Article

Structure of Microbial Nanowires Reveals Stacked

Hemes that Transport Electrons over Micrometers
Graphical Abstract Authors
Fengbin Wang, Yangqi Gu,
J. Patrick O’Brien, ..., Allon I. Hochbaum,
Edward H. Egelman, Nikhil S. Malvankar

Correspondence
[email protected] (A.I.H.),
[email protected] (E.H.E.),
[email protected] (N.S.M.)

In Brief
Stacked heme filaments form the
structural basis for long-range electron
transport in bacterial nanowires.

Highlights
d Geobacter nanowires are made up of micrometer-long
polymerization of cytochrome OmcS

d All hemes are closely stacked (<4–6 Å), providing a

continuous path for electron flow

d We show that these are the same filaments that were earlier
thought as type IV pili

d This structure explains the molecular basis for electron

conduction in protein wires

Wang et al., 2019, Cell 177, 361–369

April 4, 2019 ª 2019 Elsevier Inc.
https://doi.org/10.1016/j.cell.2019.03.029
Article

Structure of Microbial Nanowires Reveals Stacked

Hemes that Transport Electrons over Micrometers
Fengbin Wang,1,8 Yangqi Gu,2,3,8 J. Patrick O’Brien,2,4 Sophia M. Yi,2,4 Sibel Ebru Yalcin,2,4 Vishok Srikanth,2,4
Cong Shen,2,5 Dennis Vu,2,4 Nicole L. Ing,6 Allon I. Hochbaum,6,7,* Edward H. Egelman,1,* and Nikhil S. Malvankar2,4,9,*
1Department of Biochemistry and Molecular Genetics, University of Virginia School of Medicine, Charlottesville, VA 22908, USA
2Microbial Sciences Institute, Yale University, New Haven, CT 06516, USA
3Department of Molecular, Cellular & Developmental Biology, Yale University, New Haven, CT 06511, USA
4Department of Molecular Biophysics and Biochemistry, Yale University, New Haven, CT 06511, USA
5Department of Microbial Pathogenesis, Yale University, New Haven, CT 06511, USA
6Department of Materials Science & Engineering, University of California, Irvine, Irvine, CA 92697, USA
7Department of Chemistry, University of California, Irvine, Irvine, CA 92697, USA
8These authors contributed equally
9Lead Contact

*Correspondence: [email protected] (A.I.H.), [email protected] (E.H.E.), [email protected] (N.S.M.)

https://doi.org/10.1016/j.cell.2019.03.029

SUMMARY kar and Lovley, 2014; Malvankar et al., 2011). However, these fil-
aments’ composition, structure, and underlying conduction
Long-range (>10 mm) transport of electrons along mechanism have remained uncertain because the filaments
networks of Geobacter sulfurreducens protein fila- are difficult to solubilize for studies using traditional biochemical
ments, known as microbial nanowires, has been methods and X-ray crystallography. Geobacter sulfurreducens
invoked to explain a wide range of globally important serves as a model organism for the broader phenomenon of
redox phenomena. These nanowires were previously extracellular electron transfer because it produces these
conductive filaments (Tan et al., 2017) and has a fully sequenced
thought to be type IV pili composed of PilA protein.
genome and well-developed genetic system (Reguera et al.,
Here, we report a 3.7 Å resolution cryoelectron mi-
2005). In contrast to other electron-transferring bacteria (Marsili
croscopy structure, which surprisingly reveals that, et al., 2008), G. sulfurreducens does not use diffusing shuttle
rather than PilA, G. sulfurreducens nanowires are molecules but requires direct contact with an electron acceptor
assembled by micrometer-long polymerization of via conductive filaments for long-range extracellular electron
the hexaheme cytochrome OmcS, with hemes transfer (Reguera et al., 2005).
packed within
3.5–6 Å of each other. The inter-sub- G. sulfurreducens outer-surface c-type cytochromes, includ-
unit interfaces show unique structural elements such ing the hexaheme cytochrome OmcS (Qian et al., 2011), are
as inter-subunit parallel-stacked hemes and axial co- known to play a critical role in bacterial growth on insoluble elec-
ordination of heme by histidines from neighboring tron acceptors, including Fe (III) oxide and electrodes (Holmes
subunits. Wild-type OmcS filaments show 100-fold et al., 2006; Mehta et al., 2005). Nonetheless, previous studies
proposed that conductive Geobacter filaments were type IV pili
greater conductivity than other filaments from a
composed of PilA protein for a number of reasons: (1) electron-
DomcS strain, highlighting the importance of OmcS
transferring cells showed high levels of messenger RNA for PilA
to conductivity in these nanowires. This structure ex- (Childers et al., 2002), (2) the amino acid sequence of
plains the remarkable capacity of soil bacteria to G. sulfurreducens PilA is similar to the N-terminal sequence of
transport electrons to remote electron acceptors PilA from other type IV pili-producing bacteria (Reguera et al.,
for respiration and energy sharing. 2005), (3) genomic organization of G. sulfurreducens pilus biosyn-
thesis genes was also similar to other type IV pili-producing bac-
INTRODUCTION teria (Reguera et al., 2005), (4) a pilA deletion mutant strain lacked
filaments and could not transfer electrons extracellularly (Reg-
Conductive filamentous appendages of common soil bacteria uera et al., 2005), and (5) point mutations in pilA caused cells to
Geobacter (Reguera et al., 2005), referred to as microbial nano- produce filaments with different conductivities than wild-type
wires, play a critical role in long-range extracellular electron (WT) filaments (Adhikari et al., 2016; Tan et al., 2016a, 2017; Var-
transfer for respiration (Malvankar et al., 2011) and interspecies gas et al., 2013). Because of these extensive literature data, pre-
electron exchange (Summers et al., 2010). These nanowires vious studies, including some from authors of the present work
have been invoked to explain a wide range of globally important (S.E.Y., N.L.I., A.I.H., and N.S.M.), presumed that conductive
redox phenomena that influence carbon and mineral cycling in filaments were type IV pili composed of PilA protein.
soils and sediments, bioremediation, corrosion, and anaerobic Despite these data, there has never been any direct evidence
conversion of organic wastes to methane or electricity (Malvan- that conductive Geobacter extracellular filaments are composed

