO CIRCULAR DICHROISM

where er is the extinction coefficient for the absorption of right circularly polar-
ized light. A related unit that is often used is the mean residue ellipticity. In this
case, as mentioned above, the sample concentration is given in terms of total
numbers of residues (using the average molecular mass of an amino acid resi-
due) without reference to their identity, hence it is an average over the types of
residue present in the macromolecule.

5 Circular dichroism of biomolecules

5.1 Introduction
CD is commonly used to study biological macromolecules in one of two ways:
(1) to probe changes in the conformation of the macromolecule itself, and
(2) to probe its interaction with small molecules, especially achiral ones whose
induced CD is due solely to their interaction with the macromolecule.

5.2 CD of polypeptides and proteins

Proteins are linear polymers of well-defined sequences (the primary structure)
of amino acids that are folded in well-defined ways, with the overall shape of a
protein molecule being crucial for its biological activity. Proteins form regular
secondary structural units because the peptide 0=C-N- link between amino
acids is planar and rigid yet it has a large degree of rotational freedom about its
bonds to the rest of the protein chain (Figure 14). This both constrains the poss-
ible relative orientations of neighbouring residues and allows a variety of intra-
molecular hydrogen bonding arrangements between the C-0 of one peptide
unit and the N-H of another unit. The resulting limited set of common chiral
secondary structural units have more-or-less well defined CD signatures (Figure
15). It is this feature that makes CD so useful in the study of proteins.

5.3 Protein UV spectroscopy

UV spectra of proteins are usually divided into the 'near' and 'far' UV regions.
The near UV in this context means 250-300 nm and is also described as the
aromatic region, though transitions of disulphide bonds (cystines) also contribute
to the total absorption intensity here. The far UV (<250 nm) is dominated by
transitions of the peptide backbone of the protein, but transitions from some
side chains also contribute in this region and, especially if the protein a-helical
content is low, may give rise to erroneous protein structure determinations (see
Section 5.4).
The lowest energy transition of the peptide chromophore is an n-ir* trans-
ition analogous to that in ketones, and the next transition is TT-TT*. The n-rr*
transition has an extinction coefficient of —100 mol-1 dm3 cm-1; it occurs at
about 210-230 nm (depending mainly upon the extent of hydrogen bonding of
the oxygen lone pairs) and its electric character is polarized more or less along
the carbonyl bond. The r r - * transition (e ~ 7000 mol-1 dm3 cm-1) is dominated

121
ALISON RODGER AND MATTHEW A, ISMAIL

Figure 14 (a) The peptide bond, (b) the primary structure of a protein, and (c) the commonly
occurring L-amino acids. (Source: Circular dichroism and linear dichroism, A. Rodger and B.
Norden, 1997, by permission of Oxford University Press.)

Figure 15 Typical protein CD spectra for particular secondary structural motifs used in CD
protein structure fitting programs.

by the carbonyl ir-bond and is also affected by the involvement of the amide
nitrogen in the -rr-orbitals; its electric dipole transition moment is polarized
somewhere near the line between oxygen and nitrogen and it is centred at
190 nm. In an a-helix (see below), the electric dipole coupling of the TT—tt*

122
 INTRODUCTION TO CIRCULAR DICHROISM

transitions on neighbouring residues results in a long wavelength component of

this transition at ~208 nm.
The aromatic side chains, phenylalanine, tyrosine and tryptophan, all have
transitions in the near UV region. The indole of tryptophan has two or more
transitions in the 240-290 nm region with total maximum extinction co-
efficient emax(279 nm) ~ 5600 mor1 dm3 cm -1 ; tyrosine has one transition with
emax(274 nm) ~ 1400 mol-1 dm3 cm-1. phenylalanine also has one transition
with emax(258 nm) ~ 190 mol-1 dm3 cm-1. and a cystine disulphide bond
absorbs from 250-270 nm with emax ~ 300 mol-1 dm3 cm -1 . Although trypto-
phans have by far the most intense transitions in this region, many proteins
have few tryptophans compared with the other aromatic groups, so the region
is not necessarily dominated by tryptophan transitions. Metallo-proteins have
so-called prosthetic or extrinsic groups with additional chromophores that
often have transitions in the visible region of the spectrum as well as the UV.

5.4 Protein structure determination from CD

At the present the main use of CD in the study of proteins is as an empirical
gauge of protein structure and conformation using the CD induced into the
backbone amide transitions from ~ 190-240 nm (the far UV or peptide region of
the spectrum). In the absence of any contributions to the CD from side chain
transitions, this has proved to be a very successful approach (12-22). Distinctive
CD spectra (Figure 15) have been described for pure conformations such as the a-
helix, p-sheets (with different ones sometimes being given for parallel and
antiparallel sheets), B-turns and also the 'random' coil. At least in principle, the
CD spectrum of a native protein is then the sum of the appropriate percentages
of each component spectrum. However, relative arrangements of structural
units and motifs such as disulphide bonds contribute to the observed CD
spectrum.

5.4.1 a-Helix CD
The (right handed) a-helix is the dominant secondary structure in many pro-
teins and on average accounts for about one third of the residues in globular
proteins. It is a well-defined structural motif where the nth peptide unit forms
hydrogen bonds between its C-O and the N-H of the (n + 4)th peptide and
between its N-H and the (n - 4)th C-O; there is a 0.15 nm translation and 100°
rotation between two consecutive peptide units, giving 3.6 amino acid residues
per turn. Its helix pitch (number of residues times distance between a-Cs on
neighbouring residues) is 0.54 nm.
The CD spectra of a-helices are characterized by a negative band with
separate maxima of similar magnitude at 222 nm (the n-rr* transition) and 208
nm which is part of the rr-TT* transition (see Section 5.3 and Figure 15). The a-
helix is the only motif where the rr—TT* transition has such a long wavelength
component. The a-helix CD is also larger in magnitude than that due to other
motifs so it is apparent upon the most casual inspection of a spectrum. It is no

123
ALISON RODGER AND MATTHEW A. ISMAIL

surprise therefore that the various empirical CD fitting programs that are
available are fairly successful in determining the percentage of a-helical content
of a protein. It should, however, be noted that the magnitude of the CD signal
does vary with variations in the helix, and the helix length (the signal behaves
as if it were for a helix about four residues shorter than it in fact is, which
corresponds to the number of unanchored hydrogen-bonding groups) and the
interactions with neighbouring structural units.

5.4.2 310 helix CD

The 310 helix is a right handed helix with 3.0 residues per turn and a helix pitch
of 0.6 nm.

5.4.3 Proline like 3/1-helix CD

The polyproline type II helix is that adopted by the polypeptide chains of
collagen. It has 3.0 residues per turn and a helix pitch of 0.94 nm.

5.4.4 B-sheet CD
An alternative efficient formation of hydrogen bonds occurs between a sheet of
parallel or antiparallel runs of amino acids; these are known as a p-sheets. The
runs of amino acids face in alternate directions so that alternate amino acids
hydrogen bond to neighbouring runs on each side. The spectroscopic character-
ization of p-sheets has proved more difficult than that of a-helices due to the
practical reason that they are less soluble in solvents with a good UV trans-
mission, and due to the intrinsic reason that they are generally structurally less
well-defined: they may be parallel or antiparallel and of varying lengths and
widths. Furthermore, an extended B-sheet is usually found to show a marked
twist, rather than to be planar. Such tertiary structure influences the overall CD
spectrum.
The general characteristics of B-sheet CD may be taken to be a negative band
at about 216 nm and a positive band of comparable magnitude near 195 nm
(13). This level of characterization of the spectrum might be deemed to be not
much worse than that of a-helices were it not for the fact that what is often
called the 'random coil' but is more correctly referred to as 'other' structural
features (see below) have their CD maxima at similar wavelengths and are often
of opposite sign from those of the B-sheet. This means that an empirical fitting
program may incorrectly weight these spectra (and the B-turn components) in
an attempt to better account for the wavelength and magnitude variations that
occur.

5.4.5 B-Turn CD
Typically the strands of an antiparallel B-sheet are linked by B-turns where the
nth peptide unit forms hydrogen bonds with the (n + 3)rd peptide unit. How-
ever, the label B-turn is usually used to include all possible turns that occur, not
simply the ones that enable a single strand to become an antiparallel B-sheet.
About one quarter of the residues in globular proteins then fall into this

124
 INTRODUCTION TO CIRCULAR DICHROISM

structural group. Despite this range of structures, a 'typical' B-turn CD spectrum

has been identified which has a weak red-shifted negative n-TT* band near 225
nm, a strong positive TT-^TT* transition between 200 nm and 205 nm, and a
strong negative band between 180 nm and 190 nm.

5.4.6 'Random' coil CD

The label random coil describes the fully denatured state of the protein. How-
ever, the term is also often used collectively of the parts of the protein that are
not a-helices, p-sheets or turns. A spectrum such as that illustrated in Figure 15
results for a truly random coil. The other structural features in an average pro-
tein often have a strong negative CD signal just below 200 nm, a positive band
at about 218 nm in many systems, and perhaps a very weak negative band at
235 nm. Folded proteins have no true random coil elements.

5.4.7 Aromatic region

The intensity of the CD induced into the achiral aromatic side chains is very
dependent on their environment so, in contrast to fluorescence, it is not possible
with CD to limit consideration specifically to the tryptophans. Groups which can
be essentially ignored in the absorption spectrum may become significant in the
CD. In particular, disulphide groups (covalent bonds formed between cysteine
residues in different parts of a protein) have transitions in the aromatic region of
the spectrum. Although these transitions are weak (emax ~ 300 mol-1 dm3 cm-1 at
250 and 270 nm), they are magnetic dipole allowed and so may contribute signifi-
cantly to the protein CD in this region, often with tails stretching beyond 300 nm
(which aids their identification). As the side chains are usually isolated from one
another their CD usually arises from interactions with the neigh bouring amino
acid residues, as would be the case for a ligand bound to the macromolecule.

5.5 Determining the percentage of different structural units

in a protein from peptide region CD spectra
There are a range of different computer programs available for determining the
percentage of different structural motifs just from the CD spectrum (14). The
cost of these programs does not necessarily correlate with their reliability. Most
programs will give an accurate a-helix content, if you accurately know the
concentration of your protein (usually in terms of residues rather than
molecules). We shall outline the use of a program written by W. C. Johnson, M.
Parthasarathy and A. Toumadje for determining the structure of proteins from
CD data (15). The program uses ideas developed by many other workers and is
applicable only to protein structure determination. Small peptides require a
different approach as discussed below.
The program, CDsstr, is currently available via ftp from alpha.als.orst.edu by
logging in as 'anonymous' and using your email address as the password.
Changing directory by typing 'CD/pub/wcjohnson/CDsstr', then 'bin' to specify
binary file transfer, and then typing 'mget *.*' will retrieve the program files.

125
ALISON RODGER AND MATTHEW A, ISMAIL

CDsstr computes the percentages of secondary structures of a protein from

CD data using the method of Hennessey and Johnson (16) and the single value
decomposition algorithm of Compton and Johnson (17). Rather than using
typical spectra for each type of secondary structure motif, CDsstr uses a basis set
of 22 proteins with known secondary structure and known CD spectra. The
program self-consistently chooses a minimum sub-basis set of eight proteins for
a given problem using methods given in (12 15). The program gives percent-
ages of H-helix, 3/10-helix, extended-p-strand, B-turns, polyproline-like 3/1-helix
and 'others' which are explicitly not random coils. The advantage of this
approach over that of representing the CD spectrum as a sum of component
parts is that the CD due to aromatic and sulphur-containing side chains etc. as
well as helix length variations and interactions between neighbouring secondary
structural units are all accounted for because they are contained in the basis sets.
The current version of the program runs on a PC under Windows 95/98 or
NT4. The program consists of several files. The 22 basis set CD spectra are
contained in the file basCD.dta and the corresponding secondary structure data
resides in the file secstr.dta. The protein CD data to be analysed are read from
the file proCD.dta. The proCD.dta file should consist of one title line followed by
the CD data in units of Ae-, in 1 nm increments with each data value on a new
line and beginning at 260 nm and ending at 178 nm. Numerical values are read
in Fortran 1F10.2 format (i.e. one floating point number with a total of 10
characters, including any minus sign and decimal point, of which two lie after
the decimal point). This program apparently works reasonably well on data
truncated even at 200 nm, but more data give better answers.

Protocol 2
Application of CDsstr to compute protein secondary
structure from CD spectra
Equipment and materials
• A PC running Windows 95/98 or NT4 • CD data for the proteins to be analysed
• A copy of the CDsstr program

A. Running CDsstr for the first time

1 Work within windows.
2 Create a folder named CDsstr in the root directory of the hard disk (assumed to be
drive c: in the rest of this protocol)."
3 Copy all the CDsstr program files into the new CDsstr folder,
4 Delete any file with an .out filename extension from the CDsstr folder.
5 Delete the file named proCD.dta if it exists.
6 Verify the correct operation of the program by running the calculation on the test
files provided. To do this, open a DOS window by selecting 'Command Prompt'

126
 INTRODUCTION TO CIRCULAR DICHROISM

Protocol 2 continued

from the 'Programs' menu in the 'Start' menu. If'Command Prompt' is not present
in the 'Programs' menu, open a DOS window by selecting 'Run' from the 'Start'
menu and typing 'c:\windows\command.com' in the box then clicking 'OK'.*b
7 At the command prompt type 'c:' then 'CD\CDsstr',
8 Type 'copy proCD.tst proCD.dta'.
9 Type 'CDsstr' to initiate the test,
10 Enter the value of each variable as prompted: NbasCD = 22; Nwave = 83; Npro = 3;
ncomb = 100; icombf = 100000.
11 When the command prompt reappears, compare the values in the files anal.out
and reconCD.out with those in anal.tst and reconCD.tst. If CDsstr is functioning
correctly they should exactly match.
B. Using CDsstr for protein CD analysis
1 Delete, rename, or move any file with an .out filename extension remaining in the
CDsstr folder.
2 Delete any previously used file named proCD.dta unless you wish to use it in the
current run.
3 If it is not already available, prepare an input file containing the CD data of the
protein(s) to be analysed. Enter a title on the initial line then enter the CD data
as is (for molar residue concentration) with each value on a new line. Begin at
260 nm and continue in 1 nm increments down to 178 nm. Ensure that the
numerical values entered comply with the Fortran 1F10.2 format. If multiple
protein data sets are being analysed, begin each new set with a title line. Save the
file as c:\CDsstr\proCD.dta.c
4 If using truncated protein CD data (truncation above 200 nm is not recommended),
the basis set protein CD spectra in the basCD.dta file must also be truncated by
removing data points from each data set to match the experimental data set size,c
5 Determine appropriate values (see above) for the variables NbasCD, Nwave, Npro,
Ncomb and icombf.
6 Begin the analysis by opening a DOS window (see Section A above). Type 'c:'. Then
type 'CD\CDsstr' at the command prompt.
7 Enter values for the program variables as prompted.
8 When the command prompt reappears, view the results of the analysis by in-
specting the output files anal.out and reconCD.out.
a
Do not place this folder within any subdirectory if you wish to follow this protocol exactly.
b
CDsstr must be run in MS-DOS from within Windows, Restarting the computer in MS-DOS
mode will not permit the program to run.
c
A commercial spreadsheet is well suited to editing CD data files. Most CD data analysis
software permits a data file to be exported as text. The text file can then be imported into the
spreadsheet, easily formatted for CDsstr input and saved as undelimited ASCII text. Delete,
rather than cut, data cells to remove unwanted text/data.

127
ALISON RODGER AND MATTHEW A. ISMAIL

Upon execution, CDsstr prompts the user for the value of several program
variables. NbasCD is the number of basis set CD spectra and should be set to 22
if the default basCD.dta and secstr.dta files are being used. Nwave is the number
of wavelengths contained in basCD.dta and is 83 for the default basis set (i.e.
with data down to 178 nm). The number of protein CD spectra contained in the
proCD.dta file, Npro, should then be entered and must lie between 1 and 100
(3 for the test file proCD.tst). The value for the number of successful combina-
tions to be considered, ncomb, should usually be set to 100 except for alpha
helical proteins where the program's maximum value of 400 should be used. The
final requested parameter value is the maximum number of trial combinations
allowed before the program stops the analysis, icombf. A value of 100 000 is
usually appropriate for this parameter. The calculated secondary structure out-
put is written to the file anal.out along with the input data. A second output file,
reconCD.out contains the average reconstructed CD spectrum.
Peptide CD structure analysis differs from that of proteins as peptides are
usually a combination of a particular secondary structure and random coil (i.e.
unordered) residues. Proteins seldom if ever have random coils, so the CD of
the proteins in the basis set of the protein analysis program has no random
coil component. Further, the analysis of the CD of a peptide is not under-
determined, and so does not require a flexible method like Variable Selection. A
Convex Constraint Analysis (22) that extracts the component spectra has been
developed into a peptide structure analysis program by Greenfield (14).

5.6 Other applications of protein CD

Protein CD may also be used to probe whether the protein structure is per-
turbed in any way upon interaction with other molecules or as a function of a
variable such as pH or solvent. For example, a-helices may be stabilized or
induced by trifluoroethanol (23). Conversely a-helices may be destabilized by
dissolving them in non-polar media such as lipids (24). Such conformational
changes may be followed using CD. Either aromatic (near UV) or backbone (far
UV) CD may be used for a qualitative or empirical study of this kind. CD has also
been extensively used in protein folding studies, some of which are described in
more detail in Chapter 10. CD can also be used to give quantitative estimates of
binding constants when ligands bind to the protein using methods analogous to
those outlined for DNA-ligand systems (see Sections 5.10 and 5.11).

5.7 Membrane proteins

The absorption and CD spectra of membrane proteins are often distorted by
artefacts that arise as consequences of light scattering (25) (due to the large size
of the membrane particles relative to the wavelength of light) and absorption
flattening (26) (due to local high protein concentrations within the particles).
The best solution to these problems is to avoid them by incorporating the
membrane proteins into small particles, such as unilamellar vesicles, where the

128
 INTRODUCTION TO CIRCULAR DICHROISM

protein concentration is low. A useful system for this is described below. If no

appropriate small particle can be found for a given system, then the artefacts
must be removed in order to use the data.
The light scattering is only a problem because scattered light that does not
reach the PMT is assumed to have been absorbed. As light scattering increases
with decreasing wavelength, the effect of scattering is to cause a net increase in
magnitude of signals and a shift of maxima. The light scattering problem
disappears if the configuration of the instrument is such as to collect all the
scattered light as well as the unscattered light. Mao and Wallace (27) suggest
that a collection angle of 90° is sufficient. Our experiments (28) with a reflecting
surface designed to collect all scattered light and the Jasco J-715 instrument
equipped with the large sample compartment suggest that, in a J-715, essentially
all the scattered light is collected by moving the sample holder into the recess
before the PMT. This results in a collection angle of at least 72°.
The absorption flattening problem can only be treated at the sample prep-
aration stage. Somehow local high concentrations must be avoided. Detergent
solubilization, with detergents including Triton X-100, n-octyl glucoside, and
sodium dodecyl sulphate, generally reduces membrane particle sizes. However,
care should be taken with this approach given the use of the same reagents to
denature proteins. It cannot be assumed that the structure (and hence CD) of
the solubilized and original protein are the same. An alternative is to sonicate
a sample of a membrane protein with, for example, dimyristoylphosphatidyl-
choline (27) and thus to form small unilamellar vesicles (less than 100 nm
inside). A range of lysophospholipids and phospholipids have been used for this
purpose and for incorporating membrane proteins in micelles (29). The size of
the particles should be checked before use in a CD experiment.

5.8 DNA geometry and CD spectra

To a first approximation double helical DNA (or RNA) can be viewed as a more-
or-less vertical spiral staircase, where the steps are made of pairs of nitrogenous
bases hydrogen bonded together and the support is the backbone of alternating
phosphate groups and ribose sugars (Figure 16). In isolation from the chiral sugar
units the phosphates and the bases are achiral; however, when they are joined
together with the sugar units they become part of a chiral molecule, and their
transitions are expected to exhibit CD spectra. The helix of the standard B-DNA
(the common solution polymorph) staircase is right-handed and the steps are
approximately perpendicular to the helix axis. Some geometry parameters for B-
DNA and the other common polymorphs A-DNA and Z-DNA are given in Table 1.
Single-stranded DNAs are structurally less well-defined than duplex DNAs and
their CD signal is smaller. CD spectra of naturally occurring RNAs (Figure 17),
which are often assumed to be single stranded, clearly show naturally occurring
RNAs to be mainly well-structured with extensive A-form duplex regions (see
Section 5.9).

129
ALISON RODGER AND MATTHEW A. ISMAIL

Figure 16 la) B-, A- and Z-DNA viewed perpendicular to the helix axis and down the helix
axis, (b) Structural formula of DMA illustrating the definitions of torsion angles. (Source:
Circular dichroism and linear dicrimism, A. Rodger and B. Norden, 1997, by permission of
Oxford University Press.)

no
 INTRODUCTION TO CIRCULAR DICHROISM

Table 1 Typical geometric parameters (see Figure 8 for definitions) for standard A-, B- and Z-
forms of DNA (30-34)

Parameter A-DNA B-DNA Z-DNA

a -85° -47° 60°/160°
-152° -146° -175°/-135°
B . . ...... . . . . . , . ,
"V 46° 36° 178757°
5 83° 156° 140°/95°
e 178° 155° -957-110°
I ' ^ II' '-46° -95° -350-850
Sugar conformation Cy-endo C2-endo C3'-endo/C2-endo
Glycosidic bond anti anti anti(C, T), syn(G)
Base roll 12° o° 1°
Base tilt 20° 5° 9°
Base helical twist 32° 36° 11°/50°
Base slide 0.15 nm 0 nm 0.2 nm
Helix diameter 2.55 nm 2.37 nm 1.84 nm
Bases per helix turn 11 10.4 12
Base rise per base pair 0.23 nm 0.33 nm 0.38 nm
Major groove Narrow, deep Wide, deep Flat
Minor groove Broad, shallow Narrow, deep Narrow, deep

RNA differs from DNA in that every T is replaced by U, which has one less methyl group, and the backbone sugar
is ribose instead of deoxyribose. Double helical RNA adopts the A-form geometry. The repeat unit of Z-DNA is a
dinucleotide, necessitating two values of each angular parameter. adopts a range of values in Z-DNA.

Figure 17 CD of a long (372 nucleotide) naturally occurring mRNA. At 25 °C the RNA is

largely that of an A-form double helix; at 85 °C there is less chirality in the parts of the
molecule being probed—in this case the RNA bases. This is due to there being less base
pairing and base stacking at 85 °C. A260 nm = 1.0 at 25 °C. (Source: Circular dichroism and
linear dichroism, A. Rodger and B. Norden, 1997, by permission of Oxford University Press.)

5.9 UV spectroscopy of the DNA bases

The UV absorbance of nucleic acids from 200-300 nm is due exclusively to
transitions of the planar purine and pyrimidine bases. The backbone begins to
contribute at about 190 nm. The accessible region of the spectrum is therefore

131
ALISON RODGER AND MATTHEW A. ISMAIL

Figure 18 Probable transition polarizations for UV transitions of guanine and cytosine.

(Source: Circular dichroism and linear dichroism, A. Rodger and B. Norden, 1997, by
permission of Oxford University Press.)

dominated by -TT* transitions whose exact assignment is still a matter of

debate. All of them are polarized in the plane of the bases. Recent assignments
for guanine and cytosine are illustrated in Figure 18 (35—38). Adenine probably
has five IT—TT* transitions in the UV region (39). There may also be some n—TT*
transitions present whose intensity (even in CD) is so small that they have not
been definitively identified.

5.10 Nucleic acid CD

When we measure a CD spectrum of a polynucleotide with stacked bases, the
magnitude is larger at 270 nm and significantly larger at 200 nm than that of
the individual bases (4). The spectrum is dominated by the CD induced into
transitions of the bases from their coupling with each other when the bases
stack in a chiral (helical) fashion. When the CD is measured as a function of
temperature, a melting curve is plotted that enables the double- to single-
stranded transition to be followed. The resulting single-stranded, or approxi-
mately random coil, molecule still has significant CD at 270 nm due to the
intrinsic base CD (Figure 17).

5.10.1 Empirical structural analysis of DNA CD

The simplest application of CD to DNA structure analysis is to identify the
polymorph in a sample (4). Since the DNA CD from 200-300 nm is due to the
skewed orientation of the bases, if the DNA is untwisted or the bases are tilted,
a change from what is observed for B-DNA would be expected in the CD spec-
trum. Somewhat surprisingly, observed CD spectra often vary more with
changes hi base orientation (DNA polymorph) than as a function of the base
composition of the DNA, though they are of course a sensitive function of DNA
sequence as well.
The spectrum of calf thymus DNA shown in Figure 19 is typical of B-form
DNA. The CD signature of B-form DNA (Figure 20), as read from longer to shorter
wavelength, is a positive band centred at 275 nm, a negative band at 240 nm,
with the zero being around 258 nm. At 220 nm the CD signal is either positive
or less negative; a small negative peak is then followed by a large positive peak

132
 INTRODUCTION TO CIRCULAR DICHROISM

220 240 260 280 300

Wavelength / nm

Figure 19 CD spectra of duplex DNAs (pH = 6,8, 5 mM NaCI, 1 mM cacodylate) of varying

G-C base pair content: Clostridium perfringens (26% GC), calf thymus (42% GC), Micrococcus
lysodeikticus (72% GC), poly[d(G-C)]2 (100% GC). Also shown is Z-form poly[d(G-C)]2 (100%
G-C, pH = 6.8, 5 mM NaCI, 1 mM cacodylate, 50 uM [Co(NH3)6]3+). The 195 nm negative
signal of Z-DNA is obscured by the CD induced into the transitions of [Co(NH3)]3+ when this
molecule is used to induce Z-form DNA. (Source: Circular dichroism and linear dichroism, A.
Rodger and B. Norden, 1997, by permission of Oxford University Press.)

-20
210 230 250 270 290 310 220 240 260 280 300 320
Wavelength / nm Wavelength / nm

Figure 20 (a) Poly[d(A-T)]2 (44 uM; 1 mM phosphate) in B-form (20 mw NaCI) and condensed
(1 M NaCI and 45% v/v ethanol). (b) Poly[d(G-C)]2 as a function of increasing spermine,
[NH3(CH2)3NH2(CH2(CH2CH2)3NH3], concentration showing the B - Z transition. Spermine
concentrations are indicated in the figure. (Source: Circular dichroism and linear dichroism,
A. Rodger and B. Norden, 1997, by permission of Oxford University Press.)

from 180-190 nm. When B-form DNA is stretched to have 10.2 bases per turn
instead of 10.4, the 275 nm peak essentially disappears. This form of DNA can
be induced by methanol or by wrapping the DNA around histone cores to form
nucleosomes.
If B-DNA is compacted and the bases tilted and radially displaced from the
centre of the helix (thus creating a hole when one looks down the helix axis)
(Figure 16 and Table 1) then A-form DNA results. A-DNA is characterized by a

133
ALISON RODGER AND MATTHEW A. ISMAIL

positive CD band centred at 260 nm that is larger than the corresponding B-

DNA band, a fairly intense, comparatively sharp negative band at 210 nm and a
very intense positive band at 190 nm. The 250-230 nm region is also usually
fairly flat though not necessarily zero. Naturally occurring RNAs adopt the A-
form if they are duplex. A typical natural RNA spectrum is shown in Figure 17.
Z-form DNA (Figure 16 and Table 1) does not readily form for all sequences.
However, it is easily formed for poly[d(G-5meC)]2 in the presence of highly
charged ions and also for poly[d(G-C)]2 under appropriate conditions (40). Z-DNA
is characterized by a negative CD band at 290 nm and a positive band at 260 nm
(Figure 19). Care must be taken in using these signatures to identify Z-form DNA,
since the same DNA in the A-form has a negative band at 295 nm and a positive
band at 270 nm (41). Poly [d(I-C)]2, I = inosine, is also negative at long wave-
lengths. A more definitive signal is the large negative CD signal in the 195-200
nm region for Z-DNA, whereas B-form DNA CD is near zero or positive in this
region. For Z-DNA, the CD passes through zero between 180 and 185 nm. Con-
densed DNA also has a characteristic CD spectrum (Figure 20).
5.10.2 Quantitative structural analysis of DNA CD
With DNA, as discussed above, it is usually possible to identify qualitatively the
overall polymorph that is present, but a more quantitative empirical analysis has
defied extensive efforts. Somewhat perversely, however, theoretical analysis of
DNA CD performed by calculating the spectrum for a proposed geometry is
possible, whereas for proteins this approach has been less successful (41-43).

5.11 DNA/ligand interactions

Many DNA binding molecules (usually referred to as ligands or drugs even when
they may have little or no therapeutic value) are themselves achiral. However,
when they bind to DNA they acquire an induced CD (ICD) that is characteristic
of their interaction (Figure 21). If the ligand transitions of interest occur in the
DNA region of the spectrum we take the ICD to be the total CD of the system
minus that of the DNA at the same concentration. This means any DNA con-
formational changes that alter the DNA CD or any ICD of the DNA itself are
included in the net ICD of the interaction. As with application of CD to proteins
and DNA the ligand ICD can be used on a number of levels. The simplest use is
to note that it exists and therefore conclude that the molecule does bind to DNA.
However, more information can usually be extracted.
5.11.1 Empirical analyses of ligand ICD
It is often very useful to measure a series of spectra where some variable such as
the ionic strength, or mixing ratio (DNA:drug ratio), or temperature is changed.
If the CD intensity changes but the shape of the spectrum remains the same
during such an experiment, then it can be deduced that the ligand binding
mode is unchanged though the amount of bound ligand may have changed
(Figure 21). In such an experiment, if possible, it is best to keep the ligand
concentration constant as then any changes are more easily apparent. If the

134
 INTRODUCTION TO CIRCULAR DICHROISM

ligand:DNA
0.008:1
0.015:1
0.022:1
0.029:1
0.037:1

20
-60
220 240 260 280 300 240 260 280 300
Wavelength / nm Wavelength / nm
Figure 21 ICD of varying concentrations of anthracene-9-carbonyl-N1-spermine upon
interaction with (a) 160 uM poly[d(G-C)]2, 5 mM NaCI and low ligand:DNA:phosphate mixing
ratios, and (b) 130 mm poly[d(A-T)]2, 5 mM NaCI and a wide range of ligand:DNA:phosphate
mixing ratios. / = 1 cm. The smallest magnitude ICD signal corresponds to the lowest ligand
concentration. (Source: Circular dichroism and linear dichmism, A. Rodger and B. Norden,
1997, by permission of Oxford University Press.)

ligand binds in a single binding mode, or in a number of sites whose relative

proportions are independent of DNA:drug ratio, then we may write

where Lb is the concentration of bound ligand, and a is a proportionality con-

stant. If you can draw a plot of CD versus DNArdrug ratio that has a reasonably
straight part, followed by a curve, and finally a levelling off part, then one of a
number of the analysis methods may be used to determine equilibrium binding
constants and site sizes (44-47).
Consider the equilibrium association constant 1C

for the equilibrium

where Lf is the concentration of free ligands, and Sf is the free site concen-
tration. The total site concentration, Stot is given by

where CM is the macromolecule concentration. For DNA we usually use the

concentration of bases, in which case n is the number of bases in a binding site.
For a protein, CM is usually taken to be the concentration of protein molecules
in which case

is the number of ligand binding sites on each protein molecule.

135
ALISON RODGER AND MATTHEW A. ISMAIL

To perform a Scatchard plot (48) to analyse the data to give K directly, re-
arrange equation 31 as follows. If we know a (the proportionality constant, see
equation 30) then from ICD spectra we determine Ib and hence Lf

35

where
36

Thus, a plot of r/Lf versus r has slope -JC and y-intercept K/n. The x-intercept
occurs where r = 1/n. The Scatchard plot is a better way of averaging over a data
set than, for example, an arithmetic mean from direct application of equation 31
for a number of data points. The problem with the Scatchard plot is that one
needs to know the value of a in equation 30 to determine Ib, and hence Lf from
the ICD. Ensuring all the ligand in a solution is bound as required for a direct
determination of a is often impossible with DNA. The 'Intrinsic method' was
developed to solve this problem (Figure 22).
We write

and rearrange this to give

Thus, for two different total ligand concentrations, LJtot and Lk t , but the same
macromolecule concentration, i.e. SJot = Stot,

Figure 22 Graph illustrating application of the intrinsic method for determining a and n using
the data given in Figure 21. K has been determined using an average value from repeated
application of equation 37. Alternatively, the data could have been used to perform a
Scatchard plot, x and y are as defined in equations 40 and 41. (Source: Circular dichroism
and linear dichroism, A. Rodger and B. Norden, 1997, by permission of Oxford University
Press.)

136
 INTRODUCTION TO CIRCULAR DICHROISM

which gives a plot with slope Ltot/a and y-intercept na.

If there is a change in the shape of the spectrum as the mixing ratio or other
experimental variable is changed, this implies that there is a change in the
DNA-drug interaction as a function of the experimental variable. The change is
usually due to occupancy of more than one binding site when the drug load on

Figure 23 CD of constant concentration of 9-hydroxyellipticene (50 uM) in the presence of

calf thymus DMA (concentrations indicated in figure, 5 mM NaCI, 1 mw phosphate buffer).
(Source: Circular dichro/sm and linear dichroism, A. Rodger and B. Norden, 1997, by
permission of Oxford University Press.)

137
ALISON RODGER AND MATTHEW A. ISMAIL

the DNA increases. Therefore the above methods for determining K cannot be
used as equation 30 does not hold. We may also be observing changes in the DNA
conformation or to ligand-ligand interactions. A particularly dramatic change
in the ligand ICD is observed if as the DNA concentration is decreased (so one
would expect fewer ligands to bind and so a decrease in ICD), the ligands begin to
stack together and the ICD increases. Under these circumstances, we generally
observe a large excitonic CD. Thus spectra such as those of Figure 23 immedi-
ately tell us that significant ligand-ligand interactions are occurring on the
DNA.

Acknowledgement
This text has been adapted from Circular Dichroism and Linear Dichroism by
Alison Rodger and Bengt Norden (1997) by permission of Oxford University
Press.

References
1. Craig, D. P. and Thirunamachandran, T. (1984). Molecular quantum electrodynamics: An
introduction to radiation-molecule interaction. Academic Press, London.
2. Barren, L. D. (1982). Molecular light scattering and optical activity, p. 1. Cambridge
University Press, Cambridge
3. Grandison, A., McGinley, D., Shearer, T., Knight, L., Summers, E., Lyons, C. and
Forde, C. (ed.) (1994). Collins English dictionary (3rd edn), p. 284. HarperCollins
Publishers, Glasgow.
4. Rodger, A. and Norden, B. In Circular and linear dichroism (ed. R. G. Compton, S. G.
Davies, and J. Evans), p. 1. Oxford University Press, Oxford.
5. Bosnich, B. (1969) Ace. Chem. Res., 2, 266.
6. Moffitt, W., Woodward, R. B., Moscowitz, A., Klyne, W. and Djerassi, C. (1961) J. Am.
Chem. Soc., 83, 4013.
7. Hohn, E. G., and Weigang, O. E. (1968) J. Chem. Phys. 48,1127.
8. Rodger, A. and Rodger, P. M. (1988). J. Am. Chem. Soc., 110, 2361.
9. Lightner, D. A., Bouman, T. D., Wijekoon, W. M. D. and Hansen, A. (1985) J. Am. Chem.
Soc. 89, 5805.
10. Fidler, J., Rodger, P. M., and Rodger, A. (1994) J. Am. Chem. Soc. 116, 7266.
11. Schipper, P. E. and Rodger, A. (1986). Chem. Phys., 109, 173.
12. Fasman, G. D. (ed.) (1996). Circular dichroism and the conformational analysis of
biomolecules, p. 1. Plenum Press, New York.
13. Greenfield, N. and Fasman, G. D. (1969). Biochemistry, 8, 4108.
14. Greenfield, N. J. (1996). Anal. Biochem., 235, 1.
15. Johnson, W. C. Jr. (1999). Proteins: Structure, Function and Genetics, 35, 307.
16. Hennessey, J. P. Jr. and Johnson, W. C. Jr. (1981). Biochem., 20, 1085.
17. Compton, L A. and Johnson, W. C. Jr. (1986). Anal. Biochem., 155, 155.
18. Manavalan, P. and Johnson, W. C. Jr. (1987). Anal. Biochem., 167, 76.
19. Sreerama, N. and Woody, R. W. (1993). Anal. Biochem., 209, 32.
20. van Stokkum, I. H. M., Spoelder, H. J. W., Bloemendal, M., van Grondelle, R. and
Groen, F. C. A. (1990). Anal. Biochem., 191, 110.
21. Dalmas, B. and Bannister, W. H. (1995). Anal. Biochem., 225, 39.
22. Perczel, A, Park, K. and Fasman, G. D. (1992). Anal. Biochem., 203, 83.
23. Shiraki, K., Nishikawa, K. and Goto, Y. (1995). J. Mol. Bio!., 245,180.

138
 INTRODUCTION TO CIRCULAR DICHROISM

24. Aggell, A., Bell, M., Boden, N., Keen, J. N., Knowles, P. F., McLeish, T. C. B., Pitkeathly,
M. and Radford, S. E. (1997). Nature, 386, 259.
25. Bustamante, C., Tinoco, I., and Maestre, M. F. (1983). Proc. Natl. Acad. Sci. USA, 80,
3568.
26. Gordon, D. J. and Holzworth, G. (1971). Arch, Biochem. Biophys., 142, 481.
27. Mao, D. and Wallace, B. A. (1984). Biochemistry, 23, 2667.
28. Castiglione, E. and Rodger, A. Unpublished data
29. Rankin, S. E., Watts, A. and Pinheiro, T. J. (1998). Biochemistry, 37,12588.
30. Egli, M., Williams, L. D., Gao, Q. and Rich, A. (1991), Biochemistry, 30,11388.
31. Stryer, L. (1988). In Biochemistry (3rd edn), p. 1. W. H. Freeman and Company, New
York.
32. Calladine, C. R. and Drew, H. R. (1992). Understanding DNA: The molecule and how it
work, p. 1. Academic Press, Cambridge.
33. Beveridge, D. L. and Jorgensen, W. L. (1986). Ann. N.Y. Acad. Sci., 482.
34. Gessner, R. V., Fredrick, C. A., Quigley, G. J., Rich, A. and Wang A. H.-J. (1989). ]. Biol.
Chem., 264, 7921.
35. Zaloudek, F., Novros, J. S. and Clark, L. B. (1985). J. Am. Chem. Soc., 107, 7344.
36. Matsuoka, Y. and Norden, B. (1994). J. Phys. Chem., 86,1378.
37. Clark, L. B. (1994). J. Am. Chem. Soc., 116, 5265.
38. Fulscher, M. P. and Roos, B. O. (1995). J. Am. Chem. Soc., 117, 2089.
39. Holmen, A., Broo, A., Albinsson, B. and Norden, B. (1997)J. Am. Chem. Soc., 119,
12240.
40. Parkinson, A. (1998) PhD Thesis, University of Warwick.
41. Williams, A. L Jr., Cheong, C., Tinoco, I. Jr. and Clark L. B. (1986). Nucleic Adds Res.,
14, 6649.
42. Rizzo, V. and Schellman, J. A. (1984) Biopolymers, 23, 435
43. Lyng, R., Rodger, A. and Norden, B. (1992) Chem. Phys. Lett., 70, 17
44. Norden, B. and Tjerneld, F. (1976). Biophys. Chem., 4,191.
45. Dieber, H., Secco, F. and Venturini, M. (1987). Biophys. Chem., 26, 193.
46. Fronaeus, S. (1950). Acta Chem. Scand., 4, 72.
47. Rodger, A. (1993). In Methods in enzymology, Vol. 226, p. 232. Academic Press, London.
48. Scatchard, G. (1949). Ann. N. Y. Acad. Sci., 51, 660.

General CD references
• Barron, L. D. (1982). Molecular light scattering and optical activity. Cambridge University
Press, Cambridge.
• Craig, D. P. and Thirunamachandran, T. (1984). Molecular quantum electrodynamics: An
introduction to radiation-molecule interaction. Academic Press, London.
• Harada, N. and Nakanishi, K. (1983). Circular dichroic spectroscopy: exciton coupling in organic
stereochemistry. University Science Books, California.
• Mason, S. F. (1982). Molecular optical activity and the chiral discrimination. Cambridge
University Press, Cambridge.
• Michl, J. and Thulstrup, E. W. (1986). Spectroscopy with polarized light. VCH, New York.
• Nakanishi, K., Berova, N. and Woody, R. W. (ed.) (1994). Circular dichroism: Principles and
applications. VCH, New York.
• Richardson, F. S. (1979). Theory of optical activity in the ligand-field transitions of
chiral transition metal complexes. Chem. Rev., 79, 17.
• Rodger, A. and Norden, B. In Circular and linear dichroism (ed. R. G. Compton, S. G.
Davies, and J. Evans). Oxford University Press, Oxford.

139
This page intentionally left blank
 Chapter 5
Quantitative determination of
equilibrium binding isotherms
for multiple ligand-
macromolecule interactions
using spectroscopic methods
Wlodzimierz Bujalowski and Maria J. Jezewska
Dept. of Human Biological Chemistry & Genetics and The Sealy Centre for
Structural Biology, The University of Texas Medical Branch at Galveston,
Medical Research Building, 301 University Boulevard, Galveston,
TX 77555-1053, USA

1 Introduction
Thermodynamic studies provide information that is necessary in order to under-
stand the forces that drive the formation of ligand-macromolecule complexes.
Knowledge of the energetics of these interactions is also indispensable for
characterization of functionally important structural changes that occur within
the studied complexes. Quantitative examination of the equilibrium interactions
are designed to provide the answers to the questions: What is the stoichiometry
of the formed complexes? How strong or how specific are the interactions? Are
there any cooperative interactions among the binding sites and/or the bound
ligand molecules? Are the binding sites intrinsically heterogeneous? What are
the molecular forces involved in the formation of the studied complexes, or, in
other words, how do the equilibrium binding and kinetic parameters depend on
solution variables (temperature, pressure, pH, salt concentration, etc.)?
Equilibrium isotherms for the binding of a ligand to a macromolecule repre-
sent the relationship between the degree of ligand binding (moles of ligands
bound per mole of a macromolecule) and the free ligand concentration. A true
thermodynamic binding isotherm is model-independent and reflects only this relationship.
Only then, when such an isotherm is obtained, can one proceed to extract physic-
ally meaningful interaction parameters that characterize the free energies of
interaction. This is accomplished by comparing the experimental isotherms to
theoretical predictions based on specific binding models that incorporate known
molecular aspects, such as intrinsic binding constants, cooperativity parameters,
allosteric equilibrium constants, discrete character of the binding sites or over-
lap of potential binding sites, etc. (see below).

141
WLODZIMIERZ BUJALOWSKI AND MARIA J. JEZEWSKA

Any method used to quantitatively study ligand binding to a macromolecule

must relate the extent of the complex formation to the free ligand concentration
in solution. Numerous techniques have been developed to study equilibrium
properties of specific and non-specific ligand-macromolecule interactions in
which binding is directly monitored, including equilibrium dialysis, ultrafiltra-
tion, column chromatography, filter binding assay and gel electrophoresis (1-6).
These direct methods are very straightforward; however, they are usually time
consuming and some, like filter binding or gel shift assays, are non-equilibrium
techniques which require many controls before the reliable equilibrium binding
data can be obtained. Therefore, these direct methods are usually applied to
systems where the indirect spectroscopic approaches cannot be used, due to the
lack of suitable signal changes accompanying the formation of the complex.
Using indirect methods, the binding of the ligand is determined by
measuring the physico-chemical parameter of the macromolecule-ligand
mixture, most often a spectroscopic one, e.g. absorbance, circular dichroism or
fluorescence (7-32). The change in the physico-chemical parameter is then
correlated with the concentration of the free and bound species (11-12, 14-29).
The advantages of using spectroscopic measurements are that these can be
performed without perturbing the equilibrium and are relatively easy to apply.
In using a spectroscopic signal change, which accompanies the interactions,
to calculate a binding isotherm, it is often assumed that a linear relationship
exists between the fractional signal change and the fractional saturation of the
ligand or the macromolecule. Although this is true when one deals with equilib-
rium binding of only a single ligand molecule at saturation, in the general case,
however, in which multiple ligands bind to the macromolecule, the observed
fractional signal change and the extent of binding may not have a simple linear
relationship. For instance, this may occur if there are structurally or functional-
ly different sites on the macromolecule, each possessing a different spectroscop-
ic signal, and/or if there are cooperative interactions whose density changes with
the extent of the degree of binding and affects the spectroscopic properties of
the studied system (7, 25, 26, 29). The extent of any deviation from the ideal lin-
ear behaviour is usually unknown a priori, hence the degree of error introduced
into the isotherm and the resulting binding parameters is also unknown. In
other words, if a binding isotherm is obtained by indirect methods that involve
assumptions about the relationship between the observed spectroscopic signal
and the extent of ligand binding, then the isotherm and the interaction param-
eters that one obtains will be no more accurate than the assumptions. This may
cause particular problems if the isotherm is being used to differentiate between
alternative models for the interaction, since if one model does or does not 'fit'
the isotherm, this may be due either to the failure of the model or to the failure
of the assumptions on which the calculated isotherm is based. Therefore, it is
crucial to obtain a thermodynamically rigorous binding isotherm, independent
of any such assumptions (11, 12,14-29). Determination of a model-independent
isotherm constitutes the first step in a correct analysis of any ligand-macro-
molecule binding.

142
 MACROMOLECULE INTERACTIONS USING SPECTROSCOPIC METHODS

In this chapter, we discuss the use of spectroscopic approaches to study multi-

ple-ligand binding phenomena and their rigorous thermodynamic analyses to
obtain model-independent binding isotherms. In these approaches, one generally
monitors a spectroscopic signal (absorbance, fluorescence, circular dichroism,
NMR line width or chemical shift) from either the macromolecule or the ligand
that changes upon formation of a ligand-macromolecule complex; however, we
will limit our discussion to quantitative analysis as applied to the use of fluores-
cence. This spectroscopic technique is commonly available in a biochemical labo-
ratory, moreover, the derived relationships are general and applicable to any
physico-chemical signal used to monitor the ligand-macromolecule interactions.
The determination of the correct relationship between the spectroscopic
signal change and the degree of saturation of the ligand or the macromolecule
is a prerequisite for obtaining a thermodynamically rigorous binding isotherm.
The methods which we will discuss allow one to use a spectroscopic signal to
obtain a thermodynamically rigorous binding isotherm, even when direct pro-
portionality does not exist (14, 25, 26). The only constraint on these methods is
that they are not valid if the ligand or the macromolecule undergoes an
aggregation or self-assembly process within the experimental concentration
range used. Therefore, as in any case, knowledge of the self-assembly properties
of the ligand and the macromolecule under study are essential before a rigorous
analysis of a binding phenomenon can be undertaken.

2 Thermodynamic basis of quantitative

spectroscopic titrations
Two types of titrations can be performed in order to examine a binding iso-
therm in studies of ligand-macromolecule interactions. In one case, the macro-
molecule is titrated with a ligand and will be referred to as a 'normal' titration,
since the total average degree of binding Evj, (average number of moles of
ligand bound per total moles of macromolecule, LB/MT) increases as the titration
progresses (11, 14-28). Notice, that we use symbol Svf instead of just v to
describe the total average degree of binding, because, in the general case of a
multiple ligand binding system, the bound ligand molecules can be distributed
over T possible different bound states, all contributing to the total average
degree of binding 2vj.
In the second case, the ligand is titrated with the macromolecule. This type of
titration will be referred as a 'reverse' titration, since the binding density
decreases throughout the titration (24). Generally, the type of titration that is
performed will depend on whether or not the signal that is monitored is from
the macromolecule (normal) or the ligand (reverse). As we pointed out, the first
task in examining the ligand-macromolecule interactions is to convert a spec-
troscopic titration curve, i.e. a change in the monitored signal, as a function of
the titrant concentration, into a thermodynamically rigorous, model-independent
binding isotherm, which can then be analysed, using an appropriate binding
model to extract binding parameters.

143
WLODZIMIERZ BUJALOWSKI AND MARIA J. JEZEWSKA

The thermodynamic basis for these methods is that the total average degree
of binding, 2v(, of the ligand on a macromolecule, including all different dis-
tributions of ligands bound in possible different states 'f , is a sole function of the
free ligand concentration, LF at equilibrium (12, 33). Therefore, at a given free
ligand concentration the average value of Dv( will be the same, independent of
the concentration of the macromolecule, MT. The unique values of LF and 2vf and
the corresponding total ligand, LT, and total macromolecule, MT, concentrations
must satisfy the mass conservation equation

Therefore, if the set of concentrations (I/n, MTi) can be found for which tv{
and LF are constant, then 2v( and LF can be determined from the slope and
intercept, respectively, of a plot of LT versus MT, based on equation I.

2.1 The signal used to monitor ligand-macromolecule

interactions originates from the macromolecule
In this case, the signal monitors the progress of the saturation of a macro-
molecule and a 'normal' titration (addition of a ligand to a constant macro-
molecule concentration) is generally performed. Any physico-chemical intensive
property of the macromolecule (e.g. fluorescence intensity, fluorescence aniso-
tropy, absorbance, circular dichroism, viscosity, etc.) can be used to monitor the
binding, if this property is affected by the state of ligation of the macro-
molecule.
As mentioned above, for a total macromolecule concentration, MT, the equili-
brium distribution of the macromolecule among its different ligation states, M(,
is determined solely by the free ligand concentration, LF (12, 14, 33). Therefore,
at each LF, the observed spectroscopic signal, Sobs, is the algebraic sum of the
concentrations of the macromolecule in each state, M,, each weighted by the
value of the physico-chemical property of that state, Sf. In general, a macro-
molecule will have the ability to bind n ligands, hence the general equation for
the observed signal, Sobs, of a sample containing the ligand at a total con-
centration, LT, and the macromolecule at a total concentration, MT, is given by

where SF is the molar signal of the free macromolecule and Si is the molar signal
of the complex, Mf , which represents the macromolecule with i bound ligands
(i = 1 to n).
The mass conservation equation, which relates MF and Mj to MT, is given by

The partial degree of binding, vi ('i' moles of ligand bound per mole of macro-
molecule), corresponding to all complexes with a given number 'i' of bound
ligand molecules is given by

144
 MACROMOLECULE INTERACTIONS USING SPECTROSCOPIC METHODS

Therefore, the expression for Mi, the concentration of macromolecule with T

ligand molecules bound is

Introducing equations 4 and 5 into equation 3 provides a general relationship for

the monitored spectroscopic signal as

Subsequently, we can define the quantity ASobs as

and

Notice, ASobs = (Sobs — SpAl^/SpM? is the experimentally determined fractional

signal change observed at the total ligand and macromolecule concentrations,
LT and MT, and (ASj/i) = (S( - SF)/t is the average molar signal change per bound
ligand in the complex containing 'f ligands. The quantity ASobs = (Sobs — SpMr)/
SpM-r is referred to as the Macromolecular Binding Density Function (MBDF) (12,
14, 24-26).
Since ASj/i is an intrinsic molecular property of the system, equation 7
indicates that ASobs, is only a function of the binding density distribution, Sv(.
Therefore, the total average degree of saturation of the macromolecule and the
total average degree of binding of the ligand, 2vj, must be the same for any
value of LT and MT for which ASobs is constant. Thus, when you perform a
spectroscopic titration of a macromolecule with a ligand, at different macro-
molecule concentrations, MT, the same value of ASobs indicates the same physical
state of the macromolecule, i.e. the same degree of macromolecule saturation
with the ligand and the same Ev(. Since Evi is a unique function of the free
ligand concentration, then the value of LF at the same degree of saturation must
also be the same. The above derivation is rigorous and independent of any bind-
ing model and, as such, can be applied to any binding system, with or without
cooperative interactions and with overlapping of binding sites (14-28).

2.1.1 A practical example of quantitative analysis of spectroscopic

titration curves when the signal originates from the macromolecule
We will demonstrate the application of quantitative analysis to multiple ligand
binding to a macromolecule using the data from studies of the binding of the
ATP non-hydrolysable analogue, TNP-ATP, to the E. coll replicative helicase DnaB
protein (14). The DnaB helicase forms a stable hexamer built of six chemically
identical subunits (18, 21). The hexamer is specifically stabilized by the presence
of magnesium ions in solution (14, 18, 21). The intrinsic DnaB tryptophan fluor-

145
WLODZIMIERZ BUJALOWSKI AND MARIA J. JEZEWSKA

escence is substantially quenched upon binding TNP-ATP, hence this signal was
used to monitor the interaction. Therefore, in this case, the observed fractional
protein fluorescence quenching, iSnhs (MBDF) at a given T.NP-ATP concentration,
is given by (see equation 7)

where iSimjl = iS,/i is the average degree of protein fluorescence quenching per
bound TNP-ATP, when T nucleotide molecules arc bound per DnaR hcxnmer.

Protocol 1
Fluorescence titration of the DnaB helicase with TNP-ADP
Equipment and reagents
• Stock solution of the DnaB helicase (~5 x - TNP-ATP
10-5 M) * Fluorescence cuvette with an optical path
• Storage buffer (50 mM Tris/HCI,pH 8.1, of 0.5 cm
300 mM NaCl, 5 mM MgCLj, 25% glycerol) • Spectronuorimeter
• Binding buffer (50 mM Tris/HCl, pH 8.1,
100 mM NaCl, 5 mM MgCl2,10% glycerol)

Method
1 Diaiyse the stock solution of the DnaB helicase (~5 x 10-5 M (hexamer)) from the
storage buffer (50 mM Tris/HCi, pH 8.1, 300 mM NaCl, 5 mM MgCl2, 25% glycerol) to
the binding buffer (50 mM Tris/HCl, pH 8.1,100 mM NaCl, 5 mM MgCl2,10% glycerol),
The final concentration of the protein stock in the binding buffer will be
approximately 3 x 10"5 M (hexamer}. Measure the exact DnaB concentration using
the extinction coefficient at 280 nm, e280 = 185 000 M-1 cm -1 (hexamer)(21,22). After
dialysis, always store protein solutions on ice.
2 Prepare two stock solutions of TNP-ATP. One solution having the nucleotide con-
centration 1 x 10-4 M and the second solution having the concentration 1 x 10-3 M.
Using two (or more) solutions of the nucleotide allows you to add accurate volumes
of TNP-ATP to the protein sample and to perform titrations over a large range of
nucleotide concentrations.
3 Use a fluorescence cuvette with an optical path of 0.5 cm. This will allow you to use
a small volume (450 ul) of the protein solution and decrease the corrections for the
inner filter effect, due to the increasing absorption (at both excitation and emission
wavelengths) as a result of the increasing concentration of TNP-ATP as the titration
progresses (see below).
4 Set the excitation wavelength at 300 nm (the red edge of the tryptophan absorption
spectrum). Excitation at 300 nm will additionally decrease the inner filter effect. Set

146
 MACROMOLECULE INTERACTIONS USING SPECTROSCOPIC METHODS

Protocol 1 continued

the emission wavelength at 345 nm (the maximum of the protein emission

spectrum).
5 Set the spectrofluorimeter in the ratio mode. The signal coming from the emission
PMT is now divided by the signal from the reference PMT. This mode allows you to
eliminate any artefacts resulting from fluctuations, and/or a drift, of the lamp light
intensity during the titration. Add 450 ul of the binding buffer to the cuvette and
adjust the signal on the emission PMT to read approximately 0.01 in the ratio mode.
This means that the light intensity coming from the emission PMT is approximately
1% of the intensity read by the reference PMT.
6 Take an aliquot of 7.5 ul of the buffer from the cuvette and add a 7.5-ul aliquot of
the protein stock solution (3 x 10-5 M). Mix gently with a 200 ul automatic pipette.
The concentration of the protein in the cuvette is now ~5 x 10-7 M (hexamer). Wait
15 min for the temperature of the sample to equilibrate to the required tempera-
ture of the measurement, e.g. 20°C.
7 Titrate the protein solution with TNP-ATP using aliquots of 1 to 20 ul. Wait 2 min
between subsequent additions of nucleotide aliquots to allow the sample to reach
equilibrium. The final volume of the added nucleotide solution at the end of the
titration should not exceed 15% of the initial total volume of the protein sample.
8 In art independent titration experiment, determine the effect of the presence of
TNP-ATP on the fluorescence of the buffer alone for each titration point, using the
same set-up for the instrument as used in the titration with the protein. This will be
your background, B.
9 For each titration point, calculate the fractional fluorescence change using equation
7a. For each titration point'(' and the titration point where absorption of the sample
exceeds 0.01 at the excitation and emission wavelength correct the value of the
fluorescence intensity, Sf, for the dilution and inner filter effect using the formula,
S(CO, = (Si - Bi)(Vj/Vo)100'5Wu™ + Al""> where S^ is-the corrected value of the fluor-
escence intensity at a given point of titration i, Sf is the experimentally measured
fluorescence intensity, B, is the background, Vf is the volume of the sample at a
given titration point, V0 is the initial volume of the sample, I is the total length of
the optical path in the cuvette expressed in cm. A^ and A,-Xem are the absorbances
of the sample at excitation and emission wavelengths, respectively (14, 15, 33, 34).
This formula provides an accurate correction for the inner filter effect up to the
absorption value of ~0.3 at both excitation and emission wavelengths.
10 Plot the fractional fluorescence quenching ASobs (macromolecular binding density
function) as a function of the logarithm of the total TNP-ATP concentration.

2.1.2 Quantitative analysis of the fluorescence titrations

Fluorescence titrations of the DnaB protein (macromoleculc) with TNP-ATP
(ligand), at two different DnaB hexamer concent rations, in which the values of
ASobb have been plotted as a function of the total TNP-ATP concentration (LT) are
shown in Figure 1. Careful inspection of the plots already indicates that the bind-

147
WLODZIMIERZ BUJALOWSKI AND MARIA J. JEZEWSKA

Figure 1 Fluorescence titration of the DnaB hexamer with TNP-ATP monitored by the
fractional quenching of the protein fluorescence (macromolecular binding density function,
MBDF) in 50 mw Tris/HCI (pH 8.1, 10 °C) containing 20 ITIM NaCI, 5 mM MgCI2, and 10%
glycerol at two different concentrations of the DnaB protein (•) M1 = 5.38 x 10-7 M
(hexamer); (o) M2 = 3.17 X 10-6 M (hexamer). The solid lines are interpolations between
data points added to separate the two data sets. The dashed horizontal line connects the
points, at the same value of the MBDF, for each protein concentration. These points
determine the set of total TNP-ATP concentrations, (L1. and L2), for which the average binding
density, 2v, and the free protein concentration, Lf, are constant and independent of the total
protein concentrations, M,. Reprinted with permission from Bujalowski, W. and Klonowska,
M. M. (1993). Biochemistry, 32, 5888-5900. Copyright (1993) American Chemical Society.

ing process is complex, i.e. these spectroscopic isotherms indicate the existence
of two binding phases (see below). Also, depicted in Figure 1 is the approach by
which a single set of values of 2vf and LF can be obtained from these data. A
horizontal line which intersects both curves and which defines one constant
value of ASobs is drawn. The point of intersection of this horizontal line with
each titration curve defines the set of values of LTi, MTi for which LF and Evf are
constant, as discussed above. The example in Figure 1 only has two titrations,
hence only two sets of values of LTi, MTjare obtained (i = 1 and 2).
For the two titrations at macromolecular (DnaB hexamer) concentrations,
MT1 and M^ (MT2 > MT1), two mass conservation equations can be written in the
form of equation 1 for each set of LTi, MTj. These two equations can then be solved
for Xvj and LF, with the results given by

In this manner, model-independent values of IF and 2vfcan be obtained at any

value of ASobs, yielding a set of values for 1F and 2v,.
From a practical point of view, the accuracy of the determination will depend

148
 MACROMOLECULE INTERACTIONS USING SPECTROSCOPIC METHODS

upon the used concentrations of the macromolecule and the ligand. Thus, the
most accurate estimates are obtained in the region of the titration curves where
the concentration of a bound ligand is comparable to its total concentration, IT-
In other words, the concentration of the bound ligand must constitute a
significant fraction of the total ligand concentration in solution. In practice, this
limits the accurate determination of the degree of binding, 2v;, to the region of
the titration curves where the concentration of the bound ligand is at least
-10% of the IT. Therefore, a selection of proper concentrations of the macro-
molecule is crucial for obtaining LF and Svf over the largest possible region of
the titration curves, although the accuracy of the determination of 2v< is mostly
affected in the region of the high concentrations of the ligand approaching the
maximum saturation. Such a selection of macromolecule concentrations is
usually based on preliminary titrations which provide initial estimates of the
expected affinity between the ligand and the macromolecule.
In the first step of the analysis of a multiple ligand binding system, the
obtained values of 2vj, corresponding to given values of ASobs, are used to:
(1) determine the maximum stoichiometry of the macromolecule-ligand com-
plex, and
(2) determine the relationship between the signal change, ASobs, and the
average number of bound ligand molecules, i.e. in the considered case, the
average number of bound TNP-ATP molecules per the DnaB hexamer.
Both objectives can be achieved by plotting ASobs as a function of Svf. The de-
pendence of the quenching of the DnaB fluorescence upon the rigorously
determined number of TNP-ATP molecules bound per hexamer is presented in
figure 2. The selected concentrations of the DnaB helicase allowed us to obtain
an accurate (± 5 %) estimate of the average degree of binding up to 5.2 ± 0.3
TNP-ATP molecules bound. Although the maximum value of 2vf cannot be
directly determined, due to the discussed inaccuracy at the high ligand con-
centration region, the plateau of the fluorescence quenching ASmax, corres-
ponding to the maximum saturation can be determined with the accuracy of
±5% (see Figure 1). Therefore, knowing the maximum extent of the protein
fluorescence (ASmax = 0.78) one can perform a short extrapolation of the plot
(ASobs versus 2vf) to this maximum value. Such extrapolation of the data
presented in Figure 2 shows that, at saturation, the DnaB protein hexamer binds
six molecules of TNP-ATP, thus, establishing the maximum stoichiometry of the
DnaB-nucleotide complexes (14).
Notice, the plot of ASobs, as a function of 2vj, shown in Figure 2 is strongly
non-linear. The largest protein fluorescence quenching, up to —0.21 ± 0.03,
occurs upon binding the first TNP-ATP molecule. Average binding of the first
three molecules causes quenching of -0.55 ± 0.03, which corresponds to the
first step in the binding isotherm (see Figure 1). Binding of the next three
nucleotides, in the lower affinity phase, gives the maximum saturation of six
TNP-ATP molecules per DnaB hexamer and is accompanied by an additional
quenching of only -0.21 ± 0.05. The dashed line represents a hypothetical

149
WLODZIMIERZ BUJALOWSKI AND MARIA J. JEZEWSKA

Figure 2 The dependence of the fractional quenching of the DnaB protein fluorescence upon
the total average number of TNP-ATP molecules bound per DnaB hexamer in 50 mM Tris/HCI
(pH 8.1, 10°C) containing 20 mw NaCI, 5 mM MgCI2, and 10% glycerol (•). The average
number of nucleotides bound, at a particular value of the fluorescence quenching, has been
determined from the two titrations shown in Figure 1, using equations 9a and 9b. The
dashed line represents the theoretical situation where there is strict proportionality between
the degree of nucleotide binding and the quenching of the DnaB protein fluorescence. The
solid line is generated using the values of parameters a, 6 and c (see text) obtained from
the non-linear least-square fit, based on the third degree polynomial described by equation
15, with intrinsic binding constant K = 5.9 x 106 W1 and cooperativity parameter, o- =
0.55. Reprinted with permission from Bujalowski, W. and Klonowska, M. M. (1993).
Biochemistry, 32, 5888-5900. Copyright (1993) American Chemical Society.

situation when strict proportionality between the degree of nucleotide binding

and the quenching of the DnaB protein fluorescence exists. It is clear that very
pronounced non-linearity exists between the extent of the protein fluorescence
quenching and the degree of nucleotide binding, with the first three nucleo-
tides binding with higher affinity and causing 70% of the total change of the
protein fluorescence. Therefore, an incorrect isotherm would have been
calculated if a linear relationship had been assumed for this system (14).
In principle, as shown in Figures 1 and 2, only two titration curves are needed
to obtain values of 2vf and IF that cover the large range of the degree of
binding. This is possible for the data in Figure 1, since the signal change
(tryptophan fluorescence quenching) for the DnaB-TNP-ATP interaction is large
and the affinity under the conditions used is fairly high. In this case, the data
can be analysed using equations 9a and 9b. However, if necessary, more accurate
values of Svf and IF can be. obtained if more than two titrations are performed
and the data are graphically analysed. For a case in which 'n' titrations are
performed, then 'n' sets of values of LTi, MTi (t = 1 to n) will be obtained, one for
each titration curve that is intersected by each horizontal line. A plot of IT
versus MT can then be constructed, which will result in a straight line, according
to equation 1, from which the values of LF and 2vj can be obtained from the

150
 MACROMOLECULE INTERACTIONS USING SPECTROSCOPIC METHODS

intercept and the slope. If a plot of IT versus MT> determined in this manner, is
not linear, this may indicate one or more inconsistent sets of titration data or
aggregation phenomena associated with the ligand or the macromolecule, and
these data should be viewed with caution.

2.1.3 Application of a binding model to analyse the binding

isotherm
The analysis performed so far is completely model-independent, i.e. no specific
mechanism of the nucleotide binding to the DnaB hexamer has been assumed.
Once this analysis has been performed, one can postulate, on the bases of the
physical properties of the system, a specific binding model which describes the
binding isotherms and allows the experimenter to obtain intrinsic affinities and
parameters characterizing the cooperative interactions (if any) within the
system.
Briefly, the data show that binding of nucleotides to the DnaB hexamer
occurs in two phases. Three nucleotides bind in the high affinity phase and the
next three nucleotides bind in the low affinity step. The simplest explanation
for this behaviour is that the DnaB hexamer, built of six chemically identical
subunits, exhibits negative cooperativity between binding sites (14, 22). The
reader should consult the original papers for a full discussion of this aspect of
the nucleotide binding studies to the DnaB hexamer (14, 15, 22).
Binding of a ligand to a protein which forms a homohexamer in two steps,
each including the same number of bound nucleotides (three), can be described
in the simplest way by a hexagon model (14-16, 21). In this model, the ligand
can bind to any of six initially equivalent binding sites with the same intrinsic
binding constant K. However, the cooperative interactions, characterized by a
parameter CT, are limited to only two neighbouring sites. If we assign the
quantity x = KLF, where LF is the free ligand concentration, then the partition
function for the hexagon model, Z, is described by (14, 16, 22)
Z = 1 + 6x + 3(3 + 2a)x2 + 2(1 + 6o + 3cr2)x3 + 3(3<r2 + 2cr3)pc4 + 6crV + o"6x6 10
The total average degree of binding 2vj is defined by the standard statistical
thermodynamic formula (14, 33)

dlnIF
which provides the expression for Svf as
2i>i= [6x + 6(3 + 2<y)x2 + 6(1 + 6cr + 3o-2)x3 + 12(3cr2 + 2cr3)x4 + 30a4x5 + 6cr6x6]/Z
12
Because the rigorous thermodynamic approach described above allowed us
to obtain both 2vf and IF, we can use equation 12 to directly obtain K and a, from
the thermodynamic isotherm (Sv, as a function of LF), and to perform the optim-
ization of the binding parameters, K andCT,using a non-linear least-square fit.
Figure 3(a) shows the dependence of 2v; upon the logarithm of [TNP-ATP]Free.

151
WLODZIMIERZ BUJALOWSKI AND MARIA J. JEZEWSKA

Figure 3. (a) The total average number of nucleotide molecules bound per DnaB hexamer,
2v,, as a function of the free concentration of TNP-ATP in 50 mM Tris/HCI (pH 8.1, 10 °C)
containing 20 mM NaCI, 5 mM MgCI2 and 10% glycerol (•). The solid line is a non-linear
least-square fit, according to the hexagon model (equations 10-12) which provides the
intrinsic binding constant K = 5.9 x 105 M'1 and cooperativity parameter,CT= 0.55 (data
from 14). (b) Ruorescence titrations of the DnaB hexamer with TNP-ATP monitored by the
fractional quenching of the protein fluorescence (macromolecular binding density function,
MBDF) in 50 mM Tris/HCI (pH 8.1, 10°C) containing 20 mM NaCI, 5 mM MgCI2 and 10%
glycerol at two different concentrations of the DnaB protein: (•) 5.38 x 10~7 M; (o) 3.17 x
10-6 M (hexamer). The solid lines are computer fits of the experimental binding isotherms,
according to the hexagon model, obtained by the empirical function approach. First, the
fractional protein fluorescence change (MBDF) has been calculated using equation 15, and
using the values of a = 0.286, b = - 0.0484 and c = 3.609 x 10-3. Subsequently, the
fluorescence titration curves were subjected to the non-linear fit which provided the set of
interaction parameters, intrinsic binding constant, K = 5.9 x 105 M-1, and cooperativity
parameter, o= 0.55. Reprinted with permission from Bujalowski, W. and Klonowska, M. M.
(1993). Biochemistry, 32, 5888-5900. Copyright (1993) American Chemical Society.

Notice, the isotherm is more scattered than the original titration curve, and the
isotherm covers —75% of the total binding curve due to the inaccuracy of deter-
mining the degree of binding in the high concentration range of the nucleotide.
The solid line is a non-linear fit of the binding isotherm which provides 1C = 5.9
X 105 M-1 and o = 0.55.
In practice, higher accuracy in estimating the binding parameters could be
obtained if one analyses the entire original fluorescence titration curves (Figure 1),

152
 MACROMOLECULE INTERACTIONS USING SPECTROSCOPIC METHODS

without being limited to the region where the degree of binding can be
accurately estimated. A general method which allows such an analysis of the
entire fluorescence titration curves, for this very complex binding system, is
discussed in the next section.