Cell 177, 361–369, April 4, 2019 ª 2019 Elsevier Inc. 361

Figure 1. Structure of Microbial Nanowires Reveals Closely Stacked Hemes in an OmcS Filament
(A) Cryo-EM image of the purified wild-type electrically conductive filaments showing a sinusoidal undulation, with a pitch of
200 Å shown in (B).
Scale bar, 200 Å.
(B) The surface of the reconstruction (transparent gray) with superimposed ribbon models of the OmcS subunits with three subunits in the center in three different
colors.
(C) Each subunit contains six hemes closely stacked over the micrometer lengths of the filaments.
(D) A zoomed region of the box shown in (C) with the minimum observed edge-to-edge distances indicated between hemes numbered in circles. The
distance between two hemes in adjacent subunits (heme 1 and heme 60 ) is comparable to the distances between parallel-stacked hemes within a subunit
(heme 2:heme 3 and heme 4:heme 5).

of PilA. Instead, the filament composition was inferred from growth conditions promote production of conductive biofilms
indirect evidence, including the presence of PilA in biochemical and filaments as well as overexpression of PilA in comparison
analyses (Tan et al., 2016b), or from low-resolution imaging by with growth on soluble electron acceptors such as fumarate.
atomic force microscopy (AFM) and negative-staining transmis- (Malvankar et al., 2011). Consistent with previous studies (Tan
sion electron microscopy, which suggested filament dimensions et al., 2016b), we confirmed the presence of both PilA and
similar to type IV pili (Reguera et al., 2005). OmcS with expected molecular weights of
6.5 kDa and
On the other hand, alternative functional roles have been sug-
45 kDa, respectively, in our filament preparations using poly-
gested for PilA (Reguera et al., 2007). PilA is implicated in the secre- acrylamide gel electrophoresis (SDS-PAGE), peptide mass
tion of OmcS to the outer surface (Liu et al., 2018; Richter et al., spectrometry, and western immunoblotting (Figure S1; Method
2012), and overexpression of PilA is accompanied by overproduc- Details).
tion of OmcS and extracellular filaments (Leang et al., 2013; Sum- Cryo-EM images of filaments purified from the WT strain
mers et al., 2010). Here, we reconcile previous observations by showed a sinusoidal morphology with a period of
200 Å (Fig-
directly identifying the composition and structure of extracellular ure 1A). Averaged power spectra from multiple filaments show
appendages of wild-type G. sulfurreducens using cryoelectron mi- a meridional layer line at
1/(47 Å) (Figure S4B), establishing
croscopy (cryo-EM). These appendages are polymerized fila- that there are
4.3 subunits per turn of a
200 Å pitch 1-start
ments of OmcS with unique structural features that also provide helix (Figure 1B). Using the iterative helical real space recon-
a molecular basis for understanding long-range electronic trans- struction (IHRSR) approach (Egelman, 2000), we were able to
port in proteins. Gel electrophoresis, mass spectrometry, AFM, reach a resolution where the handedness of a helices was clearly
and conductivity measurements show that these nanowires are visible. The 1-start filament helix was left handed, with a rise per
the same filaments that were previously thought to be type IV subunit of 46.7 Å and a rotation of 83.1 (Figure 1B), substan-
pili. Data from our measurements are consistent with previous tially different than type IV pili that typically show a rise of
studies, and our results establish a previously unknown class of
10 Å and a right-handed helix (Wang et al., 2017). The tracing
protein-based nanowires based on cytochrome polymerization. of the Ca backbone of the protein subunit at this resolution re-
vealed that the asymmetric unit contained at least 380 residues,
RESULTS AND DISCUSSION which contrasts with the 61 residues present in PilA (Reardon
and Mueller, 2013; Reguera et al., 2005). Further, there was no
We grew WT cultures using anodes of microbial fuel cells as the apparent internal symmetry that would arise in the asymmetric
sole electron acceptors (O’Brien and Malvankar, 2017). These unit if it contained multiple copies of identical chains. The NMR