2.1.4 Empirical function approach

In order to analyse the original fluorescence titration curve, all ASimax molecular
parameters in equation 8b must be known. In the simplest case, all ASimax para-
meters could be the same, i.e. binding to each site would be characterized by
the same quenching parameter, ASmax. In such a case, the dependence of ASobs,
as a function of 2vi, would be strictly linear (dashed line in Figure 2) and equation
8b would reduce to
ASobs = ASmax(2Vj) 13
If ASmax is known, equation 13 can be directly used to fit the fluorescence
titration curves (see Figure 1) and to extract K and <o. As we discussed above,
obviously, this situation does not apply in this case. Sometimes, it is also possible
to determine all optical parameters in the analytical formula relating ASobs to the
molecular parameters ASjmax, as defined by equation 8b (35, 36). However, in the
general case, this is not necessary, and in the case of the DnaB protein, which
has six nucleotide binding sites and possibly multiple ASjmax for a given number
of bound nucleotide molecules, it is practically impossible. For instance, in the
case of the hexagon model for the DnaB-nucleotide system, the protein, with a
given number of nucleotide molecules bound, can exist in multiple configura-
tions, e.g. the hexamer with three ligands bound has 20 possible configurations
(see equation 10). Although some of these configurations are physically indis-
tinguishable, resulting from simple statistical effects of binding three molecules
to the initially independent six sites, there are configurations which differ by the
number of cooperative interactions, and they may also differ by the values of the
individual molecular quenching constants ASimax. As an example, equation 14 de-
scribes the minimum analytical relationship between the observed experi-
mental quenching, ASobs, the individual molar quenching parameters, ASimax,
and the interaction parameters, K and a for the hexagon model (14)
ASobs = (SAS1X + 6(3AS2l + 20AS22)x2 + 6(AS3l + 60AS32 + 302AS33)x3 +
12(30t2AS4l + 203AS42)x4 + 30a4AS5x5 + 606AS6x6)/ZH 14
For clarity, subscript 'max' has been omitted at individual molar quenching
constants ASjmax.
Equation 14 contains 12 independent parameters, two interactions and 10
optical parameters. For instance, there are three physically distinguishable
configurations of the hexamer with three ligands bound which have different
densities of the cooperative interactions; therefore, there are three possible dif-
ferent molar quenching constants, AS3l, AS32 and AS33, characterizing each of
these configurations. As we mentioned above, in order to apply equation 14 to
obtain interaction parameters, K and a from a single fluorescence titration

153
WLODZIMIERZ BUJALOWSKI AND MARIA J. JEZEWSKA

curve, all ten optical constants, ASj, must be known. It should be pointed out
that, at the saturating concentration of the ligand, the observed experimental
maximum quenching, is ASobs = AQ6.
The problem of finding all optical parameters can be avoided by using the
following empirical function approach. This approach can be used for any
ligand-macromolecule systems, where the determination of all optical constants
in an analytical equation is practically impossible. An empirical function, usually
polynomial, is found, which relates the experimentally determined dependence
of the average quenching, ASobs upon 2vj, as shown in Figure 2. This can be
achieved by using a non-linear least-square fit of ASobs as a function of 2vf. In the
case of the DnaB-nucleotide interactions, a minimum third degree polynomial
is necessary to describe this function as defined by (14)
ASobs = a(2Vi) + b(Evi,)2 + c(ESvi)3 15
where a, b and c are fitting constants. This function is then used to generate a
theoretical isotherm for a binding model and to extract the true binding
parameters from the experimentally obtained single fluorescence isotherm. To
generate the theoretical isotherm, the value of the degree of binding, Evi, is
calculated using equation 12 for a given free nucleotide concentration and initial
estimates of the intrinsic binding constant, K, and cooperativity parameter, a.
The obtained Evi is then introduced into equation 15 and the fluorescence
quenching, corresponding to this value of the degree of binding, is calculated.
These calculations are performed for the entire fluorescence titration curve.
Using equation 15, we performed the fit of the empirical plot of ASobs as a
function of Evi shown in Figure 2 (solid line) which provided parameters a =
0.286, b = - 0.0484 and c = 3.609 X 10~3. These parameters define the empirical
dependence of ASobs upon Evi. Subsequently, equation 15 with the obtained a, b
and c, combined with equation 12, which defines the selected model of the
nucleotide binding to the DnaB hexamer, is used in the non-linear fit of the
original fluorescence titration curves with only two fitted parameters, K and a.
The solid lines in Figure 3(b) are the non-linear least-square fits of the entire
fluorescence titration curves, according to the procedure described above.

2.2 Signal used to monitor the interactions originates from

the ligand
In this case, some spectroscopic property (e.g. ligand fluorescence intensity), S,
of the ligand changes upon binding to the macromolecule, hence the signal
monitors the apparent degree of saturation of the ligand and a 'reverse'
titration (addition of a macromolecule to a constant ligand concentration) is
generally performed (12, 24).
Consider the general case of equilibrium, multiple-ligand binding (total con-
centration, LT) to a macromolecule (total concentration, MT) where there can be
'i' states of the bound ligand, with each state possessing a different molar
signal, Si. The observed signal, Sobs, from the ligand solution, in the presence of
the macromolecule, has contributions from the free ligand and the ligand

154
 MACROMOLECULE INTERACTIONS USING SPECTROSCOPIC METHODS

bound to the macromolecule in any of its 'i' possible bound states and can be
expressed by
ASobs = SFIF + ESi,Li 16
where SF and IF are the molar signal and concentration of the free ligand,
respectively, and Si and Li are the molar signal and concentration of the ligand
bound in state '?, respectively.
Equation 16 is valid when the molar signal of each species is independent of
concentration (i.e. in the absence of ligand and macromolecule aggregation).
The concentrations of the free and bound ligand are related to the total ligand
concentration by the conservation of mass equation
IT = IF + ELi 17a
where
Ii = viMT 17b
where vi is the partial degree of ligand binding for the ith state. Substituting
equations 17a and 17b into equation 16 and rearranging provides
ASobs - SFIT = MT E(S, - SF)vi 18
Notice, SFlT is the initial signal from the ligand before titration with the macro-
molecule. Dividing both sides by SFLT and next multiplying by (IT/MT) yields
(Sobs - SFIT)1/Ir (Si ~ Sp)
=r =2 v, 19
SFIr \\MT
which can be rewritten as
(Sobs "" SFIr) —I
IT I = Vi(AS)jVj
/AO\ ™
20a

and
ASobs = 2(AS),v, 20b

It should be pointed out that ASobs = (Sobs - SFLT)/SFLT is the experimentally

observed fractional change in the signal from the ligand (with respect to the
signal of the total ligand free, SplT) in the presence of the ligand and the
macromolecule at total concentrations, LT and MT, and (AS), = (S; - SF)/SF is the
molar signal change from the ligand when it is bound in state 'i'. If the
fluorescence intensity of the ligand is being monitored, then ASobs corresponds
to the fractional change of the ligand fluorescence. Equation 20 is general and
independent of the spectroscopic method used to monitor the interactions.
Equation 20b indicates that the quantity ASobs(LT/MT) is equal to E(AS)(Vj, the
sum of the partial degrees of binding for all 'i' states of the ligand-
macromolecule system, weighted by the intrinsic signal change for each bound
state. The weighting factor, (AS)(, is a molecular quantity which is constant for a
particular binding state 'i', under a given set of experimental conditions
(temperature, buffer etc.). Therefore, the quantity, E(AS)iVi, is constant for a
given distribution of the degree of binding among different possible states, Ev(.

155
WLODZIMIERZ BUJALOWSKI AND MARIA J. JEZEWSKA

At equilibrium, the values of L,F and Ev1, arc constant for a given value of
ASobs(L,/MT), independent of the macromolecule concentration, Mr. Hence,
under identical solution conditions, one can obtain thermodynamically rigorous
measurements of Evi and 1F from plots of ASohs(LT/MT) versus MT for two or more
titrations performed at different total ligand concentrations. This is accom-
plished by obtaining the set of concentrations (L r; . MTl), from each titration, for
which the quantity ASobs(r,T/MT) is constant, and solving for Ev; and l.F. The
procedure is therefore analogous to the case in which a signal from the
macromolecule is monitored during the titration. The quantity, AS(lhs(lT/MT), is
referred to as the Ligand Binding Density Function (LBDF) (11, 24).

2,2.1 A practical example of quantitative analysis of spectroscopic

titration curves when the signal used to monitor the interactions
originates from the ligand
As an example of the analysis when the signal from the ligand is monitored, we
discuss a study of the interaction of a fluorescent nucleotide analogue, e;ADP
(ligand), with the E coli DnaR protein (macromolecule), where, instead of the
protein fluorescence quenching, the fluorescence of the analogue has been used
to monitor the binding (24). The analogue has its adenine modified with
chloroacety [aldehyde to provide a fluorescent e the no-derivative (37-39). As we
discussed above, the DnaB hexamcr binds six nucleotide molecules and the
binding is characterized by the negative cooperativity. The fluorescence of the
eADP is increased by --21% upon binding to the DnaB helicase (24). To increase
this signal change and, in turn, to obtain a higher resolution of the titration
experiments, acrylamide has been added to the solution. Acrylamide is a very
efficient dynamic quencher of the etheno-derivative of adcnosine in solution
(15, 24. 39). This extra dynamic quenching process (40, 41), which does not
affect the thermodynamics of the nucleotide-enzyme interactions, is much less
efficient for the nucleotide bound to the DnaB protein than for the free nucleo-
tide, leading to a much larger change in the nucleotide fluorescence upon
formation of the complex with the protein, thus, increasing the resolution of
the titration curves. The application of the differential dynamic quenching of
the ligand or the macromolecule fluorescence to increase the resolution of the
binding experiments is thoroughly discussed in (24).

Protocol 2
Fluorescence titration of eADP with the DnaB helicase
Equipment and reagents
• Stock solution of the DnaB helicase (-7 x • Binding buffer (50 mM Tris/HCl, pHS.l,
10- 5 M(hexarner)) 1OOmw NaCI, 5 mM MgCl2, 100 mm acry-
• Storage buffer (50 mM Tris/HCl, pH 8.1, lamide, 10% glycerol)
300 mm NaCI, 5 mM MgCl2, 25% glycerol • eADP
• Fluorescence cuvette with an optical path • Spectrofiuorimeter
of 0.5 cm

156
 MACROMOLECULE INTERACTIONS USING SPECTROSCOPIC METHODS

Protocol 2 continued

Method
1 Dialyse the stock solution of the DnaB helicase (~7 x 10'5 M (hexamer)) from the
storage buffer (50 mM Tris/HCl, pH 8.1, 300 mM NaCl, 5 mM MgCl2, 25% glycerol) to
the binding buffer (50 mM Tris/HCl, pH 8.1, 100 mM NaCl, 5 mM MgCl2, 100 mM
acrylamide, 10% glycerol). The final concentration of the protein stock in the
binding buffer will be approximately 5 x 10~5 M (hexamer). Measure the exact DnaB
concentration using the extinction coefficient at 280 nm, em, - 185 000 M-1 cm"1
(hexamer} (21), After dialysis, always store protein solutions on ice,
2 Prepare two stock solutions of the protein. One solution having the DnaB con-
centration 1 x 10~6 M and the second solution having the concentration ~5 x 10"5 M.
Using two solutions of the helicase allows you to add accurate volumes of the
protein to the nucleotide sample and to perform titrations over a large range of the
protein concentration,
3 Prepare stock solutions of eADP, from 3 x 10"5 to 3 x 10"4 M. Determine the nucleo-
tide concentration using the absorption coefficient at 294 nm, t-294 = 2900 M"1 cm"1
(37, 38). Using different stock solutions will allow you to change the concentration
of the eADP in the sample by adding the same volume of different nucleotide stock
solutions.
4 Use a fluorescence cuvette with an optical path of 0.5 cm. This will allow you to use
a small volume (450 ^1) of the fluorescent nucleotide solution.
5 Set the excitation wavelength at 325 nm (the red edge of the eADP absorption
spectrum). Excitation at 325 nm. far from the protein absorption maximum, will
eliminate, to a great extent, the correction for the background of the protein
fluorescence. Set the emission wavelength at 410 nm (the maximum of the eADP
emission spectrum).
6 Set the spectrofluorimeter in the ratio mode. The signal coming from the emission
PMT is now divided by the signal from the reference PMT. This mode allows you to
eliminate the artefacts resulting from any fluctuations of the lamp and the possible,
although slow, drift of the lamp intensity during the titration. Add 450 p,l of the
binding buffer to the cuvette and adjust the signal on the emission PMT to read
approximately 0.01 in the ratio mode. This means that the intensity measured by
the emission PMT is approximately 1% of the intensity read by the reference PMT,
7 Set the polarizer in the excitation channel at the vertical position and the polarizer
in the emission channel at 54.7° (magic angle). Performing titrations using 'magic
angle' conditions will eliminate the artefacts in the fluorescence intensity measure-
ments, due to the possible changes in the fluorescence anisotropy of the sample.
These artefacts, which commonly occur, result from the different sensitivities of the
emission PMT for both the vertically and horizontally polarized light. Although this
may not be a problem when the fluorescence intensity of a large DnaB hexamer is
studied, it could induce artefacts when the emission of the relatively small t-ADP
(affected by the binding to the large DnaB hexamer) is examined, as in this case
(32. 42).

157
WLODZIMIERZ BUJALOWSKI AND MARIA J. JEZEWSKA

Protocof 2 continued

8 Take an aliquot of 20 n-1 of the buffer from the cuvette and add a 20-u.1 aliquot of the
eADP stock solution (e.g. 1 x 10~4 M). Mix gently with a 200 u1 automatic
micropipette. The concentration of eADP in the cuvette is now ~4.5 x 10"6 M. Wait
15 min for the temperature to equilibrate to the temperature of the measurement,
e.g.20°C
9 Titrate the EADP solution with the DnaB protein using aliquots of 1 to 20 (ul. The
final volume of the added protein solution at the end of the titration should not
exceed 15% of the total volume of the nucleotide sample.
10 In an independent titration, titrate the buffer with the DnaB protein solution and
determine the contribution of the protein fluorescence to the observed fluorescence
of the eADP sample for each titration point. At high protein concentrations such a
contribution will become significant, in spite of the fact that the excitation is set at
325 nm (far from the protein absorption spectrum), due to the impurities in the
protein sample and the discrete band pass of the excitation and emission
monochromator. This is your background. B.
11 For each titration point calculate the fractional fluorescence increase, ASobs, using
the formula ASobs = (Sobs - SulT}/SEJ,T. For each titration point 'i' correct the value of
the fluorescence intensity, Si, for the dilution and background using the formula,
Sicor = (si - Bi)(Vi /Vo) where Sicor is the corrected value of the fluorescence intensity at
a given point of titration 'i', Si is the experimentally measured fluorescence
intensity, B( is the background, V, is the volume of the sample at a given titration
point, V0 is the initial volume of the sample.
12 Plot the fractional ligand (eADP) fluorescence increase, ASob!i, as a function of the
logarithm of the total DnaB protein concentration.

2.2.2 Quantitative analysis of the fluorescence titrations

A series of fluorescence titration curves of eADP with the DnaB protein, in 50 mM
Tris/HCl (pH 8.1, 10°C) containing 5 mm MgCl2, 100 mm NaCl and 10% glycerol,
at different nucleotide concentrations and in the presence of 100 mm acryla-
mide, is shown in Figure 4. At higher nucleotide concentrations, the curves are
shifted toward higher concentrations of the DnaB helicase as more enzyme is
necessary to saturate the increased amount of eADP in the sample. All curves
reach the same plateau of the relative fluorescence increase, AF, at saturating
concentrations of the DnaB helicase. As we pointed out above, in general, the
fractional change of the ligand fluorescence upon the macromolecule concen-
tration does not necessarily strictly correspond to the fractional ligand saturation.
This is never a priori known for any multiple ligand binding system. However,
the estimate of the degree of binding and the free ligand concentrations can be
obtained by using the LBDF; approach.
Figure 5 shows the plot of ASobs,(LT/MT) as a function of the DnaB protein
concentration, obtained using the fluorescence titrations presented in figure 4.
For the different total concentrations of sADP, at the same value of the binding
 MACROMOLECULE INTERACTIONS USING SPECTROSCOPIC METHODS

£o

-7 -6 -5
Log [DnaB (hexamer)]Tola|
Figure 4 Fluorescence titrations ('reverse titrations') of eADP, at different concentrations
of the nucleotide, with the DnaB helicase in 50 mm Tris/HCI (pH 8.1, 10 °C) containing
100 mM NaCI, 5 mm MgCI2 and 100 mm acrylamide (\ex = 325 nm, \em = 410 nm).
The concentrations of the nucleotide are: (•) 4 x 10"6 M; (*) 6 x 10"6 M; (o) 1 x 10~5 M;
(O) 2 x 10~5 M; (D) 3 x 10"5; (•) 4 x 10~5 M. The solid lines are computer fits of the
experimental binding isotherms, according to the hexagon model, using a single set of
binding parameters, K = 4 x 10s M"1, o- = 0.4 and ASmax = 2.35. Reprinted from Jezewska,
M. J. and Bujalowski, W. (1997) 'Quantitative analysis of ligand-macromolecule interactions
using differential dynamic quenching of the ligand fluorescence to monitor the binding.'
Biophysical Chemistry, 64, 253-269. Copyright (1997) with permission from Elsevier Science.

density function, the total degree of binding, Evi, and the free eADP concen-
trations, [eADP]Free' must be the same, thus, allowing for their determination
(see equation 20). As mentioned above, even though all of the 'reverse' titrations
shown in Figure 4 span the same full range of eADP fluorescence increase, they
do not span the same range of the degree of binding as seen from the LBDF plot
in Figure 5. As a result, multiple titrations at different values of LT are required to
span the full binding density range. A horizontal line, which intersects the LBDF
curves is drawn, defining a constant value of the LBDF, ASobs(LT/MT) (dashed line
in figure 5). The points of intersection of the horizontal line with each binding
density function curve determine the set of values (MTi, LTi) for which IF and Evi
are constant, as shown in Figure 5 for one constant value of the LBDF. Based on
equation 1, the average degree of binding, Evi, and Lf, can then be determined
from the slope and intercept of a plot of LT versus MT at each constant value of
ASobs(Lr/Mp). By repeating this procedure for a series of horizontal lines that span
the range of values of ASobs(lr/Mt) as a function of IF, the values of Evi can be
obtained and a binding isotherm can be constructed (11, 24).
In the construction of a series of LBDF plots, as shown in Figure 5, one should
cover as wide a range of ligand concentrations as possible; however, care should
be taken to avoid large changes in the total ligand concentration between two
successive titrations, since this may bias the determination of IF and Evi. In our
experience, six to eight titrations, using successive total ligand concentrations
that differ by a factor of 1.5-2, will generate an accurate set of data.

159
WLODZIMIERZ BUJALOWSKI AND MARIA J. JEZEWSKA

12

Log[DnaB (hexamer)]Tota|

Figure 5 Dependence of the ligand binding density function (LBDF), ASobs(LT/M,T), on the
logarithm of the total DnaB protein concentration, at different total concentrations of the
EADP (LT) in 50 mm Tris/HCI (pH 8.1, 10 °C) containing 100 mM NaCI, 5 mm MgCI2 and 100
mM acrylamide: (•) L1 = 4 x 10-6 M; (o) L2 = 6 x 10-6 M; (•) L3 = 1 x 10~5 M; (n) LA = 2
x ID'5 M; (A) Lg = 3 x 10~5 M; (*) Lg = 4 X 10"5 M. The solid lines are interpolations
between the data points which separate different data sets and have no theoretical basis.
The horizontal dashed line connects points at the same value of the LBDF, at different total
EADP concentrations (L 1 , L2, L3, L4, L5 and L6) and total DnaB protein concentrations (M1
M2, M3, M4, M5 and M6), at which [eADP]Free and the total degree of nucleotide binding, Ev,,
on the DnaB hexamer are the same. Reprinted from, Jezewska, M. J. and Bujalowski, W.
(1997) 'Quantitative analysis of ligand-macromolecule interactions using differential dynamic
quenching of the ligand fluorescence to monitor the binding.' Biophysical Chemistry, 64,
253-269. Copyright (1997) with permission from Elsevier Science.

The model-independent binding isotherm constructed from the full analysis

of the data in Figures 4 and 5, for the binding of eADP to the E. coli DnaB protein,
is plotted in Figure 6. The solid line is a theoretical isotherm, according to the
hexagon model, constructed using the intrinsic binding constant K = 4 x 105
M-1 and the cooperativity parameterCT= 0.4 (equation 12). These values of K and fl-
are, within experimental accuracy, the same as the values which have been
independently obtained using the rigorous fluorescence titration method in
which the quenching of the protein fluorescence has been used to monitor the
binding (14, 15, 22).

2.2.3 Correlation between the fractional signal change, ASobs, and

the average degree of binding, Evi
Measurements of the average degree of binding, Evi, as a function of the free
ligand concentration, enable one to determine the relationship between the
average signal change and the fraction of the bound ligand. This is not
necessary in order to obtain a binding isotherm, as discussed above (Figure 6);
however, if ASobs is found to be directly proportional to the fraction of the

160
 MACROMOLECULE INTERACTIONS USING SPECTROSCOPIC METHODS

-7

Log[eADP] Free
Figure 6 The dependence of the total average degree of binding of eADP on the DnaB
helicase hexamer, as a function of the logarithm of the free nucleotide concentrations
[eADP]Free in 50 mM Tris/HCI (pH 8.1, 10 °C) containing 100 mM NaCI, 5 mm MgCI2 and 100
mM acrylamide. The solid line is the non-linear least-square fit of the binding isotherm,
according to the hexagon model (equations 10-12), which provides intrinsic binding constant
K = 4 X 105 M-1 and cooperativity parameter o- = 0.4. Reprinted from Jezewska, M. J. and
Bujalowski, W. (1997) 'Quantitative analysis of ligand-macromolecule interactions using
differential dynamic quenching of the ligand fluorescence to monitor the binding.' Biophysical
Chemistry, 64, 253-269. Copyright (1997) with permission from Elsevier Science.

bound ligand (LB/lT), then binding isotherms can be constructed with much
greater ease from a titration at a single ligand concentration (see below). The
dependence of the fractional fluorescence increase of AADP as a function of the
fraction of nucleotide molecules bound to the DnaB protein, LB/IT = Ev((MT/IT), is
shown in Figure 7. The value of LB/Lr has been determined using the binding
density function plots shown in Figure 5.
It is clear that, in the studied binding of eADP to the DnaB hexamer using the
fluorescence of eADP to monitor the binding, there is a linear correspondence
between the relative increase of the nucleotide fluorescence, AS, and the
fraction of the ligand bound, IB/IT- in the examined range of the degree of
binding. However, it is important to check the relationship between the
observed fluorescence change (or any signal used to monitor the binding) and
IB/IT over a wide range of binding densities, since the signal change can, in
general, be dependent upon the degree of binding (see Figure 2). In the case of
the E. coli DnaB protein-eADP interactions, this direct proportionality holds for
the degree of binding up to -5.3 (24). Therefore, under these conditions, the
fractional fluorescence increase of eADP is equal to the fraction of the bound
nucleotide, i.e. ASobs/ASmax = Ig/Lj.. When ASobs is directly proportional to IB/IT,
one can determine the maximum extent of protein fluorescence quenching,
ASmax, from a linear extrapolation of a plot of ASobs versus IB/Lr to IB/IT = 1, as
shown in Figure 7. This short extrapolation (dashed line) to LB/LT = 1 gives the

161
WLODZIMIERZ BUJALOWSKI AND MARIA J. JEZEWSKA

2.5

2.0

1.5

1.0

0.50

0.0
0.2 0.4 0.6 0.8

Figure 7 Relationship between the relative fluorescence increase and the fractional
saturation of the nucleotide for sADP binding to the DnaB helicase in 50 mm Tris/HCI (pH
8.1, 10 °C) containing 100 mm NaCI, 5 mM MgCI2 and 100 mM acrylamide. The selected
concentration of sADP is 2 x 10"5 M. The concentration of the bound nucleotide has been
calculated from [eADP]Bound = (Ev,)[DnaB]Totai, where the total average degree of binding of
eADP on the DnaB protein, Ev,, has been determined by using the LBDF approach described
in the text. Reprinted from Jezewska, M. J. and Bujalowski, W. (1997) 'Quantitative analysis
of ligand-macromolecule interactions using differential dynamic quenching of the ligand
fluorescence to monitor the binding.' Biophysical Chemistry, 64, 253-269. Copyright (1997)
with permission from Elsevier Science.

value of the maximum relative increase, ASmax = 2.35, of the eADP fluorescence
upon saturation with the DnaB protein in studied solution conditions (24).

2.2.4 Generation of binding isotherms from a single titration when

ASobs/ASmax = LB/LT
The LBDF analysis allows one to rigorously determine a model-independent
binding isotherm and to determine the relationship between the ASobs and the
fraction of the bound ligand, LB/LJ-. The LBDF method is time consuming, since
six to eight titrations are required to construct a single, precise binding iso-
therm over a. wide range of binding densities, which is necessary if the relation-
ship between ASobs and LB/LT is not known a priori. However, if it is determined
from the LBDF analysis that a linear relationship exists between ASobs and LB/I/r
over a wide range of degree of binding (as in the case of the eADP-DnaB protein
system discussed in the previous section), then one can use this relationship to
determine the average degree of binding and the free ligand concentration
from a single titration curve (17, 19, 22-24). For such a simple case, equation 20
reduces to

21a
AS

162
 MACROMOLECULE INTERACTIONS USING SPECTROSCOPIC METHODS

and it follows that

- ASobs\T 2lb

and
A,S»u_\/L T \
~\ 21c
ASm
Thus, a single titration can be used to obtain Ev; as a function of IF. However, we
stress that one should not simply assume that the fractional signal change is
equal to the fraction of the bound ligand, since, if it is not true, this could lead
to significant errors. On the other hand, if direct proportionality does not exist
between the signal change and the fraction of the bound ligand over a wide
range of binding densities, the true binding isotherm can still be constructed
without any assumptions by using the LBDF analysis (see Figure 6).

3 Summary
Understanding macromolecular interactions, such as those involving ligands
and macromolecules, requires detailed knowledge of the energetics and kinetics
of the formed complexes. Spectroscopic methods are widely used in character-
izing the energetics (thermodynamics) and kinetics of ligand-macromolecule
interactions in solution. These methods are very convenient to use, do not
require large quantities of material, and, most importantly, do not perturb the
studied processes. However, spectroscopic methods are indirect, i.e., the inter-
actions are measured through monitoring changes of some physico-chemical
parameter (e.g. fluorescence, absorbance, circular dichroism) accompanying the
studied complex formation. In such studies, it is indispensable to determine the
relationship between the observed signal and the degree of binding in order to
obtain thermodynamically meaningful interaction parameters. The approaches
presented in this chapter describe general quantitative methods of the analyses
of macromolecular binding through spectroscopic measurements which allow
an experimenter to determine the thermodynamically rigorous degree of bind-
ing or the degree of macromolecule saturation and the free ligand concentra-
tion. The method enables one to obtain the total degree of binding, Evi and the
free ligand concentration, IF, hence, to construct the entire model-independent
binding isotherm. Only when the thermodynamically rigorous isotherm is
obtained, can it be analysed by using the thermodynamic models which incor-
porate the known molecular aspects of the ligand-macromolecule interactions,
like cooperativity, allosteric conformational changes, overlap of potential
binding sites, etc.
We have illustrated the application of the methods to situations where the
signal used to monitor the interactions originates from a macromolecule or a
ligand, using as examples the E. coli replicative helicase DnaB protein inter-
actions with nucleotide cofactors (14, 24).

163
WLODZIMIERZ BUJALOWSKI AND MARIA J. JEZEWSKA

The methods have been discussed as applied to studying ligand-

macromolecule interactions using the fluorescence intensity to monitor the
binding. However, these approaches can generally be applied to any ligand-
macromolecule system by monitoring the binding and using any spectroscopic
signal originating from a ligand or a macromolecule, thus, allowing the
experimenter to construct model-independent binding isotherms.

Acknowledgements
We thank Gloria Drennan Davis for help in preparing the manuscript.

References
1. Riggs, A. D., Bourgeois, S., and Cohn, M. (1970). J. AM. Biol., 53, 401.
2. Jensen, D. E. and von Hippel, P. H. (1977). Anal. Biochem., 80, 267.
3. Draper, D. E. and von Hippel, P. H. (1978) J. Mo!. Biol, 122, 321.
4. Garner, M. M. and Revzin, A. (1981). Nucl Acids Res., 9, 3047.
5. Fried, M. and Crothers, D. M. (1981). Nucl. Acids Res., 9, 6505.
6. Cassel, J. M. and Steinhardt, J. (1969). Biochemistry, 8, 2603.
7. Holbrook, J. J. (1973). Biochem. J., 128, 921.
8. Boschelli, F. (1982). J. Mol. Biol, 162, 267.
9. Kelly, R. C, Jensen, D. E., and von Hippel, P. H. (1976). J. Biol. Chem., 251, 7240.
10. Porschke, D. and Rauh, H. (1983). Biochemistry, 22, 4737.
11. Bujalowski, W. and Lohman, T. M. (1987). Biochemistry, 26, 3099.
12. Lohman, T. M. and Bujalowski, W. (1991). Methods in Enzymology, 208, 258.
13. Heyduk, T. and Lee, J. C. (1990). Proc. Notl. Acad. Sci., 87,1744.
14. Bujalowski, W. and Klonowska, M. M. (1993). Biochemistry, 32, 5888.
15. Bujalowski, W. and Klonowska, M. M. (1994). Biochemistry, 33,4682.
16. Bujalowski, W. and Klonowska, M. M. (1994). J. Biol. Chem., 269, 31359.
17. Bujalowski, W. and Jezewska, M. J. (1995). Biochemistry, 34, 8513.
18. Jezewska, M. J. and Bujalowski, W. (1996). J. Biol. Chem., 271, 4261.
19. Jezewska, M. J., Kim, U.-S., and Bujalowski, W. (1996). Biochemistry, 35, 2129.
20. Bujalowski, W. and Porschke, D. (1988). Biophysical Chem., 30,151.
21. Bujalowski, W., Klonowska, M. M., and Jezewska, M. J. (1994). J. Biol. Chem., 269,
31350.
22. Jezewska, M. J., Kim, U.-S., and Bujalowski, W. (1996). Biophysical Journal, 71, 2075.
23. Jezewska, M. J. and Bujalowski, W. (1996). Biochemistry, 35, 2117.
24. Jezewska, M. J. and Bujalowski, W. (1997). Biophysical Chem., 64, 253.
25. Jezewska, M. J., Rajendran, S., and Bujalowski, W. (1997). Biochemistry, 36,10320.
26. Jezewska, M. J., Rajendran, S., and Bujalowski, W. (1998). Biochemistry, 37, 3116.
27. Jezewska, M. J., Rajendran, S., and Bujalowski, W. (1998). J. Biol. Chem., 273, 9058.
28. Jezewska, M. J., Rajendran, S., Bujalowska, and Bujalowski, W. (1998). J. Biol.
Chem., 273,10515.
29. Holbrook, J. J. and Gutfreund, H. (1973) FEBS Lett., 31,157.
30. Witting, P., Norden, B., Kim, S. K., and Takahashi, M., (1994). J. Biol. Chem., 269, 5700.
31. Halfman, C. J. and Nishida, T. (1972) Biochemistry, 11, 3493
32. Lakowicz, J. R. (1983). Principles of fluorescence spectroscopy. Plenum Press, New York.
33. Hill, T. L. (1985). Cooperativity theory in biochemistry. Springer-Verlag, New York.
34. Parker, C. A. (1968). Photoluminescence of solutions. Elsevier, Amsterdam.
35. Bujalowski, W. and Lohman, T. M. (1989). J. Mol. Biol., 207, 249.

164
 MACROMOLECULE INTERACTIONS USING SPECTROSCOPIC METHODS

36. Bujalowski, W. and Lohman, T. M. (1989). J. Mol. Biol., 207. 268.

37. Secrist, J. A., Bario, J. R., Leonard, N. J., and Weber, G. (1972). Biochemistry 11, 3499.
38. Leonard, N. J. (1984). Crit. Rev. Biochem,, 15,125.
39. Ando, T. and Asai, H. (1980). J. Biochem., 88, 255.
40. Eftink, M. R. and Ghiron, C. A. (1981). Anal Biochem., 114,199.
41. Eftink, M. R. (1991). Biophysical and biochemical aspects offluorescencespectroscopy, Ch. 1
(ed. G. Dewey). Plenum Press, New York, London.
42. Azumi, T. and McGlynn, S. P. (1962). J. Chem. phys., 37, 2413.

165
This page intentionally left blank
 6
Steady-state kinetics
Athel Cornish-Bowden
Institut Federatif 'Biologie Structurale et Microbiologie', CNRS-BIP,
31 chemin Joseph-Aiguier, B.P. 71,13402 Marseille Cedex 20, France

1 Introduction to rate equations, first-order,

second-order reactions etc.
All of chemical kinetics is based on rate equations, but this is especially true of
steady-state enzyme kinetics: in other applications a rate equation can be
regarded as a differential equation that has to be integrated to give the function
of real interest, whereas in steady-state enzyme kinetics it is used as it stands.
Although the early enzymologists tried to follow the usual chemical practice of
deriving equations that describe the state of reaction as a function of time there
were too many complications, such as loss of enzyme activity, effects of
accumulating product etc., for this to be a fruitful approach. Rapid progress
only became possible when Michaelis and Menten (1) realized that most of the
complications could be removed by extrapolating back to zero time and
regarding the measured initial rate as the primary observation.
Since then, of course, accumulating knowledge has made it possible to study
time courses directly, and this has led to two additional subdisciplines of
enzyme kinetics, transient-state kinetics, which deals with the time regime before
a steady state is established, and progress-curve analysis, which deals with the slow
approach to equilibrium during the steady-state phase. The former of these has
achieved great importance but is regarded as more specialized. It is dealt with in
later chapters of this book. Progress-curve analysis has never recovered the
importance that it had at the beginning of the twentieth century.
Nearly all steps that form parts of the mechanisms of enzyme-catalysed
reactions involve reactions of a single molecule, in which case they typically
follow first-order kinetics:
v = ka 1
or they involve two molecules (usually but not necessarily different from one
another) and typically follow second-order kinetics:
v = kab 2
In both cases v represents the rate of reaction, and a and b are the concentra-
tions of the molecules involved, and k is a rate constant. Because we shall be

167
ATHEL CORNISH-BOWDEN

regarding the rate as a quantity in its own right it is not usual in steady-state
kinetics to represent it as a derivative such as -da/dt.
A third case that is useful to consider in catalysed reactions is the zero-order
reaction, in which v apparently does not depend on any concentration:
v=k 3
Although this behaviour is apparent rather than real it is useful to consider it
because some of the constants in enzyme kinetics are zero-order rate constants.
In each of the equations above the k is an overall rate constant. However, it is
often convenient to consider the rate with respect to one reactant, the con-
centrations of the others being maintained constant. For example, if the second-
order reaction above is considered in relation to a only, b being regarded as a
constant:
v = kb a 4
then it is evident that the rate is proportional to a, and it is said to be first-order
with respect to a. The 'constant' kb, which is only constant of course if b is
constant, is called a pseudo-first-order rate constant.
Although the individual steps in an enzyme mechanism may all be either
first-order or second-order reactions, the mechanism normally includes several
steps, with the result that the complete reaction rarely has a simple order. The
fundamental problem, therefore, is to set up conditions in which the rate is
easily measurable and then to determine how it varies with the conditions.

2 Units
All of the quantities required in steady-state kinetics are either concentrations,
normally measured in mol litre"1 or a submultiple, rates, measured in mol litre"1
s-1, or rate constants, with units that vary according to the type of rate constant:
a first-order rate constant has dimensions of reciprocal time, and is typically
measured therefore in s"1, and a second-order rate constant has dimensions of re-
ciprocal time multiplied by reciprocal concentration, and is typically measured
therefore in mol-1 litre s-1. It is obvious from elementary dimensional consider-
ations that a pseudo-first-order rate constant has the same dimensions and units
as a first-order rate constant, and that constants of different order cannot be
meaningfully compared.
By far the most common practice is to regard the rate as an intensive
quantity, i.e. to normalize it to a standard volume, so for any given conditions
the rate is the same whether it refers to a 0.5 ml cuvette or to a 100 litre fer-
menter. However, in some specialized applications it is more usual to define the
rate as a rate of conversion, i.e. as an extensive quantity measured in mol s-1. If
this is done, then the dimensions and units of all the rate constants must
change correspondingly. If all the processes occur in a single compartment,
there is not usually any reason to use extensive quantities, but it becomes much
more convenient, if not essential, when one is concerned with transport

168
 STEADY-STATE KINETICS

between compartments of different volumes. These applications will not be

considered in this chapter.

3 Basic assumptions in steady-state kinetics

The essential requirement in a steady-state experiment is to set up conditions
such that the rate is essentially constant during a period convenient for measur-
ing it (typically a period of several minutes), or that it decreases sufficiently slowly
for an accurate estimate of its value at zero time to be possible. There are several
experimental designs that will in principle produce such conditions, but only
one of them is in widespread use, and that is to work with very low enzyme con-
centrations (typically much lower than those that exist in the cell), and in par-
ticular to use a concentration much smaller than (no more than 1% of) those of
the substrate(s) and any other small molecules that may be present.
Although the need for low enzyme concentrations is primarily derived from
the need to create a steady state, it has an obvious economic advantage that
ensures that steady-state methods will continue to be widely used even though
transient-state methods may in principle generate much more information.
Transient-state methods typically involve direct observation of the enzyme or
its complexes with the reactants, and they require enzyme concentrations high
enough for direct observation, and consume large amounts of enzyme in each
experiment. To avoid wasting materials, therefore, steady-state experiments are
often used for preliminary exploration even when the ultimate aim is to study
the transient state.
Initial-rate conditions also require that there is no reverse reaction (at least in
the extrapolated case at zero time). This can be achieved without making any
assumptions about the value of the equilibrium constant by studying conditions
in which no products are present at zero time. This requirement does not ex-
clude studying the effects of products on the reaction rate as long as there is
more than one product (provided that at least one product is missing from the
reaction mixture the rate of the reverse reaction will still be zero at zero time).
It becomes more complicated for reactions with only a single product, but these
are not common and will not be considered in this chapter.
If the essential experimental requirement of a very low enzyme concentra-
tion is fulfilled, a reaction can be studied over a period in which the reactant
concentrations change so little that they can be regarded as constants, so there
is no need to distinguish between their initial and instantaneous values. In the
case of a reaction involving the conversion of a single substrate A, the equation
that describes the steady-state behaviour is the Michaelis-Menten equation:
koe0a Va
+a
Here e0 is the total enzyme concentration, i.e. the total of all forms of enzyme
including free enzyme and any intermediates in the reaction, and a is the
substrate concentration, and the parameters are ko. the catalytic constant (or turn-

169
ATHEL CORNISH-BOWDEN

over number), and Km, the Michaelis constant. The form shown in the middle is
more fundamental than that on the right, but the second form, in which koe0 is
written as the limiting rate V, is often used because the enzyme concentration is
not always known in meaningful units. V is the limit that v approaches as the
enzyme becomes saturated, i.e. when a becomes very large, and Km is the value
of a at which v = 0.5V, i.e. at which the rate is half-maximal. Notice that ko is a
first-order rate constant, Km is a concentration and V is a zero-order rate con-
stant, or in other words a rate. The ratio ko/Km is a second-order rate constant,
and is called the specificity constant and given the symbol kA: it is a more
fundamental constant than Km in the analysis of enzyme mechanisms (i.e. it has
a simpler mechanistic meaning), but the equation is usually written in terms of
V (or ko) and Km nonetheless.
One may derive equation 5 from the following model,
(C-j K2 k3

A+E - E A - E P - E +P 6
k-, k-2
which assumes that the reaction passes through an enzyme-substrate complex,
EA, that undergoes catalytic transformation to an enzyme-product complex, EP,
which then breaks down to form products. A simpler mechanism containing
only one intermediate also generates the Michaelis-Menten equation, but this is
too simple to be realistic. Conversely, there are infinitely many more complex
mechanisms that also generate it, so although the three-step mechanism pro-
vides a useful basis for discussion one can never be certain from steady-state
measurements alone that the true mechanism is not more complex.
The Michaelis-Menten parameters can be defined as follows in terms of the
rate constants shown in the above reaction:

7
k_2 + k2 + k3
k1,k2k3
8

_
111
Mk.2 + kj + ks)
Note that none of the three parameters has a simple transparent meaning. The
interpretations commonly attributed to them depend on additional simplifying
assumptions that are not always correct. For example, lCm is often said to be
equal to the equilibrium constant k-1/kj for dissociation of A from EA, but the
expression in equation 9 does not take this form unless k2 is very small, and it is
not safe to regard Km as a measure of the equilibrium dissociation constant.

4 Measurement of specific activity

According to equation 5 the measured initial rate of reaction is proportional to
the total concentration of enzyme. However, in practice, this is something that

170
 STEADY-STATE KINETICS

needs to be established by experiment rather than just assumed to be true,

because if the enzyme becomes inactivated when it is diluted, or if it exists in
multiple states of aggregation with different activities, or if the enzyme solution
is contaminated with an inhibitor, then the simple expectation may not be
realized.
Let us consider the last of these to illustrate the sort of problems that can
arise. If a concentration e0 of enzyme implies a concentration ae0 of a com-
petitive inhibitor then the rate equation is not equation 5 but equation 15 (below),
with the inhibitor concentration i written as ae0. Notice that this means that e0
appears not only as a factor of the numerator of the rate expression, but also as
a term in the denominator, so the whole expression is no longer proportional to
e0. This makes the rate expression as a function of e0 have the same form that it
has as a function of the substrate concentration a, so it approaches a limit of
kokic/Kma instead of increasing indefinitely as e0 increases. Clearly this limit
depends on all four of the quantities that it contains, but only one of them, a, is
(in principle) under the control of the experimenter. If a cannot be made small
enough to be negligible then use of the assay may need to be restricted to a
low range of enzyme concentrations where the effect of the inhibitor is
imperceptible.
Once conditions have been determined that give a good proportionality over a
useful range, these can be used as the basis of an activity assay for the enzyme.
As noted above, the rate is an intensive quantity, i.e. its magnitude does not vary
with the total size of the system considered. The quantity required from an
assay, however, is the amount of enzyme, which is an extensive quantity be-
cause it does vary with the size of the system. It can be obtained by multiplying
the rate obtained by the volume of the compartment in which the reaction takes
place. The unit for measuring enzyme activity is mol s-1, as the rate of con-
version of a particular substrate under specified conditions. For example, a
reaction rate of 5 uM s-1, or 5 umol litre-1 s-1, obtained in a 0.5 ml reaction volume
would correspond to an enzyme activity of 5 X 0.5 x 10-3 = 0.0025 umol s-1.
Notice that the definition of enzyme activity implies that all of the
conditions, including the identity of the substrate, are specified. Thus a change
of rate observed when a different substrate is used in the assay does not imply a
change in enzyme activity (though if the change of substrate is adopted as part
of the definition of the standard assay, it will imply a change in the definition of
enzyme activity).
The SI unit of enzyme activity, 1 mol s-1, may also be called the katal, with
symbol kat, but this term has not been widely adopted. It is a very large unit for
most purposes, but its submultiples the nanokatal (nkat) and microkatal (ukat)
are of a convenient size for measuring enzyme activities in the laboratory. In
the past it was usual to define a 'unit' of enzyme activity measured in p,mol
min"1 (= 16.67 nkat), and this is still used, though it is not very satisfactory
because the name is uninformative and fails to convey the incorporation of a
factor of 10-6 and a non-Si unit (the minute) in the definition.

171
ATHEL CORNISH-BOWDEN

5 Graphical determination of K and V

Equation 5 defines a hyperbolic dependence of rate on substrate concentration,
and in principle one could plot v against a, determine V from the asymptotic rate
at high substrate concentrations, and once this is known Km could be found as
the value of a at which v = 0.5V In practice, this is a bad method (and is little
used) because it is very difficult to decide accurately where the asymptote ought
to be drawn. It is nearly always drawn too close to the curve, with under-
estimation of V This is because the curve in fact approaches the asymptote very
slowly: even at a concentration of 10Km the rate is still only 91% of the limit. The
usual practice, therefore, is to transform equation 5 into one of the following
three forms, which underlie the three straight-line plots illustrated in Figures 1-3:

10
V a

11
v
12
a
The double-reciprocal plot, often attributed to Lineweaver and Burk (2), is illus-
trated in Figure 1 and is based on equation 10; it is the mostly widely used
straight-line plot, but it is also the least satisfactory, because it distorts the effect
of experimental error to such an extent that it is difficult to form any visual
judgement of where the best line should be drawn. Either of the other two plots
is better. The plot of a/v against a (equation 11) is associated with Woolf (3) or
Hanes (4) and is illustrated in Figure 2; notice that the definitions of the slope
and intercept on the ordinate are the opposite way round from those that apply
to the double-reciprocal plot. The plot of v against v/a (equation 12) is variously
attributed to Eadie (5) or to Hofstee (6) and is illustrated in Figure 3. It has the
particular advantage that the entire observable range of v values, from 0 to V, is

i/v

+C/

Slope = Km/V

I/a

Figure 1 The double-reciprocal plot. The dashed lines labelled +C and +U illustrate how the
line is displaced by the presence of a competitive or uncompetitive inhibitor, respectively.

172
 STEADY-STATE KINETICS

a/v

Slope = 1/V

Figure 2 The plot of a/v against a. Notice that the definitions of the slope and the intercept
on the ordinate are the opposite way round from those in Figure 1. The dashed lines labelled
+C and +U illustrate how the line is displaced by the presence of a competitive or
uncompetitive inhibitor, respectively.

V-

Figure 3 The plot of v against v/a. Notice that this plot maps the entire range of observable
rates (from 0 to V) onto a finite range of paper. The dashed lines labelled +C and +U
illustrate how the line is displaced by the presence of a competitive or uncompetitive
inhibitor, respectively.

plotted in a finite range; this makes it easy to judge by eye if an experiment has
been well designed. On the other hand, it has the disadvantage that v, normally
the less reliable measurement, contributes to both coordinates, and errors in v
cause deviations along lines through the origin rather than parallel with one or
other axis.
These plots are valuable for illustration purposes, but for actual estimation of
kinetic parameters it is more objective to use appropriate computer programs.
For this purpose it is not sufficient just to apply unweighted linear regression to
the straight-line plots, as this suffers from the same statistical distortions as the
plots themselves. Full treatment would require more space than is available
here, but may be found elsewhere (7). The following two equations for calculat-
ing best-fit values of Km and V give satisfactory results if the standard deviation

173
ATHEL CORNISH-BOWDEN

of v is approximately constant when expressed as a percentage of the true

value, as is usually at least approximately correct:
£v2S(v/q) - 2(v2/a)£v
m
2(v2/a2)2v - 2(v2/a)S(v/a)

S(v2/a2)2v2 - [S(v2/a)]2
S(v2/a2)Sv - S(v2/a)2(v/a)

6 Inhibition of enzyme activity

The simplest kind of enzyme inhibition is competitive inhibition, characterized by
a rate equation of the following form:

15
Km(l + i/Kic) + a
in which i is the concentration of the inhibitor and Kic is the competitive inhibition
constant. (The qualification 'competitive' and the second subscript c are usually
omitted if only this simplest kind of inhibition is being considered.) Competi-
tive inhibition arises when a molecule binds reversibly to the free enzyme in
such a way as to prevent the substrate from binding. Although this occurs most
obviously when the inhibitor is similar enough to the substrate in structure to
replace it in the substrate binding site, but not similar enough to undergo
reaction, competitive inhibition can also arise in more complex ways.
The opposite extreme from competitive inhibition is uncompetitive inhibition,
and is defined by the presence of a factor that affects the variable term in the
denominator of the Michaelis-Menten equation instead of the constant term:

16
Km + o(l + i/Kiu)
The inhibition constant Kiu is an uncompetitive inhibition constant. This type of in-
hibition is not common, but it is important as a component of mixed inhibition,
when both competitive and uncompetitive effects occur simultaneously:
_
V=
K m ( l + i / K i c ) + a(l+t/K i u )
There is no particular reason for the two inhibition constants Kic and Kiu to be
equal, and most of the mechanisms for mixed inhibition suggest that they
ought to be different. However, the case where Kic = Kiu is often given an un-
deserved prominence in discussions of inhibition, largely because experiments
done many years ago suggested that it was a more common phenomenon than
it is. It is called non-competitive inhibition and its rate equation is the same as
equation 17, but with both Kic and Kiu written simply as K1.
All of these kinds of inhibition are conveniently discussed in terms of apparent
Michaelis-Menten parameters, i.e. the parameters that replace the ordinary para-

174
 STEADY-STATE KINETICS

meters in the Michaelis-Menten equation when an inhibitor is present. In the

general case, equation 17, these are as follows:

m -
i+
Note that the first two expressions have the same form, and both simplify to
independence of i in the event that one or other inhibition term is negligible.
The expression for the apparent value of Km is more complicated, especially
when one considers how it varies with the different types of inhibition: it
increases with the concentration of a competitive inhibitor, it decreases as the
concentration of an uncompetitive inhibitor increases, it may change in either
direction as the concentration of a mixed inhibitor increases, or it is independ-
ent of inhibitor concentration if the inhibition is non-competitive. In general it
is simplest to regard kA as the parameter affected by competitive inhibition,
negligibly so when the competitive component is negligible; k0 as the parameter
affected by uncompetitive inhibition, negligibly so when the uncompetitive
component is negligible; and Km just as the ratio of the two, i.e. Km = Ko/kA.
The effects of the different kinds of inhibition on the common plots are illus-
trated in Figures 1-3 and follow naturally from equations 18-20. Any competitive
effect alters the apparent value of kA: hence it increases the slope of the plot of
1/v against I/a (figure 1), it increases the ordinate intercept of the plot of a/v
against a (Figure 2), and it decreases the abscissa intercept of the plot of v against
v/a (Figure 3). Conversely, any uncompetitive effect increases the ordinate inter-
cept of the plot of 1/v against I/a, increases the slope of the plot of a/v against a,
and decreases the ordinate intercept of the plot of v against v/a. When both
components of the inhibition are present both kinds of effects occur.

7 Specificity
Specificity is a fundamental property of enzymes, but it is often assessed in an
unsatisfactory way, by comparing the kinetic parameters for different reactions
measured in isolation from one another. In the cell, specificity must clearly
refer to the capacity of an enzyme to react selectively with one substrate when
others are present simultaneously, and any satisfactory measure of specificity
should take account of this (8). The equation for reaction of one substrate A in
the presence of a competing substrate A' follows a form similar to that for
competitive inhibition, equation 15:

v= _ _ 21
Km(l + a'/JC'J + a

175
ATHEL CORNISH-BOWDEN

with the inhibitor concentration replaced by a', the concentration of A', and the
inhibition constant by K'm, the Michaelis constant for the reaction of A' con-
sidered in isolation. The rate of reaction of A' is given by the same equation
with an obvious transposition of symbols:
v= _ _ 22
K'm(l + a/KJ + a'
Dividing one equation by the other, the ratio of rates is the ratio of substrate
concentrations multiplied by the ratio of specificity constants:
y_ k0/Km a_ _ kAa
V k'o/K'm'a' k'Ao'
This result explains the choice of the name specificity constant. Equation 23
applies at all concentrations, and the specificity constant measures specificity at
all concentrations. This point is worth emphasizing, because at very low sub-
strate concentrations the rate of an enzyme-catalysed reaction follows second-
order kinetics approximately, and the specificity constant is the second-order
rate constant in these conditions. As this property was well known long before
the relation to specificity was generally recognized it has sometimes generated
an incorrect idea that there is a similar concentration restriction for specificity.

8 Activators
Activation is the opposite effect to inhibition, and an activator is, naturally
enough, a species that increases the rate of an enzyme-catalysed reaction. In
practice, however, activation has been studied much less than inhibition, and
this section will be correspondingly brief. The most important point to make is
that various other phenomena are also sometimes called activation, and these
should not be confused with true activation. For example, certain enzymes, such
as the digestive enzyme pepsin, are initially synthesized as inactive 'zymogens',
pepsinogen in this case, which are transformed into active enzyme by a post-
translational partial proteolysis, a process that is sometimes called zymogen
activation. Another case is really a confusion: for some enzymes, most notably
ATP-dependent kinases, the true substrate is a complex ion containing a metal,
typically MgATP2", and such enzymes are sometimes said to be activated by
Mg2+ ions.
True activation is, as noted, the converse of inhibition, and can be classified
and analysed in similar ways. In general, whenever a term of the form i/Ki occurs
in an inhibition equation one can conceive of a corresponding activation type
where a term Kx/x defines the effect of an activator X. It is important to realize,
however, that the activation analogue of competitive inhibition cannot be
explained by any mechanism in which it makes sense to talk about competition
between substrate and activator, and thus a term like 'competitive activation'
makes no sense even though it is the inverse of competitive inhibition. Such
activation may occur, for example, if the substrate binds only to the enzyme-

176
 STEADY-STATE KINETICS

activator complex and not to the free enzyme, so a more meaningful term might
be compulsory activation.

9 Environmental effects on enzyme activity

As well as inhibitors and activators, enzymes are affected by various environ-
mental effects, of which the most important are pH and temperature.

9,1 pH
Mechanistically, most pH effects are inhibition and activation by one particular
inhibitor and activator, the proton, and the equations that describe them are
the same as those for inhibition and activation. There are, however, some im-
portant differences. First, the fact that the proton concentration can be varied
and controlled over a very wide range makes it possible to observe and measure
properties that would be difficult to observe if they occurred with other species.
Second, unlike most inhibitors and activators the proton usually displays both
properties, acting simultaneously as an inhibitor and an activator. This gives
rise to the familiar bell-shaped curve illustrated in Figure 4, which describes the
response of an enzyme to variation in the pH:
k
K
h K2 24
K, + ~h
In this equation k represents some kinetic constant for the reaction, such as
the catalytic constant or the specificity constant, k is a notional pH-independent

0.5 k

5 7 9 11
pH
Figure 4 Bell-shaped pH profile. The parameter k plotted in the ordinate represents either
the catalytic constant ko, or the specificity constant k0/Km, and the limit k is the value it
would have if all of the enzyme existed in the catalytically active state of protonation.

177
ATHEL CORNISH-BOWDEN

form of the same constant that defines the value it would have if all the enzyme
could be fixed in the active state of protonation, h is the hydrogen-ion con-
centration and K1 and K2 are the acid dissociation constants for two successive
deprotonation steps. This is, of course, the simplest possible simultaneous
activation and inhibition, and much more complex cases also occur. It should
be obvious from inspection of the equation that k approaches zero at both low
pH, where h/Kj is large, and at high pH, where K2/h is large, but is larger at
intermediate pH values. If Kt and K2 are well separated it may approximate to k
in this intermediate region, but may be much less.
We may expect both the free enzyme and all intermediate complexes that
occur in an enzyme-catalysed reaction to exist in multiple states of protonation,
and so we can expect any parameter of the reaction to vary with pH. In general
the pH dependence of the specificity constant kA or ko/Km provides information
about the protonation states of the free enzyme, and that of the catalytic
constant k0, provides information about the protonation states of the enzyme-
substrate complex(es). The pH dependence of Km is usually more complicated, as
it involves contributions from all of the different forms of enzyme; it is most
easily analysed by regarding Km as k0/(K0/Km).

9.2 Temperature
Temperature effects on enzymes follow the same principles as those that affect
simple chemical reactions, based on the Arrhenius treatment (9), but in practice
they are very complicated. This is because all enzyme mechanisms involve
multiple steps, and all of the rate constants describing the different steps vary
separately with temperature. It is unlikely, therefore, that useful information
will emerge from studies of temperature dependence unless the experimenters
are willing to go rather deeply into the theory. If one is working with an en-
zyme for which the mechanism is known in considerable detail, and for which
the rate constants of individual steps can be measured, then the temperature
dependence of any of these steps can be analysed just as if it were obtained in a
chemical kinetic experiment. Apart from this, temperature dependencies are
most likely to be useful when a break (abrupt change in slope) in an Arrhenius
plot coincides with a known physical phenomenon; for example, if one is
studying a lipid substrate that is known to undergo a phase transition at a
particular temperature, then a break in the Arrhenius plot at that temperature
may be helpful for identifying which form of the lipid reacts with the enzyme.
It is important to realize, however, that most of the abrupt changes in slope
that are reported are not real. The reasons for this are primarily psychological:
once a line or set of lines have been drawn over the experimental points in a
graph it is easy to be convinced that they fit the points rather better than they
do in reality, and because of this it is not difficult to 'fit' the data in an
Arrhenius plot so that the change in slope occurs in a place that is easy to
interpret. Consider the data in Figure 5. Is the break at (a) 9 °C or (b) 17.5 °C? Or is
there in reality no break at all but a gradual change of slope over the whole
range plotted, as in (c)?

178
 STEADY-STATE KINETICS

(a)
log

2
4
(b)
log A

J I
2
4
(c)
log k

3.3 3.4 3.5 3.6

1000/T, K~
Figure 5 Dependence of a rate constant k on temperature (Arrhenius plot). In all three
cases the same points are shown, but in (a) and (b) they are interpreted as lying on pairs of
straight lines and in (c) as lying on a smooth curve.

10 Cooperativity
All of the cases considered so far can be regarded as generalizations of the
Michaelis-Menten equation, (equation 5). However, although many enzymes do
behave in this way, at least as a first approximation, there are some important
exceptions. It is simple to calculate from equation 5 that if a = Km/9 then v = 0.1 V
and if a = 9Km then v = 0.9V; in other words, spanning the 10-90% range of
available rates requires an 81-fold range of substrate concentrations, almost two
orders of magnitude. Similar calculations may be done with any of the equations
of the Michaelis-Menten type, whether for additional substrates, inhibitors or
activators, and they imply that as long as enzymes follow Michaelis-Menten
kinetics, their rates vary very little with the sort of changes in substrate and
other concentrations that are likely to occur in the cell. As effective regulation
of metabolism requires that at least some enzymes respond sensitively to

179
ATHEL CORNISH-BOWDEN

changes in the concentrations of effectors, it is not surprising that Michaelis-

Menten kinetics is not universal and that some enzymes display cooperativity.
The most obvious indication of cooperativity is that the plots expected to
give straight lines give curves, and in principle one can determine the degree of
cooperativity by analysing the degree of curvature, but it is easier and more
usual to use the Hill plot (10). This exists in various forms, depending on whether
we are studying cooperativity with respect to a substrate, an inhibitor or an
activator. As cooperative inhibition is one of the most important cases in
practice, it is used in Figure 6 to illustrate the plot, which is then a plot of
log[vi/(vo - vi) against log i, where v0 is the uninhibited rate and vf is the rate at a
concentration i of inhibitor. If equation 15 applies, i.e. if there is no cooperativity,
then it is a simple matter to show that
log [vj(v0 - vi)] = log [Kic(l + o/K m ] - log i 25
i.e. that the slope of the Hill plot ought to be -1. The slope is negative because
this is an example of inhibition; the equivalent plots for effects of substrates or
activators are positive, and in all cases the slope, ignoring the sign, is called the
Hill coefficient.
As equation 25 illustrates, a Hill coefficient of 1 is characteristic of Michaelis-
Menten kinetics, or non-cooperative kinetics. A Hill coefficient greater than 1
indicates that the rate is more sensitive to changes in concentration than
expected from the simple model. Values in the range 1.5 to 4 are quite
commonly reported. Values greater than 4 are very rare, probably because of
the difficulty in evolving an enzyme structure with sufficiently tight coupling
between different sites to generate a higher degree of cooperativity. Values
between 1 and 1.5 are also uncommon, either because the relatively weak

M 0

-1

-1
log!

Figure 6 Hill plot for inhibition. As noted by the dashed lines at the two extremes, any Hill
plot should in principle approach a slope of unity at the extremes, even though in practice
the curvature may often be hard to detect and the points may appear to fit a straight line
over the whole experimental range.

180
 STEADY-STATE KINETICS

degree of cooperativity often passes unnoticed by the experimenter or possibly

because it offers no particular physiological function. Hill coefficients less than
1 also occur, indicating a rate that is less sensitive to changes in concentration
than expected from the simpler model, and this is called negative cooperativity.
The Hill coefficient is best regarded as a purely empirical value to allow the
degree of cooperativity to be measured and compared with other cases.
Contrary to what is sometimes claimed, there is no physically plausible model
that generates straight Hill plots with slopes other than 1 (or -1), and the
expectation is that any cooperative Hill plot should be curved at the extremes,
because the extreme slopes should approach 1 or — 1. That this curvature is not
often reported simply reflects the difficulty of carrying out sufficiently accurate
experiments over a wide enough range of conditions.

11 Experimental conditions for kinetic studies

In deciding the conditions for any steady-state experiment the primary object-
ive is to obtain accurate initial rates over a sufficient range of conditions to
define the kinetic parameters of interest. Ideally the rate should be constant, i.e.
the progress curve should be straight, during the period of measurement. If this
is impossible, then it is essential to realize that unless integrated rate equations
are used (for a full progress-curve analysis) all of the theory will assume initial
rates, and that therefore the slope to be used is the initial slope, not the average
slope over the period of measurement. Many experimenters report a 'linear'
phase followed by a tailing-off, but this is nearly always an illusion (cf. Figure 5,
which illustrates the same psychological problem in a different context), because
the tailing-off normally begins at zero time. What this implies in practice is that
the slope required is that of the tangent drawn to touch the progress curve at
zero time, not that of a chord drawn to fit the first few data points.
When the curvature is appreciable it may be necessary to use curve-fitting to
estimate the initial slope accurately. Some methods based on integrated rate
equations have been described, but for most purposes it is sufficient to use a
'generic' equation such as a quadratic function to describe the progress curve. If
this is done there are several pitfalls to be avoided.
In a few cases the rate may initially increase rather than decrease, and when
this occurs one should try to determine the cause, or at least conditions in
which it does not happen. No simple model predicts an initial acceleration, and
so if it occurs it implies either that more complex analysis will be required or
that there is an experimental artefact to be eliminated. An example of the first
type of explanation might be a slow conversion of the enzyme into a different
and more active state when the assay is initiated: this may well be interesting
and important and full understanding of the enzyme mechanism may require it
to be properly analysed, but this is a detailed topic that is beyond the scope of
this chapter. A more trivial explanation might be an assay in which the temper-
ature is inadequately controlled and allowed to increase during the first part of
the assay: this just needs to be eliminated by more careful experimentation.

181
ATHEL CORNISH-BOWDEN

12 Concluding remarks
Steady-state analysis of initial rates has formed the basis of most kinetic studies
of enzymes for almost a century, and as such it includes much more detail than
one can hope to cover in a single chapter. Some important topics, such as the
analysis of reactions with more than one substrate, have not been discussed at
all, and nearly all have been dealt with very briefly. A considerably more com-
plete account of all of the material in this chapter may be found elsewhere (11).

References
1. Michaelis, L. and Menten, M. L. (1913) Biochem. Z. 49, 333-369.
2. Lineweaver, H. A. and Burk, D. (1934) J, Am. Chem. Soc. 56, 658-666.
3. Woolf, B. (1932) cited by J. B. S. Haldane and K. G. Stern in Allgemdne chemie tier
enzyme, pp. 119-120. Steinkopff, Dresden and Leipzig.
4. Hanes, C. S. (1932) Biochem.]. 26,1406-1421.
5. Eadie, G. S. (1942) J. Biol. Chem. 146, 85-93.
6. Hofstee, B. H. J. (1952) J. Biol. Chem. 199, 357-364.
7. Cornish-Bowden, A. (1995) Analysis of enzyme kinetic data. Oxford University Press,
Oxford.
8. Fersht, A. (1985) Enzyme structure and mechanism, pp. 150-154. Freeman, New York.
9. Arrhenius, S. (1889) Z. Physik. Chem. 4, 226-248.
10. Hill, A. V. (1910) J. Physiol. 40, iv-vii.
11. Cornish-Bowden, A. (1995) Fundamentals of enzyme kinetics (2nd edn). Portland Press,
London.

182
 Chapter 7
Spectrophotometric assays
T. J. Mantle* and D. A. Harris1"
* Department of Biochemistry, Trinity College, Dublin 2, Ireland
f
Department of Biochemistry, University of Oxford, South Parks Road,
Oxford 0X1 3QU

1 Introduction
Chapter 1 describes the principles and practice of spectrophotometry, and in this
chapter we will consider how this technique can be used to measure the amount
of analyte in solution, when the product of a reaction between the analyte and a
'very useful reagent' produces a measurable change in absorbance. Some good
examples of this type of reaction are given together with various protocols for
routine applications such as the measurement of protein concentration.
If there are no contaminating species absorbing at the wavelength of interest,
it may be possible to measure the analyte directly simply by taking absorbance
measurements. Spectrophotometry has also proved to be fundamental for rate
measurements using a wide variety of enzyme-catalysed and non-catalysed
reactions. We will also see that it is often the case that we will use an enzyme
reaction that has gone to completion to measure the amount of an analyte that
is limiting in the reaction. Furthermore, in some cases we can get the reaction
to cycle so that the amplified rates can provide a highly sensitive assay, see, for
example, the Tietze method (1) for measuring glutathione (Section 3.3). We will
also briefly cover the use of so-called 'plate readers' as these are being used
increasingly as 'multi-cuvette spectrophotometers'. Finally, 'plate readers' will
be compared with 'centrifugal analysers', much beloved in clinical chemistry
labs. The reader should note that some methods are dealt with in detail here to
allow selected protocols to be used directly. Also, a number of useful reference
works for enzyme assays are given, including (2-4), which should be consulted.
With a little thought novel assays can often be developed based on the various
'coupling' methodologies described below.

1.1 Spectrophotometers
The design of standard spectrophotometers has been covered in Chapter 1 and
we should simply mention that most modern 'plate readers' work in essentially
the same way. The major difference is in their beam geometry. In spectro-
photometers, samples are read through cuvettes with a horizontal light path
which is normally 1 cm in length allowing the Beer-Lambert Law (see Section

183
T. J. MANTLE AND D. A. HARRIS

1.2) to be applied in its simplest form (constant light path). In 'plate readers' the
vertical light beam results in a pathlength that depends on the volume of fluid
in each well. The major drawback to this configuration is that the pathlength is
not maintained by the dimensions of the cuvette (as in a conventional
spectrophotometer) and small variations in the volume pipetted into the well
are a potential source of error. For most ELISA uses these errors are acceptable;
however, to use these instruments to read absorbances it is advisable to use a
series of dilutions of a standard analyte to set up a calibration curve using a
conventional spectrophotometer, and then to compare the readout for the same
solutions in a 'plate reader'. It is then possible to calculate the pathlength in the
'plate reader' and to make the appropriate corrections. This problem has re-
cently been addressed by Molecular Devices (www.moldev.com) where the path-
length is automatically corrected using near infrared measurements. The same
problem does not apply to 'centrifugal analysers' as the pathlength in this case
is determined by the dimensions of the cuvette. Another downside with earlier
'plate readers' was the restriction placed on the wavelengths that could be used
due to the limited range of filters supplied; however, the introduction of
interference filters has effectively overcome this problem.

1.2 Beer-Lambert Law

It will not hurt to remind ourselves that the use of spectrophotometry in assays
rests on the operation of the Beer-Lambert Law (described in Chapter 1) which
may be written as;
A =e cl 1
where A is the absorbance, and is equal to the log of the ratio of the incident (I0)
to transmitted (I) light, s is the extinction coefficient and I is the length of the
path cell (usually, but not always 1 cm). It is important to emphasize that A has
a logarithmic relationship to light transmittance and that as A increases from 1
to 2 only 10% and 1% , respectively, of the incident light is transmitted. In earlier
spectrophotometers, this low transmission could result in inaccurate readings
although this is not normally a problem with the current generation of instru-
ments. However, deviations from the Beer-Lambert law may be observed above
an absorbance of 1 for non-instrumental reasons and calibration of the system
to be used is a wise precaution.

1.3 The nature of the sample

One of the joys of spectrophotometric assays is that very often the analyte in
the sample does not have to removed from other contaminating compounds, as
long as it is the only component of the mixture that is absorbing light at the
wavelength of interest. For example, many dehydrogenases use NAD+ as cofactor.
When reducing equivalents are transferred from the oxidizable substrate, NAD+
is reduced to NADH which has a characteristic absorption at 340 nm. As NAD+
does not absorb at this wavelength (it has a major absorption peak at 260 nm)

184
 SPECTROPHOTOMETRIC ASSAYS

the reaction can be conveniently monitored simply by measuring the increase

in absorbance at 340 nm. Obviously if the reverse reaction is being followed, i.e.
the oxidation of NADH, then this is measured as a decrease in A340. This pro-
vides a good example of the experimental problems caused by the non-linear
response of a spectrophotometer to a change in concentration of a chromo-
phoric solute (see Section 1.2 above and Chapter 1). If an assay contains 200 uM
NADH, then the absorbance at 340 nm in a cuvette with a 1-cm pathlength will
be approximately 1.24 and this may exceed the linear range of the response.
Thus under these conditions a reaction in which the NADH is converted to
NAD+ may have a progress curve which appears to have a lag period followed
by an accelerating rate of decrease in absorbance. Such situations benefit from
the use of a cell with a shorter pathlength. This problem commonly arises in
experiments in which the Km of NADH (or any other chromophoric substrate or
coenzyme) is being determined where high concentrations are desired to allow
approach to V (see Chapter 6).
Although it is sometimes not necessary to separate the analyte from other
components of the mixture, the sample often requires some 'work up' to facili-
tate absorbance measurements. The most common type of interference in
spectrophotometric assays is light scattering due to the presence of particulate
material (e.g. mitochondria, red cells etc.). This leads to a decrease in trans-
mission and therefore an artefactual apparent increase in absorbance. This type
of particulate interference can often be removed by introducing a suitable
centrifugation step.

2 Some general comments on, and practical

aspects of, assay design
It is important that assays are specific, sensitive, accurate and precise. It should
go without saying that it is also important that they should also be as rapid and
as inexpensive as possible. There is no point in having a simple assay which
gives imprecise results; however, it is equally true that very sensitive assays are
commonly less robust and more expensive and should be avoided unless
required.
The specificity of biochemical assays often depends not on the chromophores
measured (these are fairly restricted e.g. p-nitrophenol, NADH, phospho-
molybdate, chloronaphthol etc.) but on the specificity of the enzyme or anti-
body employed. For example, the use of LDH to measure lactate will also give an
increase at 340 nm with a small number of related compounds. However, this is
not usually a problem as, of these structures, only lactate is usually encountered
in most biological systems.
As described above, it is often the case that an enzyme reaction is used to
monitor an analyte and the reaction will be allowed go to completion. If the
signal is not amplified in some way (see Section 3.3), then the sensitivity is limited
by the magnitude of the extinction coefficient. For example, the extinction

185
T. J. MANTLE AND D. A. HARRIS

coefficient for NADH at 340 nm is 6220 M-1 cm-1 which means that a solution of
1 uM has an absorbance of 0.006. This reflects the lowest amount of NADH that
can be measured with confidence using spectrophotometry. If greater sensi-
tivity is required for measuring NADH, then fluorimetry (excitation wavelength
340 nm, emission wavelength 460 nm) can be used.

2,1 Accuracy and precision

A repeated assay on the same sample yields a range of values that can be ex-
pressed as the mean, x, +/- the standard deviation o. For a precise assay o is
small compared with the mean (x) and it is desirable that the o/x (the % error)
should be less that 5%. If the scatter is small then the reproducibility is good and
the assay has good resolution (the ability to distinguish between two close
values). Elements that affect precision are random errors such as pipetting,
weighing etc.
An accurate assay is one where the measured value of x (x) is close to the true
value, X, i.e. X - x is small. If we are measuring a reaction spectrophoto-
metrically to an end point there are a number of possible systematic errors. For
example, the reaction may not go to completion if a concentration of reagent is
too low, or the time allowed, while appropriate for a high concentration of
analyte, is too short at lower concentrations. A trivial, but occasionally import-
ant source of inaccuracy can occur by working away from the wavelength of
maximum absorbance (Xmax). This is particularly applicable to compounds with
'narrow' absorption bands and is further exacerbated by use of wide slit widths
on the spectrophotometer (see Chapter 1).
In several experiments accuracy may be less important than precision. For
example, if during the purification of an enzyme a protein assay is used based
on albumin as a standard, the colour change observed for 1 mg of albumin may
be different to that observed with 1 mg of the enzyme under study. However, as
long as the error is systematic and not random, the method can still be used to
monitor the increasing specific activity of the enzyme preparation at different
stages of the purification.
In most cases, however, accuracy is critical. In measuring the stoichiometries
of ligand binding to proteins it is essential to know the true molecular weight
and concentration of the protein. We shall therefore now discuss a number of
practical aspects that should allow the absolute calibration of assays for macro-
molecules, particularly proteins and DNA

2.1.1 Volume
Assay mixtures contain, in principle, three components, i.e. the sample contain-
ing the material to be assayed, a diluent and a 'colour' producing reagent
(which of course may mediate an absorbance change in the UV region of the
spectrum). If the volume of the sample is changed to obtain a larger or smaller
colorimetric response, this change is balanced by using a correspondingly
adjusted volume of diluent to keep the sum of the two components constant. It

186
 SPECTROPHOTOMETRIC ASSAYS

is, however, often convenient, particularly in automated systems, to pre-dilute a

range of different concentrations of a standard for a calibration curve and then
to add a constant volume of different concentrations of sample.
The total volume in the assay is often chosen to be between 2 and 2.5 ml
which will ensure that the entire light beam passes through the solution in
most 1-cm pathlength cells. With any new instrument it is often useful to make
up a solution and add decreasing volumes (e.g. 3, 2.9, 2.8 ml etc.) until a
deviation in absorption (or fluorescence) is recorded which will indicate that not
all of the light is passing through the solution. For visualization, it may also be
useful to place a piece of paper into the cuvette holder and by applying a
wavelength in the visible portion of the spectrum to locate the position of the
beam.
Assay volumes can be scaled down to a final volume of 1 ml or below if semi-
microcells (0.2-0.4 cm wide; 1-cm pathlength) are available. However, it is
important in this case to use a narrow slit width for the light beam, and in some
cases it may be necessary to blacken the sides of the cuvette by insulation tape
or even by using a permanent marker pen. Semi-microcuvettes with blackened
sides can be purchased directly from manufacturers (e.g. Hellma).

2.1.2 Diluent
The diluent contains buffers ions and other reagents to allow colour develop-
ment. For example, the Biuret and Lowry assays for protein consists of alkali,
Cu2+ ions (to chelate to the protein) and tartrate to keep the Cu2+ in solution
under alkaline conditions. The sample volume is rarely allowed to exceed 25%
of the volume of diluent. It is especially important to consider that there may be
compounds in the sample that interact with key components of the diluent.
Common interactions that need to be watched for are chelators in the sample
that may lower the free concentration of metals required in the diluent. Useful
lists of compounds that interfere with the Lowry assay for proteins (see Section
4.3) can be found in (5) and (6).

2.1.3 Order of addition

Although the order of mixing of reagents does not often affect final colour
development it is common sense to add the more labile/sensitive reagents last.
If instability in a reagent produces some colour, then high blanks will be
observed. It is always useful to check that the order of addition does not affect
colour development.

3 End point and rate assays

3.1 End point assays
In these assays all of the metabolite/analyte is converted into some chromo-
phoric compound. These have an advantage in that any variation in factors that
may affect the time to completion (e.g. temperature, amount of coupling

187
T. J. MANTLE AND D. A. HARRIS

enzyme etc.) will not affect the final measurement, since this is made when the
reaction has ended. It follows therefore, that these reactions must be essentially
irreversible, i.e. AG° must be large and negative. If the equilibrium is poised it
may be possible to pull the equilibrium over to the product side by including a
trapping reagent in the assay. For example, in the estimation of ethanol by
alcohol dehydrogenase, the inclusion of semicarbazide in the assay traps the
aldehyde produced as the stable semicarbazone and so drives the reaction to
completion. Even if the reaction is complete, the appropriate absorbance may
not be measured if, as is often the case, the colour reagent produces a product
that is stable for a limited period of time (and hence the colour is only stable for
a limited time). In these cases, it is important to establish a maximum period for
the assay so that decay of the coloured product is not a significant problem. An
additional problem can occur with an enzyme-generated signal if the enzyme-
catalysed reaction is 'leaky'. For example, the assay for ATP involves a 'double
couple' of hexokinase and glucose 6-phosphate dehydrogenase (plus NADP+, see
also Section 6.3.2) in the presence of a high concentration of glucose. Under
these conditions, the reaction is expected to be limited by ATP; however, when
the enzymatically produced glucose 6-phosphate has been consumed there is a
slow increase in A340 (due to NADH) as glucose itself is a poor substrate for G6PDH.
In this case, it is important to extrapolate the progress curve to zero time.

3.2 Rate assays

If the substrate concentration [S] is well below the Km, then the initial rate (v)
exhibits an approximately linear relationship with [S], where the line goes
through the origin with a slope that approximates to V/Km. In contrast to end
point assays, it is vitally important that all variables, such as temperature, pH,
salt concentration etc. are strictly controlled to ensure that the only variable
affecting the rate is [S]. Rate assays can be facilitated by use of continuous flow
analysers, common in hospital pathology laboratories for the routine screening
of metabolites in serum/urine samples from patients. In these, a sample is
pumped for a short period of time into a continuous stream of reagents through
thermostatted delay coils which 'incubate' the reaction mixture at a set
temperature and for a set time before delivering the mixture to a flow-through
cell in a colorimeter or fluorimeter. The injection of sample is followed by an
injection of water or buffer as a wash period and then the next sample is in-
jected to the system. The period of the incubation can be adjusted by changing
the flow rate of the reagents or the length of the delay coil. The sensitivity of
the system may be adjusted either electronically at the detector stage or by
changing the ratio of the flow rates of the sample to that of the reagent. In
theory, the system should record a series of rectangular peaks as the samples,
separated by washes, flow through the detector. However, lateral diffusion
causes these peaks to become rounded to some extent. It is essential to ensure
that the rate of production of the product is linear for the total period of the
incubation.

188
 SPECTROPHOTOMETRIC ASSAYS

NADPH

Figure 1 A hypothetical scheme to illustrate the Tietze reaction. DTNB is represented as

oSSo and reacts with GSH to form the mixed disulphide GSSo. This appears to be a
substrate for glutathione reductase and is reduced to regenerate GSH, which then enters
another round with DTNB, continuing to produce the yellow anion TNB~ at a rate that is
dependent on the initial concentration of GSH.

3.3 Rate assays involving amplification

There are a number of cases where 'rate assays' involve amplification. A classic
example is the method of Tietze (1) to measure glutathione (GSH). This employs
the enzyme glutathione reductase which catalyses the NADPH-dependent re-
duction of oxidized glutathione (GSSG) to form GSH. Gluathione reductase also
appears to reduce the mixed disulphide formed when GSH reacts with Ellman's
reagent (5,5'-dithiobis-2-nitrobenzoic acid; DTNB). In this assay, GSH initially
reacts with Ellman's reagent (5,5'-dithiobis-2-nitrobenzoic acid; DTNB) liber-
ating the yellow 5-thio-2-nitrobenzoate anion (TNB~) (which is stabilized as the
quinonoid structure) and the mixed disulphide between glutathione and DTNB.
The assay appears to work because the mixed disulphide, a substrate for gluta-
thione reductase, is reduced back to GSH producing more of the yellow TNB".
The reaction cycles as shown in Figure 1, yielding an amplified signal (at 412 nm)
due to the release of the yellow TNB" ion.

4 Spectrophotometric assays for proteins

The measurement of protein concentration is an everyday occurrence in most
laboratories working in the bimolecular sciences. An accurately determined
protein concentration is an essential piece of information for many types of
experiments that require a term for molarity of a protein, to monitor the
various stages of a protein purification, to determine the amount of material to
be added to an ELISA or coupled assay, or to normalize a number of different
biological samples (e.g. number of cells or amount of cell extract). This type of
assay must be straightforward and robust. It is also important that the assay can
be used for a range of physically distinct protein preparations. For example, one

189
T. J. MANTLE AND D. A. HARRIS

may wish to measure the amount of protein in an clectrophoretically homo-

genous preparation, in a subcellular fraction {e.g. mitochondria! suspension),
whole cells or crude tissue homogcnate,

For rapid, and generally fairly accurate estimations of protein concentration in

soluble preparations the absorbance at 280 nm gives a good guide. This is based
on the fact that most proteins have approximately the same fraction of trypto-
phan residues (which contributes the most to the A280 for most proteins, sec
Section 63, Chapter 2). It is generally found that an A280 of 1 is equivalent to a
protein concentration of about 1 mg/ml. It is important not to conduct such
measurements in solutions that have a low M,. component that absorbs strongly
at this wavelength (e.g. ATP or ADP) or with pure proteins with abnormally low
fractions of tryptophan. Additionally, the A215 can be utilized since this
measures absorbance due to peptide bonds. In this case an A 215 of 0.066 is
equivalent to a protein concentration of 1 mg/ml. However, at this wavelength
additional contaminants absorb, limiting its general usefulness,
If the material is contaminated with a UV-absorbing material whose spec-
trum is known, it may be possible to correct for the contaminant by measuring
the absorbance at two wavelengths. A well-known example is protein contamin-
ated by nucleic acids (or nucleotides), when the following equation can be used:
rng/ml protein = 1.55A280 0,76A260 2

Protocol 1
A Bluret assay to determine protein concentration
Reagents
• Dissolve 1.5 g of CuS04.5H2O and 6 g • To this solution add 300 ml 10% (w/v)
sodium potassium tartrate (containing sodium hydroxide, mixing during the
four waters of crystallization) in 500 ml addition.
boiled and then cooled distilled water. * Finally add 200 ml distilled water.
This solution is stable for several months at room temperature.

Method
NB. The sample should contain 0.5-4 mg protein in 0.2 ml water or buffer.
1 Construct a standard curve for the assay by adding 2.3 ml of the reagent (see above)
to each of a number of samples (0.2 ml) containing between 0.5 and 5 mg BSA.
2 Add 2.3 ml of the reagent to 0.2 ml of the protein solution being assayed.
3 Leave the mixtures at room temperature for 2 h (alternatively heat to 80°C for 5
min, cool rapidly and stand at room temperature for 10 min).
4 Read the A540 of each sample against a blank of water or buffer plus reagent. Note
that 1 mg of protein gives an A 540 of approximately 0.1.

190
 SPECTROPHOTOMETRIC ASSAYS

The Biuret method

This assay is not affected by the variation of the amino acid content of proteins
as it is based on the complex of peptide bonds with Cu2+ in alkaline solution
and is robust, albeit the least sensitive method, with a useful range of 1 to 5 mg.
The colour is stable for several days if the solutions are kept dark. A number
of substances interfere with the assay including high concentrations of ammonia
(which complexes with the Cu 2 ~) and reducing agents (including reducing
sugars) which reduce the Cu 2 '' to Cu , It may be necessary to remove soluble
materials that interfere with the assay. One of the easiest methods to use is to
precipitate the protein using ice-cold trichloroacetic acid (TCA). It is usual to add
an equal volume of 20% (w/v) TCA to the sample, leave the mixture on ice for 10
min and then pellet the precipirnte by centrifugation. The pellet may be dried
in air or washed twice with ice-cold acetone to remove residual TCA and assist
the drying. The pellet from this procedure (or any other insoluble protein) may
be solubilizcd by the addition of 0,2 ml 0.25% deoxycholate in 0.1 m NaOH,
heating to 80°C for 5 min, allowing to cool and then proceeding at step 2 above.

4,3 The Lowry method

This method (7) has far greater sensitivity than the Biuret and has a working
range of 5 to 40 ug, which probably explains why this reference is a citation
classic, it suffers from greater interference (5, 6) than the Biuret method and
also greater variability between proteins. The method uses two 'reagents', an
alkaline copper reagent (similar to the Biuret reagent) and the Folin reagent
(phosphomolybdate and phosphottingstate). It appears that after the formation
of an initial copper-protein complex that there is some amino acid side chain
mediated reduction of the Folin reagent (8).

Protocol 2
A Lowry assay to determine protein concentration
Reagents
• 1% CuSO4.5H2O (A) • 2% sodium carbonate in 0.1 M NaOH (C)
• 1% sodium potassium tartrate.4H2O.(B) • Folin-Ciocalteau reagent (D)
These reagents are stable for several months at room temperature with the exception of
the Folin-Ciocalteau reagent which must he stored at 4°C, The working solutions must
be freshly prepared by mixing equal volumes of reagents (A) and (B) and then diluting
this mixture with 98 volumes of reagent (C) to give the alkaline copper solution. Reagent
(D) must be diluted 1:1 v/v with distilled water before use.
Method
NB, The sample should contain 5-40 ug protein in 0.2 ml water or buffer.
1 Construct a standard curve by adding 2.1 ml alkaline copper solution to each of
several samples (0.2 ml) of solution containing between 1 and 40 ug of BSA,

191
T. J. MANTLE AND D. A. HARRIS

Protocol 2 continued

2 Add 2.1 ml of the alkaline copper solution to 0.2 ml of the protein solution to be
assayed.
3 Leave the mixtures for 10 min.
4 Add 0.2 ml of the diluted Folin-Ciocalteau reagent to each sample and vortex
immediately.
5 Leave the mixtures at room temperature for 1 h,
6 Read the A750 of each sample against a blank of water or buffer plus reagents. Note
that 10 ug of protein gives an A750 of approximately 0.1.

As with the Biuret assays it may be necessary to remove soluble materials

that interfere with the assay. Common interfering agents include a number of
common buffers (e.g. IIEPES) and neutral detergents (e.g. Triton XI00). Indeed
the Folin method is notorious for the number of compounds that interfere
(5, 6). One of the easiest approaches is to precipitate the protein using ice-cold
trichloroacetic add (TCA) as described above under the Biuret assay. A number
of variations on the original method have been developed to overcome
interference, particularly by detergents in membrane-bound proteins (9).

4,4 The bicinchoninic assay

This is a modification of the Folin method, without the Folin reagent! In this
case the Cu' formed does not mediate reduction of the Folin reagent but is
chelated by bicinchoninic acid.

Protocol 3
A bicinchoninic assay to determine protein concentration
Reagents
• 1% bicinchoninic acid, 0.16% sodium tar- • A solution of 4% CuS04.5H2O in distilled /
trate, 2% sodium carbonate, 0.95% sodium deionized water (A)
bicarbonate in 0.1 M NaOH (pH 11.25) (B)
Solutions (A) and {B) are stable for several months at room temperature. To use, mix reagents
(A) and (B) in a ratio of 1:50 to form the bicinchoninic reagent.
Method
NB. The sample should contain 5-40 ug protein in 0.2 ml water or buffer.
1 Construct a standard curve by adding 2.3 ml bicinchoninic reagent to samples of
0.2 ml of BSA (1-40 ug) in water.
2 Add 2.3 ml of the reagent to 0.2 ml of the protein solution to be assayed.
3 Leave the mixtures at room temperature for 2 h (or heat at 60°C for 30 min).
4 Read the A562 of each sample against a blank of water or buffer plus reagents. Note
that 10 ug of protein gives an A562 of approximately 0.1.

192
 SPECTROPHOTOMETRIC ASSAYS

4,5 Dye-binding assay

On deprotonation, Coomasie Brilliant Blue G250 exhibits a blue colour that can
be measured at 595 nm This forms the basis of a quick, convenient and sensi-
tive assay, as deprotonation is stabilized when the dye is bound 10 protein (10),

Protocol 4
A dye-binding assay for the estimation of protein
concentrations
Reagents
• lOO mg of Serva Blue G dissolved in 50 ml • Concentrated phosphoric acid (85%)
ethanol), then diluted to 500 ml with diluted 1:5 v/v with water (B)
water (A)
Solutions (A) and (B) are mixed in a 1:1 ratio on the day of use to produce the working reagent.
Method
NB. The sample should contain 2-20 ug protein in 0.2 ml water or buffer.
1 Construct a standard curve by adding 2.3 ml of the working reagent to each of
several samples (0.2 ml) of BSA containing between 1 and 20 ug of protein.
2 Add 2.3 ml of the reagent to 0.2 ml of the protein solution being assayed.
3 Mix the solutions and read the A595 of each one immediately. Do not leave the
mixture for periods longer than 30 min. Note that 5 ug protein gives an A595 of
approximately 0.1.

The dye-binding assay docs not accommodate SDS or deoxycholate and is

therefore not useful for proteins solubilized in these detergents. However, the
assay will operate with 0.2% Triton X-100.

4,6 Fluorimetric assay

The sensitivity of protein assays can be increased by several orders of magni-
tude (sensitive down to 10 ng protein) by using fluorescence detection methods.
A number of compounds have been developed. One, the proprietary
NanoOrange™ (available from Molecular Probes), is a dye that binds non-
covalently to proteins with a large enhancement of fluorescence at 590 nm, and
this can be used in an assay very similar to the colorimetnc dye-binding assay
described above.
There are also a number of reagents (originally developed for detection in
liquid chromatography ) which react with organic amines with a large increase
in fluorescence, and which can thus be used to quantitate proteins in appro-
priate buffers. These include o-phthaldialdehyde (11), 3-(4-carboxybcnzoyl)-
quinoline-2-carbodialdehyde (CBQA) (12) and fluorescamine (a heterocyclic
dione) (13). All are readily adaptable for use in microplate format.

193
T. J. MANTLE AND D. A. HARRIS

Protocol 5
A fluorescence assay for the estimation of protein
concentrations
Reagents
• 7.5 mg of fluorescamine dissolved in 25 • Borate buffer pH 8.5 (3 g boric add + 4.8g
ml dry acetone (A) sodium tetraborate in 200 ml water) (B)
Solution A should be prepared on the day of use, in a dry test tube, and contact with moisture
(e.g. wet pipette tips) should be avoided,
Method
NB. The sample should contain 0.1-2 ug protein in 0,2 ml water or buffer.
1 Construct a standard curve by adding 2.1 ml borate buffer to each of several
samples (0.2 ml) of BSA containing between 0.1 and 2 ug of protein.
2 Add 2.1 ml of buffer to 0.2 ml of the protein solution being assayed.
3 Mix on a vortex mixer and, while still mixing, add 0.2 ml fluorescamine solution.
4 Read the fluorescence of each sample after 2 min, using Xex = 390 nm and \sm =
465 nm

Amines interfere with this assay; aliphatic amines yield fluorescent products
with these compounds, while ammonia, although not giving a fluorescent
product, will combine with (and hence use up) the reagent. Unlike the other
methods for protein determination, these reagents will give responses with
peptides in addition to proteins. Whether or not this is advantageous depends
on the requirements of the investigator,

5 Spectrophotometrlc assays for nucleic acids

DNA and RNA can be prepared from cells or tissue extracts using phenol as a
protein precipitant and then precipitating the nucleic acid using cold cthanol.
The nucleic acid can be dissolved in dilute salt normally 10 mM Tris/EDTA, pH
7.5. The bases of nucleic acids absorb at 260 nm and this gives a useful method
(akin to measuring protein by monitoring the A2S0) for estimating the con-
centration of nucleic acids. A solution of nucleic acid at 1 mg/ml gives an A260 of
approximately 20. It is useful to remember that a 1 mg/ml solution of protein
gives an A280 of approximately 1.
More sensitive methods are based on measuring the fluorescence enhance-
ment of nucleic acid-dye complexes such as those used for staining DNA in elec-
trophoresis gels. The dye Hoeschst 33258 (14) can detect double-stranded DNA
down to lOO ng, and YOYO-1 and the proprietary reagent PicoGreen™ dsDNA are
sensitive down to 50 pg. All are available from Molecular Probes. OliGrcen™
ssDNA has a similar high sensitivity for single-stranded DNA. Unfortunately,
there arc no dyes whose fluorescence is specifically enhanced by RNA.

194
 SPECTROPHOTOMETRIC ASSAYS

6 Enzyme-based spectrophotometric assays

6.1 Some general points on assay design
Most enzyme-based assays involve an analyte that is of low molecular weight, so
that large molecular weight material (particularly proteins, some of which may
exhibit an activity that uses the analyte) is generally removed with a precipi-
tating step. This often involves acid treatment (TCA or perchloric acid proving
particularly effective), metal ions (e.g. ZnCl2) or organic solvents (e.g. ethanol).
The addition of acid is an effective precipitant of proteins but will not dis-
criminate between the unwanted proteins and the coupling enzymes which
will then be used, so it is vital that the acid is removed. TCA can be conveniently
removed by extraction into ether or acetone, while perchloric acid can be
neutralized with KOH which provides two functions: the first is neutralization
and the second the precipitation of insoluble potassium perchlorate.
The components of the assay will normally be in a final volume of 2.5 ml
although scaling down can normally be achieved for smaller cells or sample
holders. Under these conditions it is usually desirable to dilute the sample
extensively into the assay mixture so that any small effects on buffering
capacity, ionic strength or potential inhibitors are minimized.
It is important to drive the reaction essentially to completion. For many
coupling enzymes that are linked to ATP (e.g. hexokinase) this is not a problem.
Some reactions, however, are at thermodynamic equilibrium and need some
thermodynamic assistance to be dragged over to completion. A classic example
of this type of strategy is the addition of semicarbazide to trap ketones in the
thermodynamic sink of the semicarbazone (see above). An alternative is to link
a reversible reaction with an enzyme linked reaction that will provide an
alternative thermodynamic sink.
In addition to the above, many coupling enzymes require additional factors,
such as Mg2* for most kinases and K+ for pyruvate kinase.

6.2 Amount of enzyme required

Various authors have considered the amount of enzyme required for an end
point assay (15-17). While it is advisable not to be too wasteful, most common
coupling enzymes are not too expensive and a few preliminary experiments
will soon determine the minimum amount needed to give reliable data in a
reasonable time. Putting the problem quantitatively, our ideal amount of
enzyme should lead to the reaction being 'virtually' complete (from 100% to 1%
of substrate remaining) in a reasonable time (say, 10 min). Since the enzyme
will slow down as the substrate is used up, we cannot just calculate the
amount of enzyme needed from its given activity (V) divided into the substrate
available; instead we need to consider the integrated rate equation for the
process. Remember that V is a measure of the amount of enzyme added (V =
kcate0) and so when we choose the amount of enzyme to add, we choose a V
value.

195
T. J. MANTLE AND D. A. HARRIS

For any enzyme reaction

ds Vs
v= - — = 3
dt Km + s
in which v is the rate, and s the substrate concentration. As the substrate is used
up, and the reaction approaches completion, s « Km, so that

~~dt = K^
Integration gives us the total time for given change in s

so $ Km o
or

For a decrease in s from 100% to 1% , in a time of 10 min (i.e. values we con-

sidered reasonable above ), the following expression must hold
100 V
2.3 log — = — 10 7
1
K-m

or
V/Km = 0.46 min-1 8
or
V = 0.46Km min-1 9
This equation tells us that, for a reaction to go to 'completion' in 10 min, the
amount of enzyme added should be an amount able to change the concen-
tration of substrate in the assay by 0.46Km per minute. If Km = 1 mM, for
example, for 99% completion in 10 min
V s 0.46 mM min-1 (> 0.46 mmol litre0-1 min-1) 10
The concentration of enzyme is conventionally given in U/ml, where 1 U is
the amount of enzyme that will convert 1 umol substrate/min at saturation. If
we add 1 U of enzyme to an assay volume of 3 ml, for example, then V = 1/3
umol/ml/min = 0.33 mmol litre-1 min-1. Thus, if Km = 1 mM, for V >: 0.46 mmol
litre-1 min-1, we need to add s 0.46/0.33 Units of enzyme if the assay volume is
3 ml.
These calculations are useful 'rangefinders', but a number of points should
be borne in mind. It is important to determine experimentally the activity of
the enzyme in the assay medium used. The units of activity that accompany
commercial preparations are those measured under optimal conditions (as
described by the manufacturer), and these conditions may not be those used in
the assay medium being used. Also note (as set out above) that the amount of
enzyme required will increase with increased assay volume. The reaction will,

196
 SPECTROPHOTOMETRIC ASSAYS

of course, proceed faster if you add more than the calculated amount and this
may be convenient if the cost is not too great.
Commercial preparations of many enzymes are supplied as suspensions in
high concentrations of ammonium sulphate or may be stabilized with glycerol.
It is often, though not always, necessary to remove these 'protectants' before
using the enzyme. This can be achieved by dialysis or gel fitration (this latter
method will not work with ammonium sulphate suspensions). It is important to
check the stability of the resultant enzyme stock solution. Some enzymes can
be stored frozen as aliquots that may be used once and then discarded. Some
may stand repeated freezing and thawing (this is unusual), while others may
retain considerable activity at 4°C. It is worth investing some time to establish
the best conditions for storing stock enzyme solutions and it is also advisable to
check the activity of stocks on a regular basis. This is as useful a check on the
substrates and buffers as on the stability of the enzyme solutions.
Remember, more enzyme, more activity, so that if the activity of a stock
solution is declining on storage it may still be possible to add sufficient units of
activity simply by adding more from the stock solution. This, of course, rather
depends on the use of small volumes of concentrated stock enzyme solutions in
the final assay mix.

6.3 Determination of glucose—a comparison of two

methods
6.3.1 Peroxidase coupled method
Peroxidase catalyses a number of reactions coupling hydrogen peroxide break-
down with dye formation. This reaction has found favour in immunoblotting
where chloronaphthol reacts with hydrogen peroxide to form a purple-blue
product. To use the peroxidase reaction to measure glucose an additional enzyme
is required that couples glucose oxidation to hydrogen peroxide production.
This is the flavoprotein glucose oxidase, and the hydrogen peroxide produced is
coupled to the formation of a green dye using the colourless leuco-dye 2,2'-
azino-di-(3-ethylbenzthiazoline) 6-sulphonate [ABTS]. This simple format can be
used with other enzymes that are specific for the substrate they oxidize and
produce hydrogen peroxide as a product (for example, galactose oxidase, D-
amino acid oxidase, monoamine oxidase etc.).
It is important to note that the order of addition of assay components may or
may not be important, but should always be considered (see Section 2.1.3). In
this case hydrogen peroxide is not stable so that the enzyme that produces this
(glucose oxidase) is added last, to start the reaction.
6.3.2 Glucose 6-phosphate dehydrogenase (G6PDH) coupled
method
This protocol couples hexokinase to G6PDH. The hexokinase catalyses the ATP-
dependent formation of glucose 6-phosphate from the glucose added, while the
G6PDH takes the phosphorylated product and catalyses the NADP+-dependent
formation of 6-phosphogluconolactone.

197
T, J. MANTLE AND D. A. HARRIS

Protocol 6
An assay for glucose using glucose oxidase
Reagents
• 1 mM ABTS in 0.1 M sodium phosphate pH 7 • Peroxidase (20 U/ml) in 0.1 M sodium
• Glucose oxidase (100 U/ml) in 0.1 M sodium phosphate pH 7
phosphate pH 7
These reagents are stable for one month at 4°C.
Method
NB, The sample should contain 1-10 nmol glucose in 0,2 ml.
1 Construct a standard curve by adding 2.1 ml ABTS to each of 0,2 ml samples of
known concentrations of glucose (1-10 nmol glucose in 0.2 ml water).
2 Add 2.1 ml of ABTS reagent to 0.2 ml of the sample of glucose to be assayed.
3 Add 0.1 ml peroxidase to all of the above samples.
4 Preincubate for 5 min at 37°C.
5 Start the reaction by the addition of glucose oxidase (0.1 ml),
6 Read the A420 of each sample after 10 min against a blank of reagents with no
glucose added.

Protocol 7
An assay for glucose using glucose 6-phosphate
dehydrogenase
Reagents
• 200 mM Tris/HCl pH 8, containing 2 mm • 100 mM NADP'. In contrast to NADH and
MgCl2 and 0.2 mM EDTA NADPH which should not be kept in acid
* 200 mM MgATP. This is prepared by solutions, NADP" should be stored in
dissolving ATP (1 mmol) in 4 ml MgCl2 buffer below pH 7. As phosphate inhibits
(250 mM) and titrating this solution to pH G6PDH Tris/HCl pH 6.5 is suitable.
7 using 1 M KOH. The final volume should • G6PDH (20 U/ml)
be brought to 5 ml. • Hexokinase (100 U/ml)

Method
NB. The sample should contain 40-400 nmol glucose in 0.2 ml.
1 Add 2.1 ml Tris buffer to 0.2 ml samples of glucose (either containing known
amounts between 40-400 nmol of glucose for setting up a standard curve or the
unknown sample being assayed).
2 Add 0.05 ml MgATP and 0.025 ml NADP" to each sample.
3 Add 0.1 ml G6PDH enzyme, mix the components and preincubate at 37°C for 5 min.

198
 SPECTROPHOTOMETRIC ASSAYS

Protocol 7 continued

4 Read the initial A340 and then add 0.025 ml hexokinase.

5 Monitor the reaction until there is no further change in A340
6 Take the difference between the two A340 values (AA 340 ).
7 Read the AA340 against a blank of reagents with no glucose added.
8 To calculate the amount of glucose in
2.5
|o,mol = AA340 X - l1
6.22

If non-trivial amounts of glucose 6-phosphatc arc present in the sample there

will be a significant initial A340 due to the presence of this material. If the amount
of glucose present in the same sample is relatively low, then the AA340 may be
difficult to measure against the high initial reading. In this case the peroxidasc-
couplcd reaction may be more suitable. Indeed the economics (ABTS is far
cheaper than NADP'') and sensitivity of the peroxidase-coupled reaction means
that this is often the method of choice for glucose. However, as elaborated by
Harris (18), the hexokinasc/GSPDH coupled method has advantages for measur-
ing ATP and NADP~, and can be made more sensitive by using fluorescence
measurements to detect NADPH. The cost may also be reduced by using a
bacterial G6PDH which uses NAD* as the coenzyme instead of the more
expensive NADP" (enzyme available from Boehringer Mannheim),

7 Luminescence-based assays
Measurement of luminescence provides the basis of highly sensitive assays
for ATP, NADPH and H2O2. The production or disappearance of ATP can be
measured by following light emission in the presence of fire-fly lucifernse which
catalyses the following reaction:
ATP + luciferin - O2 -» oxyluciferin PPi + CO2 + AMP + light 12
Likewise, NADPH can be measured using the bacterial luciferasc which
catalyses the following reactions:
NADPH + H' - FMN -> NADPH + FMNH2 13
FMNH2 + RCHO + 02 - H20 + RCOOH + light 14
In which RCHO is a long chain aldehyde (C8 to C12), Finally, earthworm lucifer-
ase can be used for determining H202. Further details can be found in (18).
Conventional spectrophotometers (which are configured to measure the dif-
ference between incident and transmitted light) do not normally show sufficient
sensitivity for determination of bioluminescence. Digital fluorimeters and scin-
tillation counters can often be adapted for measurement of luminescence but.

199
T. J. MANTLE AND D. A. HARRIS

most conveniently, a dedicated luminometer (which requires neither a mono-

chromator or light source) should be used.

8 Spectrophotometric assays of enzymes

8,1 Some elementary enzyme kinetics
Having established a spectrophotometric assay for enzyme activity (see below),
it is vital before setting off to do any detailed measurements of enzyme activity
to check two factors. The first of these is that true initial rates are being
measured. A plot of product (P) appearance (in this case some spectral change)
against time (t) must be linear over the period of measurement to measure true
initial rates (the slope of the [P]/t curve is, by definition the initial rate v). There
are a number of reasons that such [P]/t curves may deviate from linearity:
1. The substrate has simply been used up. This can be checked by calculating
the amount of product produced and determining whether a significant
fraction of the substrate has been used. Furthermore, some reactions rapidly
approach equilibrium (perhaps when only a small amount of substrate has
been used up) and, if this is the case, the reaction will not proceed without
some additional system for removing the product.
2. A build up of inhibitory product may cause deviation from linearity. There
are numerous examples of this phenomenon, which is hardly surprising as
by definition most products have dissociated from the enzyme surface and
are therefore capable of binding to the enzyme or one of the enzyme-
substrate or enzyme-product complexes.
3. The enzyme may be unstable under the conditions of assay. Selwyn (19) has
devised a method to check this possibility, where a plot of product con-
centration against the product of time and the enzyme concentration (et) is
made, and should give the same curve whatever initial concentration of
enzyme is used. If, however, the enzyme is unstable, the concentration of
active enzyme will be time dependent and in such cases the graphs of [P]
versus et will give different curves for each initial concentration of enzyme.
4. Another component of the assay other than the enzyme may be unstable.
This may be checked by incubating the assay mixture without each of its
components in turn for an appropriate period of time, and then to start the
reaction by the addition of the 'missing component*. If the initial rate is
identical, irrespective of which component is added last, then instability can
be ruled out. Conversely, if the initial rate is lower when one component is
included in the pre-incubation but not when it is added last, this indicates an
unstable component.
Having established that genuine initial rates are being measured it is im-
portant to establish that the initial rate (v) varies in a linear way with the enzyme
concentration, e. Doubling the enzyme concentration should double the initial
rate (see Chapter 6). With enzymes that exhibit concentration-dependent

200
 SPECTROPHOTOMETRIC ASSAYS

polymerization/depolymerization where the polymer is more or less active than

the monomer, one may see v versus e plots show significant deviation from
linearity curving upwards or downwards, respectively (20). This is not often a
problem but should be checked, particularly with enzymes that have not been
the subject of earlier detailed kinetic study. For a more detailed review on the
principles of enzyme assays and kinetic studies see Chapter 6 and (15-17).

8.2 Continuous assays

These are the most straightforward of enzyme rate assays and are applicable to
all of those enzymes that utilize NADH or NADPH. The spectral change (AA340) is
easily monitored by connecting the output from the spectrophotometer to a
recorder to give a continuous measure of the reaction rate. Although the slopes
on the recorder traces can be used directly as a measure of the initial rate (albeit
in arbitrary units) it is best to immediately translate the change in A340 per unit
time into (umol product/min/mg protein (the generally accepted units for initial
rates). Knowing the molar extinction coefficient for NADH (6.22 X 103 M-1 cm-1)
the AA340 is easily converted into a concentration (uM) and, knowing the volume
of the reaction mixture in the cuvette, into an amount in nmol or (xmol.
For some reactions where protons are produced or consumed it is possible,
by working at very low concentrations of buffer, to use an indicator as the
buffer and to monitor spectrophotometric changes in this way. To optimize the
sensitivity of the assay the pH should be close to the pK of the indicator and the
assay should be conducted at the pH optimum for the enzyme, where small
changes in pH have no significant effect on the activity of the enzyme. This may
entail an extensive screening for indicators of the appropriate pKa which do not
inhibit the enzyme. Details of this approach with phosphofructokinase can be
found in (21).

8.3 Stopped assays

It is not always possible to obtain such obliging enzymes and then one is forced
to devise alternative assays. One approach that has been particularly fruitful is
the design of non-physiological (chromogenic) assays which produce coloured
products that are stable and convenient to measure spectrophotometrically. p-
Nitrophenol derivatives of organic acids, phosphoric acid and sugars have pro-
duced excellent substrates for a variety of esterases, phosphatases and glyco-
sidases where the yellow p-nitrophenolate anion liberated is easily measured at
410 nm. The pKa of p-nitrophenol is 6.8 so that even at pH 7 this reaction can be
monitored continuously (albeit not with the maximum sensitivity). However,
for 'acid' phosphatase measurements, where the pH optimum is well below the
pJCa of the p-nitrophenol, the reaction is terminated by alkali and the fully
ionized p-nitrophenolate anion is then measured at 410 nm. In the case of the
arylsulphatases A and B, nitrocatechol sulphate has been found to be a particu-
larly good substrate, and the nitrocatechol liberated is seen as an orange-red
anion when the reaction is stopped by the addition of alkali. In these cases the

201
T. J. MANTLE AND D. A. HARRIS

addition of alkali performs two functions; one to terminate the reaction at the
required time and the second to produce the chromophure.
Another common procedure for stopped assays is the use of molybdate under
acid conditions to complex and spectrophotometrically assay inorganic phos-
phate. This provides a useful alternative to p-nitrophenyl phosphate for phos-
phatase assays, as physiologically relevant phosphate esters can be used. For
example, this assay also provides a useful method for measuring ATPase
activity. Details of the phosphomolybdate assay arc given here (however, see
Section 8.5 for methodologies that utilize plate readers).

Protocol 8
A phosphomolybdate assay for inorganic phosphate
released by ATPase activity
Reagents
200 mM MgATP. This is prepared by ' 100 mM Tris/HCl pH 8, containing 50 mM
dissolving ATP (1 mmol) in 4 ml MgCl2 KC1. 2 mm MgCl2 and 0.2 mM EDTA (A)
(250 mw) and titrating this solution to pH • 40% trichloroacetic acid (TCA) (C)
7 using 1 M KOH. The final volume should • 1% ammonium molybdate in 1 M sulphuric-
be brought to 5 ml. (B) acid (D)
Solutions (C) and (D) are stable for months at room temperature. The MgATP (B) can be
stored at —20 °C in aliquots, that should be used once and discarded. The colour reagent
is prepared on the day of use by dissolving 1 g FeSO4 VH2O in solution (D).

Method
1 Mix 0.2 ml of sample with 0.8 ml of buffer (A).
2 Pre-incubate the mixture at 30 °C for 5 min.
3 Add 0.02 ml MgATP (B) to initiate the reaction.
4 Terminate the reaction after 10 min by the addition of 0.1 ml TCA (C).
5 Add 1.9 ml of colour reagent and read the A700 after 10 min.
NB. The A700 of a blank solution made up of reagents with no enzyme added should be
measured and subtracted from the reading given in step 5 above. A standard curve may be
constructed by using known amounts of inorganic phosphate in 0.2 ml of water and treated as
above. 1 umol phosphate released should give an A700 of approximately 1,

It should he stressed for any stopped assay that, after demonstrating linear
product-time curves (i.e. to show genuine initial rates), in a preliminary experi-
ment, half-time points should be incorporated into the protocol for The highest
and lowest substrate concentration used, to demonstrate initial rate conditions
are maintained. For example, if the rate is linear up to 20 min then 'half-time'
points of 10 min should be included routinely.

202
 SPECTROPHOTOMETRIC ASSAYS

8.4 Coupled assays

In these assays, the initial reaction product is colourless, but a second enzyme is
included to convert this into a product measurable by spectrophotometry. For
example, for an ATPase assay, the product ADP can be used to drive NADH
oxidation (measurable by A340) in the presence of phosphoenolpyruvate +
pyruvate kinase (which generate pyruvate in the presence of ADP) and lactate
dehydrogenase (which converts NADH to NAD+ in the presence of pyruvate).
The procedure is similar to that adopted for metabolite assays (see Section 6.). In
these assays it is vital that the activity of the coupling enzyme does not limit the
activity of the enzyme to be measured. Similar calculations to those described
in Section 6 have been suggested (15-17), but it is generally advisable to moni-
tor the reaction with a range of concentrations of the coupling enzyme, and to
use an experimentally determined amount that is clearly not rate limiting under
any condition measured and which shows no significant lag. It is important to
note that the conditions employed will generally be those that are optimal for
the enzyme to be assayed and that under these conditions the coupling enzyme
may not be operating under optimal conditions. Thus it is important to demon-
strate experimentally that the amount of coupling enzyme added is sufficient.

8.5 Plate readers

Several established protocols have been adapted for 96-well plate readers in-
cluding catalase, hyaluronidase, acetylcholinesterase, protein phosphatases and
membrane-bound ATPases (22-26). In several instances these have involved
novel protocols that are well suited to the ELISA format. For example, a sensi-
tive, rapid microtitre-based assay for hyaluronidase activity was described by
Frost and Stern (23). The free carboxyl groups of hyaluronan are biotinylated in
a one-step reaction using biotin-hydrazide. This substrate is then covalently
coupled to a 96-well microtitre plate. At the completion of the enzyme reaction,
residual substrate is detected with an avidin-peroxidase reaction that can be
read in a standard ELISA plate reader. Because the substrate is covalently bound
to the microtitre plate, artefacts such as pH-dependent displacement of the
biotinylated substrate do not occur. The sensitivity permits rapid measurement
of hyaluronidase activity from cultured cells and biological samples, with an
interassay variation of less than 5%.
The standard protocol for measuring inorganic phosphate using variations of
the Fiske and Subbarow method were described in Section 8.3. There have been
a number of applications described recently that have adopted a plate-reader
format (see, for example, 25-27). With careful planning most assays that use
colorimetric reagents (e.g. those for measuring protein concentrations described
in Section 4) can be modified to facilitate the use of a plate reader.

8.6 Centrifugal analysers

These analysers use centrifugal forces to mix samples and reagents that are held
in wells arranged in concentric circles about the axis of rotation of a spinning

203
T. J. MANTLE AND D. A. HARRIS

disc. The contents of an inner well (e.g. sample) are moved radially outwards
passing into other wells holding reagents. After mixing, the sample/reagent
solutions continue to move outwards to cuvettes located towards the perimeter
of the disc. The disc rotates rapidly presenting each cuvette sequentially to an
optical detector. The reactions develop with time and the absorption (or
fluorescence) of each cuvette is measured each time it passes through the
detector to allow the rate of reaction or final value to be determined.
Since the earlier reviews on centrifugal analysers (28, 29) there has been
significant progress in design and formatting. Artuch et al (30) have recently
described an adaptation of some previously reported methods for the Cobas
Fara II centrifugal analyser, using the supernatant of a unique deproteinized
blood sample for the determination of four key analytes (lactate, pyruvate, p-
hydroxybutyrate and acetoacetate) in bioenergetic studies.
The advent of molecular biology and biotechnology have realized and facili-
tated analyses for DNA or RNA sequences and these are now being married to
spectrophotometric assays using centrifugal analysers. New technological ad-
vances have led to the automation of major parts of the assay process.
Automated systems have been developed for amplification and detection of
nucleic acid sequence for infectious agents, using the polymerase chain reaction,
ligase chain reaction, strand displacement amplification, transcription-associated
amplification, and nucleic acid sequence-based amplification. Development of
such automated systems are based on accumulation and integration of new
molecular biotechnology. There has appeared a fully automated PCR system
(COBAS AMPLICOR), which amplifies target nucleic acid sequences, captures the
biotinylated and amplified products on oligonucleotide-coated paramagnetic
microparticles, and detects the products with an avidin-horseradish peroxidase
conjugate. Automated systems provide improvement not only in terms of labour
efficiency but also assay accuracy. Recently, the extraction of a specific sequence
for hepatitis C virus RNA has been automated, and the RNA can be specifically
extracted from serum by hybridization with probe-coated paramagnetic micro-
particles, and then subjected to in vitro amplification (31).

9 Spectrophotometric assays for protein amino

acid side chains
There are a number of useful reagents that allow the stoichiometry and reactiv-
ity of amino acid side chains to be monitored spectrophotometrically. Although
these are not always absolutely specific for the amino acid side chain chosen, the
experimental protocol is straightforward and the inferences can normally be
checked using other methodologies. Note that these reagents are hazardous and users
should follow the appropriate laboratory safety regulations for their use.

9,1 Cysteine
The classic reagent is 5,5'-dithiobis-(2-nitrobenzoic acid) or DTNB, sometimes
known as Ellmans reagent (32), which is absolutely specific for the thiol side

204
 SPECTROPHOTOMETRIC ASSAYS

chain of cysteine residues. By using a molar excess of DTNB over the protein
thiol concentration in the presence of 1 to 10 mM EDTA at pH 8, the thiol
content of a protein can be conveniently monitored spectrophotometrically by
the release of a thionitrobenzoate ion (TNB~) which absorbs strongly at 412 nm
(s412 = 13600 M-1 cm-1). If all of the cysteine residues are accessible to the
reagent and all have a normal pKa (approximately pH 8.7), then the reaction is
rapid as long as the concentration of reagent is in excess. From the extinction
coefficient of the TNB~ ion released, it is possible to calculate the concentration
of thiol modified. Often a fraction of the protein's thiol groups are not available
for reaction because of steric hindrance to the reagent by side chains of neigh-
bouring residues. For monitoring total content, therefore, the protein is com-
monly unfolded before the reaction by the addition of SDS. However, useful
information on protein conformation/folding can be achieved in the absence of
SDS, where kinetic sets of 'fast' and 'slow' reacting thiols (reflecting cysteine
side chains in distinct environments) can be monitored. Other reagents include
4,4'-dithiopyridine that reacts with thiol groups to release 4-thiopyridone that
absorbs light at 324 nm with an e324 =19 800 M-1 cm-1.

9.2 Lysine
Another useful reagent is trinitrobenzene sulphonic acid (TNBS) which is almost
totally specific for amino groups and hence reacts well with the e-amino groups
of reactive lysine residues. It will, however, also react with the thiol group of
cysteine residues, but the S-TNP conjugate is not stable in mildly alkaline
solutions. Typically, 1 to 10 mM TNBS in buffer (pH 7.5 to 9) is used to modify
proteins. Reaction times are normally 1 to 4 h and the reaction is conveniently
monitored at 340 nm (s340 = 14000 M-1 cm-1). The large absorbance change
coupled with the fairly high degree of specificity means that this is a useful
reagent for kinetic analysis of reactive amine groups (33).
An alternative reagent for lysine residues is pyridoxal phosphate which reacts
with unprotonated amine groups to form a Schiff base. The reagent therefore
modifies 'reactive' lysyl amino groups at pH values around 7.5. The unstable
imine can be reduced using sodium borohydride (added in aliquots to a final
concentration of 1 to 10 mM after having adjusted the pH of the medium to 4.5
to 6) to form a stable adduct which absorbs at 325 nm (e325 = 9710 M-1 cm-1) and
emits fluorescence at 425 nm. Further details can be found in (34).

9.3 Tyrosine
Tetranitromethane is a useful (but extremely hazardous) reagent for modification
of tyrosine side chains to give a coloured product that is easily monitored by
spectrophotometry (35). There are a number of side reactions, including the
oxidation of cysteine and the modification of histidine, tryptophan and
methionine residues at higher pH, coupled with an inherent problem of cross-
linking via tyrosine residues. However, if a protein concentration of 2 to 10 mg/
ml can be achieved then this reaction can be carried out at room temperature

205
T. J. MANTLE AND D. A. HARRIS

adding the reagent to a final concentration of 1 to 3 mM (from a stock solution

in ethanol). The reaction is terminated by gel filtration to remove nitroformate
(a co-product), and the nitrotyrosine is measured at 428 nm (s428 4100 M-1 cm-1)
at pH values greater than 10. Alternatively, pH 6.0 can be used where the absorp-
tion spectrum has a maximum at 360 nm and an isosbestic point at 381 nm
with e381 = 2200 M-1 cm-1. It should be noted that tyrosines can sometimes be
disubstituted by this reagent and this may lead to errors in calculations of the
stoichiometry of the reaction.

9.4 Histidine
A similar spectrophotometric determination of histidine residues is based upon
the reactivity of the imidazole side chain with diethylpyrocarbonate (sometimes
termed ethoxyformic anhydride) at pH 6.0 to yield an N-carbethoxyhistidinyl
derivative (36). The reagent will also react to varying degrees (depending upon
pH) with other residues such as tyrosine, lysine or cysteine. The addition of
hydroxylamine to the modified protein will release the carbethoxy group from
modified histidines and tyrosines but not from cysteines or lysines. The reaction
with the imidazole side chain of histidine may be monitored and quantified by
an increase in absorbance at 240 nm (e240 = 3200 M-1 cm-1 (36)). Again caution
must be used as the use of excess reagent can give rise to disubstituted de-
rivatives of histidine which have higher molar extinctions than the mono-
substituted group and this may lead to over-estimation of the equivalents of
histidine in a protein (37). The carbethoxy-imidazole derivative is unstable and
is readily hydrolysed by water. The rate of the reaction increases with pH above
pH 6.0: at pH 7.0 and 25 °C the half-life of the group is 55 h and at pH 10.0 it is
18 min (38). Furthermore, the diethylpyrocarbonate reagent itself is also
hydrolysed by water and the type of buffer used affects the rate of hydrolysis of
the reagent. The half-time for the breakdown of the reagent in 60 mM sodium
phosphate buffer, pH 6.0 at 25 °C, is 24 min (36). Therefore, the rate of modifica-
tion of histidines in a protein will decrease with time due to the loss of the
functional reagent. The concentration of the reagent will need to be 'topped up'
during a modification experiment to ensure that all available residues are
reacted. Care must be taken to keep the stock reagent anhydrous and stock
solutions should be made up in dry ethanol.

9.5 Tryptophan
There are a number of reagents (e.g. 2-nitrophenyl-sulphenyl chloride and 2-
hydroxy-5-nitrobenzylbromide) which react with tryptophan side chains to give
products which can be measured spectrophotometrically. However, these require
an acid pH to allow the reaction to proceed with any degree of specificity (39,40).

10 Concluding remarks
The continuing improvement of both the available chemical reagents and
instrumentation for analytical purposes is leading to increases in sensitivity and

206
 SPECTROPHOTOMETRIC ASSAYS

decreases in the amount of sample required. Micro-colorimeters are commer-

cially available that require only a few ul of sample to fill liquid wave guide
capillary cells with pathlengths of several cm (see World Precision Instruments
http://www.wpiinc.com). The use of spectroscopic methods for the routine
determination of concentrations of small molecules or macromolecules or rates
of reaction is therefore certain to flourish for many years to come.

References
1. Tietze, F. (1969) Anal. Biochem. 27, 502.
2. Bergmeyer, H. U. (1974) Methods of enzymatic analysis. Academic Press, New York,
London.
3. Barman, T. E. (1969 and 1974) Enzyme handbook. Springer Verlag, Heidelberg, New
York.
4. Guilbault, G. G. (1970) Enzymatic methods of analysis. Pergamon, Oxford, London, New
York.
5. Ji, T. H. (1973) Anal. Biochem. 52, 517.
6. Peters, M. A. and Fouts, J. R. (1969) Anal. Biochem. 30, 299.
7. Lowry, O. H., Rosebrough, N. J., Farr, A. L and Randall, R. J. (1951) J. Biol. Chem. 193,
265.
8. Legler, G., Muller-Platz, C. M., Mentges-Hettkamp, M., Pflieger, G. and Julich, E.
(1985) Anal. Biochem. 150, 278.
9. Rodriguez-Vico, P., Martinez-Cayuela, M., Garcia-Peregrin, E. and Ramirez , H. (1989)
Anal. Biochem. 183, 275.
10. Bradford, M. M. (1976) Anal. Biochem. 82, 327.
11. Hernandez, L, Marquina, R, Escalona, J. and Guzman, N. A. (1990) J. Chromatog. 502,
247.
12. Liu, J. P., Hsieh, Y. Z., Wiesler, D, and Novotny, M. (1991) Analyt. Chem. 63, 408.
13. Bridges, M. A., McErlane, K. M., Kwong, E., Katz, S. and Applegarth, D. A. (1986) Clin.
Chem. Acta , 157, 73.
14. Labarca, C. and Paigen, K. (1979) Anal. Biochem. 102, 344.
15. McClure, W. R. (1969) Biochemistry 8, 2782.
16. Tipton, K. F. (1985) In Techniques in the life sciences (ed. K. F. Tipton) Vol. Bl/II,
Supplement BS113, p. 1. Elsevier, Limerick.
17. Tipton, K. F. (1992) Principles of enzyme assay and kinetic studies, in Enzyme assays: A
practical approach (ed. R. Eisenthal and M. J. Danson). IRL Press, Oxford.
18. Harris, D. A. (1987) In Spectrophotometry and spectrofluorimetry: A practical approach (ed.
C. L. Basford and D. A. Harris), pp. 49-90. IRL Press, Oxford, Washington.
19. Selwyn, M. J. (1965) Biochim. Biophys. Acta 105, 193.
20. Hulme, E. C. and Tipton, K. F. (1971) FEES Lett 12,197.
21. Hoffer, H. W. (1971) J. Physiol. Chem. 352, 997.
22. Cohen, G., Kim, M. and Ogwu, V. (1996) J. Neurosci. Methods 67, 53.
23. Frost, G. I. and Stern, R. (1997) Anal. Biochem. 251, 263.
24. Doctor, B. P., Toker, L., Roth, E. and Silman, I. (1987) Anal. Biochem. 166, 399.
25. Fisher, D. K. and Higgins, T. J. (1994) Pharm. Res. 11, 759.
26. Sadrzadeh, S. M., Vincenzi, F. F. and Hinds, T. R. (1993) J. Pharmacol. Toxicol. Methods
30, 103.
27. Drueckes, P., Schinzel, R. and Palm, D. (1995) Anal. Biochem. 230,173.
28. Burtis, C. A., Tiffany, T. 0. and Scott, T. D. (1976) Methods Biochem. Anal. 23,189.
29. Savory, J. (1977) Ann. Biol. Clin. (Paris) 35, 261.

207
T. J. MANTLE AND D. A. HARRIS

30. Artuch, R., Vilaseca, M. A., Farre, C. and Ramon, F. (1995) Eur.}. Can. Chem. din.
Biochem. 33, 529.
31. Albadalejo, J., Alonso, R., Antinozzi, R., Bogard, M., Bourgault, A. M., Colucci, G.,
Fenner, T., Petersen, H., Sala, E., Vincelette, J. and Young, C. (1998) J. Gin. Micrdbiol.
36, 862.
32. Ellman, G. L. (1959) Arch. Biochem. Biophys. 82, 70.
33. Freedman, R. B. and Radda, G. K. (1968) Biochem.]. 108, 383.
34. Lilley, K. S. and Engel, P. C. (1992) Eur.j. Biochem. 207, 533.
35. Sokolovsky, M., Riordan, J. F. and Vallee, B. L. (1966) Biochemistry 5, 3582.
36. Miles, E. W. (1977) Methods in Enzymology, 47, 431.
37. Avaeva, S. M. and Krasnova, V. I. (1975) Bioorg. Khim. 1,1600.
38. Means, G. E. and Feeney, R. E. (1971) Chemical modification of proteins. Holden-Day, San
Francisco.
39. Fontana, A. and Scoffone, E. (1972) Methods in Enzymology 25, 419.
40. Horton, H. R. and Koshland, D. E. (1965) J. Am. Chem. Soc. 87, 1126.

208
 Chapter 8
Stopped-flow spectroscopy
M. T. Wilson* and J. Torres
* Department of Biological Sciences, Central Campus, University of
Essex, Wivenhoe Park, Colchester, Essex C04 3SQ
T
Department of Biochemistry, University of Cambridge,
Tennis Court Road, Cambridge CB2 1QT

1 Introduction
There was a time, fortunately some years ago now, when to undertake rapid
kinetic measurements using a stopped-flow spectrophotometer verged on the
heroic. One needed to be armed with knowledge of amplifiers, light sources,
oscilloscopes etc. and ideally one's credibility was greatly enhanced were one to
build one's own instrument. Analysis of the data was similarly difficult. To obtain
a single rate constant might involve a wide range of skills in addition to those
required for the chemical/biochemical manipulation of the system and could
easily include photography, developing prints and considerable mathematical
agility.
Now all this has changed and, from the point of view of the scientist attempt-
ing to solve problems through transient kinetic studies, a good thing too! Very
high quality data can readily be obtained by anyone with a few hours training
and the ability to use a mouse and 'point and click' programs. Excellent stopped
-flow spectrophotometers can be bought which are reliable, stable, sensitive and
which are controlled by computers able to signal-average and to analyse, in
seconds, kinetic progress curves in a number of ways yielding rate constants,
amplitudes, residuals and statistics. Because it is now so easy, from the technical
point of view, to make measurement and to do so without an apprenticeship in
kinetic methods, it becomes important to make sure that one collects data that
are meaningful and open to sensible interpretation. There are a number of
pitfalls to avoid. The emphasis of this article is, therefore, somewhat different
to that written by Eccleston (1) in an earlier volume of this series. Less time will
be spent on consideration of the hardware, although the general principles
are given, but the focus will be on making sure that the data collected means
what one thinks it means and then how to be sure one is extracting kinetic
parameters from this in a sensible way.
With the advent of powerful, fast computers it has now become possible to
process very large data sets quickly and this has paved the way for the applica-
tion of 'rapid scan' devices (usually, but not exclusively, diode arrays), which
allow complete spectra to be collected at very short time intervals during a
reaction. In these circumstances there is a danger of being swamped by data.

209
M. T. WILSON AND J. TORRES

Whereas some years ago a single mixing experiment in a stopped-f low apparatus
would deliver some tens of absorbance values collected at a given wavelength,
the same experiment may now yield thousands of absorbance values at
hundreds of wavelengths and hundreds of time intervals. Analysis of these large
data sets now uses matrix algebra methods (Singular Value Decomposition,
SVD) to filter signal from noise and to assess, objectively, the number of linearly
independent spectral components which contribute to the total spectral change
accompanying the reaction, and to extract the pseudo-spectra (see later) and
their time courses. Inherent to these methods is great data compression, thus
permitting economic and efficient data storage. Once the number of compo-
nents and their time courses are known global analysis of the data allows one to
fit the kinetic processes to a chemical model by solving, iteratively, the differ-
ential equations which describe it. By so doing, not only can the rate constants
be extracted but also the true (model-dependent) spectra of any intermediates
on the reaction pathway. These analytical techniques are very powerful and
easy to implement; however, they can mislead and it is important to retain
one's chemica1/biochemical common sense when examining the output. It is on
these new techniques of rapid scan spectroscopy and the analysis of data that
we also spend some time.

2 Features of the basic instrument

The stopped-flow spectrophotometer is in essence a simple device designed to
mix two reactants rapidly (<1 ms in conventional instruments) and efficiently
and to transfer the resulting mixture, approx. 100-150 u1 in standard instru-
ments, to an optical cell as rapidly as possible (generally in about 1 ms). This is
achieved by driving together solutions of the reactants from the syringes where
they are stored, by pushing simultaneously their plungers (Figure l). The re-
actants flow through the optical cell and into a further syringe, the stop syringe.
Flow stops when the plunger of this syringe meets an immovable block. The sol-

Light
source
Filter or
monochromator
Syringes filled
with reagents
* Stopping syringe Trigger
r •--

Observation point
in flow capillary
Mixing
Push barrier
for syringe movement Detector

To computer To computer

Figure 1 Schematic diagram of a conventional stopped-flow spectrophotometer.

210
 STOPPED-FLOW SPECTROSCOPY

ution within the optical cell (pathlength, 1) now ages and the time course of the
reaction may be followed by monitoring the absorbance at a suitable wavelength
(A.). The advantage in following the change in absorbance (AA) is that it is directly
related to the concentration (c) of the absorbing chromophore(s) through the
appropriate molar difference extinction coefficient Ae^, i.e. ex pr0duct - e\ reactant

Absorbance changes report directly concentration changes and thus may be

fitted to the rate equations governing the reaction. Data capture is initiated just
prior to the plunger stopping, thus recording some of the 'flow-period', see
below. To achieve this rapid mixing and transfer the syringes are driven
pneumatically at a relatively high pressure, usually about 10 bar, and it is
essential that the system is hydraulically solid, i.e. no air bubbles should be
present as these lead to optical artefacts.

2.1 Instruments available

There are now some excellent modern instruments available that are sensitive,
stable and equipped with good, user-friendly, software; for example:
• Applied Photophysics Ltd
• Hi Tech Scientific
• Olis
(For further information see the List of Suppliers.) These instruments are pro-
vided with extensive instructions on how they should be used and the practical
protocols provided below should be read in conjunction with these. The sup-
pliers have informative websites that are worth visiting. These instruments,
with suitable attachments, can all record spectra rapidly during the time course
of a reaction, the Olis uses what amounts to a rapid scan monochromator which
has the advantage that intense white light is not passed through the reaction
mixture. The first two companies also manufacture small independent units
that can be used in conjunction with any conventional spectrophotometer that
has a reasonably fast A/D converter. These instruments can capture reaction
time-courses taking less than ~1 s to complete.
When one considers the mechanical operations involved in accelerating
stationary solutions to high velocity and the abrupt stopping of the mixture,
they all possess remarkable stability. For single wavelength measurements
using a PMT for detection these instruments are capable of determining rate
constants from absorbance changes of 0.005A. This means that mechanical
'ringing' of the system induced by mechanical vibration is minimal.
The examples we have chosen to illustrate have been obtained using the
Applied Photophysics instrument SX.18MV, which is provided with a PMT
detection system and supplemented with a diode array for collecting spectra
during a reaction time course.

211
M. T. WILSON AND J. TORRES

3 Measurement at a single wavelength

31 Setting up
It is good practice to check from time to time that the stability of the instrument
is within the manufacturer's specification. This is easily done by collecting what
in a conventional spectrophotometer would be termed a 'baseline' by mixing
buffer with buffer. In this, and all subsequent experiments described, the data
arc collected without using any electronic filters to remove noise. It is almost
always best to reduce experimental noise by signal averaging a number of data
sets rather than using damping circuits. Like most 'rules of thumb' there are
exceptions and for very slow reactions with very small amplitude, damping may
be necessary, but in these cases why use a rapid kinetic device?

Protocol 1
To test the stability of the detection system
Method
1 Fill working syringes with water or buffer,
2 Select a wavelength where the lamp output is high and the PMT is most sensitive,
usually somewhere between 500-600 nm.
3 Set the sensitivity on the recording apparatus to a medium range, say full scale 0.2A
units and set the PMT voltage (usually done automatically).
4 Wash the optical cell through up to 10 times by operating the pneumatic drive
system. This should ensure that the system is bubble-free.
5 Once a reasonably flat baseline is obtained set the sensitivity to maximum, collect
five traces and average these with the software provided.
6 Finally collect and average traces using the internal trigger, i.e. traces which do not
involve mechanical mixing, flow and the arrest of flow. These traces allow one to
monitor lamp fluctuations and other electronic noise.

Figure 2 illustrates the results of such a stability test. The electronic/lamp noise
is essentially zero while the noise generated by the mechanical mixing amounts
to -001A If the noise is above this, then one should clean the apparatus by
purging with detergent solutions or as recommended by the manufacturer.
Bubbles that can occasionally prove difficult to flush out may be removed by
passing degassed water through the instrument. Lamp 'ripples' (--2 Hz) can
sometimes be improved by slight lamp realignment.

3,2 Selecting the wavelength for a real experiment

If one is to use the stoppcd-How apparatus in the absorbance mode it is important
to know the spectra of the starting reactants and of the final products in order to
select the best wavelength at which to collect kinetic data. This wavelength is

212
 STOPPED-FLOW SPECTROSCOPY

0.00

0.00 0.02 0.04 0.06 0.08 0.10

time(s)
Figure 2 Baseline measurement to determine the stability of the instrument. Trace a is the
average of three individual time courses, with buffer (0.1 M phosphate) in the optical cell,
recorded by triggering data capture using the internal trigger. This shows the electronic and
lamp noise inherent in the system and demonstrates that this is essentially random noise
equivalent to « 0.001A Trace b shows a similar average captured on mixing buffer with
buffer. This demonstrates that mechanical/flow noise is ~0.001/4 over 100 ms.

generally the one where the absorbance change is greatest on going from
reactants to products. This may not always be the case, and one must sometimes
be prepared to work at wavelengths away from the peak in the difference
spectrum in order to work in regions of the spectrum where the apparatus is
more sensitive.

3,3 The form of a simple progress curve: Making sure the

apparatus is mixing and transferring reactants to the
observation chamber rapidly
Whatever the mathematical form of the time course of the absorbance change
that is reporting the reaction under study, it should commence with a short
time period during which the absorbance remains constant. This represents the
period of time between triggering the capture device and the time when the
plunger of the stop syringe meets the stopping block. This time is usually about
2 ms when mixing is pneumatically driven at pressures —10 bar. The 'flow
period' should not be confused with the dead time of the instrument (see
below) which is generally shorter.
The performance of the pneumatic system, mixing and transfer systems may
be tested by using any reaction with a moderately short half time, say 10-20 ms,
e.g. the reaction between 2,6-dichlorophenolindophenol, DCIP, and L-ascorbate.
Here we illustrate using the reaction between myoglobin and carbon monoxide,
but any reaction taking some 50 ms or less to complete may be substituted:
ki
Mb + CO ^ MbCO 2
K-1
Mb represents deoxy-Mb and MbCO the carboxy derivative.

213
M. T. WILSON AND J. TORRES

Protocol 2
To determine the time between initiation of the signal
recording and the stop of the flow of solution
Reagents and materials
• Prepare a solution of ferric (met) buffer to several cycles of exposure to
myoglobin by dissolving ~2 mg of the vacuum and equilibration with CO (at 101
freeze-dried protein, Sigma Chemical Co., kP pressure) accompanied by vigorous
in 10 ml of buffer (say 0.1 M sodium shaking. Gas equilibrates with a liquid
phosphate pH 7.0}, Add a few crystals (-2 phase faster the greater the interface area
mg) of solid dithionite (Na2S204, BDH) to between them, thus many small bubbles
remove oxygen from solution and to are better than a few larger ones. Draw
reduce the myoglobin to form the ferrous into a syringe a few ml of the CO solution
deoxy species. and add a little sodium dithionite (-2
• Prepare the stock CO solution, 1mM,by mg/10 ml). CO gas is toxic and these
equilibrating water (or dilute buffer) at operations should be performed in a
20 °C with an atmosphere of pure CO at 1 'fume' cupboard.
atmosphere pressure [101 kP). Subject the

Method
1 Load the Mb and the CO solution into the working syringes of the stopped-flow
apparatus. Flush the apparatus through with some five shots to ensure the
solutions have fully replaced water in the mixing chamber and optical cell.
2 Ensure there is no electronic noise filter and collect a further three to five traces,
each of some 400 time points, and average these.

3 and 4 illustrate a typical result of such an experiment. The 'flow

period' is clearly seen occupying the first 1.5 ms. Thereafter, once flow has
stopped, the formation of the carbon monoxy complex is seen.
Should this profile be distorted it may indicate that there is a problem with
the smooth and rapid flow of the reagents, possibly syringes sticking or the
viscosity of the solutions not permitting the solutions to accelerate fully TO the
desired velocity. Alternatively the pressure is low. Check the pneumatic pressure,
clean the system as recommended. Possibly the apparatus needs to be adjusted
to deliver more of each solution by moving the stopping block.

3.4 Measurement of the dead time

The first question one might legitimately ask is 'why bother'?'. Why should one
check the dead time of the apparatus, what bearing has it on the interpretation
of subsequent experimental results? The dead time is defined as the time
required for the reactants, once mixed, to travel from the mixing chamber To
the observation chamber. It is, Therefore, a period of Time in which the reaction
is progressing, but not being recorded. This being so, it must follow that the

214
 STOPPED-FLOW SPECTROSCOPY

0.15

-0.1
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20
Time (s)
Figure 3 Reaction between deoxy myoglobin and carbon monoxide. The time courses show
the binding of CO to deoxy myoglobin (10 uM) as a function of CO concentration (41, 56, 88,
150, 275 or 525 uM). The monitoring wavelength was 421 nm and the temperature 20°C.
The reaction was carried out in sodium phosphate buffer, 0.1 M, pH 7.0. The initial flat
portion of the progress curve (seen better in Figure 4) illustrates the flow period. The inset
gives the dependence of the pseudo-first-order rate constant, Kobs on CO concentration.

0.02

-0.1
0.000 0.005 0.010
Time (s)
Figure 4 Analysis of time course to measure the 'flow time' and the 'dead time'. The first
10 ms of the traces in Figure 3 are shown as individual data points. The solid lines are the
exponential fits to the data extrapolated back to true time zero. The flow period is the flat
initial portion of the curve and, in principle, may have any length and depends on the time
between triggering data capture and the flow stopping. The dead time is the time between t
= 0, given by the point of intersection of the curves, and when flow stops. In this experiment
the dead time and flow period are approximately the same, but need not be so.

215
M. T, WILSON AND J. TORRES

total amplitude of the absorbance change observed is smaller than expected

from static spectroscopy, from which it is always possible to obtain the start and
end points of a reaction, if not the kinetic progress curve connecting these. The
fraction of the amplitude lost depends upon the rate constant tor the reaction
and the length of the dead time. For very rapid reactions, the majority and
indeed the entire kinetic time course may be lost.
The principle of the method by which the dead time is determined is by
measuring the fraction of the expected absorbance change of a well-characterized
reaction that is lost in the time between mixing and observation. This is usually
achieved through the use of a suitable second-order reaction, the rate of which
can be controlled through the concentration of the reactants. This method is
illustrated using the reaction given in equation 2, but any suitable reaction may
be substituted, (e.g. the reaction between 2,6-dichloraphenolindophenol, DCIP,
1 mM and [,-ascorbate in the concentration range 2-120 mM, concentrations
before mixing (2)).

Protocol 3
Measurement of the 'dead time' of the instrument
Reagents
The myoglobin and CO solutions prepared as described in Protocol 2 may be used.
Methods
1 Mix the deoxy Mb with 1 mw CO and record at least three traces (at least 200 points
over 50 ms) and average,
2 Dilute the CO solution with buffer which has been degassed and equilibrated with
N2 and to which a small quantity of solid sodium dithionite has been added. To do
this, without exposure to air, take a known volume of the CO solution into a glass
graduated syringe (partially filling it) and connect this by its nozzle to a similar
syringe containing the anaerobic buffer by using a short plastic tube. Push a known
volume of the buffer into the CO solution. It is usually convenient to mix equal
volumes thereby diluting the CO by a factor of two.
3 Record a further set of traces and average.
4 Repeat the CO dilution step and collect data over the appropriate time range a
further three to four times. After each change of [CO) flush through some five shots
to ensure the new solution has completely replaced the previous solution.

The results of such an experiment are given in Figure 3. As expected for a

second-order reaction the rate depends on the concentration of the reactant in
excess, here CO. The reaction being completed in a shorter time at high CO
concentrations. Moreover it is seen that although the absorbance change must
be the same at all CO concentrations because the myoglobin concentration is
unchanged, the A. measured is smaller at high CO concentrations. This is
because of the loss of absorbance due to the reaction in the dead time.