362 Cell 177, 361–369, April 4, 2019

 Figure 2. De Novo Atomic Model Building of
OmcS Filaments
(A) Sequence-based alignment of OmcS and two
other c-type cytochromes with similar molecular
weight detected by mass spectroscopy (Figure S1B).
The six conserved CXXCH motifs responsible for
heme binding are highlighted in red. The histidine
residues paired with the CXXCH motifs in heme
binding are highlighted in yellow. Regions with
insertions or deletions compared to the OmcS
sequence are highlighted in blue.
(B) The per-residue real space correlation coefficient
(RSCC) plot of the atomic model against the 3.7 Å
cryo-EM map (top), with protein and ligand displayed
separately. The protein Ca trace in blue with ligands
(bottom), with N and C termini labeled.
(C and D) Zoomed view of the regions indicated in (A)
by green and black arrowheads, respectively, with the
OmcS atomic model fit into the cryo-EM map. The
green arrowhead in (C) indicates the location where
the two other cytochromes (OmcT, GSU2501) show a
three-residue insertion, not compatible with the map.
The black arrowhead in (D) indicates a region where
OmcT has a two-residue deletion and GSU2501 has a
serine and glycine rather than the tyrosine and proline
found in OmcS. The map has extra density that could
not be explained by a two-residue deletion or by
serine and glycine.

OmcT, GSU2501, or both that prevented

their sequences from being fit into the map
(Figures 2A, highlighted in blue, and S2). In
addition, the pattern of bulky amino acids
clearly established that only the OmcS
sequence was consistent with the map.
For example, the elongated density from
structure of PilA (Reardon and Mueller, 2013) also failed to fit into arginine (R) 256 is clearly seen in the map (Figure 2D). But in
the observed EM density map. Surprisingly, we found that there the sequences of OmcT and GSU2501, this residue would be a
were six hemes per asymmetric unit, with the highest densities in valine (Figure 2A, highlighted in purple) that will not be able to
the volume at the centers of these hemes, suggesting the pres- explain the map density. These studies show that the only cyto-
ence of metal atoms. Given that the protein contains at least 380 chrome, found in our filament preparations by mass spectrom-
residues and would therefore have a likely molecular weight be- etry, that is consistent with the cryo-EM map is OmcS.
tween 40 and 50 kDa, we used SDS-PAGE and cut out the stron- We built an ab initio atomic model of OmcS, which was then
gest band in the gel, which was at
45 kDa (Figure S1A), used to refine a filament model. We directly estimated its resolu-
analyzing this band by mass spectrometry. Five proteins were tion using a map:model comparison (Neumann et al., 2018;
identified, four of which had heme-binding motifs (CXXCH) and Subramaniam et al., 2016), which yielded an estimate of 3.7 Å
had masses between 45 and 49 kDa (Figure S1B). Three of these (Figure S3A). Strikingly, the ‘‘gold standard’’ measure of resolu-
proteins (OmcS, OmcT, and GSU2501) had a very similar pattern tion employed by Relion (Scheres, 2012), which is based upon
of heme-binding motifs that approximately matched the initial Ca reproducibility, yielded an estimate of 3.2 Å. Comparing the
trace (Figure 2B), and their sequences could be easily aligned Relion map (filtered to 3.2 Å) and our map generated in Spider
(Figure 2A). These three proteins had between 45% and 63% (filtered to 3.6 Å) showed no significant differences (Figures
sequence identity with each other. Another protein found in S3B and S3D), suggesting that the Relion estimate of resolution
mass spectrometry, OmcZ (Inoue et al., 2010), contained
30 is overly optimistic. Overall, the fit of the model to the map (Fig-
additional residues compared to the other three, and thus its ures S3E and S3F), the good density for many side chains, and
sequence could not be aligned to the others, and furthermore, the refinement statistics (Table 1) all validate our model of the
the pattern of eight heme-binding motifs in OmcZ did not match OmcS filament.
that found in the map. The remaining protein, OmpJ, contains no While it has been known for more than half a century that cy-
heme (Afkar et al., 2005). Of the three possible candidates, only tochromes can polymerize in ethanolic solutions (Margoliash
the OmcS sequence could be threaded through the map without and Lustgarten, 1962) and structures have been determined
any conflicts (Figures 2C, 2D, and S2). There are nine regions of for aggregates up to tetramers (Hirota et al., 2010), natural