216
 STOPPED-FLOW SPECTROSCOPY

By extrapolating each curve back into the dead time there should be a time
point where all the curves intersect, giving the true time zero of the reaction.
Figure 4 shows the extrapolation of a set of exponential curves fitted to the data
presented in Figure 3. This is typical for such an experiment. The majority of the
time courses intersect while one is to one side of this point. Nevertheless, the
intersection point for the majority of the lines is a good estimate of time zero.
The average time between this intersection point and the end of the flow period,
i.e. where observation truly starts is approx. 1.3 ms and this is the dead time of
the instrument. This may be made slightly shorter by cleaning the instrument,
making sure the syringes are running smoothly or by slightly increasing the
pressure of the pneumatic drive.

3.5 Amplitude of the signal

It is often essential to measure the amplitude of an absorbance change (AAJ
associated with a reaction in order to calculate the change in concentration of a
given species. For example, one may wish to determine the stoichiometry of the
reaction, i.e. how much product is formed from known concentrations of re-
actants. In principle, this is easy to do. The total amplitude change at a selected
wavelength is given directly by all modern instruments, i.e. from the end of the
flow period to the end point (AA,,). This may be corrected for loss of absorbance
in the dead time (td) to give the true absorbance (AAT). If the reaction follows an
exponential time course this may easily be achieved from the reaction half-time
(t,A) by using
AA,
In =ln2
AAo
which may be transformed to
, AAT
log
Equation 3 yields
In (AAo) = In (AAT) - ktd 5
that gives a straight line on plotting AAo versus k (the first-order rate constant, see
Section 4.1), the slope of which is td> the dead time. (A similar, if more complex
procedure, may be applied if the progress curve comprises more than one
exponential.)
The absorbance change (AAT) may now be converted to concentration change
via the appropriate extinction coefficient, Aex (see equation 1). However, it may
be incorrect to use the value of this coefficient reported in the literature as this
has been determined by conventional spectroscopy. It is often advisable to de-
termine the apparent extinction coefficient in the stopped-flow apparatus using
the instrumental settings employed in the experiment. This is because the
monochromator slit width settings are often much wider than those used in a
conventional spectrophotometer. This is especially the case when one is using
highly absorbing solutions and one wishes to pass more light through the
sample. The result of this is a much wider bandpass, i.e. the beam is no longer

217
M. T. WILSON AND J. TORRES

Table 1 Dependence of the M measured on the monochromator slit widths

Entrance silt (mm) Exit silt (mm) A4 measured ±t ^apparent X 10-*

0.1 0.1 0.106 19.4
0.1 1 0.102 18.7
1 1 0.096 17.6
1 2 0.081 14.9
1 3 0.078 14.3

monochromatic but contains wavelengths on either side of that chosen to

monitor the reaction. This hardly matters if the absorption band is broad, but if
it is sharp, the absorption is averaged and thus attenuated. In such cases, it is
well to measure Aex apparent at the chosen wavelength. Here we illustrate using
as an example cytochrome c but this calibration should be carried out for any
reaction where AAX is required to be known accurately.
Cytochrome c possesses a sharp band at 550 nm when the central iron is in
the ferrous state. The true Aex at this wavelength on going from the ferric
to ferrous form is 20 000 M-1 cm-1. The table reports the AA values measured at
550 nm in a stopped-flow experiment in which 10.9 uM ferricytochrome c was
mixed with 10 mM sodium ascorbate (0.1 M sodium phosphate pH 7.0). The
concentration of cytochrome c after mixing was 5.45 uM and thus the expected
AA was 0.109A. Table 1 gives the values of AA measured at a number of settings
of the monochromator slit width and reports the apparent Aex for these settings.
It is seen that as the slits open the absorbance measured falls because the
monitoring beam becomes less monochromatic. The amplitude of the signal at
the selected wavelength, once corrected for the absorbance lost in the dead
time, should be compared with that expected from static experiments, i.e. the
total absorbance change determined from the spectra of the starting materials
and the spectrum of the reaction mixture at completion. This is an Important check
because if there is a discrepancy this means that a fraction of the absorbance change has
occurred very rapidly within the dead time in a process with kinetics different from those of
the fraction that is observed.

3.6 Assigning a signal

It is not always obvious how an absorbance change monitored at a single wave-
length should be assigned. There is a danger of wrongly assigning such a signal
in the light of preconceptions regarding the nature and mechanism of the
reaction under study.
Suppose, for example, one wishes to determine the kinetics of the appear-
ance of a species, C, formed from the reaction between A and B, i.e.
A+ B ^ C 6
If the reaction is to be followed at a single wavelength, then this is chosen with
reference to the spectra of the components in order to maximize the expected
absorbance change. Suppose that absorbance increases on formation of C, and

218
 STOPPED-FLOW SPECTROSCOPY

that on mixing A and B an increase in absorbance is indeed observed that follows

an apparently sensible time course. It is tempting to assign this lime course to
the process under study, i.e. the production of C. However, one must be careful,
and if the system has not been studied previously it is wise, if one is not to make
embarrassing errors, to confirm this assignment. The way to do this is to moni-
tor the reaction at a number of wavelengths and construct a 'kinetic difference
spectrum'. If this spectrum matches in shape and amplitude that determined in
a conventional spcctrophotometer, then one is justified in assigning the time
course and can thereafter proceed with determination of the kinetic parameters.
We have chosen to illustrate this problem by reference to the reaction
between a ruthenium complex and nitric oxide (NO) because it poses a number
of problems. This reaction is of interest be cause of the role that such complexes
may play as NO scavengers in pathological conditions such as sepsis (3). The
reaction is very rapid and the absorbance change associated with NO binding is
small. This requires that the reaction be carried out at low temperature (here 7
°C) and at equal concentrations of reagents to yield the maximum absorbance
change at the lowest rate. The absorption spectrum of the compound is in the
UV region and approaches the lower end of the range accessible to instruments
equipped with standard light guides. Furthermore, in this region there is the
possibility of interference from side reactions, such as the production of nitrite.
It is essential when studying such a reaction that assignments are carefully
checked. Although this reaction is rather specific and unlikely to be widely
studied, the problem it poses is typical of many reactions and the protocol
below explains how to deal with these.

Protocol 4
Determination of a kinetic difference spectrum
Reagents
• Prepare in 50 mM potassium phosphate Equilibrate NO gas at one atmosphere
buffer, pH 7.4, an anaerobic solution of a (101 kP) pressure and 20°C with anaerobic
ruthenate chelate, here potassium water (see Protocol 2 above for CO). This
chloro [hydrogen(ethylenedinitrilo)rathen- yields a solution ~2 mM in NO.
ate], and dilute this in the same anaerobic Dilute, to a final concentration of 50 mM,
buffer to a final concentration of 50 uM of the stock NO solution with the anaerobic
the complex (Solution A), buffer, using the procedure described for
CO above (Protocol 3),

Method
i Fill two optical cuvettes with the anaerobic solution of the ruthenium complex, and
using one as a sample and the other as a reference take the baseline between 220
and 400 um. Gently bubble a little NO gas through the sample solution and take the
spectrum again. This is the difference spectrum between the Ru-NO complex and
the Ru-chelate (see Figure 5),

219
M. T. WILSON AND j. TORRES

Protocol 4 continued

Transfer the remainder of the solutions to the stopped-flow apparatus that has
been previously flushed through with anaerobic buffer.
Set the monochromator to the wavelength of the maximum absorbance difference
in the spectrum (here -290 nm) and record a kinetic trace (Figure 6).
Record time courses at 5-10 nm intervals, being sure to pass through the isosbestic
point indicated in Figure 5.
Plot the absorbance change (corrected at each wavelength) against wavelength, as in
Figure 5. It is clear that the spectral distribution of the absorbance changes follows
the absorption spectrum (there is a factor of two between the amplitude that is
accounted for by the twofold dilution that occurs in the stopped-flow mixing).

Wavelength (nm)

Figure 5 The spectral transition observed on mixing NO with a potassium

chloro[hydrogen(ethylenedinitrilo)ruthenate). The solid line is the difference spectrum
between the complex in the absence and presence of excess NO. The square points
represent the amplitude of the absorbance changes observed in the stopped-flow experiment
mixing the Ru complex with NO and monitoring at the wavelengths depicted. The factor of
two between the static and kinetic constants comes from dilution in the mixing experiment.

4 Determining rate constants

Time courses can follow a number of mathematical forms depending on mech-
anism. The software that runs stopped-flow spectrophotometers provides fitting
programs to extract rate constants. The operator must however decide the
likely form of the curve and the time range over which the curves should be
fitted.

220
 STOPPED-FLOW SPECTROSCOPY

figure 6 Changes in absorbance of an anaerobic solution of the Ru-NO complex in the

presence and absence of NO. The Ru-NO complex (25 uM) in the absence (a) and presence
(b) of NO (25 (uM) in phosphate buffer (50 mM), pH 7.4 at 7,3°C. The fit is to the appropriate
second-order rate equation 14. The trace is the average of six individual time courses.

4.1 First-order processes

Reactions of the form

give rise to exponential time courses of the following form.

[A] = [A]0e-(kobs)t [B] = [A]0(l - e-<k°bs)t) 8
where kobs = kj + k^; kt and k_t are the first-order rate constants and have units
s-1 and the equilibrium constant, K, is given by K = K1/k-1 As the concentrations
and absorbances are linearly related the absorbance changes used to monitor
the reaction are described by the same equations.
Where the equilibrium lies far to the right, the value of k-t may be ignored
and kobs ~ kj. An example of such a reaction is provided by pH-dependent tran-
sitions, common in proteins, and which may be studied by 'pH jump' experi-
ments. In such experiments, a pH-dependent equilibrium is perturbed by mix-
ing the system equilibrated at one pH in a weak buffer with a strong buffer at a
different pH. An example is afforded by the spectral change induced in cyto-
chrome c by increasing the pH. Here we illustrate with a chemical derivative of
cytochrome c in which the methionine that is normally coordinated to the iron
is carboxymethylated so that it can no longer play this role (4). In this
modification the ferrous haem c exhibits a pH dependent spin state transition
accompanied by a large spectral change.

221
M. T. WILSON AND J. TORRES

Protocol 5
An experiment to determine the effect of a 'pH jump' on
the absorption of carboxymethylated horse heart
cytochrome c
Reagents
• Prepare a solution of ~5 ml of the protein Prepare a stronger buffer at the desired
(—10 uM) in a weak buffer (say 10 mM pH value, e.g. 0.1 M sodium phosphate
sodium acetate pH 5.5), pH7.

Method
i Fill one drive syringe with the protein solution and fill the other drive syringe with
the strong buffer solution.
Set the wavelength to give the maximum absorbance difference for the pH-
dependent change, determined separately as described above.
Discharge several 'shots' to flush out old reagents/buffer and then record and
average the data from several reactions.

The time course approximates closely to an exponential increase in absorption

at 417 nm and has been fitted as such to yield a pscudo-first-order rate constant,
of 35 s-1. In Figure 7 the difference (multiplied by ten and moved upwards by
0.1A) between the experimental time course and the fit is also shown. The re-
siduals are small, indicating a good fit, but also show a non-random distribution
which amounts to a small ripple -~10- 3 A in amplitude which is instrumental in
origin, see Figure 2, The initial few ms are distorted by the 'flow' period.

0.25

0.2

0)
0.15
o
c
n
€o
0.05 -

000 002 0.04 0.06 0.08 0,10

time(s)

Figure 7 Optical transition observed on mixing 10 uM ferrous Cm-cyt c at pH 5.5 (0.01 M

sodium acetate) with sodium phosphate (0,1 M} buffer pH 7, temperature 20°C. The fit is to
a single exponential and the residuals are shown multiplied by a factor of ten and moved by
0.1A for clarity.

222
 STOPPED-FLOW SPECTROSCOPY

4.1.1 Multi-phasic time courses composed of the sum of

exponentials
It is very often the case that time courses are not simple but comprise more than
one exponential phase. This may be because there are parallel first-order reac-
tions proceeding simultaneously. In principle it is very easy to extract the sepa-
rate rate constants and the amplitudes of the absorbance change associated with
each reaction. All instruments are provided with software that will fit experi-
mental data to the sum of exponentials. However, this simplicity of the fitting
procedure belies the underlying difficulty inherent in this analysis. This may be
illustrated by reference to the reaction of Cm-cyt c described above but at pH 9.
Now the time course is the sum of two first-order processes, one fast with t1/2 ~ 2
ms and the other, slower, with t1/2 ~ 10 ms, see Figure 8. The trace is easily fitted,
but in order to do so the user must select the time range over which the fitting
should take place. This selection should be made to ensure that approximately
equal numbers of data points representing the kinetic process are collected (best
done by using a split time base). However, this may not always be easy to judge
and it is best to be aware that rather different values for rate constants may be
obtained depending on the 'fit-range' one chooses as shown in Figure 8.

4.2 Second-order processes

Many reactions of biochemical interest are of the mechanistic form

A+B- or A+B C+ D

i.e. second-order processes.

0.12

Fit Range it fast k slow

2.5 -50ms 32016.9 96 + 1.7
2. 5 -40ms 467 ± 13 12311
2.5 -25ms 515126 13112.9
1.5 -25ms 432 ±20 12413.6
10.00 20.00 30.00 40.00 5000

-0.04
Time(ms)
Figure 8 The same experiment as shown in Figure 7 except the final pH was 9. The inset
table illustrates that with biphasic traces, here two exponentials, the value of rate constants
(k s-1) determined by the fit depend critically on the fit range chosen.

223
M. T. WILSON AND J. TORRES

Integration of the rate equations for the above yield complex and unwieldy
expressions and if true second-order reversible reactions are to be analysed it is
often best to use global analysis to retrieve the rate constants, see below. How-
ever, in many instances the back reaction may be neglected and the reaction
considered as quasi-irreversible. Under these circumstances the rate equation
may be written:
dx
— = k, ([Ao] - x) ([B0] - x) 10

where x is the concentration of the product(s) and [Ao] and [B0] are the initial
concentrations of A and B mixed in the stopped-flow spectrophotometer. The
second-order rate constant (k1; units M-1 s-1) has the form

fct1- n
~ *)
([Ac] - [Bo]) [Ao]([B0] - x)
There are a number of ways to use this equation in the analysis of data but the
simplest way to proceed is to use this equation directly with the fitting program
provided with the apparatus. There is one problem to bear in mind, however, if
such a fitting procedure is used, namely, the routine will be executed using the
measured absorbance values and not the concentration of the reactants. This means
that the second-order rate constants will have the units of AA-1 s-1 and thus
must be converted by use of the appropriate Aex to M-1 s-1.
This equation may be considerably simplified if one conducts the experiment
under conditions in which one of the reactants, say A, is in large excess (>10-
fold over the other). Under these conditions the concentration of the com-
ponent in excess can be considered to remain constant during the reaction, i.e.
[Ao] » x. Under these circumstances x collapses to
x = [B0](l - exp(-kobst)) 12
that has the same form as that describing a first-order process. Here kobs, the
pseudo-first-order rate constant has the form Jcobs = k1[Ao], i.e. the second-order rate
constant multiplied by the concentration of the component in excess.
Under pseudo-first-order conditions ([Ao] » [B0]) reversible second-order
reactions also yield the same exponential forms as above but now
=
^obs M[AO] + k-1 13

Thus under these conditions both k1 and k-1 may be found from experiments in
which the pseudo-first-order rate constant (kobs) is determined at a number of
concentrations of the reactant in excess. The data presented in Figure 3, which
was used to measure the dead time, are the results of such an experiment. In
the inset to Figure 3 the pseudo-first-order rate constants (kobs) derived from
exponential fits to the data are plotted against the CO concentration. The slope
gives the second-order rate constant k1, ~4 x 105 M-1 s-1, and the intercept
k-1, 0.6 s-1. The latter value is rather high compared with the literature value of
0.04 s-1 and illustrates the limitation of the method when k-1 is very small (5).

224
 STOPPED-FLOW SPECTROSCOPY

The rate constant of any first-order process is related to the half-time of the
reaction (i.e. the time taken to complete half the reaction) by the expression
Kobs = 1n2/t1/2y. ~ 0.693/t,1/2.
Essentially irreversible second-order processes may also be analysed easily if
experiments are performed under condition in which the starting reactants are
at equal concentration, i.e. [Ao] = [B0]. Under these circumstances the solution of
the rate equation gives

This simple analytical equation may be used in fitting procedures (see Figure 6).
The rate constant is related to the half-time of such a reaction by the expression
k1, = l/[Ao]t1/2,

4.2.1 More complex processes

First and second-order processes are often coupled together through common
chemical intermediates to form more complex mechanisms. There are numer-
ous examples, e.g. the coupling between second-order complex formation, or
electron-transfer reactions, and first-order protein conformational changes. The
analysis of some of these is given elsewhere (see Chapter 9). Here we consider a
mechanism common to many proteases in which the enzyme and its substrate
come together in a second-order process which leads to acylation of the enzyme
and release of a product, P1. The acyl enzyme is now hydrolysed to yield a
second product, P2, and the original enzyme that enters a second cycle. This
may be written as follows:

It can readily be appreciated that if k1[S] is much larger than k2 then on mixing
E and S there is an initial rapid production of p1, the 'burst', the concentration
of P1 being stoichiometric with the initial concentration of the enzyme [E0].
Indeed as it is often easier to measure [P1]than [E0] this provides an excellent
way to measure the latter and is the basis of active site titrations. Once the
enzyme has been acylated it decays to form P2, and in the presence of excess
substrate the cycle is repeated, but in the second and subsequent cycles the
enzyme cannot react with S faster than the enzyme deacylates. Provided [S] »
[E0] this leads to a linear time course for Pt formation, with rate dependent on
the value of k2.
Where k1[S] and K2 are comparable the situation is more complex and analysis
of the mechanism leads to the conclusions depicted in Figure 9. This diagram
gives the basis of active site titrations in which the enzyme is mixed with

225
M. T. WILSON AND J. TORRES

n =:[£„](*, [S]/fc,[S] + *2)2

[p.;

Time

figure 9 Schematic diagram illustrating the 'burst' seen in the concentration of p1 on mixing a
protease with its substrate. The amplitude of the 'burst' and the rates are determined by the
kinetic parameters given in the text as are the concentrations of the enzyme and substrate.

0.09

-0.01
Time (seconds)
Figure 10 The reaction between rat tissue Kallikrein (5 (J.M) and p-nitrophenylguanidino
benzoate. Kallikrein (5 uM) was reacted with 6.25, 12.5, 25, 50, 100, 250 and 500 (uM p-
nitrophenylguanidino benzoate. The monitoring wavelength was 402 nm where p-nitrophenol,
P! absorbs. The temperature was 25°C.

226
 STOPPED-FLOW SPECTROSCOPY

increasing concentrations of S and the amplitude of the 'burst', n, is measured.

As [S] becomes very large II tends to [E0], kobs to k1[S] and the slope of the linear
portion tends to k2[Eol (6)-
An example of 'burst' kinetics is given in Figure 10 in which rat tissue kalli-
krein is reacted with a substrate, p-nitrophenylguanidino benzoate. In this
system P1 is p-nitrophenol that absorbs strongly at 402 nm allowing the reaction
to be monitored easily. The figure shows the time course at increasing concen-
trations of substrate. At high values of [S], the 'burst' phase followed by a linear
time course is seen. The maximum amplitude, together with the extinction
coefficient for p-nitrophenol, 14000 M-1 cm-1, gives the enzyme concentration
as 5 uM.

5 Multi-wavelength detection: diode array 'rapid

scan methods'
We have concentrated above on aspects of monitoring the kinetics of a reaction
by capturing and analysing time courses at a single wavelength using a photo-
multiplier as detector. In contrast to this a diode array detector can register
simultaneously absorbance changes at a large number of wavelengths. This
feature is especially valuable when a complex reaction, involving a number of
spectral species, is to be analysed. The method relies on passing white light
through the optical cell where the reaction is taking place. The emergent light
beam is now dispersed, using a diffraction grating onto a linear array of light
sensitive diodes (usually —100), such that a small range of the spectrum (~3 nm
wide) impinges on each diode. As the reaction proceeds, the changing com-
position of the reaction mixture absorbs light at those wavelengths specific for
the products, while progressively allowing light absorbed by the reactants to
pass through. These changes are recorded by the diode array, and by comparing
the voltage output of each diode with that when only water (or buffer) is in the
cell, an absorbance spectrum may be constructed. This spectrum may be up-
dated at known time intervals, thus providing the experimenter with spectra
recorded throughout the time course of the reaction. Although the precision in
the measurement of absorbance amplitudes is better in the single wavelength
mode monitored by a photomultiplier, it may be appreciated that the ability to
monitor kinetics simultaneously at some 100 wavelengths and monitor the
spectral changes throughout a reaction, all from one 'shot', holds enormous
advantages.
The data set generated by a single rapid scan experiment is very large.
Typically, for an experiment that involves the collection of 200 time points, e.g.
approximately 20 s of a reaction in which data are collected with logarithmic
periodicity, in a spectral range 370-720 nm (with one data point every 3.3 nm),
a file of around 220 kbytes will be created. The content of these raw data files
can be represented by a matrix A in which each column represents a spectrum

227
M. T. WILSON AND J. TORRES

obtained at a given time point and each row represents a time course at a given
wavelength. Capital bold letters denotes matrices. Thus A = a,
ti t2 t3 tm
•a11
a21 a22 16

X,

It is often the case that the reaction analysed is not simple and may possibly
contain more than one substrate (or product) and one or more intermediates on
the reaction pathway. This increased number of unknowns in the system can be
successfully determined with the additional information provided by a multi-
wavelength experiment. Obviously, the accuracy of this determination will de-
pend on the amount of information available, and it will be more precise the
more data points are taken in the spectral region or time period of interest.
Thus, a whole process is composed of partial reactions, each with its own rate
constant and spectral components. This information is contained in matrix A,
together with random noise inherent in any such experiment, and can be
extracted using a deconvolution procedure, based on matrix algebra methods,
called singular value decomposition (SVD), followed by global fitting. What is required
is that we analyse the matrix in such a way that we can determine the mini-
mum number of processes contributing to the overall spectral change, i.e. the
number of independent time courses and spectral components which together
can combine to form the whole data set. Then, using this information one fits
the data set to a mechanistic model of the reaction thereby yielding the rate
constants and spectra of all species.
In what follows we give a theoretical outline of the methods and then take
the reader through the steps using a practical example. For further information
and examples the reader may consult references (7-9).

5.1 SVD (singular value decomposition)

'Although SVD performs no magic, it does efficiently extract the information
contained in data sets (i.e. matrix A) with a minimum number of input
assumptions' (Henry and Hofrichter; 7).
The goal of SVD is to provide an objective assessment of the minimum number
of spectra and time courses (formally vectors) that through linear combination
can generate the whole data set. What these vectors mean in this context can be
understood by reference to the simple reaction: A —» B. At any time during the
course of the reaction the spectrum collected is a certain linear combination of
the spectra corresponding to species A and B. Thus, the concentration of A at
time t is given by: [A] = [A]e-kt (see above). Correspondingly, the concentration
of B at any time t is: [A] - [A], if [B0] = 0. Therefore:
[B] = [A] - [A] e-kt = [A] (1 - e-; 17

228
 STOPPED-FLOW SPECTROSCOPY

As the absorbance is defined as the concentration times the molar extinction,

we multiply each of the concentrations by the respective molar extinction
coefficient eA and eB, the sum of the absorbancies corresponding to A and B at
any time t is eA[AJe~te + eB[Ao](l - e~fct). As [A] is a constant, we can write eAe'^
+ eB(l - e"to). This shows that the total absorbance can be expressed as a linear
combination of eA and eB, and the minimum number of linearly independent
vectors that can be used to obtain matrix A is two. Alternatively, if the last
expression is regrouped, cAe"to + EB - eBe"kt, we obtain e"1d(eA - eB) + eB, which
again gives us two vectors (EA - eB) and eB. This time, although the number of
vectors is the same, one of them is the difference between the two.
A similar conclusion can be reached from the system:
kt k,
A->B^C 18
As before,
[A] = [A] e-k1t 19
In this case, however, the solution for [B] (or [C]) is not immediately evident:

20

and, as [A] + [B] + [C] = [A]:

[C] = [A] 1 + 21

The absorbance at any time point is expressed as the sum eA[A] + eB[B] + ec[C],
and the linearly independent vectors will be eA, EB and ec. As before, certain
combinations of these vectors, e.g. eA, eB and (ec - eB) or (eA - eB) and ss and ec,
are equally valid. In any case, the minimum number of linearly independent
vectors in this system is three.
The number of linearly independent vectors is, however, not necessarily the
same as the number of spectrally distinct species. This is made clearer by
consideration of the next system, which can be defined by three linearly
independent vectors, even though it contains four spectrally different species:

22
C^D
As in the first example,
[A] = [AJ e^'
[B] = [A] (1 - e-k1t*)
and for the second reaction,
[C] = [C] e-k2T
[D] = [C] (1 - e-K2T)
Hence, the absorbance at any time point is represented by eA[A] + eB[B] + ec[C]
+ cD[D], and substituting for the expressions above:
„ ijiQi
fc fi 1 ca~klt -L. ,, FA 1 11
i^ &DI/IQJ It — cf*-k~lt\i ~i ^CL -^2f -4- o Dl\C oJ1 M
„ [p oj1 0*" \ — fi~^2t\I oq
***J
A

229
M. T. WILSON AND J. TORRES

As the initial concentrations are constants, this can be expressed as:

eA e-fci£ + eB (1 - e^) + ec ?'** + «D (1 ~ e'"2') 26
and regrouped as:
(eA - eB) e-klt + (ec ~ «D) e'"2' + (SB + eD) 27
Hence, this system, which has four spectrally different species, can be repre-
sented by only three linearly independent vectors: (eA - eB), (ec - eD) and (eB +
ED)-
A similar example is:
A->B;C^D;E-»F 28
The absorbance at any time point is:
sA[A] + eB[B] + ec[C] + eD[D] + eE[E] + eF[F] 29
Substituting the concentrations by suitable kinetic expressions and regrouping
terms and ignoring the initial concentrations (constants) gives,
(BA - SB) e-^ + (ec - eD) e^2' + (eE - eF) e^ + (eB + eD + eF) 30
Therefore, this system can be represented by only four linearly independent
vectors, even though it has six spectrally distinct species:
(eA - SB), (ec - eD),(eE - eF) and (eB + eD + EF) 31
In other words, to define the spectrum at any time, one needs three difference
spectra, the final spectrum of the reaction mixture and their time courses.
The value of the SVD procedure is that it allows us to obtain, in an objective
manner, the number of linearly independent vectors required to describe the system,
thus providing valuable information about mechanism. Once the number of
linearly independent vectors is determined, models can be formulated, as above,
and these used to fit the time courses. Any particular mechanism is evaluated on
the basis of its ability to reproduce the original data. From this fitting procedure,
both the spectral components and the rate constants corresponding to the
partial reactions can be obtained.
Models may be eliminated by recourse to the statistics of the fitting pro-
cedure, but also by appealing to biochemical common sense, e.g. no absolute
spectrum can have negative absorbance at any wavelength.

5.1.1 The decomposition of the data set to determine pseudo

(basis) spectra, singular values and time courses (SVD)
The matrix A, corresponding to the raw data, can be expressed as a product of
two matrices. One, E, represents the linearly independent vectors referred to
above, containing spectra or difference spectra of the contributing species.
Another, CT, corresponds to the time courses of these species, the superscript
indicates that the matrix has been transposed, i.e. the columns becomes rows
and the rows columns. Thus
A = ECT 32

230
 STOPPED-FLOW SPECTROSCOPY

Our ultimate objective is to determine the contents of these two separate

matrices (i.e. to gain knowledge of all the spectra and their time courses), but
this is impossible to do directly. A useful first step is to employ SVD analysis.
This is based on a theorem that states that a matrix of numbers A (i.e. the raw
data matrix) can be expressed as a unique product of three other matrices
(figure I I ) , usually referred to as U, S and V, where U and V are orthogonal
matrices, i.e. UUT = Wt - I, the identity matrix. Thus:
A = USV1 33
r
Luckily, the matrices E and C are ultimately related to U and V. from whence
they can be obrained, as explained below. The matrix U has the same dimen-
sions as A (m X n), whereas V[ and S are square (n x n), sec figure U. When
plotted, the columns of U have features that are reminiscent of spectra and arc
termed pseudo spectra or basis spectra, a linear combination of these can give
any spectrum captured during the course of the experiment. The columns of V
are the time courses. In fact, the pairs of matrices (U, E) and (V, CT) are related,
because the entire matrices E and Ct can be generated by linear combination of
the columns in U and VT, respectively.

A - D™, * Sred * V1'™,

Figure 11 Illustration of SVD, The m by n data matrix A may be written uniquely as the
multiple of three other matrices U (m x n), S (n x n) and the transpose of V (n x n). The
chemical information is confined to the first three columns, the remainder containing noise
(increasing shading]. Once the rank has been decided as three by examination of the s
values and autocorrelation coefficients the data matrix may be reconstituted without noise by
multiplication of the reduced matrices. The significant information may be stored
economically as the reduced matrices.

231
M. T. WILSON AND J. TORRES

The minimum number of linearly independent vectors that can generate

matrix A is equal to the rank of the matrix A, which in turn is also the rank of
the matrices U and V and the number of columns in the matrices E (which are
represented by the linearly independent vectors referred to above) and CT. This
is to say that this rank is also the number of linearly independent vectors in the
matrices U and VT. In consequence, linear combinations of these are enough to
represent E and CT, respectively.
Here is where the matrix S comes into play. In the matrix S, all the elements
are zero, except for the diagonal elements, which are called singular values. The
singular values are a measure of the contribution the associated columns in the
U and V matrices make to the experimental data set in matrix A, that is, they
can be considered as a kind of weighting. These values are relatively large in the
first columns but tend rapidly to zero in the later columns. In an ideal situation,
the rank of the matrix A would be just the number of non-zero values in the
diagonal of S. In practice, however, this is not the case because of the noise in
the system and it can be difficult to locate the cut-off between values associated
with signal and those with noise.
One procedure to aid this decision is to plot the logarithm of the different
values Si as a function of i (see example later). The plot shows an abrupt change
in slope that can be used as a cut-off. Another method is the use of auto-
correlation coefficients for the columns of U and V. This coefficient is obtained
by summing over the whole column the products of xi and xi + l, where xi is one
of the elements of the column of U (orV), formally c(U) = 2Uj(Uj + 1>f from j = 1
ton — 1. Since the matrices U and V are orthogonal, the values of the elements
contained in their columns lie between 1 and -1. According to an empirical
rule, columns with autocorrelation coefficients with values >0.5 are more likely
to contain real signal rather than noise.
The rank can also be deduced from the columns in the matrix V or U. When
the columns of these matrices are represented graphically, the amount of noise
(generally) is seen to increase with the column number. The cut-off, the number
of columns before noise overcomes the signal, can also be used to estimate the
rank. Thus, the rank will be the number of rows (or columns) in which the
signal is still perceived by visual inspection (see example later). As in any noise
removal procedure, some information is lost when these columns are
eliminated. The effect of noise, which comes from many sources, is the
introduction of uncertainty about the true rank of the matrix.

5.1.2 Removal of noise and data compression

Once the rank of the matrix has been decided, i.e. the number of columns in
the U, V and S matrices that contain useful information, the remaining
columns may be discarded. This leads to two useful consequences that are
illustrated in Figure 11. Suppose one has collected, in matrix A, 200 spectra each
consisting of 100 wavelength points, i.e. we have 2 x 104 absorbance values in a
matrix comprising 100 rows (time courses) and 200 columns (spectra). Let us
further suppose that following SVD we decide that only three columns of the U

232
 STOPPED-FLOW SPECTROSCOPY

and V matrices contain significant information, the remaining containing

noise. Now all the information is contained in 3 x 100 values of U and 3 x 200
of V and 3 of S (see Figure 11). This means that we need to store only 903 data
points instead of the original 2 X 104 data points. This gives a saving of over 20-
fold in disk storage space. In addition, if we now reconstruct matrix A from the
three columns of U, V and S, then we have the experimental data filtered of noise.
The software that is provided with the commercial stopped-flow spectro-
photometers contains the SVD algorithm and the analysis takes a few seconds.
Generally about eight columns of the matrices are stored to ensure that no
useful data is discarded with the noise.

5.2 Fitting to a mechanism

The rank of the matrix, as we have seen, is the number of linearly independent
vectors that we need to reconstruct the original matrix, and is a function of
both the number of chemical components and the mechanism of the reaction.
This also means that the knowledge of the rank of the matrix will enable us to
restrict the possible mechanistic models to a handful.
The following step is the evaluation of the models represented by our chosen
rank. The procedure involves first the truncation of the matrices U, S and V, so
that they contain only the first r columns (Figure 11) and is based on the follow-
ing algebraic manipulation of the matrices.
As stated above:
A = USV1 = ECT
USV^C = ECTC
1
usyTqct)-
T r 1
= EcFqcFqr1
usv qC c)- = E
Let
H = yTqcTQ-1 35
hence
E = USH 36
substituting:
USV1 = USHCT
U^JSV1 = tfUSHC1
SVT = SHCT 37
S-'SV7 = S-'SHC7
VT = HCT
Hence:
V = CHT 38
The aim of the fitting process is to obtain a concentration matrix C* by
varying the values of the rate constants, so that the matrix V* obtained from the
relationship is as close as possible to V obtained from the data, i.e. V* - V -» 0.
For each set of estimated rate constants, a matrix V* is produced: V* =

233
M. T. WILSON AND J. TORRES

C*(C*rC*)-'CT'V, where C* is the concentration matrix obtained from the

solution of the differential equations derived from the mass action law. When
the right set of rale constants are obtained, the relation is satisfied and the
matrix H can be computed from the relationship:
H = VTC+ 39
where C1 is C*(C*TC*('1 and it is called the pseudo-inverse. H can then be sub-
stituted in E = USH to obtain E (spectral reconstruction). Instead of using the
relationship V = CH1, wo could use A - ECT so that (A* - A) — 0. However.
with this substitution the procedure is generally slower, as noise is included and
larger matrices would be involved.
A helpful hint that can help during the selection of the best mechanism is
that the plot of the V columns is characteristic of a particular mechanism. To
this end, one can construct synthetic spectra and simulate the whole data set A
for every particular mechanism. Each synthetic matrix A can then be decom-
posed into the matrices U, V and S. The plot of these synthetic V columns can

Protocol 6
Reduction of the metmyoglobin-cyanide complex by
sodium dithionite
Reagents
• Horse skeletal muscle myoglobin, • Phosphate buffer, 0.2 M, pH 6.4
supplied as the ferric form MbFe(III) • Potassium cyanide (Sigma)
(Sigma) • Sodium dithionite (BDH)

Method
1 Dissolve MbFe(III) in buffer. To measure the Mb concentration, add sodium dithio-
nite to an aliquot of the solution of MbFe(III) to obtain deoxy Fe(II} (s560 = 13800
M-' cm-1).
2 Prepare a fresh solution of KCN (1 mM) in phosphate buffer at pH 6,4.
3 Add KCN to the MbFe(M) solution to obtain the MetMb-CN complex. Final con-
centrations of 200 uM KCN and 100 uM Mb, approximately. After allowing 1 min to
allow complex formation gently deoxygenate the solution by passage of N2.
4 Prepare fresh sodium dithionite solutions adding the solid to degassed phosphate
buffer at concentrations 100, 25 and 6.2 mw.
5 Mix the MetMb-CN complex with the sodium dithionite solution in a stopped-flow
apparatus.
Syringe A: 100 uM MetMb-CN complex in phosphate buffer at pH 6.4.
Syringe B: Sodium dithionite at concentrations 100, 25 and 6.2 mM in phosphate
buffer at pH 6.4.
6 Collect the spectra in the range 400-600 nm using a diode array.

234
 STOPPED-FLOW SPECTROSCOPY

then be compared to the experimentally obtained V column. This can be used

to identify the correct mechanism.
In practice, the data is analysed using global analysis, in which all the data
(the truncated matrix A) are simultaneously fitted to a selected reaction mechan-
ism. The inputs for this analysis are the partial reactions describing the system,
in which each species is represented by one letter. The initial concentration for
each species is then introduced together with the guessed rate constants for
each one of the partial reactions. This software package is provided by
manufacturers of the stopped-flow instruments.
The output of global analysis is visualized in two ways. In one, the spectra of
the various species that compose the system are plotted. Even in the case that
the fit is good, the identification of realistic features in the spectra correspond-
ing to the different species can be used to discriminate between a good and a
bad model. The goodness of the fit can also be assessed by comparing the fitted
time courses at different wavelengths to those obtained experimentally. When
using such powerful and esoteric mathematical procedures one may feel totally
dependent on the computer. This can be dangerous and it is as well for the
experimentalist to keep a tight grip on reality. It is entirely possible to obtain
very good fits to a biochemically nonsensical mechanism—the computer just
doesn't understand this!
The theoretical steps that have been discussed above may be illustrated using
the practical example of the application of SVD and global fitting to a simple
reaction containing a spectral intermediate given in Protocols 6 and 7.

0.7

0.6

0.5-

s°
c
CO
•fi 0.3-
8
_D
< 0.2-

0.1 -

0.0-

400 450 500 550 600

Wavelength (nm)

Figure 12 The reaction between the cyanide complex of ferric myoglobin (horse heart) and
sodium dithionite. A subset of the total number of spectra collected is shown. An initial
increase in absorbance is seen at 435 nm and in the visible region, spectra displayed every
20 ms, followed by a slower decrease, only the final spectrum (thicker line) is shown for
clarity.

235
M. T. WILSON AND J. TORRES

The options for data collection are:

(1) simple, with constant time interval between spectra;
(2) split, with two different time scales: or
(3] logarithmic,
A logarithmic scale allows one to obtain data from processes that are complete
in less than a second and also from slower reactions that extend over seconds or
even hundreds of seconds, in the same experiment.
Figure 12 shows only spectra corresponding to the first observed transitions.
For the sake of clarity, spectra obtained during a second transition have been
omitted, although the final spectrum, at the end of the experiment, is also
shown. The figure shows the formation of an intermediate, with bands at 515 nm
and 585 nm in the visible region. The changes involved in the two transitions
can also be observed by inspection of single-wavelength traces, which show a
fast process followed by a slower process that leads to the formation of the final
product MbFc(II).
The whole data set can be represented by a matrix A with as many columns
as time points and as many rows as wavelengths. This matrix is analysed accord-
ing to the following protocol:

Protocol 7
The SVD procedure
Method
1 Decompose the matrix A into the matrices U, V and S using the option provided in
the stopped-flow software, Matlab or other suitable package.
2 Determine the rank of the matrix using any of the methods described above.
Depending on the rank of the matrix a mechanism is chosen to fit the data,
3 Plot the first six columns of the V matrix versus time and of the U column versus
wavelength.

Plots of the U columns (figure l3) and the V columns (Figure 14) show that the
rank is either 3 or 4. The same conclusion is reached by looking at the logs plot
(Figure 15) or at the s values and autocorrelation coefficients for the columns in
V and U (Table 2). For the remaining analysis the rank was taken as three, the
fourth s value being much smaller than the third. A very small fourth com-
ponent is in fact present and was identified as being due to a small fraction of
denatured myoglobin reacting differently from the bulk protein.
As the rank is three, mechanisms compatible with this were chosen for global
fitting to the truncated (reconstructed from the first three columns) matrix A.
Only the sequential mechanism A -- B * C gave a sensible solution, i.e. ail
kinetic and spectral profiles fitted well, and non-negative absolute spectra for all

236
 STOPPED-FLOW SPECTROSCOPY

0,40 0,4
Ucol 1
0,2
0,20
0,0

0,00 -0,2

0,4
Ucol3 Ucol 4 0,4

0,0 0,0

-0,4
-0,4

0,50 Ucol 5 Ucol 6 0,5

0,00 0,0

-0,50 -0,5

400 450 500 550 400 450 500 550

Wavelength(nm) Wavelength(nm)
Figure 13 The first six columns, pseudo-spectra, of the U matrix following SVD of the total
data set; matrix A. Columns 5 and 6 appear as noise.

VcoM
0,0

-0,1

-0,2
0,18 Vcol3 Vcol4
0,12 0,0

0,06
-0,2
0,00
-0,06 -0,4

0,30
VcolS vcoie 0,3
0,00 0,0

-0,30 -0,3
-0,60 -0,6
0,01 0,1 1 10 0,01 0,1 1 10
Log10 Time(s) Log10 Time(s)

Figure 14 The first six columns, time courses, of the V matrix following SVD of the total
data set; matrix A.

components. The results of the fitting process are shown together with the
dithionite concentration dependencies of the two processes in Figure 16 and 17.
Thus the results of this analysis may be summarized as follows:
Mb Fe(III)-CN + dithionite -* Mb Fe(II)-CN -» Mb Fe(II) + CN 40

237
M. T. WILSON AND J. TORRES

Figure 15 Plot of logs, against /, showing a break where signal gives way to noise.

Table 2 s values and autocorrelation coefficients (AC)

of the first six columns in the U and V matrices

s ACV ACU
47.9 0.973 0.998
2.74 0.913 0.982
1.14 0.898 0.978
0.14 0.888 0.605
0.06 0.193 -0.02
0.05 -0.596 -0.02

14
12
10
j 8
6
4
2
0

[Na2S2O4]1/2 (mM)1/2
Figure 16 The dithionite concentration dependence of the pseudo-first-order rate constants
of the fast and slow processes seen in the reduction of Mb-CN.

238
 STOPPED-FLOW SPECTROSCOPY

0.6-

0.4-|

0.2-

0.0
400 450 500 550 600
Wavelength (nm)

Figure 17 Global analysis of the reduced (rank = 3) matrix A to a simple sequential model
generates the spectra of the chemical species A, B and C, see text.

Dithionite first reduces the MbFe(III)-CN complex (species A) in a rapid second-

order process. The pseudo-first-order rate constant for this is directly propor-
tional to the concentration of the true reductant, S02~, which is in rapid
equilibrium with dithionite (S204 2SO 2 ), thus resulting in the linear
dependence on [S2042"]%, Figure 16. As cyanide has a low affinity for ferrous iron
the complex now dissociates in a dithionite independent process to yield,
finally, deoxyMb (species C), Figure 16. The spectrum of the short lived
intermediate is provided by the global fitting and shows the intermediate to be
a low-spin ferrous cyanide complex (species B), as expected, which on CN
dissociation gives deoxy Mb, recognized by its characteristic spectrum, Figure 17.
It may be noted that the spectra in Figure 17 are somewhat less noisy than those
in Figure 12 as the noise has partially been removed by discarding the majority
of the columns in the U and V matrices.
Acknowledgements
We thank K. Tzouvara for help with the experiments and SVD analysis of the
MbCN system.
References
1. Eccleston, J. F. (1987) In Spectrophotometry and spectrofluorimetry: a practical approach (ed.
D. A. Harris and C. L. Bashford), p.137. IRL Press, Oxford and Washington DC.
2. Tonomura, T. (1978). Anal. Biochem. 84, 370.
3. Davies, N. A., Wilson, M. T., Slade, E., Flicker, S. P., Murrer, B. A., Powell, N. A. and
Henderson, G. R. (1997) Chem. Commun. 47.
4. Brunori, M., Wilson, M. T. and Antonini, E. (1973) J. Biol. Chem 247, 6076.
5. Antonini, E. and Brunori, M. (1971) Haemoglobin and myoglobin in their reactions with
ligands. North Holland, Amsterdam and London.
6. Fersht, A. (1977) In Enzyme structure and mechanism, p.123. W. H. Freeman, Reading.
7. Henry, E. R. and Hofrichter, J. (1992) Methods in enzymology 210,129.
8. Shrager, R. I. and Handler, R. W. (1982) Anal. Chem. 54,1147.
9. Ownby, D. W. and Gill, S. J. (1990) Biophys. Chem. 37, 395.

239
This page intentionally left blank
 Chapter 9
Stopped-flow fluorescence
spectroscopy
Michael G. Gore* and Stephen P. Bottomleyf
* Institute of Biomolecular Sciences, Division of Biochemistry,
School of Biological Sciences, University of Southampton, Bassett Crescent East,
Southampton, S016 7PX
f
Department of Biochemistry and Molecular Biology, Monash University, Clayton,
Victoria 3168, Australia

1 Introduction
Biochemical reactions, such as substrate or coenzyme binding to enzymes are
usually completed in no more than 50-100 ms and thus require rapid reaction
techniques such as stopped-flow instrumentation for their study. Fortunately,
many such reactions can be followed by changes in the absorption properties of
the substrate, product or coenzyme, and examples of these have been described
in Chapters 1, 7 and 8. An alternative possibility is that during the reaction
there is a change in the fluorescence properties of the substrate, coenzyme or
the protein itself. Some reactions, particularly those involving the oxidation/
reduction of coenzymes, involve both changes in absorption and changes in
fluorescence emission intensity. In many cases, the fluorescence properties of
the ligand or protein itself may change when a complex is formed, even in
the absence of a full catalytic reaction occurring, e.g. the protein fluorescence
emission of most pyridine or flavin nucleotide-dependent dehydrogenases
is quenched when NAD(P)H or FADH (respectively) binds to them, due to
resonance energy transfer from the aromatic amino acids of the protein to the
coenzyme. Conversely, the fluorescence emission from the reduced-coenzymes
is usually enhanced on formation of the complex with these enzymes (1-3).
The principles behind both fluorescence and stopped-flow techniques have
been described in preceding chapters (2 and 8, respectively) and therefore
readers should familiarize themselves with these chapters for some of the back-
ground information. In this chapter, we discuss the use of stopped-flow fluor-
escence spectroscopy and its application to a number of biochemical problems.

2 Instrumentation
A typical stopped-flow system is assembled from modular components of a
conventional spectrophotometer/fluorimeter, a device permitting rapid mixing
of the components of a reaction and a data recording system with a fast re-
sponse. Commercially available instruments offer facilities for the observation
of changes in absorption and/or fluorescence emission after rapid mixing of the

241
MICHAEL G. GORE AND STEPHEN P. BOTTOMLEY

PMT1

Cell Cut off filter

© or PMT
2
MCM 2
I
MCM
1

Figure 1 A diagram of the optical arrangement of a stopped-flow system capable of

simultaneous observation of changes in absorbance and fluorescence. The light from the
xenon lamp is diffracted by monochromator 1 (MCM1) to select the excitation wavelength.
Usually quartz optical fibres conduct the light to the observation cell and absorption is
detected at 180° and fluorescence emission (wavelength selected by a cut-off filter or
MCM2) is detected at 90° relative to the incident light.

reagents. These measurements can often be made simultaneously due to the dif-
ferent optical requirements of the two spectroscopic techniques. Figure 1 gives a
generalized diagram of the geometry of a stopped-flow system able to simultane-
ously measure changes in absorption and fluorescence intensity of a reaction.
Note that the flow-cell is illuminated by light selected by monochromator 1
(MCM1) and that the change in light absorption at this wavelength is detected at
180° relative to the light beam. Changes in fluorescence intensity are detected at
90° relative to the excitation light beam via a second monochromator (MCM2), a
narrow band pass filter or a cut-off filter to select the emission wavelength.

2.1 Data collection

Since the reactions under study are very rapid, the short period of time avail-
able for data sampling and recording places extra demands upon the system.
However, the high sampling rate of current on-line computerized data logging
devices , typically 100 kHz, does permit a sample of the signal to be collected
approximately every 10 uS allowing several samples to be averaged per ms and
going some way to improving the quality of the detected signal. The number
of signal samples taken and averaged (n) per stored data point is described by
n = ft/d, in which t is the observation period in seconds, d is the number of data
points used to describe the reaction profile and f is the sampling rate of the A/D

242
 STOPPED-FLOW FLUORESCENCE SPECTROSCOPY

converter. For example, if 1000 data points are used to describe an observation
period of 100 ms and the effective sampling rate permitted by the A/D converter
is 100 kHz, then 10 samples of the signal will be averaged for each datum point.
The signalnoisc ratio will improve by the square root of the number of samples
in the average, and thus it is clear that a reaction progress curve will have less
noise the slower the reaction and, therefore, the greater the number of signal
sampling events for the same number of data points (1000 in this example}. For
very fast reactions the result must usually be improved by averaging the data
from several reactions. However, the amount of noise observed on a reaction
profile will also depend upon the amplitude of the signal change upon reaction
and how careful the experimenter has been in selecting the conditions for the
experiment, the selling up of the instrument and the preparation of the re-
agents. All of the recommendations made in previous chapters regarding the preparation
of solutions arc particularly important in stopped-flow fluorescence studies.

2.2 Instrument calibration, stability and dead time

It is important to test the reliability of the stopped-flow instrument using con-
trol experiments that test a range of parameters such as the dead time, mixing
efficiency and signal output. In general, these tests will be the same for the
instrument in the configuration for fluorescence studies as that for absorbance
studies, and have been discussed in Chapter 8.

2.3 Measuring mixing efficiency

It is recommended that the mixing efficiency of the instrument be checked
periodically to ensure that the reagent lines, typically 2 mm (i.d.), have not be-
come obstructed by precipitated reagents such as proteins. This can be readily
assessed by mixing together two buffer solutions of different pH (one containing
a pH indicator) with a suitable pK,, so that the final pH is somewhere between
those of the starting solutions. The change in ionization of the indicator will
occur over a very short time period relative to the mixing time of the instru-
ment and can be detected by absorbance or fluorescence changes. An example
of the latter is 4-methylumbelliferone (pKa approximately 7.6; see Figure 2)
which emits fluorescence at 450 nm when excited by light at 390 nm.

Protocol 1
To check the mixing efficiency of the stopped-flow
instrument
Equipment and reagents
• 0,1 Msodium phosphate buffer pH 6.5 1 ml of a stock solution of 10 mM 4-
(buffer A) methylumbelliferone in water
• 0.1 M sodium phosphate buffer pH 8.5 Disposable 5-ml syringes and a few glass
(buffer B) beakers
• Pipettes and measuring cylinders

243
MICHAEL G. GORE AND STEPHEN P. BOTTOMLEY

Protocol 1 continued

Method
1 Mix 50 ml of buffers A and B to give 100 ml of buffer, approximately pH 7.1 to give
C.
2 Add 0,1 ml of the stock solution of 4-methylumbelliferone to 19.9 ml of the mixed
buffer C to give D.
3 Add 0.2 ml of the stock solution of 4-methylumbelliferone to 19.8 ml of buffer A to
give E.
4 Equilibrate all solutions at the operating temperature of the stopped-fiow
observation cell.
5 Use solution C to wash out the contents of the two drive syringes of the stopped-
flow apparatus. Refill the two syringes with C and discharge five or more reactions
to wash out and fill the observation cell,
6 Set the excitation wavelength to 388 nm and use a cut-off filter to select emitted
light at 410 nm and above.
7 Flush out the system with solution D and adjust the sensitivity of the instrument to
give a working fluorescence signal. The working signal should be the same as that
obtained in the following procedure.
8 Replace the contents of one drive syringe with solution E and that of the other
drive syringe with solution B.
9 Discharge five or more 'shots' to replace the solutions in the observation cell.
10 Then discharge several more and record any change in signal intensity with time.
Results
On mixing the two solutions E and B, the pH of the reaction mixture will change to that
of solutions C and D (pH 7.1) within the dead time of the instrument, and the degree of
ionization of the fluorophore (originally at pH 6,5) will alter. Observation of the fluor-
escence intensity of the solution at 410 nm and above should yield an 'instantaneous'
increase in fluorescence intensity (within the dead time of the instrument) followed by a
stable signal intensity. Any change in fluorescence intensity over the first 10-50 ms after
mixing suggests that poor mixing is occurring.

2.4 Sample preparation

Thorough sample preparation is critical to the success of any spectra photo-
metric experiment whether using absorbance or fluorescence detection systems.
Even the rapid mixing of two samples of buffer in a sensitive instrument can
result in what appears to be a reaction progress curve. This may be caused by
simple effects such as the rapid compression and decompression of an air
bubble in the flow path, the mixing of two solutions at different temperatures
or the effect of the stopping process upon small dust panicles present in the
solutions. Therefore it is essential that solutions be prepared thoroughly before

244
 STOPPED-FLOW FLUORESCENCE SPECTROSCOPY

4.5 6.5 8.5 10.5

PH

figure 2 The effect of pH on the fluorescence emission intensity of 4- methylumbelliferone.

10 uM 4-methylumbelliferone was dissolved in 50 mw potassium phosphate buffer at the
indicated pH and its fluorescence was excited by light at 388 nm and detected at 450 nm.
The inset shows the fluorescence emission spectrum of 4-methylumbelliferone at pH 10 (a)
and at pH 4.5 (b).

use. Degas buffers to remove dissolved air that may otherwise come out of
solution following rapid decompression of the reaction solutions and filter all
solutions using an appropriate filter, e.g. Millipore 0.45 um. Sometimes sample
loss may occur during filtration in which case centrifugation of the sample may
help eliminate suspended particles without compromising the sample.

2.5 Artefacts
It is important to verify that the signal change being monitored is due to the
reaction under study and not due to artefacts. Comparison of signal changes
noted under equilibrium conditions should be compared with the kinetic differ-
ence spectrum obtained in the stopped-flow (see Chapter 8). If deviations are
noted then the observed change in fluorescence may well be due to an artefact
and not reflect the reaction under study.

2.6 Temperature effects

Good temperature control is essential to the success of any experiment, particu-
larly those using fluorescence techniques since the quantum yield of a fluoro-
phore is very sensitive to temperature (see Chapters 2 and 12). The quantum
yield of fluorescence decreases with increasing temperature due to the in-
creased molecular vibration leading to loss of excited state energy as heat rather
than light. Furthermore, the pH values of several buffers (especially Tris) are
dependent upon temperature, and it is always advisable to adjust the pH of the

245
MICHAEL G. GORE AND STEPHEN P. BOTTOMLEY

buffer to be used at the temperature required in the experiments. Table 1 in

Chapter 13 gives a list of different buffers and the effects of temperature on
their pKa value. The temperature of the reaction solutions in the stopped-flow
apparatus should be maintained at all times using a water bath system that
circulates the water rapidly, so that there is fast heat exchange between the
water bath and the observation cell.
Finally, the mixing of some solutions may lead to a rapid change in tempera-
ture taking place, e.g. if a solution of protein in 6 M Gdn-HCl is mixed with
aqueous buffers to initiate refolding of a protein, a decrease in temperature
occurs due to the dilution of the denaturant. This decrease may cause a small
increase in fluorescence emission intensity from the protein that will reduce as
the mixed solution returns to the equilibrated temperature of the reaction
vessel. This may lead to an observable short-lived artefact in the reaction
progress curve, which may be minimized by opting to mix the solution of
protein in denaturant 1:10 v/v with aqueous buffer.

2.7 Density differences between the two solutions

Mixing two solutions of different density can produce turbulence effects. Again
this is a common problem in protein unfolding studies in which a viscous
solution of 6 M Gdn-HCl is mixed with a protein solution. Varying the mixing
ratio can again limit the problem. Alternatives are to make sure that the two
solutions are matched, for example, using an inert salt. The problem may be
identified as described in Protocol 1 by including the salts of interest with the pH
indicator, e.g. 6 M Gdn-HCl in one buffer. It is important that the 'flow path' be
thoroughly rinsed with distilled water after experiments including high con-
centrations of salt. Any minute leaks around valves or syringe pistons will lead
to the formation of crystals that may scratch PTFE seals and valves or block the
narrow flow tubes.

3 Factors affecting the sensitivity of the optical

system
3.1 Slit width
The quantum yield of fluorescence from a fluorophore is the ratio of the num-
ber of photons of light emitted to the number of photons of light absorbed by
the molecule. Thus it follows (at the level of light intensity available in labora-
tory fluorimeters) that the intensity of the fluorescence emission will be directly
related to the intensity of light available for the excitation process. Since it is
not usually possible to routinely alter the nature of the light source, the light
intensity falling upon the reaction flow-cell can only be varied by adjustment of
the slit width of the excitation monochromator. However, this has repercussions
on the range of wavelengths allowed to fall upon the reaction mixture (see
Figures 11-12, Chapter 1). In addition, xenon lamps (the most common form of
lamp for use in stopped-flow fluorescence instruments) do not emit equal light

246
 STOPPED-FLOW FLUORESCENCE SPECTROSCOPY

intensities at all wavelengths. Thus a shift in the wavelength selected by the

excitation monochromator may well be accompanied by a change in light
intensity with a resultant change in sensitivity (see Figure 6, Chapter 2).

3.2 Selection of wavelength of emitted light

For fluorescence detection it is necessary to select both the excitation and the
emission wavelength. To achieve this it is usual to use a monochromator to
select the excitation wavelength and a narrow bandpass or a cut-off filter to
select the emission wavelength. The latter can, of course, be selected by a second
monochromator but this option is not often chosen since these transmit light
less efficiently than a filter leading to a significant loss of sensitivity. However,
success relies upon the availability of a suitable filter for the desired wavelength
required, and on the difference in wavelength between the excitation and the
emission wavelengths of the fluorophore (Stake's shift). If this is very small,
then the task is more difficult since overlapping excitation and emission band-
widths will lead to increased transmission of scattered light. A cut-off filter is
usually used that absorbs light at the excitation wavelengths (typically these
will have an absorbance of about 4 at the non-permitted wavelengths) but
allows emitted light at higher wavelengths to pass through. The limitation of
such a filter is that any wavelength above the cut-off value is allowed to pass.
This may lead to a high background light intensity from stray light, upon which
the fluorescence emission is superimposed. An alternative is to use a filter with
a very narrow bandpass at an appropriate wavelength as far removed from the
excitation wavelength as possible. Such filters usually have a bandpass between
1 and 20 nm. However, suitable bandpass filters for non-routine experiments are
not always readily available in the laboratory and are usually more expensive to
buy than cut-off filters. The manufacturers of stopped-flow devices supply
suitable filters for their instruments, or filters can be obtained from specialist
companies, e.g. Omega Optical Inc. (http://www. omegafilters.com).

3.3 Voltage applied to PMT

The sensitivity of a PMT depends upon the voltage applied to its dynodes. Figure
3 shows the amplified signal coming from a PMT when it is illuminated by
fluorescence emission at 360 nm arising from 15 um warfarin in 100 HM phos-
phate buffer, pH 7.4 (excited at 330 nm). The observed signal is plotted against
the voltage applied to the PMT. It can be noted that the output signal is very
dependent on the voltage applied to the PMT and on the fluorescence intensity
allowed to fall upon the PMT (controlled indirectly in this case by changing the
slit width of the excitation monochromator).
It is sometimes found that a non-linear response to light intensity may occur
if the light intensity is very high and only a low dynode voltage is required to
obtain a reasonable working signal. Under these circumstances it is better to
reduce the concentration of the fluorophore or, if this is not possible, to excite
the fluorescence at a wavelength that is lower than the excitation maximum

247
MICHAEL G. GORE AND STEPHEN P. BOTTOMLEY

0.5-

300 350 400 450 500

Voltage on PMT
Figure 3 The relationship between fluorescence signal (volts) and the high voltage applied to
the PMT. The observation cell contained a solution of 15 uM warfarin in 100 mw potassium
phosphate buffer at pH 7.4, 20°C. The fluorescence signal in volts (V) at 360 nm is plotted
against the voltage applied to the PMT of an Applied Photophysics SX.17MV stopped-flow
fluorimeter. The fluorescence was excited by light at 330 nm through 0.5 mm (»}, 1.0 mm
(•) and 1.5 mm (A) slit widths of MCM1 (see Figure 1).

(use of a higher wavelength may lead to increased light scattering problems).

Alternatively, the same result may be achieved by observation of the reaction at
a higher wavelength than that of the emission maximum. If high light intensity
falls upon a PMT then 'saturation' may occur, i.e. there is no further increase in
output signal from the PMT when the light intensity increases. At high levels of
light, permanent damage to the PMT may be caused. In general, PMTs should never
be exposed to daylight particularly while a voltage is applied to its dynodes. The linearity
of response of the PMT/system to light intensity may be tested by flowing very
dilute solutions of a fluorophore through the observation cell. The fluorescence
signal obtained should be proportional to the concentration of the fluorophore.
Note that corrections for the inner filter effect may have to be made if the
concentration of the fluorophore (hence absorption of the excitation and
emitted light) is high (see Chapter 2). This correction will depend upon the
geometry of the observation cell and the optical pathlength for the excitation
and emission light.
Figure 4 shows the noise levels of fluorescence signals of similar magnitude
arising from warfarin in phosphate buffer (the same solutions described in
Figure 3 above). The different traces were obtained by using two different inten-
sities of excitation light (adjusted by changing the slit widths of the excitation
monochromator) and different voltages applied to the PMT. The lower trace has
the least noise and was generated by using an 8-nm bandwidth and a relatively

248
 STOPPED-FLOW FLUORESCENCE SPECTROSCOPY

BlsSrPr^

1
E ° Time(s)
Figure 4 Signal intensity and noise level; the effect of slit width and voltage to PMT. A plot
showing the recorded traces from an Applied Photophysics SX.17MV stopped-flow fluorimeter
when 20 uM warfarin (final concentration) in 10 mM potassium phosphate buffer at pH 7.4,
20°C, was mixed with buffer. The fluorescence intensity at 360 nm was measured using
an excitation wavelength of 330 nm. The sensitivity of the instrument was adjusted to give
a similar signal intensity by varying the slit width and the high voltage to the PMT (see
Figure 3).

low dynode voltage (350 V). This is generally the case and hence the amount of
light available and the optical alignment of the instrument are very important
considerations for achieving high sensitivity.

4 Selection of reporter group

The choice of a suitable fluorescent reporter group is essential to the success of
any experiment. The ideal reporter group should have a high molar extinction
coefficient and a high quantum yield (so that it is easily detected by the system)
and should allow faithful monitoring of the reaction. It must also report a change
in its environment/polarity or pH and must respond to these immediately and
proportionally. Often constraints are placed on the choice of reporter due to the
nature of the sample being studied. However, when a choice is available, the
time taken to chose the best reporter will be well spent and maximize the
chances of successful experimentation. The choice of reporter group falls into
two main categories, intrinsic reporter groups and extrinsic reporter groups.

4.1 Intrinsic reporter groups

Intrinsic reporter groups are already present in an essential component of the
reaction under study. They may be located within the protein, a prosthetic
group or cofactor, such as flavins (3), NADH (2), pyridoxal-5'-phosphate (4), or a
substrate or other ligand (5).
The indole group of tryptophan (Trp) has a higher molar extinction coefficient
than the phenolic and phenyl side chains of tyrosine (Tyr) and phenylalanine
(Phe), respectively (see Table 1, Chapter 12). Thus although its quantum yield is
similar to that of Tyr, its fluorescence emission is much more intense. Further-
more, its excitation spectrum overlaps the emission spectrum of Tyr and
therefore fluorescence resonance energy transfer (FRET, see Chapters 2 and 3)
from Tyr to Trp occurs readily when both residues are in close proximity (i.e.
located in the same protein molecule) and favourably orientated. The intrinsic
fluorescence of a protein is therefore dominated by the contribution from Trp

249
MICHAEL G. GORE AND STEPHEN P. BOTTOMLEY

residues, although proteins lacking Trp residues or with many Tyr will show
fluorescence emission from Tyr residues.
The quantum yield and emission maximum of the Trp residue is highly de-
pendent upon the microenvironment of the residue. Trp residues in the buried,
apolar interior of a protein can have emission maxima as low as 308 nm and
those exposed to aqueous solvent have an emission maximum close to 350 nm
(see Section 4.3 and references 37, 38, and 40 in Chapter 12). It is this sensitivity
to environmental changes that makes the Trp residue such an ideal reporter
group. Furthermore, the judicious use of site-directed mutagenesis, particularly
with the aid of structural knowledge of the protein, allows the removal of
unwanted Trp residues or the addition of unique Trp residues to a protein or to
a domain of a protein (6-8). Recent developments in molecular biology now
allow modified Trp residues (e.g. 5-hydroxy-L-tryptophan (9)) with red-shifted
excitation and emission spectra to be located at key positions in proteins. These
residues can be selectively excited, permitting spectral changes to be identified
from specific parts of the protein without having to risk some destabilization of
the protein by the removal of other Trp residues by mutagenesis. Thus Trp
residues can be used as extremely sensitive conformational reporter probes.

4.2 Examples of the use of protein fluorescence to follow

binding reactions
4.2.1 Protein folding/unfolding
Proteins usually unfold upon the addition of denaturant; the time taken ranges
from ms to hours depending upon the concentration of the denaturant used
and the stability of the protein. During the unfolding process, endogenous Trp
residues become exposed to solvent, usually resulting in red shifts in their
emission wavelength maxima. Typically, this shift is accompanied by a simul-
taneous increase or decrease (10-12) in fluorescence intensity depending upon
the environment of the Trp residues before and after unfolding of the protein.
For example, proximity to a disulphide bridge quenches the fluorescence
emission of Trp residues and therefore unfolding, resulting in movement away
from the disulphide bridge, may result in an increase in quantum yield (10).
Refolding reaction progress curves can take place over a few uS to a few
minutes. The observed order of the reaction will depend upon the stage at
which the change in fluorescence occurs and whether, for example, association
of subunits is a rate-limiting step preceding a change in fluorescence.
A plot of the logarithm of the rates of unfolding and folding against the
concentration of denaturant (see Figure 5) can give valuable information in
protein folding studies (10). Deviation from linearity in either the refolding or
unfolding part of the curve suggests intermediates may be present in these pro-
cesses. Comparison of the two curves in Figure 5 suggests that an intermediate
may exist in the unfolding of the protein consisting of three consecutive short
consensus repeat domains of human complement receptor 1 (SCR1-3), but not
for the unfolding of a single domain protein (SCR3) derived from the third

250
 STOPPED-FLOW FLUORESCENCE SPECTROSCOPY

4.5

4.0

3.5

3
'°
2.5

2.0

1.5
0 1 2 3 4 5 6 7
GdnHCI [M]

Figure 5 The dependence of the rates of folding and unfolding of non-reduced short
consensus repeat (SCR) domains from human complement receptor 1 on the concentration
of Gdn-HCI. Unfolding reactions were achieved by mixing 1:10 v/v 50 uM of a protein
consisting of the first three domains, SCR1-3 (•), or a protein consisting of the single third
domain, SCR3 (•) of complement receptor 1 with various concentrations of Gdn-HCI in
20 mM potassium phosphate buffer, pH 7.4 at 25°C. Refolding reactions were achieved by
mixing the proteins in 3-7 M Gdn-HCI 1:10 v/v with buffer containing various concentrations
of the denaturant to give the final concentrations of Gdn-HCI shown above. This figure has
been reproduced from Clark, N. S., Dodd, I., Mossakowska, D. E., Smith, R. A. G. and Gore,
M. G. Protein Engineering (1996) 9, 10, 877 by permission of Oxford University Press.

domain of complement receptor 1. The minima of the curves occur at the con-
centration of denaturant where the rate of refolding is the same as the rate of
unfolding. That is the concentration of denaturant that gives the midpoint of an
unfolding/folding transition where half of the protein is in each of the unfolded
and folded forms (see Figure 4, Chapter 12). These rates should be compared,
whenever possible, to rates of unfolding determined by stopped-flow CD
measurements (far UV) which reflect changes in the backbone structure of the
polypeptide chain (see Chapters 4,10 and 12).

4.2.2 Protein conformational changes and protein-protein or

protein-ligand interactions
The use of quenching of protein fluorescence to monitor ligand binding is a
very sensitive technique. This high sensitivity arises from the fact that proteins
in general are highly fluorescent, not only because they may contain Trp
residues, but also because Tyr residues absorb light and fluoresce or pass on the
energy to Trp residues by FRET. Furthermore, aromatic side chains are usually
buried in the interior of a protein and are shielded from deactivating collisions
with other ions or polar species resulting in high quantum yields (see Section
4.3, Chapter 12). If a ligand, with an appropriate absorption spectrum, binds to
the protein then significant decreases in protein fluorescence may occur due to

251
MICHAEL G. GORE AND STEPHEN P. BOTTOMLEY

FRET. However, the fractional quenching of protein fluorescence by FRET to a

bound ligand is usually a complex relationship. The saturation of the binding
sites for the ligand on the protein is not linearly related to the decrease in
protein fluorescence if the protein has more than one ligand binding site per
molecule. Holbrook (1, also see 2) has shown that the protein fluorescence of an
oligomer with multiple equivalent binding sites (n) for a ligand is proportional
to (1 — a(l - x))n where a is the fractional saturation of the ligand binding sites
and x = Fp1/n where Fp is the protein fluorescence when a = 1. Exceptions to a
non-linear relationship occur if no FRET occurs between individual subunits of
an oligomer and they therefore behave spectroscopically as monomers; if bind-
ing of a ligand induces the protein to dissociate into subunits; or if the ligand
binding exhibits strong positive or negative cooperativity, i.e. the binding of one
ligand molecule results in all of the other sites for the ligand on that protein
molecule becoming saturated or the binding of one ligand molecule prevents
the binding of others to the same protein molecule, respectively.
Therefore, reaction curves obtained by monitoring the changes in protein
fluorescence may not directly correlate with the saturation of the binding sites
by ligand. Thus, it follows that experiments in which ligand binding is detected
by excitation of protein fluorescence at 280 nm and observation of the
fluorescence emission from the bound ligand (excited by FRET) will also be
subject to the same limitations.
Nevertheless, many different examples (8, 13-15) exist of the successful use
of stopped-flow techniques based upon changes in protein fluorescence to
monitor conformational changes or binding reactions. For example, site directed
mutagenesis was used to place Trp reporter groups into the L-lactate dehydro-
genase from Bacillus stearothermophilus in order to examine its reaction mech-
anism (8). The enzyme catalyses the interconversion of L-lactate and pyruvate
coupled to the reduction/oxidation of NAD+/NADH, respectively. The rate-limit-
ing step in the mechanism is a conformational change induced by coenzyme
and substrate binding, involving the closure of a peptide loop over the active
site. A Trp reporter group (Trp1O6) was placed into the peptide loop region and
the rate of the conformational change was determined from the changes in the
fluorescence emission properties of the Trp as it moved with the peptide loop
from a polar to a less polar environment.
Studies using changes in protein fluorescence have also revealed detailed
information about protein-protein binding mechanisms (13,14). Figure 6 shows
the change in fluorescence intensity from a reaction between a 78-residue single
immunoglobulin (Ig)-binding domain of protein L (SpL) from Peptostreptococcus
magnus and human Ig-kappa light chains. The two curves shown were obtained
using two mutant forms of SpL with engineered unique Trp residues at positions
39 (upper trace) or 64 (lower trace). On forming complexes with kappa
chain both mutants of SpL show rapid increases in fluorescence intensity at
335 nm that reflect the formation of initial 'encounter complexes'. The
observed rates of these reactions depend upon the concentration of the ligand
(in these experiments the concentration of SpL was varied and that of kappa

252
 STOPPED-FLOW FLUORESCENCE SPECTROSCOPY

Figure 6 The binding curves obtained on mixing K chains with a mutated immunoglobulin-
binding domain of protein L (SpL) from Peptostreptococcus magnus. The curves show the
changes in protein fluorescence intensity from unique tryptophan residues (W39 or W64) as
each domain binds to K chains at pH 8.0 in potassium phosphate buffer, 20°C. The
concentration of K chain was 3 (AM and the concentration of each mutant of SpL was 30 uM.
These data were kindly provided by Dr J. Beckingham and N. Housden.

chain kept constant). In each case, the rapid phase of change in fluorescence is
followed by a slower, single exponential process with a rate that is independent
of the total concentration of SpL used. This is usually indicative of a structural
change occurring in the complex or in either reactant.
There are several possible models that may, in principle, be able to explain
these results. In order to analyse what is happening in this or any reaction we
must first identify the step or steps in the reaction that lead to signal changes;
in this case the stages at which the rapid and slow changes in fluorescence
occur.
One possible model (a) is given below;

(a) SpL ?* SpL* + K <?* SpL*.K

In this model, an equilibrium exists between two conformations of (SpL and

SpL*); only SpL* is able to bind to K chain; there is a change in fluorescence
intensity as SpL isomerizes to SpL* (an increase for W39 or decrease for W64
mutants); an increase of fluorescence occurs (for both mutant proteins) when
SpL* binds to K chain. If we vary the concentration of K but keep it in large
excess over the total concentration of SpL (+ SpL*) present and if kf « k1[K],
then the observed initial reaction will follow an exponential curve of rate =
k1[K]+ k-i providing some SpL* exists at the time of mixing. As SpL* is removed
from the equilibrium with SpL, by formation of the complex with K chains,

253
MICHAEL G. GORE AND STEPHEN P. BOTTOMLEY

some SpL will isomerize to SpL*. This gives rise to a further change in
fluorescence intensity, an increase for the reaction involving W39 mutant and a
decrease for the reaction involving the W64 mutant. If kf is not much smaller
than k1 [K] then more complex analysis is needed. Figure 7 gives an example of
reaction conditions that permit observation of the second phase of the reaction
in virtual isolation from the initial phase. In this experiment a high concen-
tration of SpL (a non-mutated domain containing no Trp residues) was reacted
with K chain and the change in fluorescence from Tyr residues was monitored
at 305 nm (X.ex = 280 nm). The high rate of the initial phase of the reaction
(shown by a decrease in fluorescence intensity noted when this domain
binds to K chain) is virtually completed at the point indicated by the arrow
(approximately 200 ms). These conditions effectively separate out the two parts
of the reaction, and allow the rate of the slower second phase to be deter-
mined without much contribution from the fast phase of fluorescence change,
thus facilitating curve fitting algorithms (see Chapter 8, Section 4.1.1 and
Figure 8).
The exact form of binding curves for multi-phase reactions such as this will
be complex. Amongst other considerations, the relative amplitudes of the two
phases of signal change will depend upon the position of the equilibrium be-
tween SpL and SpL*. If SpL predominates then the amplitude of the slow change
in fluorescence will be larger than if SpL* predominates. Furthermore, the

1.96

Time (s)

Figure 7 The reaction progress curve obtained on mixing K chains with a non-mutated SpL
domain lacking a Trp residue. The curve show the changes in protein fluorescence intensity
at 305 nm (excited at 280 nm) from tyrosine residues as the domain binds to K chains at
pH 8.0 in potassium phosphate buffer, 20°C. The concentration of K chain was 4 uM and
the concentration of SpL was 70 uM. These data were kindly provided by N. Housden.