Cell 177, 361–369, April 4, 2019 363

Table 1. Refinement Statistics for the OmcS Model in heme binding motifs, are highly conserved amino acids in
OmcS (Ashkenazy et al., 2016). The lack of structural homology
OmcS model
seen with these other c-type cytochromes is consistent with the
Helical symmetry
observation that there is no conserved fold for this family of
Rise (Å) 46.7 proteins (Bertini et al., 2006).
Rotation ( ) 83.1 The model shows that each OmcS subunit contacts only one
Resolution (Å) 3.7 subunit on either side, so that all connectivity in the filament is
Model to map CC 0.82 along the left-handed 1-start helix (Figure 1B). The interface be-
Clash score, all atoms 12.8 tween adjacent subunits is extensive, with
2,600 Å2 of surface
area buried per subunit (Figure 3A). In addition to the buried sur-
Protein geometry
face area, interactions between adjacent subunits incorporate
Ramachandran favored (%) 88.2
additional stabilizing elements unique to the filament structure.
Ramachandran outliers (%) 0.4 Histidine 16 in each subunit coordinates the iron in heme 5 of
Rotamer outliers (%) 0.0 an adjacent subunit (Figure 3B). Furthermore, heme pairs at
Cb deviations > 0.25 Å 0 the interface are parallel rather than T-shaped, with
4 Å
RMS deviations edge-to-edge distance (Figures 1D, 3B, and 3C). This parallel
Bond (Å) 0.01
stacking and inter-subunit coordination of heme may contribute
substantially to the stability of the protein-protein interface. In
Angles ( ) 1.36
addition, the presence of parallel-stacked hemes at the interface
PDB ID 6EF8
suggests facile transport of electrons between monomers.
EMDB ID EMD-9046 There is no precedent for such seamless micrometer-long
polymerization of hundreds of cytochromes to our knowledge.
Based on previous studies that have shown that cytochromes
polymerization of the type we observe here has not been previ- could form tetramers, the OmcS polymerization could be due
ously described to our knowledge. Stacking arrangements of ar- to successive domain swapping, where the c-terminal helix
omatic rings generally prefer parallel (offset face-to-face) or can be displaced from its original position in the monomer and
perpendicular (T-shaped) conformations (Janiak, 2000). The par- histidine-heme coordination can be perturbed significantly (Hir-
allel stacking yields the highest electronic coupling, which max- ota et al., 2010). We have determined the structure for the
imizes electron transfer (Jiang et al., 2017), whereas the T-shape OmcS protomer within the filament, but the structure of isolated
enhances structural stability (Janiak, 2000). Hemes in the OmcS OmcS monomer is needed to provide insight into this surprising
nanowires form parallel-stacked pairs, with each pair perpendic- polymerization process.
ular to the next, forming a continuous chain over the entire length The atomic structure was solved with filaments from conduc-
of the filament (Figure 1D). The minimum edge-to-edge dis- tive biofilms of electrode-grown cells that require long-distance
tances is 3.4–4.1 Å between the parallel hemes and 5.4–6.1 Å be- electron transport. However, fumarate-grown cells can also pro-
tween the perpendicular stacked pairs (Figure 1D). For all hemes duce conductive filaments (Ing et al., 2017; Leang et al., 2010,
in OmcS nanowires, two histidines axially coordinate iron at the 2013; Malvankar et al., 2011; Reguera et al., 2005). Filaments
center of each heme, and the vinyl groups of each heme form co- purified with fumarate-grown cells showed similar structure
valent thioether bonds with cysteines (Figure 3). The bis-histidine to filaments purified from electrode-grown cells (Figures S4A
axial ligation of the heme iron atoms are consistent with the co- and S4B). Moreover, previously published images of intact
ordination found in other multi-heme c-type cytochromes G. sulfurreducens filaments attached to cells (Leang et al.,
(Clarke et al., 2011), and the cysteine linkages are consistent 2013) showed structural features similar to purified OmcS fila-
with the c-type hemes reported to occupy six heme-binding mo- ments such as an identical helical rise of 47 Å (Figure S4C). These
tifs of OmcS (Qian et al., 2011). results showed that purified OmcS filaments are similar in di-
Our OmcS model filament has a low percentage of a and 310 mensions and structure to cell-attached filaments that were
helices (
13%) as well as b strands (
6%), leaving
81% of previously thought to be type IV pili (Leang et al., 2013). Impor-
the model as turns and coil, which is consistent with previous tantly, these studies also indicate that the formation of OmcS fil-
secondary-structure studies of OmcS (Qian et al., 2011). We aments is a natural process and not due to artificial preparation
compared the OmcS protomer within the filament with a group or pH conditions that can cause cytochrome c to form filamen-
of three crystallographic structures of other multi-heme c-type tous structures under extremely denaturing conditions (Haldar
cytochromes (PDB: 1OFW, 3UCP, and 3OV0). These structures et al., 2015).
showed 45%, 49%, and 60% turns and coils, respectively, with The cryo-EM images also showed another filamentous struc-
uniformly hydrophobic cores surrounding the hemes and heme- ture that was thinner than the OmcS filament (Figure S5). The
binding residues. Cores in our model of OmcS also included averaged power spectrum of these filaments showed similar
buried charges (arginine at locations 333, 344, and 375) that layer lines to the OmcS filament but with a slightly different axial
lack proximal compensating charges, as well as buried side- rise of
57 Å and rotation of
160 , suggesting that this thinner
chain hydroxyls (tyrosine at locations 186, 231, and 385). In addi- filament is also not a type IV pilus and could potentially be
tion, our model of OmcS has a salt bridge between protein another cytochrome filament. Due to lower abundance of this
chains, aspartate 407 to arginine 151, which along with cysteines filament in our cryo-EM images, it was not possible to build an