254
 STOPPED-FLOW FLUORESCENCE SPECTROSCOPY

change in fluorescence signal as SpL isomerizes to SpL* will almost certainly be

of different amplitude than the signal change that occurs when SpL* interacts
with K chain.
Alternatively, let us consider another possible model; that the rapid increase
in fluorescence occurs on the formation of the binary complex SpL.K and that
the slow change in fluorescence occurs as SpL.K isomerizes to [SpL.K]*. This
model is described by scheme (b) below and has been proposed for a number of
protein-ligand systems (e.g. see 3, 14). Note that it is not known which com-
ponent in the complex changes conformation or whether both proteins are
involved.

(b) SpL + K ^ SpL.K v* SpL.K* 2

Here, k1 (M-1 s-1) is the second order rate constant for the formation of the binary
complex and k-1 (s-1) is the first order rate of dissociation of the complex. The
rate of the fast change of fluorescence intensity is proportional to the con-
centration of K chain and represents the rate of formation of SpL.K. Providing
that the rate of formation of the encounter complex SpL.K is rapid compared to
the unimolecular isomerization step (i.e. the two parts of the process can be
considered as two separate reactions), k1 and k-1 may be derived from the slope
and intercept, respectively, of a plot of kobs against the final concentration of K
chain. The Kd for the encounter complex is approximated by k -1 /k 1 -
The observed forward rate for the conformational change is the sum of
the forward and reverse rates of the conformational change between SpL.K
and SpL*K. This is given by fkf + kr in which f is the fraction of the SpL present
as SpL.K. This is readily calculated by considering the initial part of the reaction
i.e.

SpL + K ^ SpL.K
k-i
= [SpL][K] = k_! 3
d
[SpLK] k,

In which [SpL], [K] and [SpL.K] are the concentrations of these components at
equilibrium. Hence [SpL.K] = [SpL][K]/Kd and the fraction of SpL present in the
complex
[SpLK]
[SpL] + [SpL.K]

Substituting [SpL.K] in the above equation by [SpL][K]/Ka and simplifying we get

the fraction of SpL in the form SpL.K

= [K] ^
+ Kd 5

255
MICHAEL G. GORE AND STEPHEN P. BOTTOMLEY

Thus the observed forward rate of the conformational change is given by

Examination of this equation indicates

(1) If Kd is very low, then kobs = kr + kf.
(2) If the concentration of K chain = 0, then kobs = k,..
Thus, from (2) above we can see that if a preformed complex of SpL.K* is rapidly
diluted by buffer so that the free concentration of K chain is negligible, then the
observed reaction will depend upon the relative values of rates k-1 (the dis-
sociation of the encounter complex SpL.K) and kr (the reverse conformational
change from SpL.K* to SpL.K). In this example, kt is rate limiting and is therefore
observed.
In general, if the observed rate of dissociation of a complex is shown to be
independent of the final free concentration of ligand (in this case K chain) then
it is indicative that the dissociation process is rate limited by a conformational
change and supports model (b) above for the SpL - K chain reaction (16).
In the reaction between SpL and K chain, kt can be measured directly by 'dis-
placement' techniques. In these, preformed complexes of W39 SpL and K chain
are rapidly mixed with a large excess of wild-type SpL (which lacks Trp residues
and therefore gives no change in fluorescence intensity at 335 nm on binding to
K chain). Thus the decrease in fluorescence at 335nm noted from the reaction of
wild-type SpL with the W39 mutant complex is due to the dissociation of the
W39 mutant from its binary complex with K chain. Providing that the concen-
tration of the SpL.K species is very low then the observed rate for this reaction
equals KT, since in these experiments, the reaction in the forward direction (k f )
involving mutant proteins is competitively inhibited by a large excess of the
wild-type SpL that lacks Trp residues. By subtracting the value of kT,. from the
value of kobs for forward the isomerization reaction (equation 6) it is possible to
calculate the value of kf.
For this system, the Kd for the encounter complex is k -1 /k 1 and at equilibrium,
the macroscopic Kd for the reaction (Kdeq) is dependent upon all four rate
constants and is described by equation 7 (14).

k1 1+
Which is another form of the equation (3)

+ k r /kf
Note that if kf is very large compared to k,. then these equations simplify to

256
 STOPPED-FLOW FLUORESCENCE SPECTROSCOPY

Resolving whether scheme (a) or (b) is the correct model for such a reaction is
often difficult because one can rarely be certain where the changes in fluor-
escence actually occur. Other supporting experimental data are often required.
It should be emphasized that as the rates of a biphasic process become closer
then the method of analysis is more complex and readers should refer to other
texts (see Section 4.3 and references therein).
Diffusion controlled rates of bimolecular association reactions are usually
around 107-108 M-1 s-1 in agreement with theoretical modelling studies. These
high rates are often not obtained experimentally which implies that something
else, such as an isomerization in either the reactant or the complex, is limiting
the rate. However, there are other possible reasons for association rates lower
than expected and these have been reviewed in (17,18).
Finally, a cautionary note. Reaction progress curves exhibiting more than one
rate may not be due to complex kinetic behaviour in the reaction under in-
vestigation, but may be caused by a combination of real spectral changes due to
the chemical reaction under study plus spectral changes due to another un-
known process. For example, it is very common for a protein to become photo-
degraded when illuminated by intense UV radiation such as that given out by a
xenon lamp. This is particularly evident in fluorescence experiments where
high intensity excitation light is used to increase the sensitivity of the system;
thus, slow decreases in fluorescence intensity may be noted over long observa-
tion periods. Alternatively, of course, the ligand may be similarly unstable.
Control experiments in which each reagent is mixed with buffer in the stopped-
flow should always be carried out to check for such degradation processes and
to determine the expected fluorescence intensities of the control at the start
and end point of the reaction.

4.3 Use of ligand fluorescence to monitor binding

reactions
Reduced pyridine-nucleotide coenzymes (NADH and NADPH) are fluorescent
molecules excited by light at 340 nm (absorption by the reduced nicotinamide
ring) and fluorescing at approximately 460 nm in aqueous buffers. However,
when bound to proteins, the fluorescence emission is usually shifted to
other wavelengths and becomes more intense (2). The bound NADH can act as
a reporter for the binding of another ligand to form a ternary complex
(19-21). There are many other examples of fluorescent ligands.
Figure 8 shows data gained from stopped-flow fluorescence experiments in
which warfarin's (W) fluorescence emission (excited at 330 nm and detected at
360 nm and above) is used to study its binding to human serum albumin (A).
This reaction proceeds by two, consecutive reversible steps. A fast, diffusion
limited bimolecular reaction by which the initial complex is formed (A.W),
followed by a unimolecular conformational change in the protein following
complex formation (A*W).

257
MICHAEL G. GORE AND STEPHEN P. BOTTOMLEY

The critical differences between this reaction and model (b) for the reaction
between SpL and K chain (described above) is that the change in fluorescence is
limited to the unimolecular step (5) and the forward rate of the conformational
change (kf) is relatively fast. The inset to Figure 8 shows that as the concentration
of warfarin is increased, the observed rate of change of fluorescence also
increases. Over this range of concentration of ligand, the bimolecular reaction
is slower than the unimolecular conformational change and is therefore
rate limiting. However, at high concentrations of warfarin, the rate of the
bimolecular step exceeds that of the conformational change and the observed
rate of reaction becomes limited by the latter step.
In a two step reaction such as this with only one source of signal change
at the isomerization step the observed rate of the reaction is as described by
equation

100 150 200 250

Warfarin (UM)

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0

Figure 8 A reaction progress curve obtained when 20 JJLM warfarin was mixed with 1 um
HSA. In this reaction, 1 um albumin was reacted with 10 uM warfarin (final concentrations) in
100 mM potassium phosphate buffer at pH 7.4 also containing 0.8% NaCI, at 25 °C. The
fluorescence of the warfarin was excited at 330 nm, and the emission intensity at 360 nm
and above was selected by a cut-off filter. The inset shows a plot of the pseudo-first-order
rates, kobs, for the reactions between various concentrations of warfarin and 1umHSA
against the final concentration of warfarin used. We are indebted to Miss Sue Twine for the
provision and use of these data.

258
 STOPPED-FLOW FLUORESCENCE SPECTROSCOPY

in which Kd, = f -1 /k, and [W] is the concentration of warfarin. From equation 11 it
can be seen that a plot of fe,,^ against |W] gives a hyperbolic curve as shown in
figure K. The maximum value of kobs at high concentrations of warfarin can be
obtained from the plot shown in the inset to Figure 8 and equals Kf + k,.,
Rearranging equation 11 shows that a plot of l/(k obs , - Kr) against 1/|W] gives a
straight line of slope K d /k f and an intercept on the ordinate of 1/k,-. The value of kr
can be determined by a displacement reaction using a non-fluorescent com-
pound that binds at the same site on albumin as warfarin, e.g. phenylbutazone,
For a detailed kinetic analysis of this reaction see (5).

Protocol 2
Determination of kobs for the binding of warfarin to human
serum albumin
Equipment and reagents
• A stock solution of 100 um human serum * A 360 nm cut-off filter to select the
albumin (HSA) in 100 mM sodium phos- emission wavelength
phate buffer pH 7.4 containing 0,8% NaCl • Some disposable 5-ml syringes and some
• A stock solution of 3 mM warfarin in the small beakers/tubes
same buffer

Method
1 DiJute some of the warfarin solution to 50 umusing the phosphate buffer.
2 Dilute the HSA to 2 um in phosphate buffer.
3 Set the excitation and emission wavelengths to 330 nm and 360 nm (and above),
respectively.
4 Fill both of the stopped-flow drive syringes with buffer and discharge 5-10 'shots'
of these to wash the observation cell.
5 Then replace the buffer in one drive syringe with the 50 uM warfarin. After dis-
charging five 'shots' to flush out the buffer in that side of the system, discharge five
more. After these adjust the slit width of the monochromator selecting the
excitation wavelength and the high voltage on the PMT to obtain a reasonable
working signal. Record the signal voltage (unbound HSA). Note that this will be
lower than that obtained when warfarin is in the presence of HSA.
6 Refill the syringes with phosphate buffer and discharge 10 'shots' and record the
signal voltage (buffer baseline).
7 Now fill one drive syringe with warfarin and the other with 2 uM HSA, discharge
five to six 'shots' to flush out the system. During the last one or two reactions you
should notice that the signal voltage increases as the HSA and warfarin mixture
reaches the observation cell.
8 Discharge five reactions and record the reaction progress curves. These should be
single exponential curves with kobs about 35 s-1 at 25°C.

259
MICHAEL G. GORE AND STEPHEN P. BOTTOMLEY

Protocol 2 continued

9 The warfarin solution tan then be replaced with another at a higher concentration
and the process repeated to obtain a curve such as that shown in the inset to Figure 8.
Remember that if the system has a 1:1 v:v mixing ratio then the final concen-
trations of the reagents will be half that used to fill the syringes.

A cautionary note: It is very important to realize that in reversible reactions

involving more than one step, the observed rates of reaction may contain several
contributing rate constants. Furthermore, there are several possible variables to
be identified, e.g. the relative values of forward and reverse rate constants, and
at which stage or stages of a reaction the signal change occurs that is being used
to follow the reaction progress. These all give rise to different analytical require-
ments beyond the scope of this chapter and readers should refer to other
specialist texts for more information (22-24).

4 4 Extrinsic probes
Extrinsic probes must be used when the system under study has no useful
intrinsic reporter groups or a reaction produces no fluorescence change from
intrinsic reporter groups. In such cases one may use an appropriate fluorescent
group which is added to the system to report the reaction. Extrinsic reporter
groups come in many forms such as non-covalently and covalently bound
fluorescent labels. Many of the latter can be covalently attached to the protein
of interest at a specific site.

4.4.1 Non-covalent modification of proteins

The most widely used compounds for non-covalent attachment to proteins are
1-anilinonaphthalene-8-sulphonate (1-8 ANS) and bis-ANS, These compounds
are virtually non-fluorescent in aqueous solution (Oj. ANS - 0.004}, but fluoresce
intensely in apolar solvents (<t>,. ANS = 0.4 in ethanol) or when partitioned into
hydrophobic clefts or the core of a protein (<1>. ANS = 0.7 (25)), The emission
wavelength maxima (around 480 nm and 500 nm in methanol for ANS and bis-
ANS, respectively) arc also highly dependent upon the polarity of the en-
vironment. They have found use in protein folding/unfolding studies (26, 27),
especially in the detection of molten globule-like intermediates, which are able
to bind more equivalents of the dye because their expanded structure gives
increased access to the hydrophobic core. ANS and bis-ANS have also been used
to study protein conformational changes [28, 29) and ligand binding (30),

4.4.2 Covalent modification of proteins and ligands

Chemical modification reagents for labelling reactive amino acid side chains
(predominantly cysteine or lysine) have been available for many years and used
to probe for residues close to or in the active site of enzymes. Although fluor-
escent derivatives of these reagents have been used less frequently, they
occasionally reward persistent experimentation by offering more detailed

260
 STOPPED-FLOW FLUORESCENCE SPECTROSCOPY

information than simple chemical modification and the identification of

reactive residues.
Both proteins and ligands can be labelled by fluorescent compounds to report
binding interactions either directly, or in the case of fluorescent ligands, by
inclusion in competition experiments. For example, the binding properties of
the liver fatty acid binding-protein and the activity of phospholipase A2 have
been investigated by the use of ll-(dansylamino)undecanoic acid (DAUDA), a
dansylated analogue of a fatty acid (31). This binds to fatty acid binding-protein
and fluoresces intensely. Other, non-labelled fatty acids added, or liberated from
phospholipids by the action of phospholipase A2, will compete for the site and
this can be monitored by a decrease in fluorescence as the DAUDA leaves the
site. Dansyl sarcosine and dansylglycine have been used to study similar binding
reactions with albumin (32-34).
Reactive groups on proteins, such as the thiol side chains of cysteine residues
and the e-amino groups of lysine residues, are ideal for reaction with fluorescent
reagents, although guanidino, imidazole, alcohol (serine and tyrosine) and
carboxyl side chains may also be labelled by various reagents. Suitable chemical
reagents for the modification of cysteine side chains include maleimide de-
rivatives of, for example, pyrene, fluorescein and eosin, and iodoacetamide
derivatives of, for example, fluorescein or NBD (7-chloro-4-nitro-2,l,3-benzo-
oxadiazole). The fluorescence of these groups is strongly quenched by water and
is therefore a very sensitive probe for conformational changes in which in-
creased accessibility to solvent occurs. Isothiocyanate and succinimidyl esters of
most fluorophores are good reagents for labelling the e-amino groups of lysine.
o-Phthaldialdehyde is a very reactive compound that cross-links proteins by
reacting with one equivalent each of a thiol and a suitably placed amino group,
and provides a fluorescent isoindole group. Alternatively, it can be used to label
an amino group in a protein and the reaction completed by the addition of a
soluble thiol compound such as p-mercaptoethanol. The spectral properties of
these fluorophores are given in Table 1, Chapter 12. An immense range of well-
characterized fluorescent reagents are commercially available and interested
readers should consult specialist companies such as Molecular Probes
(http://www.probes.com).
Such labelling techniques can be very useful as long as the label allows the
binding reaction to be monitored and does not significantly affect the structure
or activity of the protein. It must be remembered that few chemical reagents
are absolutely specific for one type of amino acid side chain and that side re-
actions will most probably occur. Thus it is recommended that some attempts
are made to characterize the stoichiometry and site(s) of modification on a pro-
tein. Furthermore, some caution must be used in the interpretation of reaction
progress curves obtained using proteins labelled with fluorescent reagents. Care
must be taken to ensure that these reagents, usually with hydrophobic fluor-
escent groups, have not partitioned into hydrophobic regions of a protein. Such
binding will lead to inaccurate calculations of the stoichiometry of labelling
and, unless prevented (or the offending reagent removed after the modification

261
MICHAEL G. GORE AND STEPHEN P. BOTTOMLEY

reaction), may lead to spurious signal changes during experiments. It has been
found in this laboratory that passage of a labelled protein down a short gel-
filtration column to remove excess reagent is preferable to dialysis, as many of
the fluorescent reagents will adsorb to the column matrix and leave the protein.
The presence of fluorescent groups such as pyrene, fluorescein, eosin etc.
may also give characteristics to the protein or ligand that were previously not
present. For example, such groups on ligands generally decrease water
solubility and sometimes cause the compound to adsorb to the walls of plastic
containers and cuvettes.

4.5 Examples of reactions monitored by changes in

fluorescence of covalently attached fluorophores
NBD-Cl reacts at pH 7 with a cysteine residue of the Ca2+-Mg2+-ATPase from
rabbit muscle sarcoplasmic reticulum. It was found that the fluorescence
emission from the NBD label increases 17% on addition of saturating Ca2+ ions,
and this feature has been exploited in experiments to examine the binding
mechanism of Ca2+ ions to its complex with the protein (35). Several studies
giving further details of the interaction of this protein with various ligands have
been achieved using various coumarin reagents (36).
The pyrene group of pyrene-maleimide becomes fluorescent (Xex = 330 nm,
^em = 378 nm and 400 nm) on reaction with cysteine residues. When 3 (uM
reagent is incubated with 3 uM recombinant bovine brain inositol mono-
phosphatase, one equivalent of Cys218 becomes modified over 10-15 min at pH
8.0 at 20 °C without loss of enzyme activity (37). Pre-equilibrium and equilibrium
fluorescence studies showed that in this position the pyrene group is able to
report the binding of essential Mg2+ ions at the active site of the enzyme (37, 38).
Figure 9 shows that a plot of the observed rates, kobs, against the final
concentration of the Mg2+ ions produces a straight line of slope kj M-1s-1 and
intercept on the ordinate of k- s-1. The reaction is characterized by a high value
for k-1 which makes the approach to equilibrium (kobs) very fast (since kobs =
k,[Mg 2+ ] + K-1,).
It can be noted from Figure 9 that the slopes of the lines (reflecting k-1) are very
similar, whereas the intercepts on the ordinate (giving values of L-1) increase
with decreasing pH. The Kd for the equilibrium is therefore very sensitive to pH
(38), and this is almost entirely due to changes in the value of k-1. The data
suggest that the binding of the metal ion to its coordinating ligands on the
protein is not in direct competition with protons and that the rate of binding is
limited by something else, possibly a conformational change. However, the
reverse process, described by k-1, is affected by the proton concentration, and a
plot of k-1 against pH yields an approximation of the pka of the amino acid side
chain(s) to which the metal binds. The determination of the rate of dissociation
of the metal ions from the complex at various pH values can be achieved by
rapidly diluting a preformed complex, of pyrene-labelled enzyme and Mg2+ ions
at pH 8.0, with buffer (1:10 v/v) at the experimental pH containing 10 mM EDTA.

262
 STOPPED-FLOW FLUORESCENCE SPECTROSCOPY

500 -

430 -

7 360 -

290

220 -

150 -

80 160 240 320 400

Mg2+ [UM]
Figure 9 The effect of pH on k1 and k-1 for the binding of Mg2+ to pyrene-labelled
recombinant inositol monophosphatase. Enzyme (3 uM final concentration) was reacted with
various concentrations of Mg2+ ions in 50 mM Tris-HCI buffer at pH 9.0 (•), pH 8.0 (•),
pH 7.3 (n) and pH 6.5 (O) at 20°C. The kabs for each reaction is plotted against the final
concentration of Mg2+ ions used. Reproduced with permission from Thorne, M. R., Greasley,
P. J. and Gore, M. G. (1996) Biochemical Journal, 315, 989-994. © Biochemical Society.

The advantage of this method is that the protein is kept at pH 8.0 and only
exposed to low or high pH solution for a few ms.
It is interesting to note that in this case it is the high rate of dissociation of
the metal ion (k-1) that limits the ability of the stopped-flow technique to deter-
mine k1 at high concentrations of Mg2+ ions at pH 7.0 or below. The data in
Figure 9 suggest that k-1 is approximately 360 s-1 and k1 is approximately 3.7 x
105 M-1 s-1 at pH 6.5. Therefore in the presence of 300 uM Mg2+ ions the forward
rate (k1[Mg2+]) will be 111 s-1 and kobs will be approximately 461 s-1. Thus the
half-time of the reaction will be only 1.5 ms, and much of the reaction will take
place within the dead time of the instrument.

References
1. Holbrook, J. J. (1972) Biochem.j. 128, 921.
2. Holbrook, J. J. Yates, D. W., Reynolds, S. J., Evans, R. W., Greenwood, C. and Gore, M.
G. (1972) Biochem.j. 128, 933.
3. Chaiyen, P., Brissette, P., Ballou, D. P., and Massay, V. (1997) Biochem. 36, 2612.
4. Ikushiro, H., Hayashi, H., Kawata, Y., and Kagamiyama, H. (1998) Biochem. 37, 3043.
5. Wilting, J., Kremer, J. M. H., Ijzerman, A. D. P. and Schulman, S. (1982) Biochim.
Biophys. Acta 706, 96.
6. Gore, M. G., Greasley, P. J., Knowles, M. R., Gee, N., McAllister, G. and Ragan, C. I.
(1993) Biochem.j, 236, 3, 811.
7. Waldman, A. D. B, Clarke, A. R., Wigley, D. B., Hart, K. W., Chia, W. N., Barstow, D.,
Atkinson, T., Munro, I. and Holbrook, J. J. (1987) Biochim. Biophys. Acta 913, 66.
8. Waldman, A. D. B, Hart, K. W., Clarke, A. R., Wigley, D. B., Barstow, D., Atkinson, T.,
Chia, W. N. and Holbrook, J. J. (1988) Biochem. Biophys. Res. Comm. 150, 752.

263
MICHAEL G. GORE AND STEPHEN P. BOTTOMLEY

9. Ross, J. B. A., Senear, D. F., Waxman, E., Kombo, B. B., Rusinova, E., Huang, Y. T.,
Laws, W. R. and Hasselbacher, C. A. (1992) P.N.A.S. USA 89, 24, 12023.
10. Clark, N. S., Dodd, I., Mossakowska, D. E., Smith, R. A. G., and Gore, M. G. (1996) Prot.
Eng. 9, 10, 877
11. Bottomley, S. P., Popplewell, A. G., Scawen, M., Wan, T., Sutton, B. J. and Gore, M. G.
(1994) Prot Eng. 7,12, 1463.
12. James, E., Whisstock, J., Gore, M. G. and Bottomley, S. P. (1999) J. Biol. Chem. 274,14,
9482.
13. Beckingham, J., Bottomley, S. P., Hinton, R., Sutton, B. J. and Gore, M. G. (1999)
Biochem.]. 340,143.
14. Wallis, R., Moore, G. R., James, R. and Kleanthous, C. (1995) Biochem. 34,13743.
15. Lee, A. G., Baker, K., Khan, Y. M. and East, J. M. (1995) Biochem. J. 305,1, 225.
16. Benson, S. W. (1960) The foundation of chemical kinetics. McGraw-Hill, New York.
17. Pontius, B. W. (1993) Trends Biochem. Sci. 18,181.
18. Schreiber, G. and Fersht, A. R. (1993) Biochemistry 32, 5145.
19. Piersma, S. R., Visser, A. J. W. G., de Vries, S. and Duine, J. A. (1998) Biochem. 37, 3068.
20. Spencer, P. D., Slade, A., Atkinson, A. and Gore M. G. (1990) Biochem. Biophys. Acta.
1040,130.
21. Clarke, A. R., Wigley, D. B., Chia, W. N., Barstow, D., Atkinson, T. and Holbrook, J. J.
(1986) Nature 324, No. 6098, 699.
22. Strickland, S., Palmer, G. and Massey, V. (1975) J. Biol Chem. 250,11, 4408.
23. Gutfreund, H. (1975) Enzymes: physical principles. Wiley, London.
24. Fersht, A. (1984) Enzyme structure and mechanism (2nd Edn). W. H. Freeman, New York.
25. Slavik, J. (1982) Biochim. Biophys. Acta. 694,1,1.
26. Jones, B. E., Beecham, J. M. and Mathews, C. R. (1995) Biochem. 34,1867.
27. Teschke, C. and King, J. (1995) Biochem. 34, 6915.
28. Walmsley, A. R., Martin, G. E. M. and Henderson, P. J. F. (1994) J. Biol. Chem. 269,
17009.
29. Gibbons, D. L. and Horowitz, P. M. (1995) J. Biol. Chem. 270, 3, 7335.
30. Lin, 7., Scharz, F. P. and Eisenstein, E. (1995) J. Biol. Chem. 270, 3,1011.
31. Wilton, D. C. (1990) Biochem.j. 266, 435.
32. Muller, N., Lapique, F., Drelon, E. and Netter, P. (1994) J. pharm. Pharmacol 46, 300.
33. Sudlow, E., Birkett, D. J. and Wade, D. N. (1975) Mol. Pharm. 11, 824.
34. Epps, D. E., Raub, T. J. and Kezdy, F. J. (1995) Anal. Biochem. 227, 342.
35. Henderson, I. M. J., Starling, A. P., Wictome, M., East, J. M. E. and Lee, A. G. (1994)
BtochemJ. 297, 3, 625.
36. Dalton, K. A., East, J. M., Mall, S., Oliver, S., Starling, A. P. and Lee, A. G. (1998)
Biochem.j. 329, 3, 637.
37. Greasley, P. J., Hunt, L. G. and Gore, M. G. (1994) Eur.J. Biochem. 222, 453.
38. Thorne, M. R.., Greasley, P. J., and Gore, M. G. (1996) Biochem. ]. 315, 989.

264
 Chapter 10
Stopped-flow circular
dichroism
Alison Rodger* and Michael J. Careyf
* Department of Chemistry, University of Warwick, Coventry CV4 7AL
Applied Photophysics Ltd., 203/205 Kingston Road, Leatherhead,
Surrey KT22 7PB

± Introduction
As is apparent from previous chapters (Chapters 6, 8 and 9), understanding the
kinetics of chemical and biological processes is extremely important. Questions
we often consider, explicitly or implicitly, include: Has something happened
'instantaneously' or will it take 20 years? Does changing the conditions or
available reagents affect either the end product or the rate of a process? What
intermediates are produced during a reaction? Can we characterize any inter-
mediates? Do we need to remove them to prevent side reactions?
If some or all of the reactants or products are chiral, then circular dichroism
(CD) detection may be the ideal tool for following the kinetics of a reaction,
and if the half-life of the reaction is of the order of milliseconds to seconds or
even minutes then stopped-flow mixing of the reagents will almost certainly
be the appropriate choice of sample handling method. For reactions with half-
lives of a few minutes to tens of minutes the reagents can be mixed by hand in
a normal cuvette and the signal monitored at an appropriate wavelength. CD is
not well suited to kinetics on timescales of hours due to the baseline drift that
does occur (see Chapter 4, Section 2.5). Some CD spectropolarimeters have the
useful facility of being able to perform a wavelength scan at pre-set intervals as
well as monitoring continuously (except during the wavelength scan) at a
chosen wavelength, thus facilitating the characterization of any intermediates.
In this chapter we shall highlight some of the considerations of the stopped-
flow technique that are particularly relevant to CD experiments. Particular
problems may be encountered when performing CD (as opposed to other
detection methods) stopped-flow experiments. The measured signals are very
small (typical CD intensities are 0.1% or less of the absorbance signal), and the
noise level observed is particularly sensitive to any inhomogeneities or turbu-
lence in the samples. Also, as one of the main applications of stopped-flow CD is
in the study of protein folding and unfolding, samples are often very viscous
and/or corrosive, have significant absorbances due to buffers etc., and the
experiments often require wide and variable mixing ratios.

265
ALISON RODGER AND MICHAEL J. CAREY

2 Instrumentation considerations
In establishing a stopped-flow CD system for a particular laboratory a number of
different factors need to be taken into consideration. The most important of
these are:
(1) available budget;
(2) sensitivity required—the ease of averaging over multiple scans is an
important factor here;
(3) minimum dead time required;
(4) corrosiveness or adsorbance of the samples to be studied;
(5) type of flow system: stepper motor or pneumatic driven;
(6) size and design of optical cell;
(7) number of syringes and mixing stages required;
(8) mixing ratios required and whether these need to be varied from
experiment to experiment;
(9) sample viscosity;
(10) wavelength range for detection, wavelength scanning and bandwidth; and
(11) software.
In addition to the above, there is also the important question of whether the
instrument will be a stand-alone stopped-flow CD system, or a wavelength scan-
ning CD spectropolarimeter with stopped-flow attachment, or a stopped-flow
instrument where one of the detectors is a CD detector. Recent instrument
advances have complicated rather than simplified the decision about instru-
mentation. The importance of the above factors will be summarized before a
brief discussion of particular manufacturers and their current instrumentation
is given below.

2.1 Available budget

Budget is often the starting point in instrument choice. The cheaper options
involve a stopped-flow accessory attachment to existing wavelength scanning
instrumentation. (One has almost certainly been performing steady state CD
measurements if one is considering stopped-flow experiments.) The main dis-
advantages of this approach is that access to multi-purpose instruments is
almost always more restricted than to dedicated instruments and changing the
instrument from one mode to another may be non-trivial. The question of
whether sensitivity and accuracy are lost by coupling two instruments is not a
simple one, since, at least until recently, the optics of wavelength scanning CD
instruments were significantly better than those of dedicated stopped-flow CD
instruments. In addition, wavelength scans on samples used for kinetic deter-
minations were significantly better from a combined instrument. It is no longer
clear that this is the case.

266
 STOPPED-FLOW CIRCULAR DICHROISM

2.2 Sensitivity required

Usually, the required signal-to-noise ratio can be achieved by averaging together
a number of kinetic records (though it is important to remember that the noise
only decreases with the inverse square of the number of data accumulations).
However, ultimately, the optics of the instrument and the delivery of a homo-
geneous sample into the light beam will be limiting factors. The mixing issue is
more important for stopped-flow CD than for other stopped-flow spectroscopic
techniques, as turbulent flow increases the noise significantly and may even be
birefringent thus giving a false CD signal. Chiral reagents are also typically
more expensive than achiral ones so longer term budget considerations may
favour a more expensive instrument to conserve required sample volume by
minimizing the number of data accumulations. A typical system requires 200 ^,1
of sample per shot plus of the order of 2 ml to prime a reagent line; sample
concentrations are as for normal CD, which usually means an absorbance of ~1
in the stopped-flow cuvette, see Chapter 4, Sections 2.2 and 2.3.

2.3 Minimum dead time required

The minimum dead time of a stopped-flow system determines the starting point
of a kinetics experiment. The dead time depends on certain physical aspects of
the system, such as how fast the drive mechanism is able to deliver the reagents
to the stopped-flow mixer (i.e. final velocity). In addition, the physical distance
between the point of mixing and the point of observation is of paramount impor-
tance together with the type of flow that occurs immediately after mixing (i.e.
how much turbulence is created). Submillisecond dead times may be claimed by
manufacturers for their instruments but invariably such claims refer to ideal or
special conditions. Dead times for CD kinetic measurements may be considerably
longer, especially if a high viscosity reagent or asymmetric mixing is involved.

2.4 Corrosiveness and adsorbance of the samples to be

studied
Acids, bases and organic solvents are the obvious samples where the inertness
of the component parts of the stopped-flow system needs to be considered.
Whether reagents might adsorb to tubing and other component parts is also
important, not only for concentration accuracy in a particular experiment, but
also to avoid contamination in subsequent experiments if new reagents prove
to be more effective than the cleaning solvents at removing adsorbed species.
These problems have been fully addressed in the context of high performance
liquid chromatography, so almost any HPLC sales representative will be able to
tell you what components are required for any particular application. For
example, PEEK (polyether ether ketone) has excellent chemical resistance, is
not particularly expensive, is flexible and is biocompatible. It is, therefore,
suitable for flow tubing and other components that will come into contact with
reagents. However, PEEK is attacked by some chemicals including concentrated
acids, bromine, chlorine and fluorine. Titanium is a possibility for these sub-

267
ALISON RODGER AND MICHAEL J. CAREY

stances. The flowlines in some stopped-flow systems are made of stainless steel.
Although medical grade stainless steel is relatively inert, nonetheless, one
should avoid using corrosive or reactive reagents. Some biochemical measure-
ments can be compromised by the presence of ferrous ions. Whilst considering
the interaction between reagents and stopped-flow components, one should
also think about the material used in the loading and drive syringes. Plastic may
adsorb many analytes and there is also the possibility of plasticizers and other
species leaching out, especially in acidic media. Glass has similar problems and
is particularly prone to leaching sodium.

2.5 Type of flow system: stepper motor or compressed air

driven
A key component of any stopped-flow system is the drive unit for injecting
reagents into the reaction/observation cuvette. The consensus of opinion seems
to be that pneumatic drive systems provide the more rapid acceleration and
stopping of reagents with greater reproducibility. However, the disadvantage of
such systems is that only a limited range of mixing ratios is currently possible
without changing syringes and their volume. Manufacturers of stepper motor
systems claim the same dead times for their systems as pneumatic driven ones,
and that the ratios of reagents may be changed extensively and with direct con-
trol from the computer. This allows significantly more flexible experiment
design but is unnecessary for most applications. It should be noted that in order
to produce a final velocity commensurate with a low dead time, there can be
very significant pressure forces exerted on the stopped-flow cell during the
stopped-flow drive and in particular when flow is stopped. These forces can
result in small distortions of the cell window thus introducing birefringence
which will affect the efficiency with which the CD signal can be measured. If
birefringence effects are present, they will undoubtedly become more severe in
the far UV (i.e. below 230 nm). Certainly, it is always advisable to acquire suitable
reagent blanks which can then be used to compensate for baseline shifts due to
pressure effects.

2.6 Size and design of the optical cell

The optical cell is where the chemical change whose kinetics are to be followed
by CD takes place. All stopped-flow CD cells are flow-through cells, usually with
a vertical orientation. A small optical flow cell (width, length and depth) reduces
the system dead time and also reduces the amount of sample required to ensure
the cell is filled. However, it reduces the distance available for observation and
also requires a small light beam, both of which reduce sensitivity. In addition,
mixing and flow constraints preclude the use of a cell where the bore is too
narrow whereas a wide cell will have more back-mixing problems. The con-
sensus seems to be that ~2 mm path-length is optimal. The length of a cell
needs to be such as to ensure that there is no back or forward mixing during the
kinetics run.

268
 STOPPED-FLOW CIRCULAR DICHROISM

The issue of forward mixing during a kinetic data acquisition is particularly

important when high concentrations of one reagent are in use (i.e. experiments
with 8 M urea or 6 M guanidine hydrochloride). If measurements are to be made
below 200 nm then, just as for steady-state CD measurements, the cell compart-
ment must be nitrogen purged so as to optimize light throughput.
Although -5 mm length cell is often assumed to be adequate to avoid for-
ward or back mixing, 10 mm is certainly safer. Such a cell requires (for a 2 mm
square cross section cell) —40 ul of sample in the cell. It also requires the light
beam to be focused to about 1 mm by 8 mm in the centre of the cell to avoid
light scattering off the sides of the cell. If the beam is not properly focused onto
the centre of the cell and the light is scattered, then the signal will be distorted
as discussed in Chapter 4. In practice, this means that a wavelength scan per-
formed in the stopped-flow cuvette may show spectral features shifted in
wavelength. The orientation of the cell and flowline input to the cell should also
be such that gravity effects (i.e. high density reagent being able to run downhill
to produce a concentration change in the cell after a few seconds) are avoided.
Such an event can have a marked effect on the shape of the kinetic record.

2.7 The number of syringes and mixing stages required

Most kinetics experiments can be designed with two reagents, one or both of
which may itself be a multi-component mixture. However, at least an additional
syringe to contain washing solvent is required. More flexible and interesting
experiments will require three or more reagent syringes. Also consider the vol-
umes of the syringes and how often they will need to be reloaded. If each run
requires —200 fil reagent, then 1 ml only allows five accumulations with no
provision for washing lines.

2.8 Mixing ratios required and whether these need to be

variable from experiment to experiment
For single mix experiments, ratios of up to 10:1 are typically employed. These are
routinely available on commercial instruments regardless of drive type. For cer-
tain multi-sequential experiments (e.g. 10:1, 1:1, 1:10) the flexibility of stepper
motor systems may be the only option. The programmability of stepper motor
systems can also help during experimental design. Some experiments require the
concentrations of one reagent to be varied over a continuous wide range (e.g. urea
concentrations in certain protein folding experiments). A stepper driven system
may allow such experiments to be performed more quickly, as the concentration
ramp is generated by altering the relative volumes delivered from the drive
syringes, rather than the time consuming preparation of a whole range of
reagent concentrations and their subsequent loading into the instrument.

2.9 Sample viscosity

If the samples are viscous, mixing efficiency can be affected. Incomplete mixing
will result in signal changes unrelated to chemical events in the early part of

269
ALISON RODGER AND MICHAEL J. CAREY

the kinetic record. Mixer options or even premixers can be employed to counter
such difficulties but are likely to compromise the instrumental dead time.

2.10 Wavelength range for detection, wavelength scanning

and bandwidth
Several key aspects of instrument operation relate to spectral performance. For
example, quality magnesium fluoride optics and nitrogen purging is required if
the wavelength range for detection is required to extend below 200 nm. CD
spectrometers require the use of a high intensity light source which has a sig-
nificant far-UV content. For CD scanning instruments, the xenon light source,
with its intense continuum, works adequately. For kinetic measurements at
certain specific wavelengths, a mercury-xenon lamp may offer a worthwhile
improvement in signal-to-noise ratio. However, such a lamp is not ideal when
recording steady-state CD spectra nor does it provide any improvement for
kinetic measurements below about 215 nm.
Wavelength scanning capability serves two purposes for the CD kineticist:
first, it enables the CD spectrum of the solution at the end of the reaction to be
compared to that of the starting material, and, second, it is the easiest way to
confirm correct cell alignment with the optical beam and to calibrate the
instrument. Incorrect cell alignment can produce effects due to light scattering,
such as distortion of spectral features in a scanned CD spectrum. Proteins
containing tryptophan residues (e.g. lysozyme) have sufficiently resolved peaks
in the aromatic region (~280 nm) to be used for this purpose. It should be noted
that scattering problems may be masked if the bandwidth of the instrument is
very wide.
For kinetic measurements, a wider spectral bandwidth than normally em-
ployed for steady-state measurements is usually required to improve the signal-
to-noise ratio of the kinetic records. Without the increased bandwidth, a very
large amount of signal averaging may be needed to produce clean enough
kinetic data from which reliable kinetic information can be derived. This is a
very real problem when precious materials are being used. Conversely, a wide
bandwidth may obscure kinetics particularly where different components of
the sample have relatively small spectral separations.

2.11 Software
The software question relates to ease of use, degree of control over the various
instrument functions, whether the required extent of data analysis is possible,
ease of access to data, and ease of transfer of data in a format that can be readily
used on other computer platforms and software packages. For kinetics in par-
ticular, automation of repetitive procedures such as multi-shot averaging is
desirable. Flexibility in the control of data acquisition allows kinetic records to
be optimized according to the event being studied. For example, a full range of
linear and logarithmic time bases covers the requirements of simple kinetic
changes to more complicated multi-step reactions. In addition, by making full

270
 STOPPED-FLOW CIRCULAR DICHROISM

use of the ADC's (analogue to digital converter) conversion rate (oversampling),

data signal:noise quality can be significantly improved, especially as the
acquisition period increases.

3 Currently available instrumentation

The following five manufacturers are active in stopped-flow CD.
1. Applied Photophysics Ltd, Leatherhead, Surrey UK (http://www.apltd.co.uk/).
2. Aviv Instruments Inc., Lakewood, New Jersey, USA (http://www.avivinst.
com/).
3. Biologic-Science Instruments SA, Claix, France (http://www.bio-logic.fr/).
4. Hi-Tech Scientific Ltd, Salisbury, UK (http://www.hi-techsci.co.uk/).
5. OLIS, Bogart, Georgia, USA (http://www.olisweb.com/).
The manufacturers should be consulted to find details of current instrument-
ation. What follows is a brief discussion of some of the systems available in the
context of the above 'considerations' list. It should not be considered definitive
as advances are constantly being made but, hopefully, it may provide a starting
point for making comparisons. The biggest decision (especially as regards to
price) is whether to have a stand-alone stopped-flow system (either new or an
upgrade of an existing one) or to add a stopped-flow sample attachment to
existing wavelength scanning instrumentation. A summary of instrument costs
is given in Table 1.

3.1 Stopped-flow attachment for CD spectropolarimeters

Before going ahead with the purchase of an attachment for a wavelength scan-
ning instrument, it is advisable to check that the electronic response of your
instrument is fast enough for millisecond time constant acquisition of kinetic
data—the Jasco J-720, for example, needs a minor modification to remove a

Table 1 Approximate costs of instrumentation discussed in the text. Systems with an entry
in the 'attachment required' column must be fitted to another instrument

System Supplier Attachment required Price range

RX 2000 Applied Photophysics Wavelength CD $
SHU-61CD Hi-tech Jasco J-700 series $$
SFM-3 Bio-Logic Wavelength CD $$-$$$
SF305 Aviv Aviv wavelength CD
Olis module Olis Olis wavelength CD
CD.2C Applied Photophysics SX.18MV $$$
MOS-400 + SFM-3 Bio-Logic $$$$
TT*-180 Applied Photophysics $$$$$

$ = less than 5k USD; $$ = 5k to 25k USD; $$$ = 25k to 50k USD; $$$$ = 50k to 100k USD; $$$$$ =
100k to 125k USD.

271
ALISON RODGER AND MICHAEL J. CAREY

stabilizing feedback feature (which in fact seems to have no effect on normal

instrument sensitivity).
The most cost effective (see Table 1) stopped-flow accessory for use with a
standard spectropolarimeter is the RX.2000 from Applied Photophysics. This
accessory is available with a manual or a pneumatic drive. The observation cell
with integral mixer is at the end of a flexible and thermostatted umbilical. It is
designed to fit sample holders that accommodate 12 X 12 mm standard cuvettes.
Asymmetric mixing is possible by choosing drive syringes of an appropriate
size. The dead time of this unit is in the range of 5 to 10 ms depending on type
of drive and the drive syringe ratio in use.
An alternative stopped-flow attachment, designed for use with Jasco J-700
series spectropolarimeters, is the HI-TECH Scientific SHU-61CD with pneumatic
drive. Its temperature range is 5-80 °C, its pathlength is 2 mm, it has a 3.5 mm
window size and a cell volume of 19 ul. The mixer is integral to the cell. 200 (ul
is required per shot with 2 ml to prime the flowlines. Drive syringes may be
altered to allow variable mixing. Hi-Tech do not offer a dedicated stopped-flow
circular dichroism system.
The Bio-Logic SFM-3 system is a three-syringe unit with each syringe being
driven by an independent stepper motor thus enabling independent program-
ming of each syringe and easy variation of mixing ratios. For samples of
differing viscosity/density, the specially designed mixer unit (HDS) is claimed to
perform extremely well. It has a dead time of a few milliseconds (dependent on
the flow rate chosen for the experiment). The system has been designed for
Jasco, Aviv and Jobin-Yvon CD spectropolarimeters. A range of cuvettes may be
used, though the 2 mm pathlength, 54 u1 volume one is recommended for CD
applications. The instrument may be manufactured from inert materials.
The approach adopted by Aviv for CD kinetic measurements was to design a
dedicated module (SF 305 stopped flow system) to work with their existing CD
spectropolarimeter. The SF 305 uses a servo motor drive with a range of motor
options, which allow up to triple mixing with delay lines as required. It is
designed to operate with samples of differing viscosities. Absorbance and
fluorescence stopped-flow facilities may be integrated into the unit. A dead time
of 1 ms is claimed with the T mixer. A range of mixing ratios are possible using
the programmable stepper motor drive. It has a temperature range of 10 °C to
60 °C. The standard pathlength is 1 mm with a cell volume of 10 ul.
OLIS produces a CD version of its spectrometer which uses a novel RSM
monochromator design to deliver rapidly changing monochromatic light to the
sample. In addition, they claim increased sensitivity because of their ability to
use both beams produced by the polarizing beam splitter, whereas other manu-
facturers use only one of the beams. Furthermore, OLIS state that this phase-
coherent design removes the need to periodically check instrument calibration
since absolute differential absorption of left and right circularly polarized light
is correctly measured at any point in time. Whilst it is capable of operating in
rapid-scanning, slow-scanning and fixed wavelength modes, the instrument's
real strength is in its rapid scanning capability where, using real-time signal

272
 STOPPED-FLOW CIRCULAR DICHROISM

averaging, CD spectra are seen to 'come out' of the background noise. For CD
stopped-flow, OLIS attaches its stopped-flow module to the RSM instrument.
Although this combination will perform adequately where CD changes are of
good intensity, one should be cautious if the need is to measure fast reactions
accompanied by small changes in CD.
In addition to the RX.2000 stopped-flow accessory referred to above, Applied
Photophysics provides a CD detection option, CD.2C, as an upgrade to its
SX.18MV research stopped-flow spectrometer. In this instrument, the standard
optical light guide coupling between the monochromator and the sample hand-
ling unit (used in the absorbance and fluorescence configurations) is replaced by
a purpose built optical module comprising lens coupling, calcite polarizer,
photo-elastic modulator and sample housing. A second monochromator is
added to the spectrometer in order to optimize performance in the 225 nm
region. The stopped-flow CD cuvette is at the end of a short umbilical which is
attached to the SX.18MV stopped-flow sample handling unit. The umbilical
mounted cell slides in or out of the sample holder thus allowing ready inter-
change between the stopped-flow cell and rectangular spectrometer cuvettes. A
mercury-xenon lamp can be substituted for the standard xenon lamp used in
the SX.18MV spectrometer, so as to improve CD kinetic performance at selected
wavelengths. The stopped-flow sample handling unit uses a pneumatic drive.
Ratio mixing thus requires selection of appropriately sized drive syringes. The
viewport on the CD stopped-flow cell has the dimensions 1 0 X 3 X 2 mm (i.e. 60
ul). The dead time for the instrument varies according to the drive ratio (about 4
ms for a 1:1 mix and 8 ms for a 1:10 mix). For a research group already using an
SX.18MV spectrometer, the CD.2C provides a convenient upgrade path for the
measurement of CD kinetics.

3.2 Integrated stopped-flow CD systems

The Bio-Logic system MOS-400 plus the SFM-3 system is the simplest stand-alone
stopped-flow CD system. It is capable of being configured as a multi-functional
spectrometer allowing absorption, fluorescence and CD measurements to be
made. Although the instrument is optimized for single wavelength kinetics, the
addition of motorized monochromators permits steady-state wavelength scan-
ning. When configured for stopped-flow operation, the stopped-flow observation
chamber can be removed and replaced by a simple 1 X 1 cm spectrometer
cuvette holder. Simultaneous absorption and CD measurements can be made
and, for samples where it is permissible to use the same wavelength for fluor-
escence excitation as for the CD measurement, simultaneous fluorescence and
CD data acquisition is allowed.
The Applied Photophysics n*-180 kinetic spectropolarimeter, is a fully inte-
grated kinetic and steady state CD instrument. The -n-*-180 instrument is
designed in a modular format being built around a high-performance direct
coupled optical bench with built in control and acquisition electronics. Alternate
sample handling units, both kinetic and steady state, can be coupled to its

273
ALISON RODGER AND MICHAEL J. CAREY

optical interface. The principal difference between the above mentioned CD.2C
and the n*-180 instrument is that the n*-180 is purpose designed both for high-
performance steady-state CD scanning and quality CD kinetics as well. The
optical design allows CD measurements to be made down to 180 nm. It uses
the same stopped-flow cell as the SX.18MV instrument (10 X 2 X 1 mm; 20 ul
volume) which is mounted in a thermostatted, rotatable cell block so that either
2 mm or 10 mm optical paths are readily accessible (changing the optical path in
this way conveniently allows the same protein concentration to be used for
kinetic measurements in the near UV as for the far UV). Fluorescence may be
directly observed from the stopped-flow cell via a suitable cut-off filter and both
fluorescence and CD detectors can be mounted at the same time. Research
groups interested in protein studies will find switching between CD and fluor-
escence, by software alone, particularly convenient. The dead time on this
instrument is approximately 1 ms for a 1:1 mix and 3 ms for a 10:1 mix. The base
instrument also provides high quality absorbance detection and a range of inter-
esting upgrade paths, including fluorescence emission scanning and anisotropy.
The software suite provides complete instrument control as well as many novel
digital optimizations to improve data quality and acquisition efficiency. The
software is multi-tasking so that activities such as data manipulation and
analysis can be pursued whilst the next set of data is being collected.

4 Additional experimental considerations

4.1 Parameters
The choice of parameters for kinetic experiments and particularly for stopped-
flow experiments are based on the same considerations as for wavelength scans
(see Chapter 4). Signal to noise ratio increases with: \/n, where n is the number
of times the data (in this case the kinetics plot) is accumulated and the data
averaged; vA, where T is the time over which the machine averages each data
point; and \/I, where I is the intensity of the light beam (which is in turn in-
fluenced by the bandwidth). For kinetics, in contrast to wavelength scans, a new
sample is required for each new scan, thus increasing n may be expensive as
well as time consuming. Ideally each kinetics run should be reviewed before
being included in an average, as experiments may be prone to 'glitches' due to
poor mixing or air in the cell. T should be chosen as large as possible consistent
with the kinetics of your run—thus if millisecond resolution is required, choose
T to be 1 ms (0.001 s).
Almost all stopped-flow CD experiments are performed at a single wave-
length. The wavelength chosen should be one where a significant change in CD
is observed between reactants and products. Usually one chooses a (positive or
negative) maximum or a minimum of one species. It is thus difficult to design a
stopped-flow CD experiment without access to a wavelength scanning instru-
ment. Unless the bands are very narrow, a large bandwidth, b, (say 2 nm or
larger) can be chosen to optimize the intensity of light incident on the sample.

274
 STOPPED-FLOW CIRCULAR DICHROISM

The data interval, d, should be chosen to be approximately the same as T

the time constant. If this means you are collecting too many data points for
the length of kinetics run you wish to perform, then either reconsider your
choice of T or use a variable (such as logarithmic) time axis. If d is signifi-
cantly longer than T, then only every few data points measured is recorded. If it
is significantly shorter, then more-or-less identical data points are being
recorded.

4.2 Zero-time and dead-time calibration

The time t = 0 value chosen for a stopped flow kinetic experiment may be
determined by a number of different methods, none of which is perfect. The
simplest is to assume that the first measurement is the appropriate value. How-
ever, one has to be sure that the mixing and flow characteristics of the system
are not giving a spike at the beginning of the data record. If the kinetics are first
order, then it does not matter if t = 0 is incorrectly defined. However, if the
kinetics are not known to be first order, then an alternative approach is re-
quired. Unfortunately, measurement of the CD signal of the reactants and then
estimating the expected t = 0 value does not seem to work. If necessary, the
best option seems to be to mimic the kinetics experiment but remove a key
reagent each time and replace it with a non-reacting non-CD active species
(ideally of the same viscosity etc.) then add the component CD intensities. On a
multi-function instrument, a good estimate of the true t = 0 point can be
determined by using a reaction monitored by absorption or fluorescence (see
Chapter 8).

4.3 Baseline
The acquisition of blank (baseline) runs for comparison with the kinetic data set
is to be recommended. Data obtained from suitable blank runs should be sub-
tracted from the main kinetic data in order to correctly define the CD change of
interest (i.e. to correct for any CD offset together with any pressure induced
artefact present immediately after mixing).

4.4 System tests

Before performing stopped-flow kinetic measurements, the instrument calibra-
tion should be checked using the spectral bandwidth that is to be used for the
set of experiments. An aqueous solution of (+)-10-camphorsulphonic acid (CSA)
is used for calibrating in the near-UV at 290.5 nm and D-(-)-pantoyllactone can
be used to calibrate the far-UV at 219 nm. A simple and convenient reaction for
performing initial checks on a stopped-flow CD instrument is the hydrolysis of
glucuronolactone (1 mg/ml) by sodium hydroxide (1 M). The change in CD signal
is measured at 225 nm over 0.5 s (Figure 1).
In relation to protein folding, lysozyme (Figure 2) and cytochrome c (Figure 3)
provide convenient test systems. Lysozyme, for example, is prepared in 6 M

275
ALISON RODGER AND MICHAEL J. CAREY

0.2 0.3
Time/s
Figure 1 Kinetics of the hydrolysis of glucuronolactone (Sigma G-8875, 1 mg/ml in water) by
potassium hydroxide (1 M) in 1:1 mixing ratio. Data shown is from a single run measured on
a n*-180. The pathlength is 2 mm, bandwidth 4 nm, the light source is a Xe lamp and the
detection wavelength is 225 nm. A blank has been subtracted. Data can be fitted to a single
exponential: (89.7 ± 0.6)exp[(-8.41 ± 0.13)x] - (0.37 ± 0.34).

Gdn-HCl (Figure 2) with 10 mM phosphate buffer (about pH 7). Dilution by buffer

(1 in 11, i.e. 1 volume of sample plus 10 volumes of buffer) will produce a
biphasic change in the CD signal at 225 nm as folding takes place. The signal
amplitude at 225 nm is approximately 34 mdeg when the final lysozyme
concentration is 200 ug/ml (final Gdn-HCl concentration will be 0.55 M). For
protein folding measurements (Figures 2 and 3), it is usual to average at least 16
records together before performing kinetic analysis. With some instruments, it
may be necessary to average well in excess of 16 records in order to produce a
kinetic record of suitable signal-to-noise to permit accurate data analysis. Recent
instrumentation, however, can produce an analysable kinetic record using a
single stopped-flow drive as illustrated in Figure 2.
Because CD performance drops off quite rapidly in the far-UV, it is advisable
to check the absorption spectra of the starting material together with any other
reagents known or suspected to be present during a run. Although CD signals
will increase as the sample concentration increases, if the total absorption (in-
cluding all chiral and achiral species) increases above 0.8 AU, data quality will
start to degrade rapidly as far less light will then be transmitted through the
sample. Other species that absorb at the wavelength of interest should be as
dilute as possible. This causes particular problems with some buffers and de-
naturing agents in protein folding/unfolding experiments, as one is often trying
to probe at wavelengths below 230 nm and many reagents required at quite
high concentrations absorb light in this region (see Figure 4).
All solutions for stopped-flow measurements should ideally be degassed,
though if cavitation (as shown by spikes in the signal due to 'bubbles' caught in
the flow cell) does not occur it may not be necessary. A simple degassing system
is to sonicate samples under slight vacuum as illustrated in Figure 5.

276
 1.5 -1.5 -1
Time/s Time/log(s)

10

-10

3 -20 O
o -a
-a
-30 m
o
(d)
-40
0 0.5 1 1.5
Time/s Time/s o
Figure 2 Kinetics of the folding of lysozyme from hen egg white (of Boehringer 1243004, 2.2 mg/ml in 6 M Gdn-HCI and 10 mM
o
phosphate buffer pH 6.85) diluted 1:10 by phosphate buffer (10 mm phosphate buffer pH 6.85). Pathlength is 2 mm, bandwidth 4 nm, c
the light source is an HgXe lamp and the detection wavelength is 225 nm. Data shown are from a n*-180. (a) Single shot, (b) single
shot with log time base, (c) average of eight shots, (d) eight shots also showing blank run. Data from the eight-shot run can be fitted
o
to a double exponential: (16.0 ± 0.3)exp[(-42.2 ± 1.5}x] - (11.03 ± 0.12)exp[(-2.54 ± 0.05)x] - (23.39 ± 0.033). o
I
3D
O
C/5
ALISON RODGER AND MICHAEL J. CAREY

Figure 3 Kinetics of the folding of cytochrome c from horse heart (Sigma C-2506,
2.2 mg/ml in 6 M Gdn-HCI and 10 HIM phosphate buffer pH 6.85) diluted 1:10 by phosphate
buffer (10 mM phosphate buffer pH 6.85). Pathlength is 2 mm, bandwidth 4 nm, the light
source is an HgXe lamp and the detection wavelength is 225 nm. Data shown are from a
7r*-180 averaged over 16 repeat iterations. Data can be fitted to a double exponential:
(14.48 ± 0.19)exp[(-41.4 ± 0.9)x] + (7.08 ± 0.05)exp[(-1.233 ± 0.03)x] - (28.80 ±
0.05).

250 300 350

Wavelength / nm
Figure 4 Absorption spectra of Gdn-HCI (6 M in water) of the same stated purity from
different batches and/or different suppliers. The sample with the significant long wavelength
absorbance is unsuitable for spectroscopic experiments. See, for example, (1).

5 Examples
A literature search with key words such as <stopped-flow> or <kinetics> and
<CD> or <circular dichroism> and <millisecond> together with words for
particular types of systems, such a protein, will reveal many examples of
stopped-flow experiments. One area where stopped-flow CD has been a key
technique is that of protein folding and unfolding. In this section we shall
briefly review the role of stopped-flow CD in elucidating folding pathways for

278
 STOPPED-FLOW CIRCULAR DICHROISM

Manometer

Tubing
adapter
Reduction
adapter Three-neck round-
bottomed flash
Expansion
adapter

Ultrasonic bath

Conical flask
Figure 5 A simple system for degassing solutions.

lysozyme. The chapter concludes with a final example where the kinetics are
due to the simultaneous interaction of three species one of which is a catalyst.
Varying the concentration of the catalyst in the experiments takes them from
the time domain where manual mixing is feasible to millisecond and faster
timescales.

5.1 Stopped-flow CD and lysozyme folding

The folding of lysozyme has been extensively studied by a wide range of tech-
niques and the progress that has been made in understanding this subject has
been in large part due to complementary information from different tech-
niques. Stopped-flow CD has played a pivotal role in the whole story, although
the final answer has yet to be achieved. Our first lysozyme experiments (2) gave
traces similar (but with worse signal:noise ratios) to that of Figure 2. The sharp
increase in the negative signal at short times seemed to be an artefact. This
impression was reinforced by the fact that the kinetics probed in the far-UV
region seemed to be different from that indicated by the near-UV region.
However, the result has since proved to be real and illustrates the importance of
determining the wavelength dependence of the kinetics in complex systems
(3-5 and references therein). Data from complementary techniques is also
important for determining kinetic mechanisms. In this instance, the apparent
discrepancy between the near- and far-UV kinetics data arises from the fact that
the far-UV CD (see Chapter 4, Sections 5.3 and 5.4) is dominated by the form-
ation of secondary structural units (such as the helices), whereas the near-UV
region probes mainly individual tryptophans whose CD signal arise from

279
ALISON RODGER AND MICHAEL J. CAREY

coupling with the folded protein in their neighbourhood. Therefore, the near-
UV CD is a more direct probe of tertiary structure.

5,2 DNA as a catalytic template

One of the earliest kinetic reactions to be followed by CD was the racemization
of [Fe(l,10-phenanthroline)3]2+. This takes place with a half-life of approximately
20 min at room temperature. [Co(l,10-phenanthroline)3]3+ when synthesized
under fairly rigorous oxidation conditions (we found chlorine gas was effective1
(6)) is essentially stable at room temperature. Literature claims of rapid race-
mization, e.g. (7), are in fact the result of the presence of catalytic amounts of
the labile Co(II) complex; electron transfer occurs readily between the different
oxidation states of the tris-phenanthroline complexes.
When enantiomerically pure [Co(l,10-phenanthroline)3]3+ and racemic
[Co(l,10-phenanthroline)3]2+ are mixed in the presence of DNA, the DNA causes

14 uMCo(lll)
•o
28 uMCo(lll), k=8.1 S~1
I -10 ,
O 56 uMCo(lll), fc=5.6s-1
83 uMCo(lll), *=4.3s~ 1

-15 110 uMCo(lll), k=4.0S~1

140 uMCo(lll), k = 3.0s-1

100 200
Time/s

Figure 6 Single-shot stopped-flow kinetics traces from the mixing of [Co(l,10-

phenanthroline)3]3+ with a solution of [Co(l,10-phenanthroline)3]2+ and calf thymus DNA
(450 uM and 158 uM, respectively, in the reacting mixture). The +3 charged Co(lll) complex
binds much more strongly to DNA than does the Co(ll) complex, hence its smaller
concentrations. The third syringe was filled with water. Data were collected using a Jasco J-
720 spectropolarimeter with a Biologique SFM-3 attached between the sample compartment
and the photomultiplier tube. The light beam was focused with a single lens. Data were
collected with a band width of 2 nm, response time of 1 s (so comparatively slow for
stopped-flow) at 307 nm just outside the DNA absorption region.

1
The racemization of [Fe(l,10-phenanthroline)3]2+ is temperature dependent and the enan-
tiomers can be resolved with antimonyl tartrate at 0°C. Antimonyl d-tartrate is commercially
available and most resolutions of metal complexes using it involve collecting the precipitate as
one enantiomer, and precipitating what remains in solution as the perchlorate to give the other
enantiomer. The enantiomeric purity of the more soluble isomer is usually much lower than
that of the less soluble salt. Potassium antimonyl i-tartrate is not commercially available but it
may be synthesized by heating an aqueous slurry of Sb2O3 + KOH + 1-tartaric acid (in molar ratio
1:1:2) under reflux for 2-3 h. Slow evaporation in a fumehood gives the required compound in
crystalline form. [Co(l,10-phenanthroline)3]3+ may also be resolved using antimonyl tartrate
isomers.

280
 STOPPED-FLOW CIRCULAR DICHROISM

a significant increase in the catalytic activity of Co(II) (8). The CD change as a

function of time in an individual run follows a simple exponential decay (Figure 6);
however, the dependence on the concentration of each reagent is not straight-
forward. To probe the concentration dependence of the racemization with only
three syringes requires one syringe to be pre-loaded with two of the reagents.
Some illustrative data are shown in figure 6. As the concentration of the chiral
reacting species, Co(III), increases, the initial CD signal increases but the rate
decreases. The Co(II) and DNA concentrations also affect the reaction in a man-
ner that does not correlate simply with their concentration. These observations
can be understood when it is realized that the DNA is forming a template to
facilitate the electron transfer reaction between two cobalt tris-phenanthroline
complexes of different oxidation states. The DNA may also be playing a more
active role in the electron transfer. Increased concentrations of Co(II) make such
encounters on the DNA more likely, until it is in such excess that it is inhibiting
Co(III) binding. Conversely, decreasing the DNA concentration increases the
likelihood of an encounter between the reacting species molecules until the
same saturation of the kinetics occurs.

References
1. Mo, J, Holtzer, M. E. and Holtzer, A (1991) Biopolymers 31, 1417-1427.
2. Radford, S. and Rodger, A. (1988) Unpublished data.
3. Kulkarni, S. K., Ashcroft, A. E., Carey, M., Masselos, D., Robinson, C. V. and Radford, S.
E. (1999) Protein Science, 8, 35.
4. Weissman, J. S. and Kim, P. S. (1991) Science, 253,1386.
5. Evans, P. A. and Radford, E. S. (1994) Current Opinion in Structural Biology 4,100.
6. Lee C. S. (1966) Inorg. Chem. 5,1397.
7. Rehmann, J. P. and Barton, J. K. (1990) Biochemistry 29,1701.
8. Bates, P. J. and Rodger, A. (1992) Unpublished results.

281
This page intentionally left blank
 Chapter 11
Spectrophotometry and
fluorimetry of cellular
compartments and intracellular
processes
C. LINDSAY BASHFORD
Department of Biochemistry and Immunology, Cellular and Molecular Sciences
Group, St George's Hospital Medical School, Cranmer Terrace, London SW17 ORE

1 Introduction
Optical spectroscopy, Spectrophotometry and fluorimetry can be used to monitor
processes occurring in living cells provided that suitable chromophores are
present which 'report' on the events in which they participate. The advantages
of optical techniques are manifold. Firstly they can be fast—with appropriate
apparatus events in the pico- and nano-second domains can be studied by fluor-
escence spectroscopy. Secondly they are continuous—instant feedback from the
experimental system can guide the most complex of experimental protocols,
and allow the experimenter to adjust system parameters as necessary. Thirdly
they are convenient, and most laboratories have access to equipment that can
provide quantitative analysis of optical signals; examples include conventional
spectrophotometers/fluorimeters, dedicated instruments (e.g. for fluorescence
lifetime and polarization measurements), cameras, microscopes and plate
readers. Significantly detectors from one apparatus can often be used on others
to open up new experimental protocols. Fortunately the principles underlying
the use of such a diverse array of optical devices are straightforward and
universal—they apply just as much to laboratory 'work-horse' instruments as
they do to the most specialized, laser-illuminated fluorescence microscope. The
availability of fast laboratory computers with large storage capacities means that
most modern spectrometers are microprocessor controlled and digitization of
signals opens up the full range of possibilities of data accumulation, storage,
analysis and interpretation.
The main problem with optical measurements is not the acquisition but
rather the interpretation of the data obtained. Straightforward analysis of the
results depends on the clarity of the experimental design and the appropriate
choice of chromophore. This chapter describes some of the problems that can
be addressed by spectroscopic techniques and attempts to give guidance on
good experimental design.

283
C. LINDSAY BASHFORD

2 Experimental design
2,1 Apparatus
Optical spectroscopy requires either spectrophotometers, to measure absorb-
ance, fluorimeters, to measure fluorescence, or microscopes, which can measure
fluorescence or absorbance of single cells or small groups of cells. Fluorimeters
and spectrophotometers usually require solutions or suspensions of material in
conventional cuvettes; microscopes provide two-dimensional images from
smears, slices or surfaces. Other devices that record signals resolved in two-
dimensions include gel scanners and microplate readers. Essentially these de-
vices sample the 'object' in an organized manner (detectors can be set up to
record absorbance or fluorescence) and information is stored in an electronic
array that maps precisely the physical layout of the original object.
Fluorimeters and spectrophotometers may have attachments that permit
recording of signals from surfaces, or can be adapted to record such signals
using fibre optic 'light guides'. In this case, the guide (a bundle of optical fibres)
conducts the appropriate illumination to the surface and both reflected and
fluorescent light are collected by other fibres and conducted to a detector,
usually a photomultiplier (1). The reflected light has two components: specularly
reflected light, this is the mirror-like reflection, and diffusely reflected light.
Only the latter contains information (analogous to absorbance) concerning the
chromophores present at the surface.
Light output from the image plane of an optical microscope can be assessed
with the light metering system of a camera attached to the microscope. The
time taken correctly to expose a film of given speed (ASA/DIN) is inversely
related to the brightness of the image. The exposure time, that is the interval
between the opening and the closing of the shutter, is measured with a stop-
watch. For accurate measurements the exposure time should lie in the 10-100 s
range; shorter exposures are difficult to time with appropriate precision and
longer exposures are compromised by stray light entering the objective (see
Section 2.1.2 for discussion of stray light) and by photobleaching. The correla-
tion between intensity and reciprocal exposure time is linear over a very wide
range (1), so this procedure provides a handy 'rule-of-thumb' comparison of the
brightness of images. The main disadvantage of this simple approach is that no
allowance is made for the differing brightness of objects in the field of view,
indeed most of the background may be black. Most conventional camera meters
weight light in the centre of the field of view more heavily than that in the
periphery.
A digital approach can resolve the difficulty of signal variation across the
image and CCD (or similar 'digital') cameras are now routinely found in the
camera port of optical microscopes. They provide digital images that can be
analysed pixel by pixel. A wide range of image analysing software, not all of it
specific to particular commercial packages, is available. In such processing, the
key decisions that have to be taken are the zero level, sometimes called the
'black level', and the gain employed. If zero and gain are not chosen correctly a

284
 FLUORIMETRY OF CELLULAR COMPARTMENTS AND INTRACELLULAR PROCESSES

very distorted impression of relative levels can be achieved. At high magni-

fications photobleaching, that is a time-dependent loss of intensity, usually
occurs and it is necessary to arrange for short exposure times and minimal
exposure of the field to the excitation beam.