364 Cell 177, 361–369, April 4, 2019

Figure 3. Subunit Interface Interactions within OmcS Filament
(A) The large interface in the filament (
2,600 Å2 per subunit) is due to the complementarity between the upper portion of bottom subunit (red) and the lower
portion of the top subunit (green). Residues in one subunit strongly interact via hemes shown in the dashed circle and rectangle, and corresponding zoomed
images are shown in (B) and (C), respectively.
(B) Histidine 16 of the bottom subunit is coordinating the iron atom in heme 50 of the top subunit. The cryo-EM densities corresponding to histidine 16, histidine
3320 , and heme 50 are shown in a mesh.
(C) The stacking of heme 60 from a top subunit with heme 1 from the subunit below.

atomic model or to determine its composition. No filaments with measurements of WT filaments (Adhikari et al., 2016), further
power spectra consistent with type IV pili were observed in puri- suggesting that OmcS filaments are similar to the WT
fied filaments preparations or in previously published images of nanowires discussed in previous studies as type IV pili. DC-con-
intact, cell-attached filaments (Leang et al., 2013). ductivity measurements of individual DomcS filaments showed a
To evaluate the contribution of OmcS to the conductivity of very low conductivity that was more than 100-fold lower than
these filaments, the direct current (DC) conductivity of individual OmcS filaments (Figures 4F and 4G). Our conductivity measure-
OmcS filaments of WT strain was compared with sparse filaments ments thus show that OmcS is required for the high conductivity
produced by a DomcS strain (Figure 4B). G. sulfurreducens forms a of these
4 nm-thick filaments.
variety of filaments in response to genetic mutations (Klimes et al., Our finding that WT G. sulfurreducens nanowires are OmcS fil-
2010). Therefore, it is not surprising that the DomcS strain also aments is consistent with previous physiological studies. These
forms filaments that were previously thought to be conductive studies highlight the importance of OmcS in extracellular elec-
type IV pili (Leang et al., 2010). AFM (Figures 4A–4E) revealed tron transfer (Holmes et al., 2006; Leang et al., 2010, 2013;
distinct structural features for WT G. sulfurreducens filaments Mehta et al., 2005; Summers et al., 2010). OmcS is one of the
versus the type IV pili of other species (Wang et al., 2017) and fila- most abundant cytochromes found in the proteome for elec-
ments from the DomcS strain (Leang et al., 2010). In contrast to the tricity-producing G. sulfurreducens and is required only during
linear and smooth-surfaced structure of DomcS filaments, extracellular electron transfer to insoluble electron acceptors
WT OmcS filaments exhibited an axial periodicity with a 20 nm such as Fe (III) oxide (Holmes et al., 2006; Mehta et al., 2005).
pitch (Figure 4E), consistent with the helical pitch determined by It is also critical for direct interspecies electron transfer between
cryo-EM (Figure 1B). Filaments of the DomcS strain showed no syntrophic Geobacter co-cultures as evolved co-culture overex-
apparent axial periodicity under AFM. Moreover, DomcS filament pressed OmcS and deletion of the omcS gene inhibited bacterial
thickness (
1.7 nm height measured by AFM) was half that of WT ability to exchange electrons (Summers et al., 2010).
OmcS filaments (
4 nm) (Figures 4D and 4E). This substantial OmcS also plays a critical role in electron transport to
thickness difference and distinct axial periodicity observed for electrodes in current-producing biofilms. Both microarray and
OmcS filaments was used to confirm that OmcS filaments studied quantitative reverse transcription polymerase chain reaction
for electrical measurements are the same OmcS nanowires char- (qRT-PCR) analysis have demonstrated that cells show greatest
acterized by cryo-EM. increase in transcript levels for OmcS during the early stages of
AFM was further used to locate an individual filament bridging growth on electrodes (Holmes et al., 2006). Furthermore, immu-
two gold electrodes (Figure 4F, inset). Our DC-conductivity mea- nogold localization has shown that OmcS is distributed
surements of individual OmcS filaments fully hydrated in buffer throughout conductive G. sulfurreducens biofilms (Leang et al.,
yielded values (Figures 4F and 4G) comparable to previous 2013) and that deletion of the omcS gene inhibits their

Cell 177, 361–369, April 4, 2019 365

 Figure 4. Electrical Measurements Show that
OmcS Is Required for Filament Conductivity
(A–C) AFM height image of filaments from (A) WT
strain, (B) DomcS strain, and (C) zoomed image of
region shown in (B). Inset in (A) is the phase image
overlaid on height image that shows the repeating
pattern. Scale bars, 20 nm (A and C) and 100 nm (B).
(D) The height profile for filaments of WT and DomcS
strains using lateral cross section (dashed lines in [A]
and [C]).
(E) Longitudinal height profile for filaments of WT
and DomcS strain (solid lines in [A] and [C]).
(F) Current-voltage profile for individual filaments of
WT and DomcS strain compared to buffer alone.
Inset: AFM image for filament by DomcS strain
across gold electrodes. Scale bar, 500 nm.
(G) Comparison of DC conductivity for individual
filaments of WT and DomcS strains. Error bars
represent SEM of three biological replicates.

by overproduction of OmcS and filaments

(Leang et al., 2013; Summers et al., 2010),
further suggesting that PilA is involved in
secretion of OmcS filaments that may
explain previous correlations found be-
tween PilA and biofilm conductivity
(Malvankar et al., 2011). The requirement
of PilA for the synthesis of OmcS filaments
thus also explains the inability of DpilA cells
production of electricity under some conditions (Holmes et al., to grow on insoluble electron acceptors such as Fe (III) oxides
2006). However, the role of OmcS in conductivity of nanowires (Reguera et al., 2005) and electrodes (Reguera et al., 2006). A
was overlooked because DomcS biofilms were conductive and number of other bacteria have also been shown to require non-
produced high current densities (Malvankar et al., 2011). These filamentous type IV pilins for the secretion of extracellular pro-
biofilm results might be due to a reciprocal relationship (Park teins (Hager et al., 2006). One possibility is that PilA is a pseudo-
and Kim, 2011) between the expression of OmcS and that of pilin as a part of a type 2 secretion system (T2SS) (Nivaskumar
OmcZ, a cytochrome essential for current production (Nevin and Francetic, 2014) and that OmcS is exported by this T2SS.
et al., 2009). The DomcS biofilms may compensate for the loss Previous studies have shown that T2SS pseudopili can be
of OmcS by increasing the production of OmcZ. secreted outside the cell (Nivaskumar and Francetic, 2014;
Using immunogold localization, previous studies found that Vignon et al., 2003). A similar mechanism could explain the pres-
OmcS is associated with filaments (Leang et al., 2010, 2013; ence of non-filamentous PilA in our filament preparations.
Summers et al., 2010). As cytochromes were not known to This potential ability of PilA to regulate the secretion and as-
form filaments before our work, AFM images of filaments (Mal- sembly of the OmcS nanowires, as well as that of other multi-
vankar et al., 2012) as well as these antibody-labeling results heme cytochromes (Liu et al., 2018; Richter et al., 2012), could
(Leang et al., 2010, 2013; Summers et al., 2010) were interpreted explain how point mutations in pilA caused cells to produce fila-
as showing isolated OmcS monomers binding to the surfaces of ments with different conductivities than WT filaments. Filaments
the PilA filaments rather than showing antibodies directly binding produced by cells with point mutations in pilA showed very
to OmcS filaments. In light of the result presented here, a reinter- different morphology than WT filaments (Tan et al., 2016a). For
pretation of these previous studies suggests that the antibodies example, substitution of two residues in PilA with tryptophan
may have been directly binding to the subunits of the OmcS yielded mutant cells with filaments that surprisingly had half
filaments. the diameter of WT filaments (Tan et al., 2016a). Based on our
While no evidence of PilA was found in the structure of any of finding that conductive filaments are composed of cytochromes,
the purified filaments, non-filamentous PilA was present in our this very large structural change in the filaments of mutant strains
samples in lower abundance compared to OmcS (Figure S1A), suggests that the observed change in conductivity could be due
consistent with prior studies (Tan et al., 2016b). Multiple studies to filaments of different cytochromes or of different conforma-
have shown that PilA is required for secretion of OmcS to the tions of OmcS. Further structural studies on the filaments pro-
extracellular environment, as pilA deletion eliminated the pres- duced by these pilA mutants are needed to fully reconcile these
ence of OmcS in outer-surface preparations (Liu et al., 2018; studies with our finding of OmcS filaments functioning as
Richter et al., 2012). Overexpression of PilA is also accompanied nanowires.