2.1.1 Artefacts
Optical investigations of biological systems are beset with hazards and it cannot
be too strongly emphasized that very rigorous criteria must be applied before
interpreting the experimental data. Artefacts are more common in fluorescence
experiments and a few of the most easily avoided ones are considered here.

2.1.2 Stray light and turbid solutions

Stray light is a general term for any light that reaches the detector which is
irrelevant to the experiment in progress, e.g. either not of fluorescence origin or,
in an absorbance experiment, of a wavelength different from that intended. It
may include ambient light and 'leaks' through the monochromators and filters
from the source. It is avoided by correct apparatus construction, the correct
choice of blocking filters (to exclude higher order diffraction and specular re-
flection from the grating in the monochromator) and the appropriate use of dark
rooms and/or black curtains. It is important to understand that most detectors
provide an output (usually a voltage) even when no light is present. Operators
should aim to make this 'dark current' equivalent to 'zero light' (0% trans-
mission) or the background ('black') level. With some devices this selection of
true zero may be affected by automatic regulation of gain and/or contrast; in
such cases, the most precise measurements may only be possible in the manual
operation mode where the observer can choose for themselves the correct
setting.
Suspensions of biological materials are turbid. It is especially important in
measurements of fluorescence to resolve fluorescence from scattered light. Since
scattered light usually exceeds fluorescence by a substantial margin, careful
choice of filters and slit widths is essential to record just fluorescence (2). It is
important to be aware that substantial changes in turbidity, and hence in the
absolute value of fluorescence or absorbance, occur when membrane vesicles
are exposed to sudden changes in osmolarity or are exposed to permeant salts
(3). In order to maximize the chances of 'capturing' the emitted/transmitted
photons, the cuvette is usually placed as close as possible to the detector.
Turbid solutions can be useful when the chromophore has low extinction or
can only be used at very low concentration. This arises because the multiple
internal reflections amongst the particles greatly increases the effective path-
length of the absorbance experiment and hence increases the amount of light
absorbed (a consequence of Lambert's Law). In such circumstances, the dual
wavelength optical arrangement is strongly recommended: absorbance at one
wavelength is measured in the same cuvette with respect to light of a slightly dif-
ferent wavelength; changes in turbidity affect both beams equally and are thus
cancelled out (1,4). The ideal situation is for the reporter group to show a large

285
C. LINDSAY BASHFORD

response at one wavelength and a small (or oppositely directed) one at a neigh-
bouring wavelength; this is most likely to occur in compounds which have
intense, narrow ('sharp') absorbance bands.

2.1.3 Inadequate excitation

The excitation light may be absorbed by non-fluorescent material, such as occurs
when monitoring the surface fluorescence of tissues or organs. Blood com-
pletely absorbs the light used to excite fluorescence due to pyridine nucleotides
and flavoproteins, hence an increase in the amount of blood in the field of view,
due to vasodilation, will decrease fluorescence and lead to an erroneous con-
clusion that the oxidation-reduction state of the tissue mitochondria has
changed. This problem is obviated by recording the ratio of the pyridine nucleo-
tide (PN) and flavoprotein (FP) fluorescence: as mitochondria become reduced
PN fluorescence increases and FP fluorescence decreases, as the mitochondria
become oxidized the changes of fluorescence are reversed. Hence the FP/PN
fluorescence ratio is a sensitive indicator of the mitochondrial oxidation-
reduction state (5). It is not severely affected by masking problems as both
signals are affected equally so that their ratio remains unaltered. An alternative
procedure for correcting for the interference of blood is to obtain the 'corrected'
fluorescence, namely the difference between the measured fluorescence and
the measured reflectance (of the excitation light). In this mode, loss of excita-
tion, due to absorbance by the blood pigments, leads to a loss of reflectance
which compensates for the loss of fluorescence. The adequacy of this procedure
has been demonstrated in perfused systems where the blood can be replaced by
a transparent perfusate (6), but its applicability in vivo is less certain. In many
applications the ratiometric technique is the method of choice to overcome the
problem of heterogeneous distribution of the indicator molecules. In very high
resolution optical microscopy, the slight difference in refraction of excitation
beams of differing wavelength may mean that an image perfectly focused for
one wavelength is marginally out of focus for the other. Fortunately this
discrepancy is usually sufficiently small to lie within the range of experimental
error of most quantitative measurements.
The excitation light may be absorbed so strongly by the fluorophore that
fluorescence is reduced along the excitation path, one aspect of the inner filter
effect (1, 2). Such an effect should be suspected if dilution of a sample leads,
initially, to a fluorescence increase. This could arise, for example, if the detector
'observes' a part of the sample reached by the excitation beam only in dilute
solutions. The expected linear relationship between fluorescence and con-
centration occurs only with dilute (<10~3 M) preparations (1). Remember that it
is not only the fluorophore that may absorb the excitation and that it is the
overall absorbance that must be kept low to avoid any inner filter effects.
Inner filter effects can be reduced by altering the geometry of the fluor-
escence experiment: fluorescence from the surface of highly absorbing material
is much less sensitive to the effect, because neither the excitation nor the
emission has to traverse an opaque medium in order for a fluorescence signal to

286
 FLUORIMETRY OF CELLULAR COMPARTMENTS AND INTRACELLULAR PROCESSES

be recorded. Indeed, if a microscope with an epifluorescence attachment is used

to excite the material, the effective pathlength is of the order of a few microns
and little filtering occurs. This can lead to confusion if results obtained with a
microscope are compared with those obtained in a conventional fluorimeter.
Cyanine dyes are avidly accumulated by mitochondria in vivo (7) with a resultant
quenching of their fluorescence when assessed in cell suspension (8); however,
in the fluorescence microscope the mitochondria appear as brightly fluorescent,
filamentous structures (7). Similarly acridine orange and 9-aminoacridine are
accumulated by acidic endosomes with a quenching of their blue fluorescence;
in the fluorescence microscope the vesicles appear as yellow or orange vesicles,
this being the characteristic emission of very concentrated solutions of these
dyes (1).

2.2 Light sources

The selection of light source is determined by the nature of the experiment. For
most determinations of fluorescence the most important factor is providing
sufficient excitation to generate enough fluorescence to detect. The most intense
sources are lasers, and they have the added advantages that they have very well-
characterized wavelengths containing essentially no 'stray' light of other wave-
lengths and their output level is very steady. The latter is particularly important
when monitoring small changes of fluorescence with time—ultimately the
sensitivity of such experiments is limited by the time-dependent variation in
excitation level of the source. The disadvantage of lasers is that they only
produce a limited range of excitation wavelengths which are not suitable for all
potential fluorophores. The widespread availability of 488 nm, 543 nm and 610
nm lasers means that many potentially useful fluorophores have been deliber-
ately synthesized to have spectral properties that match those wavelengths. 488
nm and 543 nm are the wavelengths used for the 'fluorescein' ('green' channel)
and 'rhodamine' ('red' channel) of fluorescence microscopes.
After lasers the next most intense sources are arc-lamps, usually mercury or
xenon. Mercury arcs provide a very intense output at 366 nm with reasonable
levels at other wavelengths; xenon arcs provide a more balanced output across
the ultraviolet and visible regions of the spectrum, although users should be
aware that there are a number of significant peaks even in the xenon arc
output. Arc lamps run at high temperatures and generate ozone so they must be
well-ventilated and cooled. In addition the ultraviolet output is so strong that
they should not be observed with the naked eye—indeed no high intensity light
source should be so observed. For some fluorescence applications pulsed arc
lamps are particularly useful if the detector is synchronized to the source. This
arises because the power output during the pulse is very high while overall
output is much lower. A 30 watt pulsed xenon source can be as effective as a
150 watt continuous source for this reason.
In absorbance experiments where transmission is monitored in the presence
and absence of sample a very steady light output is required, often not of

287
C. LINDSAY BASHFORD

particularly high intensity. Filament lamps are usually perfectly satisfactory in

this regard.

2.3 Wavelength selection

Choice of the correct wavelength is essential for all optical measurements.
Usually the procedure is straightforward: for absorbance choose the wavelength
available from your source that most closely matches the absorbance maximum
of the material you are monitoring; use the same criterion when selecting an
excitation wavelength in a fluorescence experiment. If the molecule you wish to
study has spectral properties that overlap with other material in your sample, it
may be necessary to choose an alternative wavelength, as a rule of thumb it is
wise to select an absorbance/excitation where your probe contributes at least
90% of the total absorbance at that wavelength. Remember that when single
wavelengths are employed a change in signal may arise either from a change in
the absolute level of absorbance/fluorescence or from a shift in the absorbance/
fluorescence spectrum. Indeed it is always good practice, where possible, to
record the spectrum of the response to help in the interpretation of the
results.
In fluorescence measurements, the emission wavelength must be sufficiently
different from the excitation wavelength that scattered and 'stray' light do not
contribute to the output. Again inspection of excitation and emission spectra
will guide the choice of wavelength. Fluorophores with very high quantum
yields, such as fluorescein, have strongly overlapping excitation and emission
spectra. To provide enough excitation and emission without the problems of
scattered/stray light it is usually necessary to choose an excitation wavelength
to the blue of the excitation maximum and an emission wavelength to the red
of the emission maximum.

2.4 Chromophore selection

Useful chromophores may be endogenous, or 'intrinsic', that is, already present
in the system of interest, or exogenous, 'extrinsic', pigments provided by the
observer. In the former category, pigments which absorb light in the visible
region of the spectrum (400-700 nm) are the most useful because they are
restricted both in abundance and in the range of processes in which they
participate. Porphyrins, flavins and carotenoids exhibit characteristic spectra
with rather narrow, intense absorbance bands. Thus the role of cytochromes in
mitochondrial electron transport was appreciated long ago by Keilin (9), as a
result of his studies of insect flight muscle using a simple spectroscope. In
addition the cofactors most commonly found in biological oxidation-reduction
reactions, namely the pyridine nucleotides (NAD, NADP) and the flavins are
fluorescent in either their oxidized or their reduced forms. In photosynthetic
systems, in addition to chlorophyll, the carotenoids, secondary 'antennae pig-
ments', serve as 'molecular voltmeters' since the positions of their absorption
peaks alter as a function of the electrical potential across the photosynthetic

288
 FLUORIMETRY OF CELLULAR COMPARTMENTS AND INTRACELLULAR PROCESSES

membrane (10). This property is useful for studying the mechanisms of energy
transduction in photosynthesis, and stimulated an extensive search for exo-
genous pigments with similar properties which could be used for studies of
membrane potential in a wide variety of systems (11).
In addition to exogenous pigments which register membrane potential, dyes
have been employed to monitor many cellular properties; for example, cyto-
plasmic and organelle pH, cytoplasmic Ca2+ and Mg2+ concentration, membrane
microviscosity, fate of ingested particles. More recently probes have been
developed to report on processes such as apoptosis, critical to cell survival. In
these cases the molecule in question is specifically designed: (i) sensitively to
register the parameter of interest; and (ii) simply to reach its ultimate destina-
tion. Probes of cytoplasmic Ca2+ or pH exhibit characteristic fluorescence or
absorbance depending on the cations in their environment. Membrane-permeant
esters of the indicators diffuse to the compartment where endogenous enzymes
catalyse their hydrolysis to generate the active species in situ. More specific
delivery may be accomplished by exploiting natural endocytotic and/or bio-
synthetic processes. The range of possible probe experiments is so great that it
is impractical to describe each in detail. The publications of Molecular Probes
(their catalogue, the series of leaflets called BioProbes and their very user-
friendly web site http://www.probes.com) provide the most extensive intro-
duction to the types of probe experiments available and the literature from
which those approaches were drawn. It is not always appreciated how simple
many of the experiments are to perform using equipment available in most
laboratories.

2.4.1 Criteria for chromophore selection

The primary decision, and often the most difficult, is the choice of a suitable
chromophore which will provide information pertinent to the problem under
consideration. In the case of intrinsic chromophores this must be a choice of
system, namely the one in which the chromophore occurs and in which it
exhibits the desired properties. For example, the carotenoid pigments which
record the light-dependent membrane potential associated with photosynthesis.
Such pigments occur in many photosynthetic membranes, but their use as
'molecular voltmeters' is most easily accomplished in preparations, known as
chromatophores, from certain photosynthetic bacteria such as Rhodobacter
sphaeroides or Rhodopseudomonas capsulata. Again, studies of oxygen consumption
and delivery in vivo require highly vascularized tissues which contain many
mitochondria, for example, brain, liver or cardiac muscle (although in the latter
case the myoglobin contribution cannot be resolved from that of haemoglobin).
In each case, however, the oxygen carriers tend to mask the much less
abundant mitochondrial pigments. It is not always easy to resolve contributions
from endogenous pigments in cell suspensions as the cell density required to
give a suitable output is often too large to be compatible with provision of an
adequate oxygen supply.
Whether the final decision is to use an endogenous or an added chromo-

289
C. LINDSAY BASHFORD

phore there are a number of factors that always have to be taken into account
(6,12).

1. Instrument compatibility. Only chromophores with characteristics compatible

with the instruments available are suitable. The main features to consider are
excitation wavelength, availability of appropriate emission filters and the sensi-
tivity of the detector. For example, many dyes with red fluorescence are usually
adequately excited by 543 nm laser light and detected via the 'rhodamine' filter
sets provided with commercial fluorescence microscopes; likewise dyes with
green fluorescence are suited to the 'fluorescein' set up.

2. Colour. Fluorescence instruments may be able to register signals from more

than one dye in the same experiment. Thus it is possible, for fluorescence micro-
scopy, to include dyes with green and red fluorescence and monitor their out-
put using successively, in the same sample, the 'fluorescein' and 'rhodamine' set
ups. If the images are stored separately, it will subsequently be possible to
combine them, in which case regions containing both red and green fluor-
escence will appear orange and it is possible to tell whether the molecules
segregate to the same or different regions of the image. This can be particularly
useful for following the movement of different materials through intracellular
compartments.

3. Signal intensity. In absorbance experiments, dyes with high extinction co-

efficients provide the largest signals. In fluorescence experiments, the output
depends on both the efficiency of absorbance (of the excitation beam) and of the
fluorescence emission (usually characterized by the quantum yield (Q): a Q, value
of 1 indicates that every photon absorbed is re-emitted as fluorescence, a Q
value of 0.1 indicates that one in every ten photons absorbed is re-emitted as
fluorescence). Good compromises can often be achieved using highly fluor-
escent dyes with less than optimal excitation or by using weaker emitters but
with very efficient excitation.

4. Stability. Nearly all chromophores will be 'bleached' by persistent high in-

tensity excitation. This is because the excited state of the chromophore usually
has enhanced chemical reactivity compared to the ground state, so strong
excitation promotes formation of non-fluorescent/non-absorbing material,
especially if oxygen is present.

5. Background interference. All biological materials absorb light at one wavelength

or another. Membranes contain components which absorb near-ultraviolet and
scatter blue light strongly. Probes with fluorescence/absorbance in the red region
of the spectrum are usually the most useful. Remember, however, that many
photometers are less sensitive to red light than they are to blue light; so that the
advantages gained by choosing a probe with little optical 'overlap' with native
pigments are offset by the decreased sensitivity of the apparatus. The fluor-

290
 FLUORIMETRY OF CELLULAR COMPARTMENTS AND INTRACELLULAR PROCESSES

escence of extrinsic probes will be severely quenched if the emission of the

fluorophore overlaps with intense absorbance bands of intrinsic chromophores.
Cellular autofluorescence (due, at least in part, to flavins) is green/yellow and
can interfere with experiments using fluorescein-based chromophores. There is
little autofluorescence in the orange/red region of the spectrum where rhoda-
mine derivatives emit. In favourable circumstances, changes in fluorescence
may be observed even when absorbance changes are masked by endogenous
pigments (13). It is always good practice to view materials without your probe
present to check that endogenous signals are unlikely to complicate the
interpretation of your experiment.

2.4.2 Delivery of exogenous probes

Optical probes used to study membranes are usually lipophilic and can be stored
as stock solutions in solvents such as ethanol or dimethyl sulphoxide. The stock
solution is diluted at least 200-fold with the membrane preparation and the dye
reaches its destination by diffusion. Such labelling is non-specific and in multi-
membrane systems such as intact cells, the probes partition unevenly among
the various compartments. This is helpful if the dye concentrates in the region
of interest, as cyanine dyes do in mitochondria (7) and weak bases do in acidic
endosomes (14, 15). However, if cytoplasmic pH (Section 3.3.2) or plasma
membrane potential (Section 3.2.1) are the object of interest, these dyes cannot
be used in a straightforward way unless measures are taken to suppress re-
sponses from other compartments. Interpretation is more complex when
different systems are labelled. For example, the cyanine dye diO-C3-(6) has been
reported as an excellent mitochondrial stain, when used at 'low' concentration
(less than 10-7 M (7)), and as a stain for endoplasmic reticulum (16) when used at
'high' concentration (greater than 10-6 M). The reason for this is that high
concentrations of diO-C3-(6) poison mitochondria (they inhibit complex 1 of the
respiratory chain (17)), dead mitochondria release accumulated dye which is
now available to accumulate in endoplasmic reticulum.
Slightly more specific labelling can be achieved by delivering an inactive
(usually an ester) form of the dye which is locally converted to the active form
by endogenous esterases. This is the technique used to probe cytoplasmic Ca2+
and H+ levels with fluorescent analogues of EDTA (18,19). Specificity is achieved
by the location of active esterases in the compartment of interest (see Section
3.3.2).
Specific labelling can be achieved by coupling the chromophore covalently to
a vehicle, usually an immunoglobulin or a hormone, that binds specifically to
cellular receptors. This is the technique widely used to identify specific sub-
cellular compartments/locations. Antibody specificity is coupled to the sensitivity
of fluorescence to permit the highlighting of particular objects for which
specific, often monoclonal, antibodies are available. Polyclonal antibodies specific
for determinants on the monoclonal antibody are available with covalently
attached chromophores, whose spectral properties match the excitation and
emission criteria available in most 'fluorescence' microscopes.

291
C. LINDSAY BASHFORD

2.4.3 Toxic actions of chromophores

Chromophores which are non-toxic at low concentrations may become potent
inhibitors if they are concentrated in specific compartments. Cyanine dyes
severely inhibit respiration at site 1 of the mitochondrial respiratory chain, pro-
viding that the inner mitochondrial membrane potential is substantial (inside
negative) (18). In living cells, the inner mitochondrial membrane potential is
about 180 mV and the plasma membrane potential is about 60 mV (both inside
negative), and extracellular cyanine at 10-7 M is at electrochemical equilibrium
with mitochondrial cyanine when the latter reaches 10-3 M. Very low concen-
trations of cyanines may thus adversely affect cellular respiration and energy-
dependent processes (18).

2.4.4 Complex formation

The dye chosen may interact with elements of the system with a consequential
change in its properties. Thus the pKa of a pH indicator in solution may not
necessarily have the same value when the indicator is bound to cellular or other
constituents (see, for example, (20, 21)); indeed the spectra of the bound dye
may differ from those of the free dye, making any simple evaluation of pH
problematical. It may be that the dye is useful as a qualitative indicator of the
parameter under investigation but cannot be calibrated because it forms com-
plexes with agents required in the calibration protocol. It is always good
practice to use several, independent calibration procedures to check for internal
consistency of the dye response. Thus when calibrating optical probes of mem-
brane potential (22, 23) the same values should be obtained when valinomycin/
K+ or uncoupler carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP)/
H+ are introduced.

2.5 Characterization and calibration of optical signals

Once the choice of chromophore and system has been made it is essential to
characterize their optical properties. The spectrum of the chromophore (absorb-
ance or fluorescence) in situ may contain unexpected contributions which will
limit the scope of the experiment. Such investigations establish the signal-to-
noise ratio for the intended observations (noise being used in its most general
sense).
It is unfortunately the case that the calibration of optical signals, be it with
pH, transmembrane potential or cation concentration, often destroys the
experimental system. For example, ionophores irreversibly modify the mem-
brane potential during the calibration of dye indicators. Likewise, entrapped
indicators of cell pH or Ca2+ are calibrated only after the cells have been
completely permeabilized. Once a suitable calibration procedure is estab-
lished, it is possible to record the desired information from the experimental
system.

292
 FLUORIMETRY OF CELLULAR COMPARTMENTS AND INTRACELLULAR PROCESSES

3 Examples
31 Oxidation-reduction state of tissue mitochondria
The properties of mitochonctrial suspensions have long been the centre of
intense study and the characteristics of the pigments of the respiratory chain
have been resolved by dual-wavelength spectroscopy (4; Section 2.4.1). However,
mitochondrial preparations may not reflect all the properties found m vivo.
Direct recording of the respiratory pigments in situ, by absorbance and/or
fluorescence spectroscopy. provides one way of resolving this dilemma. In
anaesthetized animals spectroscopy of exposed tissues (24, 25) reports the
oxidation-reduction state of The electron transport chain and the oxygenation
of haemoglobin/myoglobin. Unfortunately the rapid fluctuations of 02 delivery
and consumption make it difficult to establish convenient steady states. An
alternative procedure is to freeze the tissue rapidly and then to obtain spectra
more leisurely at low temperatures.

Protocol 1
Reflectance spectra of freeze-trapped rat kidney
Equipment and reagents
• liquid nitrogen Wavelength scanning, dual-wave length
• Aluminium tongs for freeze-damping spectrophotometer
('Wollenberger' tongs (26)) Anaesthetized rat

Method
1 Dissect the anaesthetized rat to reveal the kidneys.
2 Chill the Wollenberger tongs in liquid nitrogen. A polystyrene ice bucket is a con-
venient vessel in which to keep the liquid nitrogen; it is easy to break dewar vessels
with aluminium tongs.
3 Bring the cooled tongs close to the kidney.
4 Remove the kidney, place on the cooled tongs and squeeze the tongs together,
5 Keeping the tongs closed return them to the ice bucket containing liquid nitrogen.
6 Open the tongs, remove the crushed, frozen kidney with plastic forceps and store
it in suitably labelled aluminium foil in liquid nitrogen,
7 To record reflectance spectra unwrap the foil surrounding the kidney—in liquid
nitrogen using plastic forceps.
8 Chill the common terminal of the optical fibre connected to the spectrophoto-
meter(l),
9 Under liquid nitrogen press the common terminal of the optical fibre against the
smooth frozen face of the kidney.

293
C. LINDSAY BASHFORD

Protocol 1 cor

10 Record the reflectance spectrum—repeat steps 8-10 for several different regions of
the frozen tissue.
11 Use the white surface of the polystyrene ice bucket as a 'baseline' for computing
the absolute reflectance spectra of the kidney.
12 Use different areas of the frozen kidney surface for computing difference spectra.

Figure 1 shows reflectance spectra of freeze-trapped rat kidneys. Three differ-

ent kidneys were assessed: a normally perfused kidney, and kidneys rendered
ischaemic by tying a ligature round their arterial supply, and freeze-trapped
cither 1 min or 10 mill later. The main features of the visible reflectance spectra
arise from oxyhaemoglobin (peaks at 542 and 577 nm), haemoglobin (peak at
555 nm) and ferrocytochromes aa-, (maximum at 605 nm) and c (maximum at
550 nm). The excess of haemoglobin means that in absolute spectra the ferro-
cytochrome contributions appear as shoulders rather than discrete maxima. In
normal kidney (figure I, trace a) only oxyhaemoglobin is observed indicating
adequate oxygen supply and adequate mitochondrial respiration. After 1 min of
ischaemia (figure 1, trace b), the oxyhaemoglobin content is much reduced and
the shoulders at 605 and 550 nm indicate reduction of the mitochondrial
respiratory chain as the supply of oxygen to the point of consumption, fails. The
general features of the spectrum after 10 min of ischaemia are similar to those

0.50r

500 550 600 650

Wavelength (nm)

Figure 1 Reflectance spectra of freeze-trapped rat kidney. Kidneys from anaesthetized rats
were freeze-trapped (26) before, 1 min and 10 min after occluding the renal blood supply.
Reflectance spectra were recorded with a bifurcated light guide attached to a scanning, dual-
wavelength spectra photo meter (1, 24) using 700 nm as the reference wavelength. The
baseline was the spectrum of white polystyrene. Spectra labelled a and b were recorded
from the normally perfused and 1 min ischaemic kidneys, respectively. Spectrum c is the
difference between the 1 min and 10 min ischaemic kidneys. Reproduced with permission
from C. L. Bashford and M. Stubbs (1986) Biochemical Society Transactions, 14,
1213-1214. © The Biochemical Society,

294
 FLUORIMETRY OF CELLULAR COMPARTMENTS AND INTRACELLULAR PROCESSES

found after 1 min and the difference between these two states (Figure 1, trace c) is
featureless and gives an indication of the baseline for this experiment. Inter-
estingly, oxyhaemoglobin is not absent from the ischaemic tissue suggesting
that there are metabolic compartments where oxygen is delivered but not
consumed (27, 28).
Similar experiments with freeze-trapped gerbil cerebral cortex (24, 29; and
see Figure 2) indicate that only in extreme anoxia does reduction of the mito-
chondrial pigments occur. Figure 2 shows the difference reflectance spectrum
obtained when the reflectance spectrum from a sample obtained from an
animal after breathing 5% oxygen for 1 min is subtracted from that taken from
an animal which had been breathing 0% oxygen for 1 min. In both situations
the haemoglobin was largely deoxygenated and the difference spectrum of the
whole tissue closely resembles that of reduced minus oxidized mitochondria.
This shows that the respiratory chain still has enough oxygen under conditions
where tissue haemoglobin is almost completely deoxygenated; or put another
way the large difference in affinity for oxygen exhibited by mitochondria and
haemoglobin in vitro is still apparent in vivo.
The experiments illustrated in Figures 1 and 2 employed a microprocessor-
controlled, scanning, dual-wavelength spectrophotometer which was capable of
storing and manipulating spectra.

3.2 Membrane potential of cells and organelles

Most biological membranes have the ability to sustain electrostatic potential
differences between the aqueous phases which they separate. This potential,

0.02

0.01

AA

-0.01

500 550 600 650

Wavelength (nm)
Figure 2 Difference spectrum of freeze-trapped cerebral cortex. Gerbil brains were freeze-
trapped (43) 1 min after the anaesthetized animal began breathing either 5% 02, 90% N2,
5% C02 or 95% N2, 5% C02. Reflectance spectra were recorded with a bifurcated light guide
attached to a scanning, dual-wavelength spectrophotometer (1, 24) from the right cerebral
cortex at 77 K, and the difference spectrum was computed from the spectra from each
sample. Figure redrawn from data originally published in (29).

295
C. LINDSAY BASHFORD

usually called the membrane potential, can play a significant part in membrane
activities; it is, for example, critical for signalling in excitable cells and makes a
contribution to the accumulation of nutrients driven by the electrochemical
Na+ gradient across the plasma membrane of many cells. In subcellular organ-
elles, such as chloroplasts and mitochondria, the membrane potential is an
important 'high energy' intermediate in the synthesis of ATP. In only a limited
number of circumstances can the potential be recorded directly and other
methods such as the use of dyes with specific sensitivity to membrane potential
are available (30).
Many dyes change either their absorbance or their fluorescence in response
to a change in membrane potential (11). However, it is often difficult accurately
to calibrate the signal with known potentials. In one procedure, the membrane
permeability to a particular cation, usually K+ or H+, is increased by adding a
reagent called an ionophore. In the presence of sufficient ionophore, such that
all other permeabilities are relatively small, the membrane potential (V) will
approach that predicted by the Nernst equation:
V=(-RT/f)ln([M + ] i /[M + ] 0 ) 1

0.02
i) Val [K+]mM ii) Val [K+]mM
1 40
1 U 1
/ 33
0.01 / 20 /**"

^630-590
UUP 9J

0
[
-^d *^J

-0.01 —H 5 minutes p~

Figure 3 Measurement of Lettre cell membrane potential using oxonol-V. (i) 5 x 106 Lettre
cells/ml were suspended in 5 mM HEPES, 1 mw MgCI2, 2.2 x 10-6 M oxonol-V, pH adjusted
to 7.4 with NaOH at 33°C and 155 mM NaCI plus KCI with the final K+ concentration
indicated. Valinomycin was added as indicated to give a final concentration of 1 x 10-6
g/ml. Traces from four separate experiments are superimposed such that the signal before
the addition of valinomycin is identical. The extracellular K+ concentration at which
valinomycin would give no change in absorbance was estimated to be 7.1 mM by
interpolation. The intracellular K+ concentration was 73.3 mM giving a membrane potential
(see equation 1) under these conditions of -62 mV. (ii) 5 x 106 Lettre cells/ml were
suspended in 150 mM NaCI, 5 mM KCI, 5 mM HEPES, 1 mM MgCI2, 2.2 x 10-6 M oxonol-V,
pH adjusted to 7.4 with NaOH at 33°C. Valinomycin (1 x 10-6 g/ml) and KCI to give the
concentration indicated were added as shown by the arrows. The extracellular K+
concentration at which valinomycin would give no change in absorbance was estimated to be
7.1 mM by interpolation, the intracellular K+ concentration was 63.3 mM giving a membrane
potential (see equation 1) under these conditions of -60 mV. Reproduced with permission
from Bashford, C. L, Alder, G. M., Gray, M. A., Micklem, K. J., Taylor, C. C., Turek, P. J. and
Pasternak, C. A. (1985) J. Cell. Physiology, 123, 326.

296
 FLUORIMETRY OF CELLULAR COMPARTMENTS AND INTRACELLULAR PROCESSES

where M" is the relevant ion, R the gas constant, T the absolute temperature
and F Faraday's constant. At 37°C this relation simplifies to:
V(mV)= -61.5log(|M'y[M-| 0 ) 2
which shows that, for a 10-fold change in cation gradient, you obtain a 61,5 mV
change in potential. The cation gradient can be manipulated by altering the
medium concentration of M' and hence a calibration curve of optical signal and
potential can be obtained in the presence of ionophore.

3.2.1 Plasma membrane potential of animal ceils

The membrane potential of whole cells is more difficult to assess than that of
isolated organelles with optical indicators because cells contain many sub-
cellular compartments, and it is necessary to isolate the response of the plasma
membrane from that of any other cell membrane. For example, the intense
staining of mitochondria by cyanine dyes (7). which are membrane-permeant
cations, means that if you choose a cyanine dye T or other lipophilic cation, to
measure plasma membrane potential it is important to establish conditions
where the mitochondrial potential is disabled. If an anionic dye is chosen, this
problem is overcome but there is, then, very little entry of dye into the cells
unless the dye is itself reasonably lipophilic. Thus, for studies of whole cells, we
employ an oxonol dye with a phenyl substiluent. The procedures for measuring
plasma membrane potential with this dye, oxonol-V (31), are illustrated in Figure
3 and Protocol 2.

Protocol 2
Measurement of plasma membrane potential using
oxonol-V and cells in suspension
Equipment and reagents
• Isotonic saline; e.g. 0.15 M NaCl, 0.005 M Stock solutions of oxonol-V, valinomycin
KC1, 0.005 M HEPES, 0.001 M MgCl2 pH and FCCP, 0.001 M in DMSO or ethanol
adjusted to 7.4 with NaOH. Do not include Stirrer
serum or other albumin-containing media
1-cm square cuvettes
because albumin/protein will sequester
the oxonol and none of the dye will Recording spectrophotometer with a
interact with the cells thermos tatted cell holder
« Pipettes

Method
1 Wash the cells in the saline and resuspend them at a cytocrit of about 0,5-1% v/v.
2 Transfer 1 ml of the cell suspension to a cuvette and place this in the spectrometer,
select the appropriate wavelength(s) and then zero the absorbance/fluorescence
reading.

297
C. LINDSAY BASHFORD

Protocol 2 continued

Stir in oxonol-V to give a final concentration of 1-5 x 10-6 M; record the signal until
it reaches a steady value—this takes up to 10 min.
Then stir in valinomycin or FCCP to give a final concentration of 1-5 x 10-6 M;
record the change in signal.
Then stir in KC1 (valinomycin-treated cells) or acid/base (FCCP) treated cells to
double/halve the relevant cation concentration; record the change in signal.
Use the changes in signal induced by valinomycin/FCCP to estimate the K + / H k
concentration at which addition of valinomycin/FCCP would no give no change in
signal (the 'null-point').
Determine the internal (cytosolic} K ' / H ' concentration and calculate the mem-
brane potential using equation 1.

The first protocol is to incubate the cells in media of differing K' content and
to assess the effect of valinomycin addition (Figure 3(i)). The value of the K null
point is obtained by interpolation. This procedure works best with cells that
have rather low K' permeability; cells more permeable to K + depolarize in high
K' media and the subsequent addition of valinomycin has little effect.
In the second protocol, after valinomycin has been added to the cells and a
new steady reading obtained, increase the concentration of K1 in the medium
by adding successively (Figure 3(ii)) small aliquots of concentrated KC1, about 3 M
is the highest manageable concentration. After taking readings at a few differ-
ent extracellular K' concentrations, obtain the value of the null point by inter-
polation (as in method 1), In both methods it is sufficient to interpolate by
joining data points rather than by using sophisticated least-squares protocols,
especially as the theoretical form of the dye absorbance versus potential plot is
unknown,
In either procedure, once the 'null-point' is known it is necessary to measure
cytoplasmic K' concentration to calculate the potential according to equation 7,
This can be achieved by pelleting the cells through a mixture of di-n-butyl-
phthalate (2 parts) and dinonylphthalale (1 part) with a density of 1.02; cells
pellet through such an oil within 10 s in a Beckman microcentrifuge B. Cell
cations can be determined (after the water content; wet - dry weight) by atomic
absorption spectrometry. The spillover of medium into the cell pellet is usually
less than 0.05%, but this can be verified by including an extracellular space
marker, such as [' 4 C]inulin, in the incubation medium.
There is a danger that the lipophilic valinomycin/K' complex may precipitate
lipophilic anions such as the oxonol dyes and it is advisable TO check on the
calibration described above with another ionophore. The uncoupling agent
FCCP is useful in this respect as it selectively increases the H* permeability of
membranes. The experimental protocol for measuring the potential using FCCP

298
 FLUORIMETRY OF CELLULAR COMPARTMENTS AND INTRACELLULAR PROCESSES

is identical to that using valinomycin except that medium pH rather than K+ is

varied by addition of acid or base (33). Calculation of the membrane potential
using the pH null-point requires a knowledge of cell pH (usually close to 7. 1)
which has to be determined separately, for example, by using indicators (see
below).
A more direct optical method has been introduced whereby a membrane-
permeant, charged dye equilibrates across the membrane of interest. Ideally the
dye will equilibrate rapidly and show minimal binding to materials on either
side of the membrane (34). Using quantitative confocal fluorescence micro-
scopy, the fluorescence of the dye on either side of the membrane is measured;
as fluorescence intensity is linearly proportional to fluorophore concentration,
the measured intensities can be substituted for the ion concentrations ([M+]j and
[M+]0) in equation I and the potential obtained (34). This technique is ultimately
limited by the spatial resolution of the confocal microscope—it works well for
large compartments in whole cells. Recent computational approaches to image
analysis have allowed the spatial resolution to be increased to the level where
mitochondrial membrane potential can be monitored by such means in
individual cells (35).

3.3 pH of cellular compartments

3.3.1 pH of endosomes
Most eukaryotic cells internalize components of their plasma membrane and
associated exogenous material by endocytosis. The subsequent fate of the
endosomes may be complex but it usually involves a step in which the vesicle
milieu is acidified (36). The pH of such an endosome can be monitored directly
if a pH-sensitive dye can be attached either to the relevant piece of mem-
brane or to a suitable extracellular vehicle which is endocytosed. The dye
most commonly used is fluorescein because it is easily visualized with a
fluorescence microscope and because its isothiocyanate derivatives can be
covalently conjugated to the amino groups in proteins (for receptor-mediated
endocytosis) or to dextran. Furthermore the fluorescence of fluorescein and its
conjugates is sensitive to pH in the range 4.5-7.5 (20). The endosomes can be
loaded with the fluorescein derivative by incubation in a suitable physiological
medium and the extracellular dye is subsequently removed by washing
(Protocol 3). The calibration of the fluorescence signal is illustrated in Figure 4(a):
after recording the signal, the pH of all the cellular compartments is set equal
to the bulk pH by adding the ionophore monensin; the bulk pH is then
systematically reduced until the original signal is restored and the final value
of bulk pH equals that originally present in the endosomes. The major
difficulty of this technique arises from cellular autofluorescence which may,
itself, be pH-dependent.

299
C. LINDSAY BASHFORD

Protocol 3
Labelling of endosomes with fluorescein conjugates
Equipment and reagents
• Culture medium • Fluorimeter or fluorescence microscope
• Fluorescein conjugate • Media for mounting cells—slides and
cover slips

Method
1 Incubate the cells in culture medium containing the fltiorescein conjugate (present
at ~l mg/ml) at 37°C. The conjugate enters cells rapidly (within 10 min) and is
subsequently transferred to secondary lysosomes (20). Endocytosis and intracellular
membrane traffic are markedly reduced at lower than physiological temperatures.
At 4°C only binding of ligands to cell surface receptors occurs, so receptors can be
'loaded' at low temperature and endocytosis subsequently 'triggered' by raising the
temperature.
2 Remove the extracellular conjugate by washing the cells with cold (4°C), unlabelled
medium. The residual fluorescence arises from within the cells.
3 Place the labelled cells in a fmorimeter/fluorescence microscope and record the
fluorescence intensity. Cells attached to a solid substratum can be monitored with
light guides (1), or by fitting the substrate diagonally in a fluorescence cuvette and
using a conventional fluorimeter (the final optical arrangement resembles the
'front face' system used to record fluorescence from opaque or strongly absorbing
samples). It is important not to move the substrate during signal calibration.
4 Calibrate the signal with extracellular pH using monensin and the protocol
illustrated in Figure 4(a).

A more qualitative estimate of endosome pH can be obtained by monitoring

the distribution of a permeant. fluorescent amine such as 9-aminoacridine or
acridinc orange whose fluorescence varies with concentration. Such amines
accumulate in acid compartments because the protonated form of the dye is
much less membrane permeant than the free base. Indeed, if the latter equili-
brates readily across membranes then the pH within the vesicle is given by the
relationship (15):

normally pH(m, < pK - 1 so that [AH*] = [AH'j + |Aj. 9-Aminoacridine (10 mM)
exhibits a yellow (1) and acridine orange a red rather than the blue fluorescence
seen at low dye concentrations. Cells incubated with 0.01 mM 9-aminoacridine
may have many yellow vesicles when examined by fluorescence microscopy
using the fluorescein settings; this indicates that such vesicles have accumu-
lated the dye by at least two orders of magnitude and that the internal pH must
be in the range 5-6.

500
 FLUORIMETRY OF CELLULAR COMPARTMENTS AND INTRACELLULAR PROCESSES

(a) ^P' Tris HCI

(c)
EGTA
100 r-"J
i I
pH8 Ca2+
f i" ' ^ l
*~^ EGTA

io.2A EG.TA
PH7.3
Sa^o.i Jti
50 HCI HCI , —J {Ca2* — -
I
.PH.4 i
T
t '
k— ^ t pH*?! Triton [
HCI i
Triton i <1 nM
+ EGTA Ca2+
jPj^sl i^pH6
+ EDTA
0 •
Monensin PH 4

Figure 4 Calibration of intracellular probes of pH and Ca2+. (a) Cells labelled with
fluorescein-dextran (Protocol 3; 20} in Hanks balanced salts solution buffered with 20 ITIM
HEPES, pH 7.0 (NaOH) were excited at 493 nm and fluorescence monitored at 520 nm. The
initial signal (6 units) was calibrated by adding as indicated: 1 ug/ml monensin (final
concentration) to equalize the pH of all compartments in the system (20) followed by HCI to
bring the solution pH to the values indicated. The original signal from the cells corresponds
to that found at pH 4.8 in the presence of monensin indicating that the fluorescein-dextran
occupied a compartment whose pH was 4.8. Records should be corrected for cellular
autofluorescence (fluorescence of unlabelled cells) and, after calibration, the supernatant
fluorescence should be measured and found to be <10% of that found in the presence of
cells, (b) Thymocytes labelled with quene-1 (19) were suspended at 5 x 106 cells/ml in a
medium containing the inorganic salts of RPMI 1640 (without phenol red) and supplemented
with 11 mM glucose and 10 mw HEPES, pH 7.3 (NaOH) at 37°C. Fluorescence at 530 nm
was excited at 390 nm; the signal (49 units) was calibrated by permeabilizing the cells with
0.05% Triton X-100 (Triton) in the presence of 0.5 mM EGTA and 0.5 mM EDTA. More than
90% of the dye is released by the Triton and it is titrated by the addition of Tris and HCI to
give the pH values indicated. The signal before permeabilization corresponds to that found
for pH 7.1 in the presence of Triton indicating that cytoplasmic pH was 7.1 when the
medium pH was 7.3. (c) Lymphocytes labelled with quin-2 (18) were suspended at 5 x 106
cells/ml in 130 mM KCI, 20 mM NaCI, 1 mM CaCI2, 1 mM MgCI2, 5 mM glucose, 10 mM
MOPS pH 7.4 (NaOH). Fluorescence at 500 nm was excited at 339 nm. The signal was
calibrated by permeabilizing the cells with 0.05% Triton X-100 (Triton) and subsequent
titration of the medium with EGTA, MgCI2 and KOH to give 0.2 and 0.1 x lO"6 M Ca2+ as
indicated, with 1 mM free Mg2+, pH 7.05 (18). 'Zero' Ca2+ (<1 nM) was achieved by adding
2 mM EGTA and Tris to give a pH > 8.3. Autofluorescence from unlabelled cells has been
subtracted from each reading and the fluorescence of unpermeabilized lymphocytes
corresponds to a Ca2+ level of 150 nm. All traces in this figure are redrawn from data
originally published in (18-20).

3.3.2 pH or [Ca2+] of cytoplasm or isolated membrane

vesicles
Trapped dyes will also record the pH or calcium ion concentration in the
cytoplasm or sealed compartments. For this purpose indicators of a type similar
to quene-1 (19) or quin-2 (18) are convenient.

301
C. LINDSAY BASHFORD

Protocol 4
Measurement of cytoplasmic Ca2 concentration or pH
using acetoxymethylester-linked indicators
Equipment and reagents
• Isotonic saline; e.g. 0.15 M NaCl, 0.005 M • Stock solutions of acetoxymethylester-
KC1, 0,005 M HBPES, 0.001 M MgCl, pH linked indicator, 0.01 M in DMSO, and
adjusted to 7,4 with NaOH. It helps to Triton X-100, Ca2i. EDTA, EGTA, acid and
include up to 0.025 M NaHCO3 in the base, in water, for calibration titrations
medium (replacing some of the NaCl) so • 1-cm square cuvettes
that intracellular pH remains slightly
• Recording spectra photo meter with a
alkaline chiring hydrolysis of the
thermostatted cell holder
acetoxymethyl ester
• Pipettes • Water bath thermostatted to 37°C
• Stirrer

Method
1 Wash the cells in the saline and resuspend them at acytocrit of about 0.5-1% v/v.
2 Transfer 1 ml of the cell suspension to a cuvette and place this in the spectrometer,
select the appropriate wavelength(s) and then zero the absorbance/fluorescence
reading.
3 Stir in acetoxymethylester-linked indicator to give a final concentration of 1-5 x
10'° M.
4 Follow the hydrolysis of the ester by monitoring the change in absorbance/
fluorescence.
5 When ester hydrolysis is more than 75% complete, this may take 30-90 min, pellet
the cells, resuspend them in unlabelled isotonic saline and incubate them for a
further 30 min at 37 °C.
6 Then pellet the cells once more and resuspend them in isotonic saline for
absorbance/fluorescence measurements.
7 Calibrate the optical signals with extracellular pH or Ca2" after permeabilizing the
cells with Triton X-100 (see Figure 4),

A word of caution about the cleaning of cuvettes is appropriate here; the use
of chromic acid may introduce fluorescence quenchers. These can usually be
removed by soaking the cuvettes in solutions of dictating agents such as EDTA.
Unfortunately some cells seem not to retain the acid form of these pH and
Ca J ~ indicating dyes very long even though the free acid is supposed to be
membrane impermeant. It is very important, therefore, to check that the signal
observed does indeed originate from within cells; this is best done by pelleting a
sample of the suspension and verifying that the supernatant lacks dye (check
the supernatant fluorescence). If leakage is particularly troublesome, reasonable

302
 FLUORIMETRY OF CELLULAR COMPARTMENTS AND INTRACELLULAR PROCESSES

data can be obtained by loading the cells for only a short time (much less than
the time taken for complete hydrolysis of the ester), rapidly washing the cells
and immediately making the pH or Ca2+ determination. Unfortunately this
leads to an inevitable wastage of rather expensive starting material.
It is also possible to record pH or [Ca2+] directly using the appropriate
indicators and quantitative confocal microscopy in a similar manner to that
used to measure plasma and mitochondrial membrane potential directly (see
Section 3.2.1). The great advantage of the 'direct' technique is that it removes
the requirement for the destructive titration/calibration procedures used by the
other methods. This opens up the possibility of following dynamic changes in
pH and cell calcium directly.

3.4 Membrane cycling and recycling

3.4.1 Endocytosis and exocytosis
During endocytosis dyes, such as FM1-43 which are non-fluorescent in aqueous
media but brightly fluorescent when membrane bound, accumulate in the inner
leaflet of the endocytotic vesicles where there fluorescence is readily detected.
This provides a useful fluorescence assay of endocytosis. Furthermore if dye is
removed from the medium the cell-associated fluorescence diminishes as the
endocytotic vesicles undergo further rounds of exocytosis—the dye washing
away into the bath. Indeed the rate of decrease in fluorescence intensity is a good
measure of the rate of exocytosis (37). Recently quantitative confocal fluorescence
microscopy (38) has revealed that FM1-43 is released 'quantally' during exo-
cytosis, confirming that this is an essentially 'all-or-none' process. Intriguingly
the size of the fluorescence 'quantum' is time-independent, strongly suggesting
that in these cells, hippocampal neurons, endocytotic vesicles undergo exo-
cytosis before they have the chance to 'mix* with newly formed vesicles arising
from intracellular membranes. Once again the power of quantitative image
analysis coupled with the exquisite resolution and sensitivity of the confocal
microscope has opened up cell biology to the quantitative approaches of physical
chemistry.
In our own studies with tumour cells in culture we have found a completely
different pattern of FM1-43 behaviour (39). Here the dye accumulates in vesicu-
lar bodies within the cell quite rapidly—as if endocytosis is occurring. However,
once in the cells the dye remains there for extended periods (up to a week)
suggesting that there is not the same tight link between endo- and exo-cytosis
in this material.

3.4.2 Trafficking of intracellular proteins

With the advent of very high resolution fluorescence microscopes it is perhaps
not surprising that new approaches have been developed to study the move-
ment of materials through cells. In particular, the gene for the naturally fluor-
escent 'green fluorescent protein' (GFP) of the jellyfish Aequoria victoria has been
cloned, isolated and subjected to deliberate genetic modification. GFP has been

303
C. LINDSAY BASHFORD

fused with a viral glycoprotein so that the synthesis and processing of the
glycoprotein can be monitored in real time and space (40). These results show
how newly synthesized material moves between the endoplasmic reticulum
and the Golgi apparatus. More recently the fluorophore of GFP has been
mutated to make it sensitive, in one study, to the putative pH values of cellular
compartments (41) and, in another, to cytoplasmic Ca2+ levels (42). In each case,
the advantages of the system are that the photosensitive material is guided
directly to the arena of interest by the normal biosynthetic activities of the cell.
Astonishingly it appears that adding the GFP to endogenous material does not
significantly alter its pattern of processing.

4 Future prospects
The elegance and simplicity of optical experiments ensures that considerable
efforts will continue to be expended to develop photometric and fluorometric
assays of cellular membrane function. The power to direct sensitive chromo-
phores precisely to interesting destinations and the ability to assess them
quantitatively at the single cell level promises to open up the dynamics of
membrane biology to the rigorous analyses of chemistry and physics. The new
indicators of pH and cell calcium will help to unravel the role of these cations in
cell stimulation and differentiation, particularly at the single cell level. Finally,
existing procedures will be used, on account of the wide applicability of the
techniques, on systems inaccessible to more elaborate experiments permitting
a proper investigation of the full range of biological diversity.

Acknowledgements
I would like to thank Mrs B. J. Bashford for her patient preparation of the
diagrams and the Royal Society, Biotechnology and Biological Sciences Research
Council and the Cell Surface Research Fund for financial support. This chapter
is an extensive revision of Chapter 7 in Biological membranes: A practical approach
(ed. J. B. C. Findlay and W. H. Evans), published by IRL Press in 1987; any material
reproduced is by permission of Oxford University Press.

References
1. Harris, D. A. and Bashford, C. L (eds) (1987) Spectrophotometry and spectrofluorimetry. A
practical approach. IRL Press, Oxford and Washington, DC.
2. Miller, J. N. (1981) Standards influorescencespectrometry. Chapman and Hall, London.
3. Bangham, A. D., De Gier, J. and Greville, G. D. (1967) Chem. Phys. Lipids, 1, 225.
4. Chance, B. and Williams, G. R. (1955) J. BioJ. Chem., 217, 395.
5. Chance, B., Schoener, B., Oshino, R., Itshak, F. and Nakase, Y. (1979) J. BioL Chem., 254,
4764.
6. Harbig, K., Chance, B., Kovach, A. G. B. and Reivich, M. (1976) J. Appl. Physiol, 41. 480.
7. Johnson, L. V., Walsh, M. L., Bockus, B. J. and Chen, L. B. (1981) J. Cell Biol, 88, 526.
8. Waggoner, A. S. (1979) Methods Enzymol, 55, 689.
9. Keilin, D. (1925) Proc. R. Soc. Lond. B, 98, 312.

304
 FLUORIMETRY OF CELLULAR COMPARTMENTS AND INTRACELLULAR PROCESSES

10. Jackson,]. B. and Crofts, A. R. (1969) FEES Lett., 4,185.

11. Cohen, L B., Salzberg, B. M., Davila, H. V., Ross, W. N., Landowne, D., Waggoner, A. S.
and Wang, C. H. (1974) J. Membr. Biol, 19, 1.
12. Molecular Probes (1998) Bioprobes, 29,16.
13. Kauppinen, R. A. and Hassinen, I. E. (1984) Am. J. Physio!., 247, H508.
14. Reijngoud, D. J. and Tager, J. M. (1973) Biochim. Biophvs. Acta, 297,176.
15. Rottenberg, H. (1979) Methods Enzymol, 55, 547.
16. Terasaki, M. (1989)Meth. Cell Biol, 29,125.
17. Montecucco, C., Pozzan, T. and Rink, T. J. (1979) Biochim. Biophys. Acta, 552, 552.
18. Tsien, R. Y., Pozzan, T. and Rink, T. J. (1982) J. Cell Biol, 94, 325.
19. Rogers,j., Hesketh, T. R., Smith, G. A. and Metcalfe, J. C. (1983)J. Biol. Chem., 258,
5994.
20. Geisow, M. J. (1984) Exp. Cell Res., 150, 29.
21. Barber, J. (1986) In Ion interactions in energy transfer biomembranes (ed. G. C.
Papageorgiou, J. Barber and S. Papa), p. 15. Plenum Press, London.
22. Akerman, K. E. O. and Wikstrom, M. K. F. (1976) FEBS Lett., 68,191.
23. Bashford, C. L, Alder, G. M., Gray, M. A., Micklem, K. J., Taylor, C. C., Turek, P. J. and
Pasternak, C. A. (1985)J. Cell. Physio!., 123, 326.
24. Bashford, C. L, Barlow, C. H., Chance, B., Haselgrove, J. C. and Sorge, J. (1982) Am. J.
Physiol., 242, C265.
25. Jobsis, F. F., Keizer, J. H., LaManna, J. C. and Rosenthal, M. (1977) J. Appl. Physiol., 43,
858.
26. Wollenberger, A., Ristau, O. and Schoffa, G. (1960) Pflugers Arch. ges. Physio!., 270, 399.
27. Epstein, F. H., Balaban, R. S. and Ross, B. D. (1982) Am. J. Physiol, 243, F356.
28. Bashford, L. and Stubbs, M. (1986) Biochem. Soc. Trans., 14,1213.
29. Bashford, C. L, Barlow, C. H., Chance, B. and Haselgrove, J. (1980) FEBS Lett., 113, 78.
30. Bashford, C. L (1981) Btosci. Rep., 1, 183.
31. Smith, J. C., Russ, P., Cooperman, B. S. and Chance, B. (1976) Biochemistry, 15, 5094.
32. Hoffman, J. F. and Laris, P. C. (1974) J. Physiol (Lond), 239, 519.
33. Bashford, C. L. and Pasternak, C. A. (1984) J. Membrane Biol, 79, 275.
34. Loew, L. M. (1998) In Cell biology: A laboratory handbook (ed. J. E. Celis), Vol. 3, p. 375.
Academic Press, San Diego.
35. Fink, C., Morgan, F. and Loew, L. M. (1998) Biophys. J., 75,1648.
36. Geisow, M. J. and Evans, W. H. (1984) Exp. Cell Res., 150, 36.
37. Betz, W. J. and Bewick, G. S. (1992) Science, 255, 200.
38. Murthy, V. N. and Stevens, C. F. (1998) Nature, 392, 497.
39. Schaefer, M., Djamgoz, M. B. A. and Bashford, C. L. (1999) in preparation.
40. Presley,]. F., Cole, N. B., Schroer, T. A., Hirschberg, K., Zaal, K. J. M. and Lippincott-
Schwartz, J. (1997) Nature, 389, 81.
41. Miesenbock, G., De Angelis, D. A. and Rothman, J. E. (1998) Nature, 394,193.
42. Miyawaki, A., Llopis, J., Helm, R., McCaffery, J. M., Adams, J. A., Ikura, M. and Tsien, R.
Y. (1997) Nature, 388, 882.
43. Kerr, S. E. (1935) J. Bio!. Chem., 110, 625.

305
This page intentionally left blank
 Chapter 12
Use of optical spectroscopic
methods to study the
thermodynamic stability of
proteins
Maurice R. Eftink and Haripada Maity
Department of Chemistry, 112 Coulter Hall, The University of Mississippi,
MS 38677, USA

1 Introduction
The biophysical characterization of globular proteins will almost always include
some type of study of the unfolding of protein to obtain thermodynamic para-
meters. The basic idea is that a transition between a native and unfolded state,
induced by temperature, pH, or denaturant concentration, can serve as a stand-
ard reaction for obtaining a thermodynamic measure of the stability of the
native state. For example, the free energy change for the unfolding reaction can
be used to compare the stability of a set of mutant forms of a protein (1-4).
This type of analysis is based both on assumptions of the thermodynamic
model for the unfolding process and on assumptions in the way the data are
analysed; some of these assumptions and their limitations will be discussed
below.
There are a variety of methods that can be used to monitor an unfolding
process. A common method is differential scanning calorimetry, DSC, which
measures the variation in the specific heat of a protein-containing solution as a
protein is thermally unfolded (5-7). DSC is a popular method for this purpose, but
optical methods can also provide suitable information for tracking the unfolding
of a protein The spectroscopic signals for the native and unfolded states of a
protein can give some insight regarding the structure of the states, and often can
provide advantages of economy, ease of measurement and amenability to a wide
range of sample concentration. The optical spectroscopic methods that have
been used most often for this purpose are absorption spectroscopy, circular
dichroism and fluorescence, which will be discussed in this chapter. A key to each
of these methods and their use in protein unfolding studies is that the signal is a
mole fraction weighted average of the signals of each thermodynamic state. That
is, the observed signal, S, can be expressed as

307
MAURICE R. EFTINK AND HARIPADA MAITY

where Xj is the mole fraction of species i and si is the intrinsic signal of species i.
In order for a particular spectroscopic signal to be useful for tracking a N <—» U
transition of a protein, the signal must be sufficiently different for the N and U
states.
In this chapter we will first discuss thermodynamic models, and their
assumptions, for protein unfolding reactions. Then we will discuss advantages
and limitations of these three individual optical methods. Since most of the
fundamentals of optical spectroscopy are covered in other chapters in this
volume, emphasis will be placed on thermodynamic models and practical
matters related to the application of the spectroscopic methods.

2 Basic thermodynamic principles

Globular proteins are almost always considered to exist, under some conditions,
as a native state, N, that has a defined three-dimensional structure and gener-
ally has some type of biological activity. If the N state is subjected to high (or
low) temperature, extreme pH, high concentrations of chemical denaturants
(e.g. urea or guanidine-HCl (Gdn-HCl)), or pressure, the organized native struc-
ture can be lost in a more or less cooperative manner to form an unfolded state,
U. A thorough discussion of an unfolded state is beyond the scope of this
chapter (see, for example, the article by Dill and Shortle (8)), but it is considered
to have a much more random and fluctuating structure, with amino acid side
chains much more exposed to solvent than they are in the native state. We use
the term 'unfolded' in this chapter, as opposed to 'denatured', to stress that the
unfolded state can refold to the native state if the perturbing condition is
removed. We refer to 'denaturation' as a process that yields an altered structure
that cannot readily refold to the native state.
The simplest model for a protein unfolding reaction is a two-state model,
N «—» U. An important issue is whether or not the unfolding process really
follows this model or involves one or more equilibrium folding intermediates
(i.e. N <—» I <—» U). If one assumes an incorrect model, the resulting thermo-
dynamic parameters will have little meaning (5, 9). However, it is very difficult
to determine experimentally that there is one or more intermediate states, and
the standard approach is to assume the process to be two-state, unless the data
clearly indicate otherwise.

2.1 The two-state model

For a two-state transition between a native state, N, and an unfolded state, U,
the equilibrium unfolding constant, Kun, partition coefficient, Q, mole fraction
of each state, Xf, and standard free energy change for the unfolding reaction are
defined as follows.
N ^ U 2

Kun = 23 3
d
[N]

308
 USE OF OPTICAL SPECTROSCOPIC METHODS

5
6
The value of Kun (or AG^) will depend on the extent of progress along the
perturbation axis (temperature, T, pressure, P, pH, etc.). The following sections
give various functions that are generally accepted as describing the dependence
of AG°un on these perturbation axes. One of the equations below, when com-
bined with those above and equation I can describe spectral data as a function of
temperature or chemical denaturant, in terms of the two-state model.

2.2 Thermal unfolding

The accepted relationship for the temperature induced unfolding of a protein is
AG°un (T) = AH°un - TAS°un 7a
AG°n (T) = AH°,un + ACP (T - T0) - T[AS°,un + ACP In (T / T0)] 7b
where AH°un and AS°un are the enthalpy and entropy changes for unfolding.
Both AH°un and AS°un may be temperature dependent, when the heat capacity
change, ACp, has a non-zero value. In this case, equation 7b (the Gibbs-Helmholtz
equation) applies, where the AH°0,un and AS°oun are values at some reference
temperature, T0 (e.g. 0 or 20 °C). If the reference temperature is selected to be
the high temperature unfolding temperature, TG, then equation 7b can be recast
as
,T,
(T) = AHG,U T G - T + Tln(—;
ir.

In this case, AHG,un (the value of AH°un at T = TG), ACp and TG would be fitting
parameters, whereas in equation 7b the fitting parameters would be AH°oun,
AS°0,un and ACp.
The heat capacity change for unfolding of proteins has been found to be
positive and to be related to the increase in solvent exposure of apolar side
chains (1-3). In other words, a positive ACp is a result of the hydrophobic effect
and a consequence is that the AG^fTJ for unfolding of a protein will have a
non-linear dependence on temperature, reaching a maximum at some
temperature and showing both high temperature and low temperature induced
unfolding.

2,3 Denaturant induced unfolding

The following equation applies to this process
AG°un(fdjj = AG°,un - m [d] 9
where AGoUn([dJ) is the free energy change for unfolding at the reference
condition in the absence of denaturant and m is a denaturant susceptibility
parameter (= -d&G°UIJd[d]). This equation, referred to as the linear extrapolation

309
MAURICE R. EFTINK AND HARIPADA MAITY

model, LEM, is the most widely used relationship for describing denaturant
induced unfolding (10,11). Admittedly, this is an empirical relationship, but it
appears to adequately describe the pattern for denaturant induced unfolding of
a number of proteins. There are other models (and equations) that have been
suggested to describe denaturant induced unfolding, including a binding model
(12,13) and a solvent partitioning model (12), and a combination of the two (14),
but the advantage and popularity of the LEM model is that it has only two
fitting parameters, AGo,un and m, for describing data. The AG° un is the para-
meter of prime interest, since it is a direct measure of the stability of a protein
in the absence of denaturant, and this parameter applies to the ambient solvent
conditions, which can be moderate temperature and pH (e.g. 20 °C and pH 7). By
contrast, it is more difficult to determine AGo,un at 20°C from thermal unfolding
studies, since it requires precise determination of AH° un , ASo,un and ACp and
use of these parameters to extrapolate back from the unfolding temperature.
(which is usually above 50 °C) to 20 °C.
Whereas the AGoiUI1 is a measure of thermodynamic stability of a protein, the
m value also provides structural insights, m values have been suggested to cor-
relate with the change in solvent-accessible apolar surface area upon unfolding
of the protein (12,15). That is, a relatively large m value (which corresponds to a
high susceptibility of the unfolding reaction to denaturant and a steep plot of Xu
or XN versus [d]) has been taken to indicate that there is a large change in the
exposure of apolar side chains on unfolding. This might apply to a protein that
has an extensive core of apolar side chains that are exposed upon denaturation.
Conversely, a small m value (and a less steep dependence of Xu on [d]) would
indicate that there is a lesser change in the exposure of apolar side chains,
which might apply to a protein that has few buried apolar side chains in the N
state (or a case in which the apolar side chains are already on the surface in the
N state, so that there is little change in their exposure on unfolding).

2.4 Acid induced unfolding

A relationship for acid induced unfolding is given by equation 10. This is the
simplest relationship of this type (16), and assumes that there are n equivalent
acid dissociating groups on a protein that all have the same pK aU in the
unfolded state, and that they are all perturbed to have a pKa,N in the N state that
is at least 2 pH units lower than pKa,U. If the pKa shift is not at least 2 pH units,
equation 11 should be used. If there are more than one type of perturbed amino
acid residue and/or if the residues are perturbed by less than 2 pH units, then
the equation must be expanded further to include several more fitting
parameters, which usually makes fitting data untractable.
AG°un(pH) = AG°0,un - RT In [(1 + [H+]/Ka>u)n] 10
+ n
L + [H ]/Ka,u)
11
+ [H+]/Ka,Nf
Equation 10 includes as fitting parameters the free energy of unfolding at neutral
pH, AGo,un, the number of such assumed residues, n, and their pKa,U in the

310
 USE OF OPTICAL SPECTROSCOPIC METHODS

unfolded state. Presumably n should be an integer and pKa,u should be approxi-

mately equal to the values for such amino acids as glutamate, aspartate (for
which pKa,U should be about 4-4.3) or histidine (for which pKa,U should be
around 6.5).

2.5 Pressure induced unfolding

The relationship for pressure, P, induced unfolding of proteins is given by
equation 12, where AGo.un is again the value of the free energy change at 1
atmosphere pressure and AVun = Vu — VN is the difference in volume of the
unfolded and native states.
AG°un(Pj = AG°,un - AVun (P - P) 12
Pressure induced unfolding studies are more rarely performed since a special-
ized high pressure cell is required.

2.6 Simulations
Shown in Figure 1 are simulated plots of XN versus perturbant axis for various
types of perturbations. Shown in Figure 2 are corresponding simulated plots of
AG^ versus perturbant axis, all for the two-state unfolding reaction and for
thermodynamic parameters given in the legend to Figure 1. The symbols
represent simulated data points for which XN is 0.05 to 0.95, the range over
which this value can be determined with reasonable accuracy. The plots illus-
trate the length of the extrapolation needed to determine AG°0,un from typical
unfolding data.

2.7 Unfolding of oligomeric proteins

There have been several studies of the stability of homodimeric or other oligo-
meric proteins (30-35). These proteins are interesting as models for understand-

10 20 30 40 60 6
Temperature (°C) [urea] (M) pH

Figure 1 Simulated plots of the mole fraction of the native state, XN, versus perturbation
axis for a two-state transition described by the following thermodynamic parameters. The
circles indicate the anticipated range of data that is usable for calculating the unfolding
equilibrium constant (or AGSn). The curves were simulated for a two-state model using the
following parameters and equations 3-9 and 11. AHS,un = -20000 cal/mol at T0 = 0°C,
AS° un = -90 cal/mol-K, ACP = 200 cal/mol-K, m = 4.0 kcal/mol-M, pKa,u = 4.5,pKa,N =
2.5, and n = 4.

311
MAURICE R. EFTINK AND HARIPADA MAITY

0 10 20 30 40 50 60 70 1 2 3
Temperature (°C) [urea] (M)

Figure 2 Simulated plots of AGJn versus perturbation axis for the simulations in Figure 1.
The circles are the anticipated range of data over which reliable values of AGun can be
determined.

ing intermolecular protein-protein interactions. For a dimeric protein there is a

general question of whether the protein unfolds in a two-state manner, D <—> U,
or whether there is an intermediate state, which might be either an altered
dimeric state, D', or a folded (or partially folded) monomer species, M. Models
for these two situations are as follows:
D ^ D' ^ 2U
12a
D ^ 2M ^ 2U
The latter situation is particularly interesting because the intermediate state, M,
may or may not have a structured conformation similar to that of the mono-
meric units of the dimer.
As an example, consider the lambda phage protein Cro, which is a homo-
dimeric protein having subunit molecular weight of 7351 (36). This protein, like
many other dimeric DNA binding proteins, has two helix-turn-helix reading
heads and the two major domains of the protein are interlinked by a beta-sheet
formed by their C-terminal segments. Shown in Figure 3 are urea induced
unfolding data determined by monitoring a CD signal. A cooperative transition
is observed that can be well fitted by a two-state model, N <—> U. However, the
data can also be fitted by a D <—» 2 U model. If the latter model is correct, then
one would expect to see a protein concentration dependence of the unfolding
data. In fact, this is observed, as shown in Figure 3. For a D <—» 2U. model, the
relationships between the observed CD signal, Sexp, the mole fractions of dimer,
XD, and unfolded monomer, Xu, and the unfolding equilibrium constant (kun =
[Xu]2/[XD]) will be given by:

13
XTI=-
4[P]0
where [P]0 is the total protein concentration (expressed as monomeric form), St
is the relative CD signal of species i and where Kun will depend on the per-

312
 USE OF OPTICAL SPECTROSCOPiC METHODS

-10 -

J -100
2 3
[Urea], M
Figure 3 Urea induced unfolding of Cm represser as a function of total protein
concentration, [P]0, and as measured by changes in CD signal (at 222 nm for the less
concentrated protein sample and at 232 nm for the more concentrated sample). The solid
curves are a global fit of the D ^ 2U model with the parameters given in the text.

turbant as given by one of the above equations. That is, the transition should
depend not only on the value of Kun (or AG° il]n ) and m (in the case of urea or Gdn-
HCl induced unfolding), but also on the total subunit concentration, [PI,,.
The data in Figure 3 are for total monomer concentrations of [P]0 - 3.5 H.M and
27 u.M. As protein concentration is lowered, the dimer becomes destabilized. The
solid lines through the two data sets are global fits of equation 13 to the multiple
data sets with a single value of AG°Mm - 11.2 kcal/mol and m= 1.6 kcal/mol-M.
The following paragraphs give a detailed experimental description of the
unfolding study with Cro.

Protocol 1
Observation of the unfolding of a protein
Equipment and supplies
• A fresh, filtered solution of 20 mM A stock solution of the protein dialysed
NaH2P04, pH 7,0 against (or, if lyophilized, dissolved in} the
• Since urea slowly decomposes to form same buffer. The concentration will
cyanate. stock solutions should be depend upon the optical techniques being
prepared fresh every day or two used. For CD studies, a 50-200 uM protein
• Access to a CD instrument, such as an Aviv stock solution should allow a minimal
62DS spectropolarimeter, equipped with a volume to be added to the denaturant
thermoelectric (or thermojacketed. with a solution to give a working concentration
circulating water bath) cell holder to in the 5-20 uM range
maintain constant temperature Pipettes
• Quartz cuvettes
MAURICE R. EFTINK AND HARIPADA MAITY

Protocol 1 continued

Method
1 Set up a series of solutions (5 ml) of buffer containing urea concentrations between
0-8 M urea.
2 Take 2 ml of each of these solutions and add to each an aliquot of 100-200 u1 of the
stock solution of protein, thus forming a series of solutions having the same pro-
tein concentration and varying denaturant concentration,
3 Incubate the solutions overnight.
4 Calculate the concentration of urea in each solution by taking a small sample of
the different urea solutions, measuring the difference between the refractive index
of the urea solution and the buffer, AN. and using the following equation (10):
[urea] = 117.66AN + 29.753AN2 + 185.56AN3
(It is usually best to measure the refractive index of the solutions to which protein
has not been added and then make a small correction of dilution. However, in cases
where the protein concentration is low, it is acceptable to make the refractive
index measurement with that solution.}
5 Measure the CD signal at a chosen wavelength for each sample in the series. The far
UV region of 220-230 nm is sensitive to secondary structure changes. Lower
wavelengths can be used, but one must make certain that light absorbance by the
denaturant does not become a problem.
6 Plot the data as described in this chapter.
See Pace and coworkers for a thorough description of this general protocol (10).

Protocol 2
Observation of the unfolding of a protein, e.g. lambda Cro
Equipment and supplies
• Afresh, filtered solution of 20 mM A solution of 8 M urea in 20 mM NaH2PO4,
NaH2PO4, pH 7.0 containing the desired pH 7.0 also containing the same final
final concentration of the protein (e.g. protein concentration (from addition of
prepared by addition of an aliquot from a an aliquot of concentrated stock). This is
concentrated protein stock solution). the concentrated denaturant solution.
Prepare approximately 2,0 ml of this zero- Depending on the midpoint urea
denaturant solution concentration tor 50% unfolding, one will
• Access to a CD instrument, such as an Aviv need from 2 ml (for a low midpoint) to 6
62DS spectropolarimeter, equipped with a ml of this solution (for a high midpoint)
thermoelectric {or thermojacketed, with a Quartz cuvettes
circulating water bath) cell holder to Pipettes
maintain constant temperature and a
magnetic stirrer and flea for stirring

314
 USE OF OPTICAL SPECTROSCOPiC METHODS

Protocol 2 continued

Method
1 Incubate both the zero denaturant and the concentrated denaturant solutions
containing protein at the desired temperature to achieve thermal equilibration.
2 Place 1.8 ml of the zero denaturant solution into the cuvette and then into the
spectropolarimeter, the cell block of which is controlled to the desired tempera-
ture.
3 Measure the CD signal of this first solution as in Protocol 1.
4 Remove an aliquot (e.g. 50-200 ul) of the solution from the cuvette and then add
the same volume of the concentrated denaturant solution to the cuvette.
5 Mix and allow 1-5 min for chemical equilibration. (If after this incubation period
the CD signal is still changing with time, then a longer incubation is needed.)
6 Measure the CD signal of the solution.
7 Repeat steps 4-6 until the concentration of urea in the cuvette has reached a point
where the signal no longer changes.
8 Plot the data as described in this chapter.
If a computer interfaced syringe pump is available, this titration procedure can be
automated, as described below and in (19). This reference also discusses an acquisition
routine for determining when the signal has equilibrated.
See Figure 3 for an example of data.