366 Cell 177, 361–369, April 4, 2019

 In summary, our findings show that conductive G. sulfurredu- der Cooperative Agreement Number W911NF-18-2-0100 (with N.S.M.). This
cens filaments are polymerized chains of OmcS. The filament research was supported by NSF Graduate Research Fellowship awards
2017224445 (to J.P.O.) and DGE-1321846 (to N.L.I.) and the Air Force Office
structure has hemes closely stacked along the micrometer
of Scientific Research, grant FA9550-14-1-0350 (to A.I.H.). Research in the
length of the filament, establishing the molecular basis for Malvankar laboratory is also supported by the Charles H. Hood Foundation
electronic conductivity in these nanowires. Functional character- Child Health Research Award and the Hartwell Foundation Individual Biomed-
ization of conductivity in individual filaments shows that OmcS is ical Research Award. The cryo-EM imaging of filaments from fumarate-grown
required for the filament conductivity. The structure presented cells, conducted at the Molecular Electron Microscopy Core facility at the Uni-
here provides insights into supramolecular protein nanowires, versity of Virginia, was supported by the School of Medicine and built with NIH
grant G20-RR31199. The Titan Krios and Falcon II direct electron detector
explaining the remarkable capacity of soil bacteria to transport
were obtained with NIH S10-RR025067 and S10-OD018149, respectively.
electrons to extracellular electron acceptors for respiration (Mal-
The cryo-EM imaging of filaments used to solve the atomic structure was con-
vankar et al., 2011) and for sharing of energy and nutrients ducted at the Yale University’s Cryo-EM resource facility.
with syntrophic partners (Summers et al., 2010) that are
hundreds of micrometers away. The advances in understanding AUTHOR CONTRIBUTIONS
of the structural basis for conductivity in microbial nanowires
presented here can provide design principles for development F.W., under the supervision of E.H.E, performed the image analysis, recon-
of future bioelectronic interfaces between living cells and structed the OmcS filament structure, generated, refined the filament model,
devices. and imaged and biochemically analyzed filaments from fumarate-grown cells
that yielded similar helical parameters as filaments from electrode-grown cells
used to build the atomic model. Y.G. prepared and optimized cryo-EM grids,
STAR+METHODS collected data used to build the atomic model, and carried out AFM imaging on
wild-type filaments. J.P.O., V.S., and C.S. grew cells for studies on cell-
Detailed methods are provided in the online version of this paper attached filaments and purified filaments used for building the atomic model
and for functional studies. Y.G. and V.S. identified cell-attached and purified
and include the following:
filaments by negative stain electron microscopy. S.M.Y. carried out biochem-
ical analysis and, with D.V., performed AFM and conductivity measurements of
d KEY RESOURCES TABLE
filaments. S.E.Y. performed AFM imaging of cell-attached filaments of the
d CONTACT FOR REAGENT AND RESOURCE SHARING
DomcS strain, compared their height profiles with filaments of WT strain,
d EXPERIMENTAL MODEL AND SUBJECT DETAILS and interpreted morphological differences. D.V., Y.G., and C.S. prepared
B Bacterial strains and growth conditions nanoelectrodes. N.L.I., under the supervision of A.I.H. prepared filaments
d METHODS DETAILS from fumarate-grown cells. N.S.M. supervised the sample preparation, char-
B G. sulfurreducens filament preparation and biochem- acterization, and data collection used to build the atomic model; conceived
ical characterization and designed functional studies; and wrote the manuscript with input from
all authors.
B Filament preparation from fumarate-grown cells
(Figure S4A)
DECLARATION OF INTERESTS
B Cryo-EM sample preparation conditions
B Cryo-EM data collection conditions The authors declare no competing interests.
B Cryo-EM image analysis
B Model building of OmcS filaments Received: November 26, 2018
B Atomic force microscopy Revised: January 14, 2019
B Direct current conductivity measurements Accepted: March 11, 2019
Published: April 4, 2019
B Conductivity calculations
d QUANTIFICATION AND STATISTICAL ANALYSIS
REFERENCES
d DATA AND SOFTWARE AVAILABILITY
Adhikari, R., Malvankar, N., Tuominen, M., and Lovley, D. (2016). Conductivity
SUPPLEMENTAL INFORMATION of individual Geobacter pili. RSC Advances 6, 8354–8357.
Afkar, E., Reguera, G., Schiffer, M., and Lovley, D.R. (2005). A novel Geobac-
Supplemental Information can be found with this article online at https://doi. teraceae-specific outer membrane protein J (OmpJ) is essential for electron
org/10.1016/j.cell.2019.03.029. transport to Fe(III) and Mn(IV) oxides in Geobacter sulfurreducens. BMC Micro-
biol. 5, 41.
ACKNOWLEDGMENTS Ashkenazy, H., Abadi, S., Martz, E., Chay, O., Mayrose, I., Pupko, T., and Ben-
Tal, N. (2016). ConSurf 2016: An improved methodology to estimate and visu-
N.S.M. thanks Derek Lovley for providing strains; Eric Martz, Ruchi Jain, Daniel alize evolutionary conservation in macromolecules. Nucleic Acids Res. 44
Shapiro, Atanu Acharya, and Yong Xiong for helpful discussions; and Shenp- (W1), W344–W350.
ing Wu as well as Marc Llaguno for help with cryo-EM. We also thank Hao
Bertini, I., Cavallaro, G., and Rosato, A. (2006). Cytochrome c: Occurrence and
Jiang and Qiangfei Xia for providing nanoelectrodes and Olivera Francetic
functions. Chem. Rev. 106, 90–115.
for helpful discussions about the T2SS.This research was partially supported
by the Career Award at the Scientific Interfaces from Burroughs Wellcome Cheung, M., Kajimura, N., Makino, F., Ashihara, M., Miyata, T., Kato, T.,
Fund (to N.S.M.), the National Institutes of Health (NIH) Director’s New Inno- Namba, K., and Blocker, A.J. (2013). A method to achieve homogeneous
vator award (1DP2AI138259-01 to N.S.M.), NIH R35GM122510 (to E.H.E.), dispersion of large transmembrane complexes within the holes of carbon films
and the National Science Foundation (NSF) CAREER award no. 1749662 (to for electron cryomicroscopy. J. Struct. Biol. 182, 51–56.
N.S.M.). Research was sponsored by the Defense Advanced Research Pro- Childers, S.E., Ciufo, S., and Lovley, D.R. (2002). Geobacter metallireducens
jects Agency (DARPA) Army Research Office (ARO) and was accomplished un- accesses insoluble Fe(III) oxide by chemotaxis. Nature 416, 767–769.