A second approach, which works well for rapidly unfolding/folding proteins

(and/or for cases in which a researcher has a computer interfaced syringe
pump), is to make measurements on a sample as denaturant is incrementally
(or continuously) added.
When performing denaturant titrations using fluorescence, either Protocol 1
or 2 can be followed. When monitoring the fluorescence of tryptophan residues
in a protein, excitation and emission wavelengths of 290 nm and 350 nm,
respectively, are typical. It is important to also measure the 'fluorescence' signal
from the denaturant solution and to subtract this 'blank' signal from the
fluorescence signal from the protein solutions,
In acquiring the data in Figure 3 we prepared two protein samples, each with
the same protein concentration (by dilution from a more concentrated protein
stock solution), but one with and one without the highest concentration of de-
naturant (e.g. around 7-8 M urea). We placed 1.8 ml of the 'without' sample and a
magnetic stirring flea into a standard 1 cm quartz cuvette. This was then placed
in an Aviv 62DS spectropolarimeter, equipped with a thermoelectric cell holder
(with temperature set at 20 °C) and a magnetic stirrer. Approximately three
times the volume of the 'with denaturant' sample was loaded into a syringe
pump, with the delivery tip directed into the sample cuvette. We then recorded
the CD signal at 222 nm, with averaging of the raw ellipticity signal over abotit
MAURICE R. EFTINK AND HARIPADA MAITY

15 s and digital storage. An aliquot of the 'with denaturant' sample was then
delivered (following the removal of an equal volume of sample from the cuvette
by the second channel of the syringe pump, in order to maintain a constant total
volume in the cuvette; we find that this strategy of ensuring a constant solution
height in the cuvette works best for good mixing of the solutions by the stirring
flea, and the removal of a volume is also needed to avoid overfilling the cuvette
in cases where large total aliquot volumes are required). Following a waiting
period for mixing and chemical equilibration (e.g. 1-5 min), the signal is again
recorded. (See (19) for more details and for a description of a program for
controlling the data acquisiting and for determining when the signal has
equilibrated.)
The raw signal versus denaturant concentration data was then analyzed by
fitting a combination of equations 1, 9, and 13 to the data using a nonlinear
least-squares program (19). (In this particular case, the total protein con-
centration is also needed for equations 13, since this is a dimeric protein; for the
unfolding of a monomeric protein, the fitting equation for a simpler N ^ U
model would be a combination of equations 1, 5, 6, and 9.) In this fitting
procedure, the baseline slopes for the native signal and unfolded signal regions
were assumed to be linear, with these slopes being allowed to vary as fitting
parameters. The two other fitting parameters were the AGo,un (or Kun) and m in
equation 9.

3 Practical considerations and deviations from the

two-state model
3.1 Existence of an equilibrium intermediate(s)
What if there is one or more unfolding intermediate (i.e. the transition is not
two-state)? As mentioned above it is often very difficult to make this deter-
mination. Elsewhere (9, 17) we have shown extensive simulations to illustrate
the difficulty in distinguishing between two-state and three-state transitions
when performing denaturant induced unfolding studies.

3.2 Kinetic considerations

What if the transition is reversible, but equilibrium is not reached during the
course of the experimental measurement? For example, if one measures a
spectroscopic signal to track protein unfolding as a function of temperature,
what will be the effect of a relatively slow approach to equilibrium as temper-
ature is scanned? Lepcock and coworkers (18) have presented interesting
simulations pertinent to this question for differential scanning calorimetry
data. Such simulations are relevant to spectral thermal scanning studies as well.
In general, if the unfolding transition is slow compared to the thermal scan
rate, the appearance of the transition will be skewed. In general, the kinetics of
the conformational transition will be slower at the lower temperature side of
the transition curve and the kinetics will speed up at higher temperature. As a

316
 USE OF OPTICAL SPECTROSCOPIC METHODS

result, what would otherwise be a nearly symmetrical plot of XN versus T (i.e. if

the scan rate were infinitely slow) will be skewed to appear to be sharper than it
should be. If this effect is not realized, a researcher fitting raw data would find
apparent AH G,unand TG values that are larger in magnitude than they should be.
Likewise, for other types of perturbations, if the kinetics of the unfolding
reaction are relatively slow and if a scanning or real-time titration procedure is
employed, then it is important to demonstrate that the scan or titration does
not occur too rapidly so as to cause a skewing of the apparent transition. A way
to experimentally demonstrate that a scan or titration is being performed
slowly enough to allow equilibration is to perform experiments at two or more
scan rates and then choose a slow enough scan rate that gives unchanging
values for the apparent mid-point (and steepness) of the transition. Alterna-
tively, we have a data acquisition routine that enables us to acquire spectral
data for a protein, add titrant (or change temperature) and then automatically
pause for various intervals until the spectroscopic signal is no longer changing
with time. This procedure is described in Ramsay et al. (19) for denaturant
induced unfolding reactions, using CD and fluorescence to track the unfolding
of a protein.

3.3 Irreversibility
What if the unfolding process is irreversible? A common problem and concern
in protein unfolding studies involves irreversibility of the process. This may be
observed as an aggregation under denaturing conditions and/or an inability to
recover biological activity or characteristic spectroscopic or other physical
properties of the native state when the perturbing conditions are removed. For
example, a common way to observe such irreversibility is to notice a hysteresis
in plots of a signal versus temperature during upward, followed by downward
scans.
Such irreversibility may be modelled as
kd
N ^ U -» D
where D is a permanently denatured form of the protein and kd is the apparent
rate constant for the irreversible process. If, for example, kd has a steep depend-
ence on temperature, acid or base, or if kD depends on protein concentration,
then the irreversibility of the process will increase as the perturbing condition is
made more extreme. That is, it still may be possible to characterize the N ^ U
process by avoiding extreme perturbing conditions. If the U —» D reaction
cannot be avoided, the result will be a skewing of the shape of the transition,
making the recovery of valid thermodynamic data unreliable (18).
Another model for an irreversible process is one in which there is an equili-
brium intermediate, N ^ I ^ U, and the I species reacts in an irreversible
manner. There has been some discussion lately of this mechanism with the idea
that an intermediate may have a greater exposure of apolar side chains and that
this might lead to an enhanced tendency of the intermediate species to self-
aggregate (20).

317
MAURICE R. EFTINK AND HARIPADA MAITY

3.4 Baseline considerations

When spectroscopic signals are used to track the unfolding of a protein, there
are always minor problems associated with baseline trends. Whether measuring
absorbance, CD or fluorescence signals (especially the latter), it is usually found
that the intrinsic spectroscopic signal of the N or U state of a protein will
depend on the perturbing condition. That is, the absorbance, CD or fluorescence
intensity of a protein will usually depend on temperature, chemical denaturant
concentration, pH or pressure, so that there will be non-zero baseline slopes.
The following paragraphs will comment on effects of temperature and chemical
denaturant, these being the most commonly used perturbants, on these
spectroscopic signals.
The absorbance of any solution will depend slightly on temperature, due to
thermal expansion (and hence dilution) of a chromophore. Also, changes in re-
fractive index with temperature can lead to apparent absorbance changes. With
pure samples of chemical denaturants, such as urea and Gdn-HCl, there is
usually a very small dependence of baseline absorbances on denaturant con-
centration, provided that the excitation wavelength is one where the denaturant
is transparent.
The intrinsic temperature dependence of CD signals of a native and unfolded
protein is difficult to assess, but the trends appear to be small. For the proteins
studied in our laboratory we find that the far-UV CD signal of an unfolded
protein decreases (becomes more negative) with increasing temperature and
becomes more positive with increasing concentration of urea or Gdn-HCl (21).
We note that other laboratories seem to find similar patterns (22-23).
The fluorescence intensity of tryptophan (or N-acetyl-L-tryptophanamide), the
principle fluorophore in proteins, increases slightly with increasing urea con-
centration (24, 25). The trend is a bit larger with Gdn-HCl. Our experience is that
such baseline trends with these chemical denaturants are larger for the U form
of a protein than for the N form. The largest fluorescence baseline trends occur
with temperature. While these baseline trends can be significant, we find that
this does not cause a serious problem in using fluorescence to study thermal
unfolding of proteins, if there is a large enough difference in the fluorescence
signal of the N and U states. The temperature dependence of the fluorescence
intensity of either the N or U states is frequently treated as a linear slope.
Invariably, the slope of the N state will be less than the slope of a U state and
these slopes will always be negative (i.e. fluorescence decreases with increasing
temperature due to thermally activated quenching processes, see Chapter 2).
Although a linear assumption is almost always made by researchers, we actually
know from model system studies (e.g. the fluorescence of tryptophan or indole
as a function of temperature) that plots of fluorescence intensity versus temper-
ature are not linear. Instead, the fluorescence quantum yield, <I>F (which is
directly proportional to fluorescence intensity, F), is more accurately described
by the relationship (26, 27)
kf
F
~ kf +fcnfo+ 2Anfl exp (-

318
 USE OF OPTICAL SPECTROSCOPIC METHODS

where kf is the rate constant for radiative decay (which is assumed to be

independent of temperature), knfo is the rate constant for temperature-
independent non-radiative processes (e.g. intersystem crossing to the triplet
state), and Afn and Eainfi are the Arrhenius. factor and activation energy for
quenching of fluorescence. If the £a value is relatively small, it will be difficult to
tell that a plot of <f>F (or f) versus temperature is not linear. Fluorophores that
are buried in the apolar interior of a protein tend to have smaller £a values for
thermal quenching. Likewise, solvent exposed fluorophores, as in an unfolded
state of a protein, will tend to have a larger temperature dependence (28).
A conventional way to analyse protein unfolding data is the graphical
approach in which linear baselines are drawn through the regions where the
spectral signal is dominated by the N or U states. This, of course, can introduce
human biases, particularly when the baseline region is limited. After drawing
the two baselines, the mole fraction of the N and U states are calculated from
the extent of the displacement of a data point (in the transition region) from
these two baselines. The preferred method is to avoid the straight edge and to
use nonlinear least-squares computational methods to fit equations to data,
including slopes for the baseline regions as fitting parameters (11, 29).

3.5 Interfering substances

The presence of an interfering substance can lead to spurious changes in optical
spectroscopic signals as well as perturbations of the chemical equilibrium. The
latter situation, such as the case of metal ions that preferentially bind to either
the native or unfolded state, are particular chemical problems associated with
each system and there is not much that can be said other than 'researcher be
aware'. The controlled variation in the concentration of these chemically per-
turbing species can reveal the problem and can sometimes lead to understand-
ing of coupled equilibria.
Cases where an interfering substance contributes to the apparent spectro-
scopic signal, without directly interacting with the protein, can also be a serious
problem. If, for example, there is a contaminant that has a constant absorbance,
ellipticiry or fluorescence signal (as a function of the perturbant axis of temper-
ature, denaturant concentration, etc.), then the contaminant will not be much of
a problem, other than the fact that it diminishes the relative amplitude of the
signal from the protein. If a contaminating substance undergoes a change in
signal as a function of the perturbant axis, this can lead to difficult problems,
such as giving the appearance that a protein unfolding reaction is multi-state,
when it is actually two-state. It is, of course, the responsibility of a researcher to
identify, eliminate or minimize such contaminating signals, but sometimes the
presence of a contaminating signal can be surprising. For example, a contamin-
ating fluorescence signal can alter apparent absorbance or ellipticity measure-
ments, since photomultipliers measure photons whether they are transmitted
directly or luminesced (placement of the detection photomultiplier further from
the cuvette minimizes this measurement of fluorescent photons). Also, a strong
absorbance signal from a contaminant (the signal of which changes along the

319
MAURICE R. EFTINK AND HARIPADA MAITY

perturbant axis) can distort the apparent fluorescence or ellipticity signal of a

protein. The distortion of the fluorescence signal by an absorbing contaminant is
a consequence of the inner filter effect (see Chapter 2). The distortion of an
ellipticity signal is a consequence of the fact the CD instrument usually measures
a difference in absorbance divided by a total absorbance. For example, when
using CD to perform a pH induced unfolding of a protein, we were almost mis-
lead on one occasion by the contribution from the chelating agent, EDTA. This
chelator does not have an intrinsic CD signal itself, so the standard procedure of
substracting a blank did not correct for the problem. The absorbance of EDTA in
the 220-230 nm range varies greatly with pH, as its carboxylate groups undergo
proton dissociation, thus causing a perturbation of the relative ellipticity of the
protein sample in an indirect manner. Scattered light, due either to contaminants
(which usually should be removable by micropore filtration or centrifugation) or
the self-association of the protein under study, can also diminish a CD signal, or
perturb a fluorescence signal as Rayleigh scattering.

3.6 Global analysis

A very useful strategy when performing spectroscopic studies of the unfolding a
proteins is to do a global analysis over multiple data sets (see Chapter 8) (19, 29).
For example, shown in Figure 4 are data for the Gdn-HCl induced unfolding of
Staphylococcal nuclease A, in which changes in fluorescence intensity and circular
dichroism ellipticity were measured simultaneously on the same sample using a
multi-dimensional spectrophotometer. The solid line is a global nonlinear least-
squares fit of a two-state unfolding model to both data sets.

4 Advantages of different spectroscopic signals

4.1 Absorbance
When using absorbance spectroscopy (difference spectroscopy) to monitor con-
formational transitions in proteins, the measurements usually focus on absorb-

-20 a>
Nuclease A I
-40

-60 o
LL

0.0 0.5 1.0 1.5 2.0

[Guanidine-HCI], M
Figure 4 Data for the Gdn-HCl induced unfolding of Staphylococcal nuclease A, monitored by
changes in CD signal at 222 nm and by the fluorescence intensity at 340 nm (excitation at
295 nm). The solid curves are a global nonlinear least-squares fit of a two-state model to
the data with fitting parameters AG°un = 5.32 kcal/mol and m = 5.83 kcal/mol-M.

320
 USE OF OPTICAL SPECTROSCOPIC METHODS

Table 1 Spectral properties of some intrinsic and extrinsic protein probes3

Chromophore Absorbance Fluorescence

E ^max *,
(nm) (x lO^M-^nr1) (nm)
Tryptophan 280 5.6 355 0.13
Tyrosine 275 1.4 304 0.14
Phenylalanine 258 0.2 282 0.021
FAD 450 11 530 0.03
Pyridoxamine phosphate 325 8.3 390 0.14
Dansyl chloride 330 3.4 510 -0.1
1,5-IEDANS 360 6.8 480 '-0.5
PRODAN 364(342)" 14.5 531(401)''" -
Fluorescein derivatives 495 42-85 516-525 0.3-0.5
Rhodamine derivatives 560 12 580 -0.7
Pyrene derivatives 342 40 383 0.25
f*
1,8-ANS 355 515 0.004
1,8-ANS-apoMb 374 ' "e.s 454 0.98
Coumarin derivatives -390 26-30 -460 -0.7

See (38) for original references for values.

" Values are for neutral aqueous solution at ~20°C, unless stated otherwise. The values for the dansyl
chloride, IAEDANS, fluorescein, rhodamine, pyrene and coumarin derivatives are approximate values for
protein adducts in aqueous solution.
b
Values are for hexane as solvent.

ance changes in the aromatic region of the spectrum. In Table 1 is shown

general information of the spectral properties of the three aromatic amino
acids, tryptophan, tyrosine and phenylalanine. Since tryptophan and tyrosine
have a much large molar extinction coefficient than does phenylalanine, the
discussion will be restricted to the former pair.
Both the absorbance spectra of the indole ring of tryptophan and the phenol
ring of tyrosine show a sensitivity to solvent polarity. With increasing solvent
polarity there is usually a red shift in the absorbance of indole and phenol.
Consequently, one would expect that the unfolding of protein, and increase in
solvent exposure of tryptophan and tyrosine residues, would usually lead to a
blue shift in absorbance. For tryptophan, it is typical to monitor the absorbance
at 291-294 nm when studying conformational changes in proteins (there is also
a significant difference spectrum around 284 nm, but this can be obscured by
tyrosine in proteins). That is, one will usually see a decrease in the absorbance at
293 nm upon unfolding of a tryptophan-containing protein with denaturants, pH
or temperature. Likewise, for tyrosine there is a peak in the difference spectrum
at 285-288 nm, which can be used to study protein unfolding in tyrosine-
containing proteins. It should be noted in particular that tryptophan's absorb-
ance is also sensitive to the local electrostatic field, so that changes in indole-
charge interactions can cause either red or blue shifts upon protein unfolding.

321
MAURICE R. EFTINK AND HARIPADA MAITY

Absorbance measurements will require 0.01 to 0.1 mM protein solutions to

achieve reasonable signal-to-noise in 1 cm pathlength cells. Thermal scans are
fairly easy to perform, with thermoelectric cell holders or thermal-jacketed cells
and a circulating bath. For denaturant induced unfolding studies, the common
procedure is to prepare a series of solutions having the same protein concen-
tration and varying denaturant concentrations. When either urea or Gdn-HCl is
used as denaturant, the absorbance signals show a significant and fairly linear
baseline slope. These slopes are probably due to changes in refractive index of
the solution caused by the addition of denaturant. In thermal scans, the base-
line slopes are smaller and the temperature dependence of the absorbance is
usually not linear, making analysis of such data a bit more precarious.
The advantages of absorbance measurements are the ready availability, ease
of use and low cost of the instrumentation. The biggest disadvantage is that it is
less sensitive than the other two methods described below.

4,2 Circular dichroism

CD is a very commonly used method for studying protein conforrnational
changes. The CD spectrum of a protein includes the far-UV region from 180-250
nm, which is dominated by the absorbance of peptide bonds. The CD signal in
this region senses the secondary structure (primarily a-helix) of a protein. The
CD spectrum in the aromatic UV region from 250-300 nm senses the chirality
around the aromatic side chains. Hence there is usually a structured aromatic
CD spectrum (or signature) for the native state of a protein and an absence of
spectral features for the unfolded state; changes in the aromatic CD region are
thus thought to reflect changes in tertiary structure of a protein.
CD is a moderately sensitive method in the far-UV spectral region, requiring
in the range of 0.01 mM solutions. Modern CD instruments can be purchased
with thermoelectric cell holders for thermal scans and with automated titrator
syringe pumps for chemical denaturant titrations. Also, the instruments will be
capable of data averaging and on-line data acquisition.
When performing CD measurements it is necessary to pay attention to the
buffer and salts being used, particularly if one wishes to make measurements
below 200 nm, since various buffers and salts can absorb a significant amount
of light in the far-UV. Schmid (24) has provided a very useful table of the absorb-
ances of common buffer components. Also, chemical denaturants will absorb
light in the far-UV below 210 nm, which must be considered when designing
unfolding studies.
The baseline slopes of CD signals versus temperature or chemical denaturant
concentration are usually assumed to be linear. In the far-UV region we find the
baseline slopes for the native state of a protein are close to zero for both tem-
perature and chemical denaturant. For unfolded proteins we usually find the
baseline slopes to the negative for temperature and positive for urea and
Gdn-HCl.
Aromatic-UV CD signals can be used for studying unfolding transitions, but

322
 USE OF OPTICAL SPECTROSCOPIC METHODS

the CD signals are much smaller, thus having lower signal-to-noise as compared
to far-UV CD or fluorescence data.
The advantage of CD measurements over the other optical methods is that
the far-UV signals observe changes over the entire protein (i.e. its secondary
structure). Also, the nature of the signal change can be related to structural
changes (e.g. loss of ellipticity at 222 nm can be attributed to loss of a-helix). A
disadvantage is that CD instruments are moderately expensive and less avail-
able than standard absorbance and fluorescence instruments. However, CD
instruments are marketed primarily for the biochemistry-protein-peptide re-
search community and an excellent selection of accessories for protein
unfolding studies is available.

4.3 Fluorescence
When using steady-state fluorescence to study protein unfolding, the most
commonly used fluorophores are the intrinsic tryptophan and tyrosine residues.
Since tryptophan has a larger molar extinction coefficient, a redder absorbance,
and is found in most proteins, this fluorophore is the one most frequently em-
ployed in fluorescence studies. The sensitivity of fluorescence detection is such
that concentrations as low as 10-8 M can be studied with commercial fluori-
meters, and, by selecting smaller cell pathlengths or longer excitation wave-
lengths, concentrations as high as 10-4 M can be studied. A very important
property of tryptophan fluorescence is that it is very dependent on the
environment of the indole side chain, making tryptophan fluorescence
responsive to the structure of a protein.
The emission maximum of tryptophan in proteins can range from 308 nm
for a residue buried in a completely apolar environment (as in the single trypto-
phan variant of azurin (37)) to 350 nm for a fully solvent-exposed residue
(38-40). Due to their apolar character, tryptophan residues are usually either
fully or partially buried in the three-dimensional structures of proteins and are
expected to have emission maxima from 320 to 340 nm. Unfolding of a protein,
and concomitant increase in the solvent exposure of tryptophan residues will
invariably lead to a red-shift in the fluorescence of a tryptophan-containing
protein.
The fluorescence intensity, or quantum yield <I>F, of a tryptophan residue in a
protein also varies over a wide range. The $F of single-tryptophan-containing
proteins can range from 0.01 to 0.4 in globular proteins (40). The wide range
appears to be due to a combination of factors, including the closeness of the
indole side chain to intramolecular quenching groups, such as peptide bonds
and certain amino acid side chains (41-43). There does not appear to be such a
wide dispersion of <£F values for unfolded proteins. The consequence of this
variation in $F, along with the above mentioned red-shift, is that the fluor-
escence intensity of tryptophan-containing proteins will almost always change
at either the red or blue edge of the emission spectrum when a protein unfolds.
The fluorescence intensity (either at a single wavelength, or integrated over

323
MAURICE R. EFTINK AND HARIPADA MAITY

the entire emission envelope) is a signal that follows equation 1 and can be used
to extract thermodynamic information. The apparent emission maximum of a
protein sample does not follow equation 1, so one must exercise caution in its
use. As we have shown by simulations elsewhere (25), if there is a significant
increase or decrease in 4>F of a protein upon unfolding (as well as a red shift),
the measurement of the apparent emission maximum will not give a true
reflection of the change in population of native and unfolded states, so that any
recovered thermodynamic parameters will be under- or over-estimates of the
actual values. While this problem is probably recognized by most people using
fluorescence, there is still a strong desire to be able to use emission maxima in
studying protein unfolding, since the maxima are not dependent on protein
concentration (except possibly for associating systems) or lamp intensity and
thus would make it more convenient in the measurement of a series of solu-
tions. A way to use emission maxima in thermodynamic studies is to do curve
fitting of the composite spectra. That is, it should be possible to fit the fluor-
escence spectra at intermediate points along a transition (e.g. when XN = 0.1 to
0.9) as a linear combination of the basis spectra of the native and unfolded
species times their relative population.
Another steady-state fluorescence signal that can be measured fairly routinely
is the fluorescence anisotropy, r. The anisotropy value for the fluorescence of a
tryptophan residue will depend on its rotational freedom (as expressed by its
rotational correlation time, 4>) and its fluorescence decay time, T, with im-
mobilized and short-lived excited states having largest r values. Since there is a
range of fluorescence lifetime and rotational correlation times for tryptophan
residues in native proteins, and since these 4> and T values will usually have a
narrower range for the unfolded state, it follows that r values will usually
change upon unfolding a protein. However, as we have discussed elsewhere
(25), r values do not follow equation 1 , but instead follow

15

That is, the observed r value depends not only on the rt value and mole fraction,
Xj, for the N and U states, but r also depends on the quantum yield of each state,
<J>i. If one of the states has a larger quantum yield than the other, the result will
be a skewing of the plot of r versus perturbation axis toward the more domi-
nant fluorescing state. In principle, if the <5F (or relative intensity) of the native
and unfolded states are known, one can still use r values to obtain thermo-
dynamic information. We have shown elsewhere that plots of fluorescence
anisotropy versus denaturant concentration for the unfolding of a protein can
give the same thermodynamic parameters as fluorescence versus intensity data,
if the above equation is used (17). Also, it must also be realized that anisotropy
measurements have more noise than simple intensity measurements, so that it
is preferable to just make intensity measurements, both because of the better
quality of the latter data and the more straightforward analysis.
If a protein contains tyrosine but not tryptophan residues, the fluorescence of

324
 USE OF OPTICAL SPECTROSCOPIC METHODS

tyrosine can also be used for unfolding studies. The emission maximum of
tyrosine is very insensitive to environment (and protein conformational state)
and its <t>F is also relatively insensitive to environment. These factors, coupled
with the lower extinction coefficient of tyrosines, make it less valuable as a
reporter group. The advantage of tyrosine fluorescence is that there are usually
several such residues in a protein, which partially overcomes the lower extinc-
tion coefficient. Besides tyrosine and tryptophan, extrinsic fluorescence probes
can also be used. For example, dansyl groups can be covalently attached to
cysteine or lysine residues. (See (44) for a detailed description of several extrinsic
probes.)
In cases where there is not a significant change in fluorescence intensity
upon unfolding of a protein, it may be possible to enhance the signal change
by adding a solute quencher or by changing the solvent to D2O. A solute
quencher, such as acrylamide or iodide, will usually be able to quench the
fluorescence of exposed fluorophores in the unfolded state of a protein to a
much greater extent that for the same fluorophores in a native protein.
Alternatively, heavy water is known to increase the fluorescence intensity of
some fluorophores and may have different degrees of enhancement for buried
and exposed fluorophores.
Fluorescence instrumentation is relatively widely available and can be
purchased with various configurations and accessories over a wide range of
prices. As with CD instruments, some fluorescence instruments are designed
to allow automated thermal scans and/or titrations, usually with digital
acquisition.
To summarize, the advantages of fluorescence methods for studying protein
unfolding reactions are the wide concentration range that can be measured and
the sensitivity of the signal to the microenvironment of the fluorophore. Also,
fluorescence signals of the native and unfolded state can provide some low
resolution structural information about these states (at least with respect to the
microenvironment of the fluorophores).

5 Concluding remarks
Optical spectroscopic techniques provide several advantages in studies of the
kinetics and thermodynamics of protein unfolding. This chapter has attempted
to summarize many of these advantages, as they apply to equilibrium studies,
and to also highlight the assumptions involved and to point out some practical
considerations.

Acknowledgements
We acknowledge the contributions of recent graduate students and post-
doctoral associates to our work in this area. In particular, the contributions of
Glen Ramsay and Roxana lonescu are noted. Some of the unpublished work
presented in this chapter was supported by NSF grant MCB 9808635.

325
MAURICE R. EFTINK AND HARIPADA MAITY

References
1. Schellman, J. A. (1987( Annu. Rev. Biophys. Biophys. Chem. 16,115-137; Becktel, W. J.
and Schellman, J. A. (1987) Biopolymers 26,1859-1877.
2. Privalov, P. (1989) Annu. Rev. Biophys. Biophys. Chem. 18, 47-69.
3. Robertson, A. D. and Murphy, K. P. (1997) Chemical Physics 97,1251-1267.
4. Shortle, D. A., Meeker, A. K. and Freire, E. (1988) Biochemistry 27, 4761-4768.
5. Lumry, R., Biltonen, R. and Brandts, J. F. (1966) Biopolymers 4, 917-944.
6. Sturtevant, J. M. (1987) Annu. Rev. Phys. Chem. 38, 463-488.
7. Grikko, Y. V., Privalov, P. L, Sturtevant, J. M. and Venyaminov, S. Y. (1988) Proc. Natl
Acad. Sci. USA 85, 3343-3347.
8. Dill, K. A. and Shortle, D. (1991) Annu. Rev. Biochem. 60, 795-825.
9. Eftink, M. R. and lonescu, R. (1997) Biophys. Chem. 64,175-197.
10. Pace, C. N. (1986) Methods Enzymol. 131, 266-280; Pace, C. N., Shirley, B. A. and
Thomson, J. A. (1989) In Protein structure and function: A practical approach (ed. T. E.
Creighton) pp. 311-330. IRL Press, Oxford.
11. Santoro, M. M. and Bolen, D. W. (1988) Biochemistry 27, 8063-8068; Min, Y. and Bolen,
D. W. (1995) Biochemistry 34, 3771-3781.
12. Tanford, C. (1970) Adv. Protein Chem. 23,1-95; Aune, K. and Tanford, C. (1969)
Biochemistry 8, 4586-4590.
13. Makhatdze, G. I. and Privalov, P. L (1992) J. Mol. Biol. 226, 491-505.
14. Staniforth, R. A., Burston, S. G., Smith, C. J., Jackson, G. S., Badcoe, I. G., Atkinson, T.,
Holbrook, J. J. and Clarke, A. R. (1993) Biochemistry 32, 3842-3851.
15. Myers, J. K., Pace, C. N. and Scholtz, J. M. (1995) Protein Sci. 4, 2138-2148.
16. Barrick, D. and Baldwin, R. L. (1993) Biochemistry 32, 3790-3796.
17. Eftink, M. R. (1998) Biochemistry (Moscow] 63, 276-284.
18. Lepcock, J. R., Ritchie, K. P., Kolios, M. C., Rodahl, A. M., Heinz, K. A. and Kruuv, J.
(1992) Biochemistry 31.12706-12712.
19. Ramsay, G. D., lonescu, R. and Eftink, M. R. (1995) Biophys. J. 69, 801-707.
20. Safar, J., Roller, P. P., Gajdusek, D. C. and Gibbs, C. J., Jr (1993) J. Biol. Chem. 268,
20276-20284.
21. Inoescu, R. M. and Eftink, M. R. (1997) Biochemistry 36,1129-1140.
22. Kuhlman, B. and Raleigh, D. P. (1998) Protein Sci. 7, 2405-2412.
23. DeKoster, G. T. and Robertson, A. D. (1995) J. Mol. Biol. 249, 529-534.
24. Schmid, F. X. (1989) In Protein structure. A practical approach (ed. T. E. Creighton), pp.
251-285. IRL Press, Oxford.
25. Eftink, M. R. (1994) Biophys. J. 66, 482-501.
26. Kirby, E. P. and Steiner, R. F. (1970)J. Phys. Chem. 74, 4480-4490.
27. Robbins, R. J., Fleming, G. R., Beddard, G. S., Robinson, G. W., Thistlethwaite, P. J.
and Woolfe, G. J. (1980) J. Am. Chem. Soc. 102, 6271-6279.
28. Burstein, E. A., Vedenkina, N. S. and Ivkova, M. N. (1973) Photochem. Photobiol. 18,
263-279.
29. Ramsay, G. D. and Eftink, M. R. (1994) Methods Enzymol. 240, 615-645.
30. Jaenicke, R. (1991) Biochemistry 30, 3147-3161.
31. Gittleman, M. S. and Matthews, C. R. (1990) Biochemistry 29, 7011-7020.
32. Eftink, M. R., Helton, K. J., Beavers, A. and Ramsay, G. D. (1994) Biochemistry 33,
10220-10228.
33. Bowie, J. U. and Sauer, R. T. (1989) Biochemistry 28, 7139-7143.
34. Pakula, A. A. and Sauer, R. T. (1989) Proteins: Struct. Funct. genet. 5, 202-210.
35. Herold, M. and Kirschner, K. (1990) Biochemistry 29,1907-1913.
36. Bolotina, I. A., Kurochkin, A. V. and Kircpichnikov, M. P. (1983) KEBS Letters 155,
291-294.

326
 USE OF OPTICAL SPECTROSCOPIC METHODS

37. Hutnik, C. M. and Szabo, A. G. (1989) Biochemistry 28, 3923-3934.

38. Eftink, M. R. (1991) Methods Biochem. Anal,. 35,127-205.
39. Lakowicz, J. R. (1983) Principles of fluorescence spectroscopy. Plenum Press, New York.
40. Beechem, J. M. and Brand, L (1985) Annu. Rev. Biochem. 54, 43-71.
41. Busheuva, T. L., Busel, E. P. and Burstein, E. A. (1975) Stud. Biophys. 52, 41-52.
42. Chen, Y. and Barkley, M. D. (1998) Biochemistry 37, 9976-9982.
43. Ricci, R. W. and Nesta, J. M. (1976);. Phys. Chem. SO, 974-980.
44. Haugland, R. P. (1996) Handbook of fluorescent probes and research chemicals (6th edn).
Molecular Probes, Eugene, OR 97402.

327
This page intentionally left blank
 Chapter 13
The use of spectroscopic
techniques in the study of DNA
stability
John Santalucia, Jr
Department of Chemistry, Wayne State University, Detroit, Ml 48202, USA

1 Introduction
Accurate determination of nucleic acid thermodynamics has become increas-
ingly important in understanding biological function as well as applications in
biotechnology and pharmaceuticals. Knowledge of the thermodynamics of DNA
hybridization and secondary structure formation is necessary for understanding
DNA replication fidelity (1), mismatch repair efficiency (2) and the mechanism
of DNA triplet repeat diseases (3). In addition, RNA folding thermodynamics are
an important aspect of understanding ribozyme catalysis, as well as understand-
ing the regulation of protein expression, mRNA stability and the mechanism of
protein synthesis by the ribosome (4). With the genome sequencing era upon us
(5), it will increasingly become important to predict the folding and hybrid-
ization thermodynamics of DNA and RNA, so that accurate diagnostic tests for
genetic and infectious diseases can be developed. Thus, there is a need to
develop a database of accurate thermodynamic parameters for different nucleic
acid folding motifs (4).
This chapter describes practical aspects of the application of UV absorbance
temperature profiles to determine the thermodynamics of nucleic acid struc-
tural transitions. Protocols and practical advice are presented for issues not
normally addressed in the primary literature but that are crucial for the
determination of reliable thermodynamics, such as sequence design, sample
preparation, choice of buffer, protocols for determining strand concentrations
and mixing strands, design of microvolume cuvettes and cell holder, instru-
mental requirements, data analysis methods, and sources of error. References to
the primary literature and reviews are also provided where appropriate. Sections
of this chapter have been adapted from previous reviews and are reprinted with permission
from the Annual Review of Biochemistry, Volume 62 © 1993, by Annual Reviews
wwwAnnualReviews.org (6) and with permission from Biopolymers © 1997, by John Wiley
& Sons, Inc. (4).

329
JOHN SANTALUCIA, JR

2 Overview of UV melting
The temperature-induced transition between native and random coil states of a
nucleic acid can be conveniently monitored by ultraviolet (UV) absorbance (see
(4, 9, 10) for reviews). The reason for this is that stacked bases have a smaller
absorption per base than unstacked bases; this is called hypochromicity (11,12)
which is defined as:

^hypochromicity = 100(Adenatured ~ A°ative) 1

•^native

where Adenatured and Anative are the absorbances at high and low temperature,
respectively. The absorbance versus temperature profile is commonly referred
to as a UV absorbance melting curve (Figure 1). As the temperature increases, the
ratio of molecules in the single-stranded versus native states increases, resulting
in an increase in the UV absorbance. The melting temperature, TM, is defined as
the temperature at which half of the strands are in the native state and half are
in the 'random coil' state. Whereas many methods such as circular dichroism
and NMR can be used to monitor thermal denaturation, UV absorbance is the
most sensitive due to the high molar absorption of the bases (6).
The simplest way to derive thermodynamic parameters from UV melting data
is to apply a van't Hoff analysis of the data by assuming a two-state model (i.e.
native and denatured states) and that the difference in heat capacities of the
native and denatured states, ACP°, is zero (13-16) (more complex models are de-
scribed in Section 5). At each temperature the absorbance can be used to
calculate the fraction of strands in the native and denatured states, thereby
allowing the calculation of an equilibrium constant (10). Thus, the absorbance
versus temperature profile is used to determine the temperature dependence of

Denatured
Upper baseline__..

Lower baseline

Native Structure
20 30 40 50 60 70 90

Temperature (°C)

Figure 1 Typical experimental UV melting profile. At a given temperature, the fraction of

strands in the duplex state, a, is given by the ratio a/(a + b), where a and b are the
respective vertical distances from the upper and lower baselines to the experimental melting
curve. The temperature at which a is equal to 0.5 is defined as the melting temperature, TM.
The data are for CGTGTCTCOGGAGTCACG at 2.62 X 10-4 M total strand concentration
dissolved in 1.0 M NaCI, pH 7 solution (2).

330
 THE USE OF SPECTROSCOPIC TECHNIQUES IN THE STUDY OF DNA STABILITY

the equilibrium constant, allowing the calculation of AH° and AS0 for the tran-
sition (10) from a van't Hoff plot (i.e. InK versus 1/T plot). AG°T is then calculated
using equation 2 (discussed further in Section 5).
AG^ = - RT In K = AH° - TAS° 2
In practice, however, to obtain accurate thermodynamics from the shape of a
melting curve it is necessary to subtract upper and lower temperature baselines
because the extinction coefficients of the denatured and folded states are
temperature dependent (9, 14, 16). The details of one commonly used program
for fitting UV melting curves were published recently (17).
For sequences that form self-complementary duplexes, the melting tempera-
ture, TM, is calculated using equation 3.
_ AH°
M 3
~ AS° + RlnC T
where R is the gas constant (1.987 cal/K mol) and CT is the total oligonucleotide
strand concentration. For non-self-complementary molecules, CT in equation 2.3
is replaced by CT/4 if the strands are in equal concentration, or by (CA - C B /2) if
the strands are in different concentration, where CA and CB are the concentra-
tions of the more concentrated and less concentrated strands, respectively.
Observing the transition with a different technique (e.g. calorimetry, NMR, or
circular dichroism) can test the two-state assumption. The observation of
isosbestic or isodichroic points in the absorbance or circular dichroism spectra,
respectively, is diagnostic of a two-state transition (6). Alternatively, the melting
profile can be monitored at a different wavelength, so that contributions from
different nucleotides are emphasized. Deriving thermodynamics from the
concentration dependence of the TM (see below) can also test the two-state
assumption. If all methods give the same results, the two-state approximation is
validated since the methods have different sensitivities to each species.
The dependence of the TM on the oligonucleotide strand concentration re-
veals the molecularity of the transition (6). For example, formation of a hairpin
structure is a unimolecular process and therefore does not depend on concen-
tration, whereas formation of a duplex is bimolecular and is concentration
dependent for oligonucleotides. Polynucleotides show little concentration de-
pendence of the TM because double-strand initiation, which is the event depend-
ent on concentration, is only a small fraction of the total free energy involved in
the transition. Higher order complexes such as triplexes and quadruplexes
show stronger concentration dependencies (10, 18), but are usually non-two-
state. The concentration dependence of the TM provides an alternative van't Hoff
method for calculating folding thermodynamics. For bimolecular reactions,
thermodynamic parameters can be derived by rearranging equation 3 to give
equation 4 and plotting reciprocal melting temperature (in Kelvins) versus
logarithm of CT (19) (Figure 2).
1 R . „ AS0

331
JOHN SANTALUCIA, JR

334

332

,-> 330
t-i
t4 328
Wj
'0 326
X 324
I
SB 322

320
318

316
-14 -13 -12 -11 -10 -9

Figure 2 Typical 1/TM versus lnCT/4 plot. The data are for the duplex
CGTCTCTCOGGAGJCACG at strand concentrations from 2.62 x 10~" M to 4.00 x 10-6 M in
1.0 M NaCI, pH 7 solution (2). The thermodynamic parameters derived from the plot are:
AG°37 = -6.90 ± 0.21 kcal/mol, AH° = -56.8 ± 0.9 kcal/mol and AS0 = -160.9 ± 2.1
e.u. The thermodynamic parameters derived from the average of the fits are: AG°37,= -6.89
± 0.10 kcal/mol, AH" = -63.7 ± 2.1 kcal/mol and AS" = -183.2 ± 6.4 e.u.

The concentration should be varied more than a factor of fifty for reliable
measurements of AH° and AS0. The general equations for two-state melting of
complexes of any molecularity are provided in an elegant paper by Marky and
Breslauer (18). If the molecularity is known from UV absorbance mixing curves,
where absorbance is measured while varying the stoichiometry of the reacting
strands (20), or from another technique such as NMR, size exclusion chromato-
graphy or non-denaturing gel electrophoresis, then equation 4 can be used to test
the two-state approximation. For example, if significant concentrations of
intermediates are present, then the AH° from equation 2 (from fitting the shape
of the curves) will be expected to be smaller than the AH° from equation 4 (21).
On the other hand, if the transition is known to be two state, differences in AH°
from equation 2 and equation 4 can reveal the molecularity. For example, a triplex
would give a AH° value from equation 2 which is 50% larger than that from
equation 4, and would indicate that equations appropriate for triplexes not
duplexes need to be used (18, 21).
It is important to note that the AH° and AS0 determined by van't Hoff analysis
are determined at the TM. If large temperature extrapolation is required (greater
than -20°C away from TM), then heat capacity effects should be accounted for.
Previous data have indicated that ACp is usually small for nucleic acids (14, 15).
Based on these data, an approximate estimate of AC° is in the range of 0-120
entropy units (cal K-1 mol) per base pair and AC£ is assumed to be temperature
independent. Given AC;, the AHj, AS0-, and AGj- can be calculated more
accurately than equation 2 with the equations 5 (14).
AH* = AH°Tm + AC; x (T - Tm) 5a
AS£ = AS°Tm + AC; X In (T/TJ 5b
AG^ = AH°Tm + AC; x (T - Tm) - TAS°Tm - TAG; X In (T/TJ 5c

332
 THE USE OF SPECTROSCOPIC TECHNIQUES IN THE STUDY OF DNA STABILITY

Due to enthalpy-entropy compensation, AG°T is relatively insensitive to even

large values of AC£ (22).
In large molecules, like tRNA or group I introns, melting is not two state so
more complicated models are required to derive thermodynamic parameters
(6). UV melting, however, can reveal information on the stepwise mechanism of
thermal denaturation. For example, UV melting of £. coli formylmethionine
tRNA shows several transitions. First, the tertiary interactions are broken and
then individual helices (secondary structure) are melted (23, 24). The transitions
are assigned with the help of other techniques such as circular dichroism,
temperature-jump kinetics, NMR or chemical modification. Since the hypochro-
micity of A«U and G»C base pairs are maximum at 260 and 280 nm, respectively
(25), the ratio of A»U and G»C base pairs broken during each transition can be
determined by measuring the hypochromicities at both wavelengths (10). For
example, the melting curve of the Tetrahymena thermophila large subunit group
I intron involves three major transitions (26). The first occurs at 36 °C and is
attributed to partial unfolding of the tertiary structure, the second transition
occurs at 68 °C with the 260 nm hypochromicity greater than the 280 nm
hypochromicity suggesting the transition involves an A»U rich region. The third
transition occurs at 72 °C with the 280 nm hypochromicity greater than the
260 nm hypochromicity, suggesting that the transition involves G»C rich regions
(26, 27). These results are consistent with the general notion that tertiary struc-
ture is less stable than secondary structure in RNA. This is one reason for the
success of secondary structure predictions for RNA (28).

2.1 Strengths and weaknesses of UV melting and

calorimetry
Microcalorimetry, including differential scanning calorimetry (DSC) and iso-
thermal titration calorimetry (ITC) (29), provides complementary information to
that provided by UV melting. UV detection offers the advantages of high sensi-
tivity so that small sample sizes are required (typically ~3 A260 units are required
for a full set of curves), and the standard errors in AH° and AS0 are typically about
5-8% (4). Due to compensating errors in AH° and AS0 (4,14, 30, 31), a van't Hoff
analysis of UV melting curves provides very precise measurements of AG°37 (stan-
dard error ±2-5%) and TM (±0.5-1.0 °C). Microcalorimetry offers the advantages
that transition enthalpy changes are directly measured, are model independent,
and recent improvements in instrument design allow AH° for nucleic acid sam-
ples to be determined with standard errors of 2-5% (32). Calorimetric methods,
however, require substantially larger sample sizes (typically 30-50 A26o units for
a full set of measurements). The enthalpy-entropy compensation of errors is not
as evident in DSC studies (33), and errors in the AG°37 and TM are often much larg-
er than those from UV melting studies. Thus, DSC measurements are usually
accompanied by UV melting experiments to determine accurate AG°37 and TM
values (34). While DSC provides model-independent thermodynamic para-
meters, often the calorimetric AH° is used in conjunction with the TM deter-

333
JOHN SANTALUCIA, JR

mined from UV melting and the oligonucleotide concentration to obtain the

entropy for the reaction from equation 3. This amounts to applying the two-state
approximation to the data. Further, while it is true that DSC provides a model
independent measurement of the heat released in the transition between two
temperatures even for non-rwo-state reactions, it does not reveal the nature of
the states involved in the transition (10).

3 Sample
3,1 Sequence design
Design of optimal nucleic acid sequences to address specific scientific questions
or to determine specific incremental contributions to overall thermodynamics
is something of an art. There are, however, a number of simple guidelines that
we routinely use that minimize the number of samples that give artefactual
data. In general, a little time spent on careful experimental design results in a
large saving in time and confusing results in the long run. The two most basic
principles are (i) careful design of individual sequences that do not have the
potential to form undesired structures and (ii) to make multiple measurements
for each unknown so that sequences with anomalous thermodynamics are
readily identified. On-line servers for secondary structure prediction of single-
stranded RNA and DNA are very useful for deducing potential alternative
structures (see http://www.ibc.wustl/~zuker and http://JSLl.chem.wayne.edu/).
Algorithms for the prediction of duplexes and higher molecularity complexes
are currently under development in my laboratory. No matter how carefully
one designs sequences, however, there are always occasional examples where
new motifs with surprising stability are discovered, for example, the excep-
tional stability of tandem GA mismatches in DNA and RNA was unanticipated
(35, 36). Sequences with runs of three or more guanines should be avoided since
they form G-quartet structures and consistently yield lower than expected AH°
parameters (34). Sequences should also be designed to have TM values close to
physiological temperature (37 °C) to minimize temperature extrapolation errors
(particularly if ACP° is large) (13). Sequences should also be designed to have TM
values between 20 to 75 °C to allow upper and lower baselines to be adequately
defined.
For hairpins, sodium concentration below 0.1 M is recommended to minimize
bimolecular self-complementary internal loop formation (37). The predicted
thermodynamics from RNA-MFOLD (28, 38) or DNA-MFOLD (39) programs are
accurate enough to deduce if the putative internal loop will form in high
population (to use MFOLD the two strands should be connected by the dummy
sequence LLL (M. Zuker and J. SantaLucia, unpublished). Simulating the pre-
dicted populations of the hairpin and duplex forms by solving the simultaneous
equilibria is also recommended (31, 36).
For bimolecular structures (duplexes), it is generally recommended that self-
complementary sequences be avoided, since they have high propensity to form
competing hairpin structures. Even non-self-complementary duplexes can some

334
 THE USE OF SPECTROSCOPIC TECHNIQUES IN THE STUDY OF DNA STABILITY

times form competing structures, such as 'slipped' self-complementary duplexes

or hairpins (31, 36, 40). To design non-self-complementary single strands that
are unlikely to form hairpins, it is recommended that they be folded with
MFOLD and that a 'dot plot' is made to search for the possibility of self-
complementary duplex formation by each strand (36). On the other hand, self-
complementary duplexes offer the advantages that only one sequence needs to
be synthesized and a given motif (e.g. mismatch or dangling-end) can be
included twice in the duplex, thus cutting in half the experimental error for the
contribution of a motif. To minimize hairpin formation for self-complementary
duplexes, it is recommended that measurements be carried out in high salt
conditions (e.g. 1 M NaCl) and to design the middle of the duplex to be G-C rich
and the ends of the duplex to be A-T rich (41). The very ends of duplexes,
however, should be G-C whenever possible to minimize 'end-fraying' artefacts
(34, 41). In studying the thermodynamic contributions of molecules with
multiple motifs that require multiple linear regression analysis, the rank of the
occurrence matrix should be tested to ensure that it is possible to solve for all
parameters uniquely; this issue has been thoroughly addressed in the literature
(42, 43) for Watson-Crick pairs and by my laboratory for mismatches (2, 31). It is
also important to realize that at salt concentrations below 1 M, melting is length
dependent and the corrections provided in (44) are recommended.
The design of more complex nucleic acid structures requires particular care.
A general principle that has been applied toward the synthesis of multi-
branched loops (45) and novel structures (46) is to maximize 'orthogonality' of
the base pairing potential of different stems. The idea is to design the different
stems with the lowest potential for forming alternative stem structures. If a
complex structure requires a hairpin, the exceptionally stable UUCG (for RNA)
or GNRA (for DNA or RNA) or GNA (for DNA) sequences are recommended (47,
48). Often, it is advisable to study the thermodynamics of complex structures,
by reducing the molecularity of the system by connecting two strands by a
stable tetraloop. This strategy has been effectively applied toward the study of
coaxial stacking interactions (49, 50). Another concern for large nucleic acids or
systems with tightly bound ligands is that they often exhibit slow kinetics.
Thus, slow heating rates are recommended for these systems.

3.2 Redundant design of motifs

Since UV detection requires so little material, it is usually possible to synthesize
multiple molecules with a given structural motif. Essentially, this allows for
multiple measurements for each unknown (e.g. thermodynamic loop contribu-
tion), so that molecules with anomalous thermodynamics due to unforeseen
formation of alternative structures are readily identified, particularly by using
resampling statistical analysis (31). This strategy also has the advantage that
statistical errors can be reduced by taking advantage of the l/(n - v)1/2 dependence
of the error propagation, where n is the number of measurements and v is the
number of parameters determined from the data (51). For example, we recently

335
JOHN SANTALUCIA, JR

combined measurements from our laboratory and others in the literature (108
sequences) to determine the 10 Watson-Crick nearest-neighbour propagation
AG°37 parameters and two initiation parameters (31). The average standard error
in the AG°37 of each of the 108 sequences is estimated as 0.4 kcal/mol; thus, the
average standard error in the 12 unknown parameters is estimated as 0.4
kcal/mol/(108 - 12)1/2 = 0.04 kcal/mol.
In our actual study (31), we rigorously propagated individual errors of experi-
mental measurements to the derived parameters using the variance-covariance
matrix during singular value decomposition (52) and the results are in good
agreement with the average 0.04 kcal/mol error calculated above. This example
dramatically illustrates the value of performing a large number of measure-
ments to reduce statistical uncertainties in derived parameters. It should be
noted, however, that the propagated errors reflect precision of the determined
parameters and not the accuracy, because of the possible presence of systematic
errors in measurements. Another important point is that the derived parameters
are also inaccurate because the imposed model itself is limited (e.g. due to next-
nearest-neighbour interactions). The issue of error propagation during regression
analysis from the measured thermodynamics to the calculated parameters is
discussed in detail in our previous work (4, 31, 53). We have also extended this
type of analysis (i.e. multiple measurements for each unknown) to the deter-
mination of mismatch contributions (31) and this methodology is readily
extended to the determination of other motifs.

3.3 Sample preparation

Accurate determination of nucleic acid thermodynamics requires that samples
are pure (typically >95% purity), free of dust and degassed. Both automated
chemical synthesis as well as in vitro enzymatic methods of DNA and RNA
synthesis are now routine (54-57). Sample purification usually consists of HPLC
(58, 59), denaturing gel electrophoresis or thin layer chromatography (60). Just
because a sample is purified by these methods does not mean, however, that the
sample is free of carried over salts and other contaminants which can affect
nucleic acid melting temperature and thermodynamics. Reliable methods for
sample desalting include: SEP-PAK purification (Waters Inc.) (Protocol 1), G25 size
exclusion chromatography and continuous flow dialysis (we recommend the
apparatus from BRL Inc.). Ethanol precipitation is also a good method for re-
moval of large amounts of salt, but generally samples still require further de-
salting by the above methods to be ready for UV melting experiments. Trace
amounts of divalent metals can sometimes lead to RNA hydrolysis or dramatic-
ally affect nucleic acid structural transitions (26, 61). The best procedure for
removal of trace divalent metals, which can cause RNA hydrolysis, is to add a few
grains of Chelex-100 to the RNA or DNA sample, and then remove them by
decanting (using a thin gel-loading plastic pipette tip) (62). Alternatively, the
sample can be dialysed against 10 mM EDTA, which does not efficiently pass
through 1000 Dalton cut-off dialysis membrane.

336
 THE USE OF SPECTROSCOPIC TECHNIQUES IN THE STUDY OF DNA STABILITY

Protocol 1
SEP-PAK desalting (7)
Equipment and reagents
• Solution A: 10 mM ammonium Solution B: 30% acetonitrile (HPLC
bicarbonate (or ammonium acetate) grade)/70 % deionized water with no added
adjusted to pH 4.5 with 1 M HC1. pH 4.5 is buffer.
critical for efficient sample loading of Nucleic acid sample
most sequences. It is important that the
buffer used is volatile so that it does not SEP-PAK column
carry over into your final sample 'Speed-Vac' evaporator or equivalent
preparation. Occasionally, we find that 1.5 ml centrifuge tubes
adjusting the pH to 7 helps sequences that 0.2 micron Millipore filters
are rich in A, C or U (T) to stick to the
column. Quartz cuvettes

Method
1 (Optional) Filter Solutions A and B through a 0.2 micron Millipore filter. This will
remove dust and sterilize the buffers. Do not autoclave the buffers because
acetonitrile and ammonium bicarbonate are volatile.
2 Add 5 ml of Solution A to the dried nucleic acid sample,
3 Attach the SEP-PAK column (C-18 reverse phase column) to a 10 ml syringe barrel
and clamp the syringe barrel to a ring stand. Equilibrate the SEP-PAK column: (i)
rinse with 10 ml of 100% HPLC grade acetonitrile and (ii) rinse with 5 ml of Solution
A. Air bubbles that can impede the solvent flow are easily removed by using a glass
Pasteur pipette to 'swish' solvent in the syringe barrel. Gravity is sufficient force for
all steps if air bubbles are removed. If desired, the flow rate can be increased by
repeatedly pushing the syringe plunger -0.5 cm into the barrel and then removing
the plunger.lt is important to never let the SHP-PAK go dry.
4 Set up a series of 1.5 ml centrifuge tubes to collect fractions (including the sample
load and wash). Directly load the sample onto the column (without letting the SEP-
PAK go dry). Wash with 5 ml Solution A (this desalts the oligomer and also removes
other small molecule impurities). Wash with 5 ml Solution B (30% acetonitrile).
5 Quantify number of A260 units (see Protocol 2) of all fractions including the sample
load and wash. Combine only the fractions eluted with Solution B that contain
large amounts of ougonucleotide. Do not combine the load and wash fractions
since they are not desalted.
6 Dry the sample either by lyophilization or by centrifugal evaporation (i.e. use the
'speed-vac' apparatus).

337
JOHN SANTALUCIA, JR

Protocol 2
Measurement of optical density units
The concept of absorbance units at 260 nm, AMO (also known as optical density units,
ODU) often causes confusion for students. By definition, one A260 unit is the amount of
sample that gives a UV absorbance at X = 260 nm of 1.0 if the sample is dissolved in 1 ml
volume and measured in a 1-cm pathlength cuvette. Since aborbance is proportional to
concentration (Beer's Law} and the volume is 1 ml, the number of A260 units of a sample
is proportional to the number of moles of strands. Knowledge of the number of A2b0 units
is important for planning UV melting experiments to optimize sensitivity.
Method
1 'Zero' the spectrophotometer by placing deionized water or buffer in a cuvette and
use the UV spectrophotometer's calibration routine,
2 Dissolve the DNA (or RNA) in 1 ml of deionized water, put into a 1-cm pathlength
raicrovolume cuvette, and measure the absorbance at 260 nm. If the absorbance is
less than 2, then the absorbance reading is the number of A260 units. If the sample
is dissolved in a volume other than 1 ml. then use the following equation:
Total A260 units = A x V/l
where A is the measured absorbance, V is the sample volume in ml and 1 is the
pathlength in cm.
3 If the absorbance is greater than 2, then the sample must be diluted (usually 10-fold
is sufficients or the absorbance measured in a short pathlength cuvette. Take 0.1 ml
of the original sample and add 0,9 ml of water and put in the 1 cm cuvette.
Remeasure the absorbance. Multiply the absorbance reading by the dilution factor,
D (a factor of 10 in this example), according to the following equation:
Total A260 units = A x D/l
4 To determine the number of moles of sample, we simply rearrange Beer's Law and
use the molar extinction coefficient at 260 nm, c, calculated from the nearest-
neighbour model (10, 63):
Moles = (Total A260 units)/e
Note that if the oligonucleotide folds at room temperature, the extinction
coefficient must be corrected for hypochromicity or the absorbance measured at
-85 °C.
5 Aliquot the volume of solution that contains the desired number of absorbance
units (typically 3.0 A260 units) of oligonucleotide for the melting experiment, and
evaporate to dryness. See Protocol 4 for a description of a procedure to accurately
mix non-self-complementary strands in equal concentration.

338
 THE USE OF SPECTROSCOPIC TECHNIQUES IN THE STUDY OF DNA STABILITY

3.4 Choice of buffer

Choice of buffer is an important yet often overlooked aspect of performing
reliable UV thermal denaturation studies. Among the most important consider-
ations are the buffer pka, buffer capacity, compatibility with divalent metals
and temperature dependence of pKa. Table 1 lists the properties of the most
commonly used buffers for UV melting experiments. A list of other non-
complexing buffers over the entire pH range has recently been described (64).
Despite the wide use in the literature of Tris as a buffer, it is not recommended
because of its large temperature dependent pKa (consider that between 20 to
80°C the pKa of Tris changes by 1.7 units!). For routine measurements near pH 7
in the absence of Mg2+, phosphate is the buffer of choice. Phosphate buffer is
not appropriate, however, for solutions that include Mg 2 ", since it forms tight
Bjerrum ion pairs (which decreases the available concentration of divalent
metal) often resulting in precipitation of magnesium phosphate (particularly at
pH > 7). Cacodylate is an excellent buffer between pH 5-7 and is compatible
with Mg2+ at pH > 6 (below pH 6, however, we have occasionally observed
unusual complexation of magnesium cacodylate with RNA; S. Varma and J.
SantaLucia, unpublished results). Cacodylate also offers the advantage that it is
antibacterial, thereby facilitating long-term storage.
Choice of a buffer with an appropriate pKa to give sufficient buffer capacity at
the desired pH is important to prevent changes in pH as a result of a nucleic
acid structural transition. Since nucleic acid concentrations for UV melting
studies are routinely 10-4 to 10-6 M and most nucleic acid structural transitions
do not involve the absorption or release of very many equivalents of acid
(C+-GC triple helices are a notable exception), a buffer capacity of greater than
1 mM is usually sufficient to prevent large fluctuations in pH during a melting
curve. For example, we and others have observed that the adenine Nl of A+-C

Table 1 Commonly used buffers for UV thermal denaturation studies3

Buffer pK, (@ 25 °C, APK./AT Compatible with References

/=0.1M) rc-1) divalent metals
Sodium cacodylate 6.27 -0.0015 Y b, c
Sodium phosphate 7.20 -0.0028 N c
MES 6.26 -0.011 Y d, c
Trisd 8.06 -0.028 N c
PIPES 6.80 -0.0085 Y c
HEPES 7.55 -0.014 Y c
Acetic acid 4.76 +0.0002 ? c

'See reference (64) for a complete list of non-complexing buffers.

"(67).
°(66).
"Tris is not recommended for thermal denaturation studies due to the large temperature dependence of its

339
JOHN SANTALUCIA, JR

mismatches has a pka of -6 and we thus performed UV melting studies at pH 5

(where adenine Nl is ~90% protonated) and pH 7 (where adenine Nl is - 90%
deprotonated) (65). In that study 20 mM cacodylate (pK, of cacodylate = 6 . 0 in 1 M
NaCl according to the Davies equation, see (66)) was used to buffer the solution,
because at pH 5 ~90% of the buffer capacity is available to prevent raising the
pH. At pH 7, on the other hand, - 90% of the buffer capacity is available to
prevent lowering the pH. Another consideration in studies that plan to study
structural transitions at widely varying ionic strengths is that the pK, values of
all buffers are ionic strength dependent. The empirical extension of the Depye-
Huckel equation by Davies is sufficient for most applications (66).

Protocol 3
Sample degassing
Equipment
• Speed-Vac evaporator or equivalent • Quartz microcuvettes
Method
1 The major dissolved gas in aqueous solutions is 02. To remove dissolved oxygen, we
recommend gently bubbling argon through the melting buffer. The buffer can then
be added to the dried nucleic acid sample without introducing gas by gentle
pipetting. Samples also spontaneously degas during the annealing step of the UV
melting experiment (Protocol 6).
2 An alternative procedure is to fill the quartz microcuvette with the desired sample
and to place the cuvette in a centrifuge tube. The sample is then briefly
(approximately 1 min) centrifuged at low speed under vacuum (a 'speed-vac' works
well for this). The only drawback to this procedure is that occasionally it is difficult
to remove the cuvette from the centrifuge tube.

Nucleic acid stability at different pH values is also a consideration. At pH < 2,

DNA is susceptible to depurination (usually not a problem for RNA), while for
RNA at pH > 9 strand cleavage by hydrolysis is an issue. RNA strand cleavage is
often catalysed by the presence of traces of divalent metals even at neutral pH,
and thus 0.1 mM disodium EDTA is added to RNA solutions for thermal de-
natural ion studies. On the other hand, many interesting RNA structural
transitions occur in the presence of Mg24 (26, 61), and in these cases it is
recommended that the RNA solutions be prepared freshly and that temperatures
greater than 80°C be avoided if possible. Also at pH < 4 cytosine and adenine
bases are protonated, usually resulting in severe destabilization of double helical
structures and thus favouring the single-stranded state. Above pH 12, DNA
damage can be a concern particularly at temperatures above 80°C. Thus, the
practical range of pH accessible to UV thermal denaturation studies of nucleic
acids is 4-9.

340
 THE USE OF SPECTROSCOPIC TECHNIQUES IN THE STUDY OF DNA STABILITY

Protocol 4
Mixing equal concentrations of non-self-complementary
strands* (8)
Equipment and reagents
• Desalted strand A 1.5 nil centrifuge tubes
• Desalted strand B Quartz cuvettes (0.1 cm pathlength)
• Melting buffer Speed-Vac evaporator or equivalent

Method
1 Aliquot -2.5 ODU of desalted strands A and B in separate tubes. Evaporate both
samples to dryness and dissolve each sample in 100 ul melting buffer (not water).
2 Determine the concentration of each strand by placing 60 ul of each sample in
separate 0.1 cm pathlength microcuvettes and measuring their absorbances at 260
nm and at high temperature (we usually use 90 °C). The concentrations, CA and CB,
are then determined from their respective absorbances, single-stranded extinction
coefficients (calculated from the nearest neighbour model (10, 63)). and using
Beer's Law. It is important to measure the absorbances at high temperature since
strands often form self-structure at low temperatures resulting in a hypochromic
absorbance.
3 Use the following equations to calculate the volumes of strands A and B to be
mixed:
VA - 160 X CB - (CA + CB)
VB = 160 X CA - (CA + Cu)
4 Mix together volumes VA and VB of strands A and B, respectively (note VA + VB =
160 ul). The resulting solution has equal concentrations of the two strands within
about 10%. It is important to use gentle pipetting back and forth to achieve good
mixing.
5 To calculate the average e for the mixed sample (which will be used later to
calculate the total strand concentration, CT). use the equation:

"This procedure yields 160 u1 of DNA with strands in equal concentration that can be used for
the UV melting dilution series (Protocol 6) experiment. If a different volume is desired, then
simply substitute the desired volume for 160 in all of the equations below.

341
JOHN SANTALUCIA, JR

Protocol 5
Sample dilution scheme (8)
The following protocol provides 10 concentrations over a 100-fold range and utilizes
custom manufactured microcuvettes" (see text).
Equipment
• Quartz microcuvettes (0.1 cm pathlength)
Method
1 Start with 4 A260 units of dry sample. Dissolve in 160 u1 of the desired melting
buffer. If a non-self-complementary duplex is melted, then use the solution result-
ing from Protocol 4. We usually perform the melting experiment at 280 nm for
Dilution series 1 and at 260 nm for Dilution series 2 so that a larger concentration
range can be studied,
2 Dilution series 1:
(i) Take 60 ul (out of the 160 ul total) and place in cell 1 (0,1 cm). Check that the
absorbance of this sample is below 1.6 at the desired wavelength, dilute with
melting buffer if necessary.
(ii) To the 100 ul remaining solution, add 70 ul of melting buffer, take 60 ul (out of
the 170 ul total) and place in cell 2 (0.1 cm).
(iii) To the 110 u1 remaining solution, add 77 u1 of melting buffer, take 60 u1 (out of
the 187 ul total) and place in cell 3 (0,1 cm).
(iv) To the 127 u1 remaining solution, add 89 u1 of melting buffer, take 120 ul (out
of the 216 u1 total) and place in cell 4 (0.2 cm),
(v) To the 96 ul remaining solution, add 67 ul of melting buffer, take 120 ul (out of
the 163 u1 total) and place in cell 5 (0.2 cm). 43 ul is leftover for the next
dilution series.
3 Dilution series 2: Combine the contents of cells 2, 3 and 4 (300 ul total) with the left
over solution from Dilution series 1 (43 (il) to give a total of 343 ul. If you suspect
sample degradation from the previous melt (as might occur for RNA in the
presence of divalent metals), then prepare a fresh sample for the second dilution
series. If sample evaporation occurred in one of the cuvettes in Dilution series 1 (as
evidenced by a large air bubble in the cuvette}, then do not combine this sample
since the salt concentration in that sample will be more concentrated than desired,
(i) To the combined volume of 343 u1 add 600 ul of melting buffer. Take 300 u1
(out of the 943 ul total) and place in cell 1 (0.5 cm). Check that the absorbance
of this sample is below 1.6 at the desired wavelength, dilute with melting
buffer if necessary.
(ii) To the 643 uI remaining solution, add 400 ul of melting buffer, take 300 ul (out
of the 1043 ul total) and place in cell 2 (0,5 cm).
(iii) To the 743 ul remaining solution, add 470 ul of melting buffer, take 600 ul (out
of the 1213 uI total) and place in cell 3 (1.0 cm).

342
 THE USE OF SPECTROSCOPIC TECHNIQUES IN THE STUDY OF DNA STABILITY

Protocol 5 continued

(iv) To the 613 ul remaining solution, add 390 u1 of melting buffer, take 600 ul(out
of the 1003 ul total) and place in cell 4 (1,0 cm).
(v) To the 403 ul remaining solution, add 230 u1 of melting buffer, take 600 u1 (out
of the 633 ul total) and place in cell 5 (1.0 cm).
"Important notes on filling cuvettes: Before filling cuvettes, visually inspect them for cleanliness
(see Protocol 8). When filling cuvettes, be sure to leave a small air space between the top of the sample
and the stopper, to allow for the ~4% volume expansion that occurs for water between 0 to
95 °C If you do not do this, then either the cap will pop off during the melt or the cuvette will
crack. It is also important to use gentle pipetting back and forth to achieve good mixing for
each, of the dilutions. Also be sure to insert a 1 cm x 1 cm strip of Teflon tape between the
stopper and the cuvette to provide a tight seal. Warning: The stopper must be inserted with
sufficient force so that it will not pop off during the melting curve. On the other hand, if the
stopper is forced in too tightly you will crack the cuvette. There is no need to run a buffer
control for each melting curve, since subtracting the buffer data from the sample data adds
noise to the data; thus, all five cuvettes are filled with sample.

4 Instrumentation
4. 1. Microvolume cuvettes and aluminium cuvette adapters
Standard 1 cm x i cm cuvettes are not appropriate for UV melting studies of
nucleic acids because they require too much sample and they are difficult to hcat
evenly. A much preferred option is to have microvolume cuvettes custom manu-
factured (both Precision Glass Inc. and Helma Cells produce high quality cuvettes).
We use microvolume cuvettes with the following pathlengths and volumes: 0.1
cm path, 60 ul; 0.2 cm path, 120 ul; 0.5 cm path, 300 ul; 0.8 cm path, 480 ul; 1.0
cm path, 600 ul With this set of cuvettes, we routinely investigate an 80-100-fold
range in oligonuclcotide concentration (Protocol 5). If millimolar concentrations are
desired (e.g. from an NMR sample), then the 0.1 cm path cells can have a 0.09 cm
thick quartz window inserted to give a net path of 0,01 cm (10). In designing the
cuvettes it is important to minimize the sample volume, minimize the amount of
quartz (since quartz is a thermal insulator), and provide a design that lasts at least
200 melting curves before stress leads to cracking. The cuvettes also need to have- a
tightly fitted Teflon stopper or screw cap, or be covered with Dow silicone oil
(Corning, 200-fluid 20 centipoise viscosity) (do not overfill to prevent spillage
during the volume expansion that occurs with temp-erature increases). Detailed
drawings of our microvolume cuvettes can be ob-tained from the following web
address: http://JSLl .chem.wayne.edu or upon request from the author.
In order to fit the microvolume cuvettes into the standard 1,25 cm x 1.25 cm
sample chamber, it is necessary to have custom adapters machined from
aluminium which properly position the microcuvette in the light beam and
provide optimal thermal contact with the thermoelectric controller. Since the
microcuvettes are different sizes, separate aluminium adapters are required for
the different pathlength cuvettes. An important consideration in the design of

343
JOHN SANTALUCIA, JR

the aluminium adapter is the exact position of the light beam and that the
adapter be snug enough to provide good thermal contact and yet have some
tolerance to allow for volume expansion during the melting curve. The adapter
should also allow for the slight variance in stopper heights. Detailed drawings of
the aluminium adapters can also be obtained from the web address given above.

4.2 Spectrophotometer
There are several instrumental requirements for precision UV melting curve
analysis. Accuracy and stability of the absorbance reading at least in the forth
decimal place is required. This is routinely achieved with double-beam instru-
ments (e.g. our Aviv 14DS is stable in the fifth decimal place) as well as well-
designed single-beam instruments such as the Gilford-250/260 or Beckman-DU
instruments. The instrument also needs to be interfaced to a temperature
controller (see below) and the absorbance and temperature readings need to be
continuously output to a computer file for later data processing. It is important
that the raw absorbance data are output, since many instruments apply 'box-
car' or other averaging methods which make melting curves look smoother, but
distort the shape of the melting curve and decrease the temperature resolution
so that subtle transitions are missed; both of these effects negatively affect the
quality of the derived thermodynamic parameters. The sample chamber should
be connected to a compressed nitrogen source with flow controller. This allows
the sample compartment to be purged with nitrogen so that water conden-
sation is prevented at low temperatures.
The decisive factors in choice of commercial instruments for UV melting
curves are the method of temperature control, method of temperature measure-
ment and the number of cuvettes in the cell compartment. Peltier effect heating/
cooling is the preferred method since circulating water baths with jacketed
cuvettes usually do not change temperature quickly enough to be convenient
(cooling is particularly slow, often taking several hours). All instruments with
Peltier-effect heaters are not equivalent; many of the commercially available
spectrophotometers can cool only to 10°C and heat to ~80°C. The Peltier-effect
heaters in the Aviv 14DS allow routine heating/cooling in the entire 0-100°C
range. The Aviv temperature controller can actually go to significantly higher or
lower temperatures than this range if appropriate solvent mixtures are used to
prevent sample freezing and/or boiling. An important maintenance tip is to
periodically (every 3 months or so) blow out the sample compartment and
thermoelectric controller with compressed nitrogen. This removes dust which
can significantly decrease the performance of the thermoelectric controller. A
thermoelectric controller with multiple cuvettes is essential if many melting
curves need to be recorded. The Aviv 14DS is equipped with a five-cuvette
thermoelectric controller, that allows us to routinely record 15 melting curves
per day. This makes studying the concentration dependence of multi-molecular
complexes routine. A design feature of the Aviv thermoelectric controller is that
the samples are rotated in a circle rather than the linear configuration used in
most commercially available instruments. This ensures uniform heating of the

344
 THE USE OF SPECTROSCOPIC TECHNIQUES IN THE STUDY OF DNA STABILITY

samples. The temperature controller also needs to be flexible and programmable

so that melting curves can be recorded at different heating rates in both the
forward and reverse directions (e.g. to check for hysteresis, which is an indicator
of non-equilibrium heating rate).
The method of temperature measurement is absolutely critical to obtaining
accurate thermal dcnatuvation profiles. In principle, the best method would be to
directly measure the temperature of each sample during the melting experiment
with a separate temperature probe inserted into each sample. In practice, this is
not easy to engineer because each of the cuvettes is sealed with a Teflon stopper
and there is very little room to fit a temperature probe. Thus, most thermo-
electric devices have a solid state temperature transducer (e.g. from Analog
Devices Inc.) mounted somewhere in the sample changer (in the Aviv 14DS the
temperature transducer is mounted in the centre spindle which receives the
cuvette holder. The reading from the transducer is calibrated periodically by
performing a mock UV thermal denaturation curve and measuring the voltage
produced by a type K thermocouple (or using a 100 ohm porcelainizecl platinum
resistance temperature detector (68)) inserted into a buffer filled cuvette, to
simulate as much as possible the conditions of an actual melting experiment.
This method of calibration provides temperature precision of - - O . 1 C and
accuracy of-0.3°C, which is sufficient for most applications. Studies of polymer
melting transitions require more rigorous temperature measurement and the
method described by Blake and co-workers is recommended (69, 70).

Protocol 6
Typical melting protocol
Equipment
• UV spectrophotometer with thermostated • Quartz cuvettes/microcuvettes
cell holding

Method
1 Turn on the instrument at least half an hour before making measurements to allow
the D2 lamp to warm-up to minimize instrument drift. Purge the sample compart-
ment with nitrogen at least 20 min before a melting curve is recorded. Place buffer
in one cuvette and 'zero' the absorbance reading.
2 Load filled microcuvettes (Protocols 4 and 5} into the aluminium cell holders and
place into the thermoelectric controller.
3 Set the wavelength to 260 mm and raise the temperature to 90°C for 5 min. This
'annealing' helps equilibrate the structure and to degas the sample. Record the
absorbances in a laboratory notebook for future calculation of strand concentra-
tions. If optimum hypochromicity is desired, then record a high-temperature
wavelength scan while at high temperature.
4 Ramp the temperature down to 0°C (we actually go to -3'C for solutions in 1 M
NaCl) over —10 min. If the wavelength for optimum hypochromicity needs to be
JOHN SANTALUCIA, JR

Protocol 6 continued

determined, then record a low-temperature wavelength scan once the sample is

equilibrated at 0°C.
Input the high and low temperature absorbance versus wavelength data into a
spreadsheet (e.g. Microsoft Excel) and calculate the hypochromicity at each wave-
length using the equation 1. The optimum wavelength has the largest hypochro-
micity. It is also important to consider that the spectrometer signal-to-noise ratio is
optimum at an absorbance of 0.454 and that the signal-to-noise ratio decreases
dramatically below 0.2 or above 1.8.
Once the samples are equilibrated at 0°C, begin ramping up the temperature at a
rate of 0,25, 0.5, 0.8 or 1.0°C/min. The absorbance reading is usually averaged over
a 5 s period and sample changing requires ~ 1 s. Thus each cell has the absorbance
read every 30 s (corresponding to 0.5°C for a l°C/min heating rate). Repeating the
melting curve at slower heating rates is a good way to check that the heating rate is
slow enough for equilibrium to be achieved.
When the melting curve is complete, note the high temperature absorbances of all
cells in your laboratory notebook. Compare these results with those measured in
step 3. If any of the absorbances are significantly higher, this indicates that the
Teflon cap probably popped off during the experiment resulting in evaporation;
data for such cells should be rejected.
Ramp the temperature down to 25 °C. Remove samples, clean and dry the cuvettes
(see Protocol 8). Load a new set of samples and go to step 1 or shut down the
spectrophotometer according to manufacturer recommendations.
Occasionally (about 1 in every 100 melting curves), sample degassing can occur
during the melting curve (or a dust particle might move) resulting in a spurious
absorbance reading at one temperature. Visually inspect the melting curve data file
and manually remove any data points that are anomalous. Save the data file on a
computer disk (give it a systematic name and record the name in your notebook).
The data are now ready for fitting and analysis (see Section 5),

Protocol 7
Sample recovery
UV melting studies typically use only 3 A260 units for a full set of curves and thus samples
are routinely discarded or archived at -70°C in case future analysis is required. If a
sample is precious, then it may be desirable to recover the sample:
1 Generally, it is recommended that the sample purity be checked by analytical HPLC
or gel electrophoresis to ensure that the sample has not degraded.
2 Typically, sample desalting by the methods above (see text and Protocol 1) is
sufficient for further experiments to be performed.
3 Treatment with Chelex-100 is crucial if the sample was previously dissolved in a
buffer containing multivalent metals.

346
 THE USE OF SPECTROSCOPIC TECHNIQUES IN THE STUDY OF DNA STABILITY

Protocol 7 continued

In the event that Dow silicone oil (Corning, 200-fluid 20 centipoise viscosity) was
used to prevent sample evaporation, sample recovery can be readily obtained by
pipetting out the entire sample (including oil) onto a Teflon dish. The oil will stick
to the Teflon as the bead of sample is rolled around the dish (10).

Protocol 8
Cleaning of quartz cuvettes
Equipment and reagents
• Plastic gel loading pipette tip • Methanol (HPLC or spectroscopic grade)
• Plastic scissor clamps • Compressed nitrogen
• 50% nitric acid
Safety precaution: Wear safety goggles and latex gloves, since nitric acid is caustic and
mutagenic,
Method
1 After a melting curve is complete, remove the sample using a plastic gel loading
pipette tip.
2 Clamp the cuvette with locking plastic scissor clamps (the scissor clamp should
grip the frosted sides of the cuvette to prevent scratching the optical surfaces).
Remove traces of your sample by rinsing the cuvettes with distilled water in a
squirt bottle. The water is most easily removed from the cuvette with a syringe that
has the needle capped with a short length of narrow gauge plastic tubing (to
prevent the syringe needle from scratching the quartz). We do not recommend
shaking the cuvette upside down to remove the liquid contents since this often
results in the cuvette slipping out of the clamps and breaking the cuvette.
3 Fill the cuvettes with 50% nitric acid (if the nitric"acid is brown it has degraded and
a fresh solution should be made). Place the filled cuvette into a 50% nitric acid bath
for~10 min. Cover the acid bath with a watchglass to prevent dust from getting in.
4 Remove the sample from the nitric acid using the plastic scissor clamps and re-
move the nitric acid using a pipette (not the syringe, since the nitric acid will
corrode the metal syringe needle).
5 Rinse several times with water and only then rinse several times with methanol
(HPLC or spectroscopic grade), Never allow methanol to come into contact with the nitric
acid. Dry the cuvette with compressed nitrogen (be sure to hold the cuvette firmly
as the nitrogen pressure can cause it to be dropped).
6 Store the cuvettes on a benchtop covered with dust-free cloth. Cover the cuvettes
with an inverted beaker to prevent dust accumulation. Cuvettes stored this way are
ready to use for the next melting experiment without further cleaning (unless you
do not trust the cleaning technique of your labmates).

347
JOHN SANTALUCIA, JR

5 Data analysis
5.1 Curve fitting to calculate thermodynamic parameters
The methods for fitting UV melting curves and determining thermodynamics
have been extensively reviewed (4, 9, 10, 14, 16-18). The optimal choice of fit-
ting method depends on a number of factors including molecularity of the
transition, melting temperature of the transition (i.e. whether sufficient upper
or lower baseline is available) and which approximations are introduced (e.g.
two-state model with AC° equal to zero). This chapter presents the methods for
fitting unimolecular and bimolecular transitions since they form the majority
of measurements made on nucleic acids. For higher molecularities the reader is
referred to the primary literature (9, 18, 40). It is appropriate to reiterate here
that careful design of oligonucleotide sequences (see Sections 3.1 and 3.2) which
minimize the potential for the formation of undesired structures significantly
simplifies data analysis and interpretation.

5.1.1 Unimolecular transitions

Unimolecular structural transitions involve the equilibrium between random
coil and ordered states such as single-stranded stacking (71) and/or hairpin
formation (48).
ARC <—* Ap 6
where ARC and AF are the random coil and folded states of sequence A. Before
analysing the melting data, it is essential that the concentration independence
of the transition be experimentally verified. This is accomplished simply by
recording the UV melting profile over at least a factor of 10 in oligonucleotide
concentration and plotting the normalized absorbance curves (see Section 5.2).
Note that it is important that the entire shape of the curve is concentration in-
dependent, not just the TM. Observation of small concentration dependence
(particularly at high oligonucleotide concentrations) usually suggests the
presence of end-to-end aggregation, non-specific aggregation, or the presence of
bimolecular internal loop formation. These artefacts can be minimized by
recording the melting curves at low salt concentration and low oligonucleotide
concentration.
To derive the enthalpy and entropy changes for a random-coil to hairpin
transition it is necessary to determine the temperature dependence of the
equilibrium constant by analysing the shape of the melting curve. The work of
Turner and co-workers indicates that to derive accurate thermodynamics from
the shape of a melting curve it is necessary to subtract the effects of sloped
upper and lower baselines (14, 16) (Figure 1). Sloped baselines arise because the
extinction coefficients of both the folded and denatured states are temperature
dependent, although the physical origin of the temperature dependence is
largely unknown. The simplest interpretation of sloping baselines is that they
indicate that some non-two-state behaviour is occurring, such as helix fraying,
single-strand stacking, or changes in hydration with temperature (71). The best

348
 THE USE OF SPECTROSCOPIC TECHNIQUES IN THE STUDY OF DNA STABILITY

procedure is to assume linear baselines with the following temperature

dependent extinction coefficients:
eF(T) = m? X T + bF 7
eRc(T) = mRC X T + bRC 8
where eF(T) and eRc(T/) are the temperature dependent extinction coefficients of
the folded and 'random coil' states and m and b are the slopes and intercepts.
For a hairpin transition assuming a two-state model, the fraction of strands
in the folded state, a, is given by:
a = [AF]/CT 9

a = —— 10
a+b
where a and b are the respective vertical distances from the upper and lower
baselines to the experimental melting curve (Figure 1}. The temperature at
which a is equal to 0.5 is denned as the melting temperature, TM. The fraction of
strands in the random coil state is given by 1 — a. Thus the equilibrium
constant at a particular temperature is given by:

By using sloping baselines, one is effectively removing from the shape analysis
those molecules in the sample with intermediate states. Hence, only those
molecules in the sample that actually melt two-state are used to calculate the
thermodynamics. Whilst technically this is somewhat alarming, there is general
agreement that the use of sloping baselines provides the most reliable thermo-
dynamic data (10, 14, 16). Assuming a two-state model with ACp° equal to zero
(i.e. the van't Hoff approximation), the AH° and AS0 can be derived from the
slope and intercept of a InK versus 1/T plot by rearranging equation 2:

In practice, only the data points for 0.15 < a < 0.85 are used due to experi-
mental uncertainty in the 1C values outside of this range (10).
An alternative and preferred method to determine thermodynamics is to
directly fit the entire absorbance versus temperature curve (9, 10, 14, 17). The
total absorbance at temperature T, A(T), is given by the sum of Beer's law
contributions of each component in the solution:
A(T) = eRC(T) X I X [ARC] + eF(T) X I X [AF] 13
where I is the pathlength. Substituting the linear approximations for the base-
lines (i.e. equations 7 and 8) and equation 9 gives:
A(T) = CT X I X [(mRC X T + bRC) X a(T) + (mF X T + bF) X (1 - a(T))] 14
According to equations 11 and 12, a(T) is determined by AH° and AS0. Thus the
experimental absorbance versus temperature curve is fit using multiple non-

349
JOHN SANTALUCIA, JR

linear regression analysis, which amounts to a six parameter fit of the

parameters: AH°, AS°, mRC, bRC, mF and bF (14). In practice, the data are truncated
to include the data within ~30 °C of the TM so that the baseline slopes in the
transition region are accurately accounted for. Consult (17) for details on a
widely used program for curve fitting. It should be noted that this method does
not work well in cases where the upper or lower baselines are not well defined
due to a high or low TM value. In cases where the upper or lower baselines are
not well defined, it is recommended that thermodynamics be derived using the
absorbance derivative method described in (72) and (18). It should also be noted
that observation of a large lower baseline slope often suggests the presence of
structural intermediates which compromise the validity of the van't Hoff
method (37). As stated in Section 2, the van't Hoff method provides two-state
thermodynamic parameters at the TM. In principle, ACp° could be added as a
seventh unknown in the curve fitting so that more accurate temperature
extrapolations are possible, though the noise level in a typical absorbance
experiment may not justify such a treatment (32, 73).

5.1.2 Bimolecular transitions

The van't Hoff curve fitting methods presented above for hairpin transitions are
general, but for higher molecularity the equilibrium expressions are different
(equations 6, 9 and II) (9, 10, 18). For self-complementary duplexes the relevant
equilibrium equations are:
2 ARC *— > A2 15
2[A2]
a=— — 16

[ARC]2 2 X (1 - a)2CT
For non-self-complementary duplexes the relevant equilibrium equations are:
ARC + BRC <— » AB 18
2[AB]
- -
uT
19

= [AB] = 2a
[ARC][BRC] (1 - «)2CT
Note the fourfold difference in equations 17 and 20 which is ultimately reflected
in the different concentration dependence of the TM for self-complementary
versus non-self-complementary sequences (equation 3).
As described in Section 2, the concentration dependence of the TM provides
an alternative van't Hoff method for determining thermodynamics. In this
method, the transition AH° and AS° are obtained by equation 4 from the slope
and intercept of a 1/TM against lnCT/4 plot. Since the TM is relatively insensitive
to non-two-state behaviour and to the choice of baselines, 1/TM versus logCT
plots provide very reliable AH°, AS0, and AG°37 parameters (16), particularly if

350
 THE USE OF SPECTROSCOPIC TECHNIQUES IN THE STUDY OF DNA STABILITY

the TM is measured over a large concentration range. It is worth noting that the
TM cannot be accurately determined from an inflection point of the melting
curve determined from the maximum of the temperature derivative of the
absorbance (9, 10). It is necessary to subtract the upper and lower temperature
baselines before performing the differentiation, since they affect the position of
the apparent TM (10, 14). The preferred method for calculating the TM is to use
equation 3 using the AH° and AS° determined from the fit of the shape of the
curve with sloped baselines, as described above (14,16).
Since curve fitting and 1/TM against lnCT/4 plot methods depend differently
on the two-state approximation, agreement of AH° obtained from both methods
provides a test of the validity of the two-state approximation (6,10). Agreement
within 10-15% of AH° parameters from a 1/TM versus lnCT plot and from fits of
the shapes of melting curves is generally regarded as indicative of a two-state
transition (4), though caution is recommended. This criterion suggests the
standard error in the van't Hoff AH° parameters from optical melting is about
5-8%. In most cases that have been studied by both optical melting and
calorimetry, the AH° values are in agreement if the two ways of analysing the
optical data give the same value within 10% (15, 16, 34). Agreement between
enthalpy changes determined by different methods is a necessary, but not a
sufficient criterion to definitively establish two-state behaviour (4, 18). For ex-
ample, the self-complementary DNA sequence (CGTTG,CGXAACG)2 yielded AH°
parameters from a 1/TM versus lnCT plot and from fits of the shapes of optical
melting curves that agree within 10% for the single strand to duplex transition.
The temperature dependence of NMR spectra and comparison of that sequence
with thermodynamics for other sequences, however, revealed that it actually
melts through a hairpin intermediate (31).
An alternative criterion for two-state thermodynamics is to compare the
enthalpy changes from optical melting and calorimetry (15, 16, 34). For tran-
sitions with large ACp°, it appears that van't Hoff analyses and calorimetric data
provide systematically different AH° parameters and the possible origins of
these differences have been recently reviewed (32, 73). Even agreement
between van't Hoff analysis of optical melting and calorimetry does not
guarantee a two-state transition. The AH° values for the DNA duplex,
(GCGTACGCATGCG'CGCATGTGTACGC), are in agreement, but the transition is
not two-state as evidenced by the melting of the individual single strands (31,
40). Known exceptions of this type are rare, however.
For two-state transitions, the AH° parameters obtained by both methods are
equally reliable, so it is best to report the average of parameters from the 1/TM
versus lnCT plot and from the fits of the shapes of curves (41, 74). Obtaining AH°
from melting curves measured at different wavelengths can be used to check
the two-state approximation (6). Importantly, agreement between enthalpy
changes determined by different methods is a necessary, but not a sufficient
criterion to establish two-state behaviour (16, 27, 38). Additional methods for
validation of the two-state model are described in Section 2.
A van't Hoff analysis of UV melting data cannot be used to reliably measure

351
JOHN SANTALUCIA, JR

the thermodynamics of molecules with non-two-state transitions. In the case

where the transition is clearly non-two-state (particularly if end-fraying or
internal loop structures are significantly populated), a statistical mechanical
approach may be able to yield reliable thermodynamics in some circumstances
(75, 76). One drawback of the statistical mechanical approach is that the data
are fitted with more parameters than justified by the signal-to-noise ratio of the
data, and thus loop parameters are introduced in an ad hoc fashion. Recently, we
derived a method for treating UV melting curves by a three-state model (31) in
which a self-complementary sequence was allowed to form a hairpin inter-
mediate, utilizing a coupled equilibrium method described previously for an
RNA with duplex intermediates (36). Such an approach is easily generalized to
non-self-complementary duplexes, with each strand able to fold into hairpin
intermediates. A better approach, however, is to carefully design oligonucleo-
tides so that the possibility for alternative structure formation is minimized.
Performing the control melting experiments of the individual single strands
reveals if hairpin states are likely to be significantly populated. It is worth
noting that actual single-strand melts should not be directly subtracted from
duplex melting curves, since in duplex melting the competition with the
duplex state significantly reduces the population of hairpin conformation (36).
Instead, a fully coupled three-state model is appropriate (31).

5.2 Presentation of normalized absorbance curves

Often it is useful to present several UV absorbance curves on the same figure to
illustrate differences in TM or hypochromicity. To display the curves on the
same scale, it is necessary to normalize the curves either to high temperature or
to both high and low temperature. To normalize different melting curves to the
high temperature absorbance, Am, equation 21 is applied to each curve and then
the data are plotted on the same graph.

=
-^Norm ~7 21
A
HT

where ANorm is the normalized absorbance at high temperature and A is the raw
absorbance data. Equation 21 preserves the shape of the melting curve without
distorting the hypochromicity. Alternatively, the data can be 'double normal-
ized' to both high and low temperatures by applying equation 22 to each curve
and then plotting the data on the same graph.
_ A/A LT -1
A ~~
Norm -
HT/ALT

where ALT is the low temperature absorbance. Equation 22 makes all curves start
at 0 and finish at 1 and preserves the shape of the melting curve, but inform-
ation about hypochromicity is lost. Equation 22 is useful for comparing the
melting curves of different sequences (at the same concentration) or comparing
the melting curves of one sequence measured at different wavelengths.

352
 THE USE OF SPECTROSCOPIC TECHNIQUES IN THE STUDY OF DNA STABILITY

5.3 Error analysis

Careful analysis of errors is an important aspect of any experimental study. The
main sources of errors in optical melting are:
(1) signal-to-noise ratio of the data;
(2) random errors due to fluctuations in sample preparation (e.g. oligonucleo-
tide concentration, purity, volume, salt concentration, pH, errors in mixing
etc.);
(3) systematic errors due to incorrect instrument calibration;
(4) systematic errors introduced as a result of poor oligonucleotide design (e.g.
a sequence that can form intermediates);
(5) systematic errors due to incorrect assignment of baselines; and
(6) systematic errors due to imposing the two-state approximation and from
assuming that ACP° is zero.
An important distinction is the difference between precision, which reflects
the experimental reproducibility of the data, and accuracy, which reflects how
well the experimental measurement agrees with the real value if a perfect mea-
surement were made (51). The first two sources of error are easy to quantify by
simply reproducing one's data or analysing the sampling error in a 1/TM versus
InCT plot or the sampling error in the fitted data. The theory for determining
sampling errors in AG°37, AH°, and AS0 from the linear regression of the 1/TM ver-
sus lnCT plot using standard statistical analysis (77) has been previously
described (30). The best method to quantify systematic errors (sources 3-6 above)
is to compare thermodynamic measurements on the same oligonucleotides that
were independently determined from different laboratories utilizing different
instrumentation and techniques. For DNA duplexes, an estimate for systematic
error sources 3 and 5 can be derived from results on three sequences
(CGATATCG, GAAGCTTC and GGAATTCC) that have been independently mea-
sured by two groups (34,41, 74). The average deviations for AG°37, AH°, AS0 and
TM for these sequences are 3%, 6%, 6% and 1.0 °C, respectively.
It is important to understand how errors in AH° and AS° propagate to give
the error in AG°37. Experimental AH° and AS0 parameters are not independently
determined, but instead are highly correlated, with a typical R2 > 0.99 (14, 22, 31).
This enthalpy-entropy compensation results in errors in AG°37 that are much
smaller than would be expected if AH° and AS° were uncorrelated. Equation 4.8
of Bevington (51) provides the equation for error propagation for a general
function x = flu,v):

Performing the appropriate differentiation of equation 2 and substitution into

equation 23 gives the equation for the propagation of error from AH° and AS°,
and o-AS=, to give the error in AG°37, o-AC<>37 (30):
2

353
JOHN SANTALUCIA, JR

whereCTAH°AS°is the covariance between AH° and AS0, and T is 310.15 K. An

alternative equation expresses the covariance in terms of the correlation
coefficient of a plot of AH° versus AS0, RAH°AS° (78):
- 2T (RAH-AS-) <TAH° ^AS» 25
Performing the appropriate differentiation of equation 3 and substitution into
equation 23 gives the equation for the propagation of <TAH° and CTAS°, to give the
error in TM, ffrm (31):

26
AH° AH° (AH0)2
This equation assumes that there is negligible error in the lnCT term. This is
reasonable because a 10% error in CT propagates according to equation 23 to a 1%
error in lnCT at oligonucleotide concentrations in the range of 10-5 M (7). The
enthalpy-entropy compensation effect is evident in the high quality of pre-
dictions made for AG17 and TM (31,41, 53).
Another way to evaluate error propagation with minimal assumptions about
the experimental errors or how these errors propagate to the derived NN
(nearest neighbour) parameters is to use a resampling analysis of the data (79).
Consider the 'unified data set' of 108 sequences with only Watson-Crick base
pairs (31). First, SVD analysis is used to determine the linear least-squares fit of
the data to obtain the 10 NN parameters and the two initiation parameters as
described (31, 41, 53). Since the unified data set contains 108 equations with 12
unknowns, the problem is overdetermined. A total of 30 resampling trails were
used. For each trial, a different set of 68 randomly selected sequences was used
in the SVD analysis to calculate the 12 unknowns. It is important to check the
column rank of the stacking matrix for each trial. The results from the 30 trials
were averaged and the standard deviations calculated for each NN parameter.
This resampling analysis was performed for AG°,7, AH° and AS0. We also per-
formed controls in which certain classes of sequences (e.g. all sequences from a
particular lab or all sequences greater than 10 bp, etc.) were systematically omitted
from the SVD analysis and found that in all cases the parameters agreed within
the propagated experimental error of the NN from the total dataset.

References
1. Petruska, J., Goodman, M. F., Boosalis, M. S., Sowers, L. C, Cheong, C. and Tinoco, I.,
Jr (1988). Proc. Natl. Acad. Sri. USA., 85, 6252-6256.
2. Peyret, N., Seneviratne, P. A., Allawi, H. T. and SantaLucia, J., Jr (1999). Biochemistry,
38, 3468-3477.
3. Gacy, A. M. and McMurray. (1998). Biochemistry, 37, 9420-9434.
4. SantaLucia, J., Jr and Turner, D. H. (1997). Biopolymers, 44, 309-319.
5. Koonin, S. E. (1998). Science, 279, 36-37.
6. Jaeger, J. A., SantaLucia, J., Jr., and Tinoco, I., Jr. (1993). Annu. Rev. Biochem., 62,
255-287.
7. SantaLucia, J., Jr (1991) in Department of Chemistry, Ph.D. Thesis. University of Rochester,
Rochester, NY.

354
 THE USE OF SPECTROSCOPIC TECHNIQUES IN THE STUDY OF DNA STABILITY

8. Allawi, H. T. (1998) in Department of Chemistry, Ph.D. Thesis. Wayne State University,

Detroit, MI.
9. Breslauer, K. J. (1995). Methods Enzymol, 259, 221-241.
10. Puglisi, J. and Tinoco, I., Jr (1989). Methods Enzymol, 180, 304-325.
11. Doty, P., Boedtker, H., Fresco, J. R., Haselkorn, R. and Litt, M. (1959). Proc. Natl. Acad.
Sci. USA, 45, 482-499.
12. Tinoco, I., Jr (1960). j. Am. Chem. Soc., 82, 4785-4790.
13. Freier, S. M., Kierzek, R., Jaeger, J. A., Sugimoto, N., Caruthers, M. H., Neilson, T. and
Turner, D. H. (1986). Proc. Natl. Acad. Sci. USA, 83, 9373-9377.
14. Petersheim, M. and Turner, D. H. (1983). Biochemistry, 22, 256-263.
15. Rentzeperis, D., Ho, J. and Marky, L A. (1993). Biochemistry, 32, 2564-2572.
16. Albergo, D., Marky, L., Breslauer, K. and Turner, D. (1981). Biochemistry, 20,
1409-1413.
17. McDowell, J. A. and Turner, D. H. (1996). Biochemistry, 35,14077-14089.
18. Marky, L A. and Breslauer, K. J. (1987). Biopolymers, 26, 1601-1620.
19. Borer, P. N., Dengler, B., Tinoco, I., Jr and Uhlenbeck, 0. C. (1974).J. Mol Biol., 86,
843-853.
20. Stevens, C. and Felsenfeld, G. (1964). Biopolymers, 2, 293-314.
21. SantaLucia, J., Jr, Kierzek, R. and Turner, D. H. (1990). Biochemistry, 29, 8813-8819.
22. Krug, R. R., Hunter, W. G. and Grieger, R. A. (1976). J. Phys. Chem., 80, 2335-2341.
23. Cole, P. E. and Crothers, D. M. (1972). Biochemistry, 11, 4368-4374.
24. Crothers, D. M., Cole, P. E., Hilbers, C. W. and Schulman, R. G. (1974). J. Mol. Biol, 87,
63-88.
25. Felsenfeld, G. and Hirschman, S. Z. (1965). J. Mol Biol, 13, 407-427.
26. Jaeger, J. A., Zuker, M. and Turner, D. H. (1990). Biochemistry, 29,10147-10158.
27. Banerjee, A. R., Jaeger, J. A. and Turner, D. H. (1993). Biochemistry, 32,152-163.
28. Mathews, D. H., Andre, T. C., Kim, J., Turner, D. H. and Zuker, M. (1997). In Molecular
modeling of nucleic acids (ed. N. B. Leontis and j. SantaLucia, Jr), Vol. 682, pp. 246-257.
A.C.S., Washington, DC.
29. Breslauer, K., Freire, E. and Straume, M. (1992). Methods Enzymol, 211, 533-567.
30. SantaLucia, J., Jr, Kierzek, R. and Turner, D. (1991). J. Am. Chem. Soc., 113, 4313-4322.
31. Allawi, H. T. and SantaLucia, J., Jr (1997). Biochemistry, 36, 10581-10594.
32. Liu, Y. and Sturtevant, J. M. (1997). Biophys. Chem., 64, 121-126.
33. Law, S. M., Eritja, R., Goodman, M. F. and Breslauer, K. J. (1996). Biochemistry, 35,
12329-12337.
34. Breslauer, K. J., Frank, R., Blocker, H. and Marky, L. A. (1986). Proc. Natl Acad. Sci. USA,
83, 3746-3750.
35. Li, Y., Zon, G. and Wilson, W. D. (1991). Biochemistry, 30, 7566-7572.
36. Longfellow, C. E., Kierzek, R. and Turner, D. H. (1990). Biochemistry, 29, 278-285.
37. SantaLucia, J., Jr, Kierzek, R. and Turner, D. H. (1992). Science, 256, 217-219.
38. Zuker, M. (1989). Science, 244, 48-52 .
39. SantaLucia, J., Jr and Zuker, M. (unpublished experiments).
40. Plum, G. E., Grollman, A. P., Johnson, F. and Breslauer, K. J. (1995). Biochemistry, 34,
16148-16160.
41. SantaLucia, J., Jr, Allawi, H. and Seneviratne, P. A. (1996). Biochemistry, 35, 3555-3562.
42. Gray, D. M. (1997). Biopolymers, 42, 795-810.
43. Goldstein, R. F. and Benight, A. S. (1992). Biopolymers, 32, 1679-1693.
44. SantaLucia, J., Jr (1998). Proc. Natl Acad. Sci. USA, 95,1460-1465.
45. Ladbury, J. E., Sturtevant, J. M. and Leontis, N. B. (1994). Biochemistry, 33, 6828-6833.
46. Seeman, N. C. (1990). J. Biomol Struct. Dyn., 8, 573-581.

355
JOHN SANTALUCIA, JR

47. Hirao, I., Nishimura, Y., Tagawa, Y., Watanabe, K. and Miura, K. (1992). Nucleic Acid
Research, 20, 3891-3896.
48. Antao, V. P., Lai, S. Y. and Tinoco, I., Jr (1991). Nucleic. Adds Res., 19, 5901-5905.
49. Walter, A. E., Turner, D. H., Kim, J., Lyttle, M. H., Muller, P., Mathews, D. H. and
Zuker, M. (1994). Proc. Natl. Acad. Sci. USA, 91, 9218-9222.
50. Peyret, N. and SantaLucia, J., Jr (1999). (In preparation).
51. Bevington, P. R. (1969). Data reduction and error analysis for the physical sciences. McGraw-
Hill, New York. pp. 58-60
52. Press, W. H., Flannery, B. P., Teukolsky, S. A. and Vetterling, W. T. (1989) pp 52-64,
498-520. Cambridge University Press, New York.
53. Xia, T., SantaLucia, J., Jr, Burkard, M. E., Kierzek, R., Schroeder, S. J., Jiao, X., Cox, C.
and Turner, D. H. (1998). Biochemistry, 37,14719-14735.
54. Milligan, J. F., Groebe, D. R., Witherell, G. W. and Uhlenbeck, 0. C. (1987). Nucleic
Acids Res., 15, 8783-8798.
55. Capaldi, D. and Reese, C. (1994). Nucleic Acids Res., 22, 2209-2216.
56. Brown, T. and Brown, D. J. S. (1991). In Oligonudeotides and analogues (ed. F. Eckstein),
pp. 1-24. IRL Press, Oxford.
57. Zimmer, D. and Crothers, D. (1995). Proc. Natl. Acad. Sci. USA, 92, 3091-3095.
58. Arghavani, M. B. and Romano, L. J. (1995). Anal. Biochem., 231, 210-209.
59. McLaughlin, L W. (1989). Chem. Rev., 89, 309-319.
60. Chou, S.-H., Flynn, P. and Reid, B. (1989). Biochemistry, 28, 2422-2435.
61. Zarrinkar, P. P. and Williamson, J. R. (1994). Science, 265, 918-924.
62. Primrose, W. U. (1993). In NMR of macromokcules: A practical approach (ed. G. C. K.
Roberts), pp. 7-34. IRL Press, Oxford.
63. Richards, E. G. (1975). In Handbook of biochemistry and molecular biology: Nucleic acids (ed.
G. D. Fasman), Vol. I, p. 597. CRC Press, Cleveland, OH.
64. Yu, Q., Kandegedara, A., Xu, Y. and Rorabacher, D. B. (1997). Anal. Biochem., 253,
50-56.
65. Allawi, H. T. and SantaLucia, J., Jr (1998). Biochemistry, 37, 9435-9444.
66. Perrin, D. D. and Dempsey, B. (1979). Buffers for pH and metal ion control. Halsted Press,
New York. pp. 6-20
67. Lewis, E. A., Hansen, L. D., Baca, E. J. and Temer, D. J. (1976). J. Chem. Soc. Perkin II
125-128.
68. Delcourt, S. G. and Blake, R. D. (1991). J. Biol. Chem., 266,15160-15169.
69. Blake, R. D., Vosman, F. and Tarr, C. (1981). In Biomolecular stereodynamics (ed. R. H.
Sarma), Vol. I, pp. 439-458. Adenine Press, New York.
70. Yen, S.-W. W. and Blake, R. D. (1980). Biopolymers, 19, 681-700.
71. Freier, S. M., Hill, K. O., Dewey, T. G., Marky, L. A., Breslauer, K. J. and Turner, D. H.
(1981). Biochemistry, 20,1419-1426.
72. Gralla, J. and Crothers, D. M. (1973). J. Mol. Biol., 73, 497.
73. Chaires, J. B. (1997). Biophys. Chem., 64,15-23.
74. Sugimoto, N., Nakano, S., Yoneyama, M. and Honda, K. (1996). Nucleic Acids Res., 24,
4501-4505.
75. Wartell, R. M. and Benight, A. S. (1985). Phys. Rep., 126, 67-107.
76. Vologodskii, A. V., Amirikyan, B. R., Lyubchenko, Y. L. and Frank-Kamenetskii, M. D.
(1984). J. Biomol. Struct. Dyn., 2,131-148.
77. Meyer, S. L. (1975). Data analysis for scientists and engineers. Wiley, New York. Chapter 19
78. Snedecor, G. W. and Cochran, W. G. (1971). Statistical methods. The Iowa State
University Press, Ames, IA. p. 190
79. Efron, B. and Tibshirani, R. (1993). An introduction to the bootstrap. Chapman & Hall,
London.

356
 List of suppliers

This core list of suppliers appears in all books in the Practical Approach series.

Aladdin Biofluorescence Center, School of Pharmacy, University of Wisconsin-

Madison, 425 N. Charter St., Madison, WI53706 USA
([email protected]).
URL:http://www.src.wisc.edu/highlights/time_flour/default.hrml.
Ambion, Inc., 2130 Woodward Street, 200, Austin, TX 78744-1832, USA
Amresco, 30175 Solon Industrial Parkway, Solon, OH 44139, USA
Anderman and Co. Ltd, 145 London Road, Kingston-upon-Thames, Surrey, KT2
6NH
Tel: 0181 541 0035 Fax: 0181 541 0623
Applied Photophysics Ltd, 203/205 Kingston Road, Leatherhead, Surrey, KT22 7PB,
UK (http://www.apltd.co.uk/).
Applied Scientific, 154 W Harris Avenue, South San Francisco, CA 94080, USA
Aviv Instruments Inc., Lakewood, New Jersey, USA (http://www.avivinst.com/).

Beckman Coulter Inc. 4300 N Harbor Boulevard, PO Box 3100, Fullerton, CA

92834-3100, USA
Tel: 001 714 871 4848 Fax: 001 714 773 8283 Web site: www.beckman.com
Beckman Coulter (U.K.) Limited, Oakley Court, Kingsmead Business Park, London Road,
High Wycombe, Buckinghamshire, HP11 1JU
Tel: 01494441181 Fax: 01494 447558 Web site: www.beckman.com
Becton Dickinson and Co., 21 Between Towns Road, Cowley, Oxford, 0X4 3LY
Tel: 01865 748844 Fax: 01865 781627 Web site: www.bd.com
Becton Dickinson and Co., 1 Becton Drive, Franklin Lakes, NJ 07417-1883, USA
Tel: 001 201 847 6800 Web site: www.bd.com
Bel-Art Products
BioDiscovery, 11150 W Olympic Blvd. Ste. 805E, Los Angeles, CA 90064, USA
Bio 101 Inc., c/o Anachem Ltd, Anachem House, 20 Charles Street, Luton, Bedford-
shire, LU2 OEB
Tel: 01582 456666 Fax: 01582 391768 Web site: www.anachem.co.uk
Bio 101 Inc., PO Box 2284, La Jolla, CA 92038-2284, USA
Tel: (+1)7605987299 Fax: (+)1 760 598 0116 Web site: www.biol01.com

357
LIST OF SUPPLIERS

Biologic-Science Instruments SA, Claix, France (http://www.bio-logic.fr/).

Bio-Rad Laboratories Ltd., Bio-Rad House, Maylands Avenue, Kernel Hempstead,
Hertfordshire, HP2 7TD
Tel: 0181 328 2000 Fax: 0181 328 2550 Web site: www.bio-rad.com
Bio-Rad Laboratories Ltd., Division Headquarters, 1000 Alfred Noble Drive, Hercules,
CA 94547, USA
Tel: (+)1 510 724 7000 Fax: (+)1 510 741 5817 Web site: www.bio-rad.com

Calbiochem-Novabiochem Corporation, 10394 Pacific Center Court, San Diego,

CA 92121 USA. Mailing Address: P. O. Box 12087, La Jolla, CA 92039-2087
Tel: 800 854 3417 (Calbiochem)/800 228 9622 (Novabiochem)/619 450 9600
Fax: 800 776 0999/619 450 9600
E-mail: [email protected] [email protected]
Web Site: http://www.calbiochem.com
Cartesian Technologies Inc., 17781 Sky Park Circle, Irvine, CA 92614, USA
Center for Fluorescence Spectroscopy, University of Maryland School of
Medicine, 108 North Greene Street, Baltimore, MD 21201 USA.
E-mail: [email protected]. URL: http://charlie.ab.umd.edu/cfs/info.html.
Cortex Technology, Hadsund, Denmark.
CP Instrument Company Ltd, PO Box 22, Bishop Stortford, Hertfordshire, CM23
3DX
Tel: 01279 757711 Fax: 01279 755785 Web site: www.cpinstrument.co.uk
Current Designs Inc., 3527 Hamilton Street, Philadelphia, PA 19104-2420, USA.

Dupont (UK) Ltd, Industrial Products Division, Wedgwood Way, Stevenage, Herts,
SGI4QN
Tel: 01438 734000 Fax: 01438 734382 Web site: www.dupont.com
Dupont Co, (Biotechnology Systems Division), PO Box 80024, Wilmington, DE 19880-
002, USA
Tel: (+)1 302 774 1000 Fax: (+)1 302 774 7321 Web site: www.dupont.com

Eastman Chemical Company, 100 North Eastman Road, PO Box 511, Kingsport,
TN 37662-5075, USA
Tel: (+)1 423 229 2000 Web site: www.eastman.com
Edinburgh Instruments Ltd, Riccarton, Currie, Edinburgh EH14 4AP, UK.
E-mail: [email protected]. URL: http://www.edinst.com.

Fisher Scientific UK Ltd, Bishop Meadow Road, Loughborough, Leicestershire,

LE11 5RG
Tel: 01509231166 Fax: 01509 231893 Web site: www.fisher.co.uk
Fisher Scientific, Fisher Research, 2761 Walnut Avenue, Tustin, CA 92780, USA
E-mail: [email protected]. URL: http://www.physics.uiuc.edu/groups/fluorescence/.
Tel: (+)1 714 669 4600 Fax: (+)1 714 669 1613 Web site: www.fishersci.com
Fluka, P.O. Box 2060, Milwaukee, WI 53201, USA
Tel: (+)1 414 273 5013 Fax: (+)1 414 2734979 Web site: www.sigma-aldrich.com

358
 LIST OF SUPPLIERS

Fluka Chemical Company Ltd, PO Box 260, CH-9471, Buchs, Switzerland

Tel: {+) 41 81 745 2828 Fax: (+) 41 81 756 5449 Web site: www.sigma-aldrich.com

GeneMachines, PO Box 2048, Menlo Park, CA 94026, USA

General Scanning Inc., 500 Arsenal Street, Watertown, MA 02472, USA
Genetic Microsystems Inc., 34 Commerce Way, Woburn, MA 01801, USA
Genetix Ltd., Unit 1, 9 Airfield Road, Christchurch, Dorset BH23 3TG
Genomic Solutions Inc., 4355 Varsity Drive Ste. E, Ann Arbor, MI 48108, USA
Genome Systems Inc., 4633 World Parkway Circle, St. Louis, MO 63134, USA
Genosys Biotechnologies Inc., Lake Front Circle, Suite 185, The Woodlands, TX
77380, USA
Globals Unlimited, Laboratory for Fluorescence Dynamics, University of Illinois at
Urbana-Champaign, Department of Physics, 1110 West Green Street, Urbana, IL
61801-3080 USA.

Hewlett-Packard, 3000 Hanover Street, Palo Alto, CA 94304-1185, USA

Web site: www.Hewlett-Packard.com
Hi-Tech Scientific Ltd, Brunei Rd, Salisbury, Wiltshire, SP2 7PU, U.K.,
(http://www.hi-techsci.co.uk/).
Hybaid Ltd, Action Court, Ashford Road, Ashford, Middlesex, TW15 1XB
Tel: 01784 425000 Fax: 01784 248085 Web site: www.hybaid.com
Hybaid US, 8 East Forge Parkway, Franklin, MA 02038, USA
Tel: (+)1 508 541 6918 Fax: (+)1 508 541 3041 Web site: www.hybaid.com
HyClone Laboratories, 1725 South HyClone Road, Logan, UT 84321, USA
Tel: (+)1 435 753 4584 Fax: (+)1 435 753 4589 Web site: www.hyclone.com

Imaging Research Inc., Brock Univesiry, 500 Glenridge Avenue, St. Catherines,
Ontario L2S 3A1, Canada
Instruments S.A., Inc. (SPEX), 3880 Park Avenue, Edison, NJ 08820 USA.
URL: http://www.instrumentssa.com/.(See their web site for international offices.)
Intelligent Automation Systems, 149 Sidney Street, Cambridge, MA 02139, USA
IVEE Development AB, Forsta Langgata 26, SE-413 28 Goteborg, Sweden
Invitrogen BV, PO Box 2312, 9704 CH Groningen, The Netherlands
Tel: (+)800 5345 5345 Fax: (+)800 7890 7890 Web site: www.invitrogen.com
Invitrogen Corporation, 1600 Faraday Avenue, Carlsbad, CA 92008, USA
Tel: (+)1 760 603 7200 Fax: (+)1 760 603 7201 Web site: www.invitrogen.com

JASCO Incorporated, 8649 Commerce Drive, Easton, Maryland 21601-9903, USA

Tel: (+)1 410 822 1220 Fax: (+)1 410 822 7526 E-mail: [email protected]
Web site: www.jascoinc.com
JASCO Europe s.r.l Via Confalinieri 25, 22060 Cremella (Co), Italy
Tel: 039 956439 Fax: 039 958642 E-mail: [email protected]

Kartell, Via delle Industrie 1,20082 Noviglio, Milano, Italy

Tel (+)39 02900121 Web site: www.kartell.it

359
LIST OF SUPPLIERS

Laboratoire pour 1'Utilisation du Rayonnement Electromagnetic (LURE), Bat

209D Centre Universitaire Paris-Sud, B.P. 34 - 91898 Orsay Cedex, France.
URL: http://www.lure.u-psud.fr/www/bienvenue.htm.
Life Technologies Ltd, PO Box 35, Free Fountain Drive, Incsinnan Business Park,
Paisley, PA4 9RF
Tel: 0800 269210 Fax: 0800 838380 Web site: www.lifetech.com
Life Technologies Inc, 9800 Medical Center Drive, Rockvitte, MD 20850, USA
Tel: (+)1 301 610 8000 Web site: www.lifetech.com

Merck Sharp & Dohme Research Laboratories, Neuroscience Research Centre,

Terlings Park, Harlow, Essex CM20 2QR
Web site: www.msd-nrc.co.uk
Microflex, PO Box 1865, San Francisco, CA 94083-1865, USA
Millipore (UK) Ltd, The Boulevard, Blackmoor Lane, Watford, Hertfordshire, WD1
8YW
Tel: 01923 816375 Fax: 01923 818297
Web site: www.millipore.com/local/UK.htm
Millipore Corporation, 80 Ashby Road, Bedford, MA 01730, USA
Tel: (+)1 800 645 5476 Fax: (+)1 800 645 5439 Web site: www.millipore.com
Molecular Devices, Unit 6, Raleigh Court, Rutherford Way, Crawley, RH10 2PD, UK;
1311 Orleans Drive, Sunnyvale, CA 94089, USA
Molecular Dynamics, 928 East Arques Avenue, Sunnyvale, CA 94086, USA
Molecular Probes, Inc., 4849 Pitchford Avenue, PO Box 22010, Eugene OR 97402-
9165, USA
MSD Sharp and Dohme GmbH, Lindenplatz 1, D-85540, Haar, Germany
Web site: www.msd-deutschland.com

National Synchrotron light Source, Brookhaven National Laboratory, P.O. Box

5000 Upton NY 11973-5000 USA. URL: http://www.nsls.bnl.gov/.
New England Biolabs, 32 Tozer Road, Beverley, MA 01915-5510, USA
Tel: 001 978 927 5054
Nikon Corporation, Fuji Building, 2-3, 3-chome, Marunouchi, Chiyoda-ku, Tokyo
100, Japan
Tel: (+) 813 3214 5311 Fax: (+) 813 3201 5856
Web site: www.nikon.co.jp/main/index_e.htm
Nikon Inc, 1300 Walt Whitman Road, Melville, NY 11747-3064, USA
Tel: (+)1 516 547 4200 Fax: (+)1 516 547 0299 Web site: www.nikonusa.com
Nycomed Amersham plc, Amersham Place, Little Chalfont, Buckinghamshire,
HP7 9NA
Tel: 01494 544000 Fax: 01494 542266 Web site: www.amersham.co.uk
Nycomed Amersham, 101 Carnegie Center, Princeton, NJ 08540, USA
Tel: (+)1 609 514 6000 Web site: www.amersham.co.uk

Olis (On-line Instrument Services), Inc., 130 Conway Drive, Suites A and B, Bogart,
GA 30622-1724, USA (http://www.olisweb.com/).

360
 LIST OF SUPPLIERS

Omega Optical Inc., P.O. Box 573, Brattleboro, VT, 05302 - 0573, USA.
Email: [email protected].

Perkin-Elmer Corporation, 761 Main Avenue, Norwalk, CT 06859 USA.

URL: http://www.perkin-elmer.com/.
Perkin Elmer Ltd, Post Office Lane, Beaconsfield, Buckinghamshire, HP9 1QA
Tel: 01494 676161 Web site: www.perkin-elmer.com
Pharmacia Biotech (Biochrom) Ltd, Unit 22, Cambridge Science Park, Milton Rd,
Cambridge, Cambs, CB4 OFJ
Tel: 01223 423723 Fax: 01223 420164 Web site: www.biochrom.co.uk
Pharmacia and Upjohn Ltd, Davy Avenue, Knowlhill, Milton Keynes, Bucking-
hamshire, MK5 8PH
Tel: 01908 661101 Fax: 01908 690091 Web site: www.eu.pnu.com
Photon Technology International, Inc. (PTI), 1 Deerpark Drive, Suite F,
Monmouth Junction, NJ 08852 USA.
E-mail: [email protected].
URL: http://www.eurotek.com.pl/ptiwebsite/welcome.html (see their web site for
international offices).
Promega Corporation, 2800 Woods Hollow Road, Madison, WI 53711-5399, USA
Tel: (+)1 608 274 4330 Fax: (-t-)l 608 277 2516 Web site: www.promega.com
Promega UK Ltd, Delta House, Chilworth Research Centre, Southampton, SO16 7NS
Tel: 0800 378994 Fax: 0800 181037 Web site: www.promega.com

Qiagen UK Ltd, Boundary Court, Gatwick Road, Crawley, West Sussex, RH10 2AX
Tel: 01293 422911 Fax: 01293 422922 Web site: www.qiagen.com
Qiagen Inc, 28159 Avenue Stanford, Valencia, CA 91355, USA
Tel: (+)1 800 426 8157 Fax: (+)1 800 718 2056 Web site: www.qiagen.com
Quantum Northwest, 9723 W. Sunset Highway, Spokane, WA 99224-9426 USA.
E-mail: [email protected]. URL: www.qnw.com.

Research Genetics Inc., 2130 Memorial Pkwy SW, Huntsville, AL 35801, USA
Research Instrumentation Shop, University of Pennsylvania School of Medicine,
79E John Morgan Building, 3620 Hamilton Walk, Philadelphia, PA 19104-6059, USA
Roche Diagnostics Ltd, Bell Lane, Lewes, East Sussex, BN7 1LG
Tel: 01273 484644 Fax: 01273 480266 Web site: www.roche.com
Roche Diagnostics Corporation, 9115 Hague Road, PO Box 50457, Indianapolis, IN 46256,
USA
Tel: 001 317 845 2358 Fax: 001 317 576 2126 Web site: www.roche.com
Roche Diagnostics GmbH, Sandhoferstrasse 116, 68305 Mannheim, Germany
Tel: 0049 621 759 4747 Fax: 0049 621 759 4002 Web site: www.roche.com

Schleicher and Schuell Inc, Keene, NH 03431A, USA

Tel: 001 603 357 2398
Shandon Scientific Ltd, 93-96 Chadwick Road, Astmoor, Runcorn, Cheshire, WA7
1PR

361
LIST OF SUPPLIERS

Tel: 01928 566611 Web site: www.shandon.com

Shimadzu, 1, Nishinokyo, Kuwabaracho, Nakagyou-ku, Kyoto 604 8511, Japan
Tel: 81 (75) 823 1111 Fax: 81 (75) 823 1361 Web site: www.shimadzu.co.jp
Sigma-Aldrich Company Ltd, Fancy Road, Poole, Dorset, BH12 4QH
Tel: 01202 733114 Fax: 01202 715460 Web site: www.sigma-aldrich.com
Sigma Chemical Company, PO Box 14508, St Louis, MO 63178, USA
Tel: (+)1 314 771 5765 Fax: (+)1 314 771 5757 Web site: www.sigma-aldrich.com
Spectronic Instruments, Inc., 820 Linden Avenue, Rochester, NY 14625 USA.
E-Mail: [email protected]. URL: http://www.spectronic.com/ (see their web
site for international offices).
Stanford Research Systems, 1290-D Reamwood Avenue, Sunnyvale, California
94089 USA.
Email: [email protected]. URL: www.srsys.com/.
Stratagene Europe, Gebouw California, Hogehilweg 15, 1101 CB Amsterdam
Zuidoost, The Netherlands
Tel: 00 800 9100 9100 Web site: www.stratagene.com
Stratagene Inc, 11011 North Torrey Pines Road, La Jolla, CA 92037, USA
Tel: (+)1 858 535 5400 Web site: www.stratagene.com
Starna
Tel: (+)44 (0) 181 501 5550 Web site: www.starna.com
Synchrotron Radiation Source, Warrington, Cheshire, England.
Email: [email protected]. URL: http://www.dl.ac.uk/SRS/.
Synteni (Incyte Pharmaceuticals), 6519 Dumbarton Circle, Fremont, CA 94555,
USA

Tel-Test, Inc., 1511 County Road 129 Po Box 1421, Friendswood, TX 77546, USA

United States Biochemical, PO Box 22400, Cleveland, OH 44122, USA

Tel: 001 216 464 9277

Varian (Head Office), Varian Australia Pty Ltd, 679 Springvale Road, Mulgrave,
Victoria 3170
Tel: (+)61 1300 658 274 Fax: (+)61 1300 658 274 Web site: www.varianinc.com
Vysis Inc., 3100 Woodcreek Drive, Downers Grove, IL 60515, USA

Whatman International Ltd., Maidstone, Kent, UK

362
 Index

A215: 190 forNADPH 199 chromophore selection 288-91

•^26o' 338 for nucleic acids 194 circular dichroism 99-139
A280: 190 phosphomolybdate 202 applications 100, 121-38,
absorption 2, 33-7 precision 186 313-15, 322-3
absorption cells, see cuvettes for proteins 189-94 baseline correction 105
absorption spectrum 2, 37 rate assays 188-9 cuvettes 103-5
addresses of suppliers 357-62 samples 184-5 degenerate coupled-oscillator
a-helix circular dichroism stopped 201-2 111-13
123-4 volume 186-7 instrumentation 101-5
amino acid absorption spectra 6 ligand binding 134-8
amino acid assays 204-6 measurement units 119-21
ANS/bis-ANS 260, 321 noise reduction 107-8
apparent absorbance 4 bandwidth 17-19, 270 non-degenerate coupled-
arc lamps 42, 85, 89-90,102, baseline correction 22-4,105, oscillator 114-15
287 212, 213, 275, 318-19 nucleic acid analysis 129-34
aromatic side chain circular Beer-Lambert law 2-3, 184 octant rule 115
dichroism 125 deviations from 3-4 parameter settings 107,109
Arrhenius plots 178-9 Beer's law 2 protein analysis 121-9
assays 183-208 (3-sheet circular dichroism 124 protein unfolding 313-15,
accuracy 186 B-turn circular dichroism 124-5 322-3
for amino acid side chains bicinchoninic acid assays 192 samples 103, 108-9
204-6 blood, 286 signal to noise ratio 106
for ATP 199 boxcar averager 87, 88 spectrometers 101-3
automated 204 'burst' kinetics 226-7 spectrum measurement 99,
bicinchoninic acid 192 101-10
Biuret method 190-1 theory 110-19
continuous 201 wavelength range 105-6
coupled 203 calcium ion concentration zeroing 105
design 185-7 301-3 circular dichroism, induced
diluent 187 calibration 24, 275-8, 292 134-8
dye-binding 193 calorimetry 307, 333-4 circular dichroism, stopped-flow
end point assays 187-8 carbon monoxide (CO) 265-81
enzyme-based 195-9 difference spectroscopy 8 applications 278-81
of enzymes 200-4 carbonyl transition 115-17 attachments 271-3
fluorimetric 193-4 carotenoids 7 bandwidth 270
for glucose 197-9 cell membrane, see membrane baseline correction 275
H202 199 headings calibration 275-8
Lowry method 191-2 cells, see cuvettes cells 268-9
luminescence-based 199-200 centrifugal analysers 203-4 data interval 275
mixing order 187 cerebral cortex 295 dead time 267, 275

363
INDEX

degassing systems 276, 279 dansyl chloride 321 enzyme activators 176-7
flow systems 268 data analysis 30-1, 209-10, enzyme activity 170-1
instrumentation 266-74 227-39, 284, 348-54 andpH 177-8
instrument manufacturers d-d transitions 117-19 and temperature 178-9
271 deactivation 35-8 units 171
lysozyme folding 279-80 dead time 214-17, 267, 275 enzyme assays 200-4
mixing 269-70 decay associated spectra (DAS) enzyme-based assays 195-9
pneumatic drives 268 75 enzyme cooperativity 179-81
protein folding 278-80 degassing 276, 279, 340 enzyme inhibition 174-5
sample properties 267-8, 8-function pulses 73 enzyme kinetics 167-82, 220-7
269-70 deoxygenation 90-1 units 168-9, 171
sensitivity 267 derivative spectra 30-1 enzyme specificity 175-6
stand-alone systems 273-4 desalting 336, 337 eosine labelling 94
stepper motors 268, 269 diethylpyrocarbonate 206 equilibrium binding isotherms
syringe numbers 269 differential scanning 141-65
time constant 274, 275 calorimetry 307, 333-4 equipment suppliers 357-62
wavelength range 270 dilution schemes 342-3 ethoxyformic anhydride 206
wavelength scanning 270 dimerization 4 excitation spectra 54-5, 56
wavelength selection 274 diode array spectrophotometers exocytosis 303
zero-time 275 13-16
circularly polarized light 99 dissymmetry factor 119
CO difference spectroscopy 8 DNA
computer controlled as a catalytic template 280-1 FAD 321
instruments 17, 31, 47, non-self-complementary first-order kinetics 167
270-1 strand mixing 341 first-order rate constant 168,
computer programs, protein stability and pH 340 221-3
structure analysis 125-8 DnaB helicase/sADP flash lamps 42, 85, 89-90, 102,
constant fraction discriminator fluorescence titration 287
88 156-8 flavine adenine dinucleotide
continuous assays 201 DnaB helicase/TNP-ADP (FAD) 321
continuous flow analysers 188 fluorescence titration flow period 213-14, 215
convolution integral 73 146-7 fluorescein derivatives 321
Coomassie Brilliant Blue 193 DNA circular dichroism 129-34 fluorescence 37-8, 69-84
cooperativity of enzymes DNA fluorescence anisotropy fluorescence anisotropy 60-1,
179-81 77 75-8, 324
coumarin derivatives 321 DNA/ligand interactions 134-8 fluorescence assay 193-4
coupled assays 203 double beam fluorescence decay 70
Cro 312 spectrophotometers 9-10 fluorescence life time
curve fitting 348-52 double monochromator distribution 75
cuvettes 25-6 methods 16, 44 fluorescence quantum yield (4>F)
for circular dichroism 103-5, DTNB 204-5 39-40, 55, 57-8, 59-60,
268-9 dual-wavelength 245, 323-5
cleaning 26, 347 spectrophotometers 10-12 fluorescence quenching 62-6
disposable 26 dye-binding assays 193 acrylamide 65-6
for fluorescence spectrometry dyes 193, 289, 292 fluorescence resonance energy
44-5 transfer (FRET) 60
microvolume 343 fluorescence spectra 39, 47-58
mixing contents of 25-6 fluorescence spectrometry,
semi-micro 25 elastic scattering 48-9 steady state 33-67
for UV melting 343 electronic transitions 110-19 applications 58-66
volume determination 25 ellipticity 120-1 computer interfacing 47
cyanine dyes 292 Ellmans reagent 204-5 cuvettes 44-5
cysteine 204-5 endocytosis 303 electronic circuitry 46-7
cytochromes 7-9, 27-8 endosome pH 299-301 excitation spectra 54-5, 56
cytoplasm pH/calcium ion end point assays 187-8 inner filter effect 47-8
concentration 301-3 energy transfer 71-2 instrumentation 41-7

364
 INDEX

instrument settings 50-1 haem proteins 7-9 Raman (inelastic) 49-50

light scattering 48-50 310 helix circular dichroism Rayleigh (elastic) 48-9
light source 42 124 light sources 42, 85-6, 89-90,
optical components 45-6 HEPES 339 102, 287-8
photon counting 46-7 Hill coefficient/plot 180-1 linear extrapolation model
principles of 33-40 histidine 206 309-10
protein fluorescence 61-2 H2O2 assays 199 Lowry assay 191-2
protein unfolding, 323-5 hypochromicity 7, 330 low temperature spectroscopy
quantum yield measurement 20-2
55, 57-8, 59-60 luminescence-based assays
reference solution 199-200
preparation 45 1,5-IEDANS 321 lysine 205
sample holders 45 inelastic scattering 49-50 lysozyme 279-80
spectral corrections 51-4 infrared radiation 1
wavelength selectors 42-3 inhibition of enzymes 174-5
fluorescence spectroscopy, inner filter effect 47-8, 286-7
macromolecular binding
stopped-flow 241-63 instrument suppliers 357-62
applications 250-63 density function (MBDF)
intermolecular quenching 63-5
145
artefacts 245 intersystem crossing 37
'magic angle' 78
data collection 242-3 quantum yield (#ISC) 70
melting curves 330-4; see also
filters 247 intramolecular quenching 62-3
ultraviolet melting of
instrumentation 241-6 isosbestic points 24
nucleic acids
mixing efficiency 243-4 isotherm, equilibrium binding
melting temperature (TM) 330,
photomultiplier tube (PMT) 141-65
331, 350-1
voltage 247-9
membrane potentials 295-9
reporter probes 249-63
membrane protein circular
sample preparation 244-5
dichroism 128-9
temperature control 245-6 katals 171
membrane vesicle pH/calcium
turbulence 246 kidney 293-5
ion concentration 301-3
wavelength selection 247 kinetic difference spectrum
mercury arc lamps 287
fluorescence spectroscopy, time- 219-20
MES 339
resolved 69-89 kinetics 167-82, 220-7
metallo-proteins 123
applications 77 first-order/second-order 167
metmyoglobin-cyanide
data analysis 78-84 limiting rate (V) 170, 172-4
complex reduction 234
electronic circuitry 86-9 reaction rate (v) 167-8,
Michaelis constant (Km) 170
instrumentation 84-9 169-70
determination 172-4
light sources 85-6 units 168-9, 171
Michaelis-Menten equation
principles of 69-78 Klett meters 4-5
169-70
fluorescence titrations 146,
Michaelis-Menten parameters
147, 156-8
170
quantitative analysis 147-51,
micro-colorimetry 207
158-60 Lambert's law 2
lasers 85, 287 microwell plate-reading
Forster's theory 71
spectrophotometers 16
FRET (fluorescence resonance least-squares fit 78-9
mitochondria, oxidation-
energy transfer) 60 lifetime distribution 75
reduction state 293-5
ligand binding 134-8, 141-65,
monochromators 43, 44
251-6
slit width 217-18, 246-7
warfarin/human serum
multi-wavelength
Gibbs-Helmholtz equation albumin 257-60
spectrophotometers 13
309 ligand fluorescence 257-60
global analysis 233-9, 320 ligands, induced circular
glucose assays 197-9 dichroism 134-8
glucose 6-phosphate light 1-2 NAD(P)H
dehydrogenase 197,198-9 light energy 2 absorption spectrum 7
green fluorescent protein light absorption 2, 33-7 assays 199
303-4 light scattering 4-6, 48-50 NanoOrange™ 193

365
INDEX

nephelometers 4 phenylalanine 321 fluorescence spectroscopy

Nernst equation 296 phosphate buffer 339 250-1, 323-5
nucleic acid phosphomolybdate assay 202 global analysis 320
absorption spectra 6-7 phosphorescence 38, 70-2 intermediates 316
denatured 7 phosphorescence lifetime/decay irreversibility of 317
hypochromicity 7, 330 71 kinetics 316-17
non-self-complementary phosphorescence quantum linear extrapolation model
strand mixing 341 yield 71, 72 309-10
stability and pH 340 phosphorescence spectroscopy modelling 308-13
nucleic acid assays 194 89-94 of oligomeric proteins
nucleic acid circular dichroism photomultiplier tubes (PMT) 46, 311-13
129-4 86, 247-9 pressure induced 311
nucleic acid ultraviolet melting photon counting 46-7, 88-9 simulations 311
329-56 photoproducts 37 temperature induced
buffers 339-40 PIPES 339 (thermal) 309
curve fitting 348-52 plasma membrane potential two-state model 308-9
cuvettes for 343 297-9 proteins, oligomeric, unfolding
data analysis 348-54 plate readers 16,183-4, 203 311-13
error analysis 353-4 polarized light, circularly 99 pseudo-first-order rate constant
instrumentation 343-5 polyether ether ketone (PEEK) 168, 225
melting protocol 345-6 267 pulse sampling 87-8
normalized absorbance polymerase chain reaction, pyrene derivatives 321
curves 352 automated 204 pyridoxal phosphate 205, 321
principles of 330-4 probes 249-63, 289, 291, 321
sample preparation 336-8, PRODAN 321
340, 342-3 progress-curve analysis 167
sample recovery 346-7 proline-like 3/1-helix circular quantum yield (<t)
sequence design 334-5 dichroism 124 fluorescence (4>F) 39-40, 55,
nucleic acid ultraviolet protein 57-8, 59-60, 245, 323-5
spectroscopy, 131-2 absorption spectrum 6 intersystem crossing ($ISC)
eosine labelling 94 70
protein assays 189-94 phosphorescence (0P) 71, 72
protein circular dichroism quenching 62-6
octant rule 115 121-9, 278-80
Olis RSM-1000: 16 protein fluorescence 61-2,
optical density 4 250-7
units 338 protein folding 250-1, 278-80 Raman scattering 49-50
optical rotatory dispersion 101, protein-ligand binding, see random coil circular dichroism
119 ligand binding 125
oxygen removal 90-1 protein phosphorescence rate assays 188-9
89-93 rate constant 167,168, 220-7
protein-protein binding 257 first order 168, 221-3
protein ultraviolet spectroscopy pseudo-first-order 168, 225
PCR, automated 204 121-3 second order 223-5
peroxidase 197, 198 protein ultraviolet-visible units 168
PH spectrum recording 29 Rayleigh criterion 20
of buffers 245-6 protein unfolding 307-27 Rayleigh scattering 48-9
of cytoplasm 301-3 absorbance spectroscopy reaction rate (v) 167-8, 169-70
ofendosomes 299-301 320-2 reflectance measurement 16
and enzyme activity 177-8 acid induced 310-11 cerebral cortex 295
jump 221-2 baseline trends 318-19 kidney 293-5
of membrane vesicles 301-3 circular dichroism 313-15, skin 16
and nucleic acid stability 340 322-3 reporter probes 249-63, 289,
and temperature 245-6 denaturant induced 309-10 291, 321
phase/modulation fluorometry differential scanning rhodamine derivatives 321
74-5, 83-4, 89 calorimetry 307 RNA stability and pH 340

366
 INDEX

sample preparation 184-5, mixing 269-70 and enzyme activity 178-9

244-5, 336-8, 340, 342-3 pneumatic drives 268 temperature control 20-2,
scan speed 20 protein folding 278-80 245-6
scattering of light 4-6, 48-50 sample properties 267-8, tetranitromethane 205-6
Raman (inelastic) 49-50 ' 269-70 thermodynamics 308
Rayleigh (elastic) 48-9 sensitivity 267 Tietze reaction 189
second-order kinetics 167 stand-alone systems 273-4 time-correlated single photon
second-order rate constant stepper motors 268, 269 counting 88-9
223-5 syringe numbers 269 time-resolved fluorescence
sensitized phosphorescence time constant 274, 275 anisotropy 75-8
yield 72 wavelength range 270 time-resolved fluorescence
SEP-PAK desalting 336, 337 wavelength scanning 270 spectroscopy 69-89
single beam wavelength selection 274 applications 77
spectrophotometers 9 zero-time 275 data analysis 78-84
singlet state excitation 35, 40 stopped-flow fluorescence electronic circuitry 86-9
singular value decomposition spectroscopy 241-63 instrumentation 84-9
228-33, 236 applications 250-63 light sources 85-6
singular values 232 artefacts 245 principles of 69-78
skin reflectance 16 data collection 242-3 time-resolved phosphorescence
specificity constant 170 filters 247 anisotropy 93-4
specificity of enzymes 175-6 instrumentation 241-6 TNBS 205
spectral bandwidth 17-19, 270 mixing efficiency 243-4 trace metal removal 336
spectral correction 51-4 photomultiplier tube (PMT) trans mission/transmittance
spectral resolution 19-20 voltage 247-9 3
spectrophotometers 284-5 reporter probes 249-63 trinitrobenzene sulphonic acid
choice of 22 sample preparation 244-5 205
computer controlled 17, 31, temperature control 245-6 triplet-triplet energy transfer
47, 270-1 turbulence 246 72
operating conditions 17-26 wavelength selection 247 Tris, buffer use 339
types of 9-17 stopped-flow spectroscopy tryptophan (Trp)
spectrophotometric assays, see 209-39 absorbance 321
assays absorbance change amplitude assays 206
spectrophotometry 2-6, 283-92 217-18 fluorescence 61-2, 321, 323
spectroscopic titrations, see baseline measurement 212, residues 249-50, 323-4
titrations 213 turbidimeters 4-6
split beam spectrophotometers data analysis 209-10, 227-39 turbid samples 29-30, 246,
9-10 dead time 214-17 285-6
Stern Volmer relationship 64 flow period 213-14, 215 tyrosine
stirrers 25-6 instrumentation 210-11 absorption spectrum 6, 321
stopped assays 201-2 kinetic difference spectrum assays 205-6
stopped-flow circular dichroism 219-20 fluorescence 321, 324-5
265-81 multi-wavelength detection
applications 278-81 227-39
attachments 271-3 'rapid scan' 209, 227-39
bandwidth 270 rate constants 220-7 ultraviolet melting of nucleic
baseline correction 275 single wavelength acids 329-56
calibration 275-8 measurements 212-20 buffers 339-40
cells 268-9 stability tests 212, 213 curve fitting 348-52
data interval 275 stray light 3, 285 cuvettes for 343
dead time 267, 275 suppliers 357-62 data analysis 348-54
degassing systems 276, 279 sychrotron radiation 85-6 error analysis 353-4
flow systems 268 instrumentation 343-5
instrumentation 266-74 melting protocol 345-6
instrument manufacturers normalized absorbance
271 temperature curves 352
lysozyme folding 279-80 and buffer pH 245-6 principles of 330-4

367
INDEX

sample preparation 336-8, v (reaction rate) 167-8,169-70 wavelength selection 17,

340, 342-3 van't Hoff analysis 330-1, 105-6, 212-13, 247, 274,
sample recovery 346-7 349-50,351-2 288
sequence design 334-5 visible spectrum 1, 2 Wood's anomaly 61
ultraviolet radiation 1, 2
ultraviolet spectroscopy 29,
121-3, 131-2 xenon arc lamps 42, 89-90,
warfarin/human serum albumin 102, 287
binding 257-60
wavelength of light 1
V (limiting rate) 170,172-4 and energy 2 zero-order reaction 168

368