Cell 177, 361–369, April 4, 2019 367

Clarke, T.A., Edwards, M.J., Gates, A.J., Hall, A., White, G.F., Bradley, J., Malvankar, N.S., Vargas, M., Nevin, K.P., Franks, A.E., Leang, C., Kim, B.C.,
Reardon, C.L., Shi, L., Beliaev, A.S., Marshall, M.J., et al. (2011). Structure Inoue, K., Mester, T., Covalla, S.F., Johnson, J.P., et al. (2011). Tunable
of a bacterial cell surface decaheme electron conduit. Proc. Natl. Acad. Sci. metallic-like conductivity in microbial nanowire networks. Nat. Nanotechnol.
USA 108, 9384–9389. 6, 573–579.
Coppi, M.V., Leang, C., Sandler, S.J., and Lovley, D.R. (2001). Development of Margoliash, E., and Lustgarten, J. (1962). Interconversion of horse heart cyto-
a genetic system for Geobacter sulfurreducens. Appl. Environ. Microbiol. 67, chrome c monomer and polymers. J. Biol. Chem. 237, 3397–3405.
3180–3187. Marsili, E., Baron, D.B., Shikhare, I.D., Coursolle, D., Gralnick, J.A., and Bond,
Egelman, E.H. (2000). A robust algorithm for the reconstruction of helical fila- D.R. (2008). Shewanella secretes flavins that mediate extracellular electron
ments using single-particle methods. Ultramicroscopy 85, 225–234. transfer. Proc. Natl. Acad. Sci. USA 105, 3968–3973.
Emsley, P., and Cowtan, K. (2004). Coot: Model-building tools for molecular Mehta, T., Coppi, M.V., Childers, S.E., and Lovley, D.R. (2005). Outer mem-
graphics. Acta Crystallogr. D Biol. Crystallogr. 60, 2126–2132. brane c-type cytochromes required for Fe (III) and Mn (IV) oxide reduction in
Hager, A.J., Bolton, D.L., Pelletier, M.R., Brittnacher, M.J., Gallagher, L.A., Geobacter sulfurreducens. Appl. Environ. Microbiol. 71, 8634–8641.
Kaul, R., Skerrett, S.J., Miller, S.I., and Guina, T. (2006). Type IV pili-mediated Mindell, J.A., and Grigorieff, N. (2003). Accurate determination of local defocus
secretion modulates Francisella virulence. Mol. Microbiol. 62, 227–237. and specimen tilt in electron microscopy. J. Struct. Biol. 142, 334–347.
Haldar, S., Sil, P., Thangamuniyandi, M., and Chattopadhyay, K. (2015). Neumann, P., Dickmanns, A., and Ficner, R. (2018). Validating resolution rev-
Conversion of amyloid fibrils of cytochrome c to mature nanorods through a olution. Structure 26, 785–795.e4.
honeycomb morphology. Langmuir 31, 4213–4223. Nevin, K.P., Kim, B.C., Glaven, R.H., Johnson, J.P., Woodard, T.L., Methé,
Hirota, S., Hattori, Y., Nagao, S., Taketa, M., Komori, H., Kamikubo, H., Wang, B.A., Didonato, R.J., Jr., Covalla, S.F., Franks, A.E., Liu, A., and Lovley, D.R.
Z., Takahashi, I., Negi, S., Sugiura, Y., et al. (2010). Cytochrome c polymeriza- (2009). Anode biofilm transcriptomics reveals outer surface components
tion by successive domain swapping at the C-terminal helix. Proc. Natl. Acad. essential for high density current production in Geobacter sulfurreducens
Sci. USA 107, 12854–12859. fuel cells. PLoS ONE 4, e5628.
Holmes, D.E., Chaudhuri, S.K., Nevin, K.P., Mehta, T., Methé, B.A., Liu, A., Nivaskumar, M., and Francetic, O. (2014). Type II secretion system: A magic
Ward, J.E., Woodard, T.L., Webster, J., and Lovley, D.R. (2006). Microarray beanstalk or a protein escalator. Biochim. Biophys. Acta - Mol Cell Res.
and genetic analysis of electron transfer to electrodes in Geobacter sulfurredu- 1843, 1568–1577.
cens. Environ. Microbiol. 8, 1805–1815. O’Brien, J.P., and Malvankar, N.S. (2017). A simple and low-cost procedure
Ing, N.L., Nusca, T.D., and Hochbaum, A.I. (2017). Geobacter sulfurreducens for growing Geobacter sulfurreducens cell cultures and biofilms in bio-
pili support ohmic electronic conduction in aqueous solution. Phys. Chem. electrochemical systems. Current Protocols in Microbiology 43, A.4K.1–A.4K.27.
Chem. Phys. 19, 21791–21799. Park, I., and Kim, B.-C. (2011). Homologous overexpression of omcZ, a gene
Inoue, K., Qian, X., Morgado, L., Kim, B.C., Mester, T., Izallalen, M., Salgueiro, for an outer surface c-type cytochrome of Geobacter sulfurreducens by single-
C.A., and Lovley, D.R. (2010). Purification and characterization of OmcZ, an step gene replacement. Biotechnol. Lett. 33, 2043–2048.
outer-surface, octaheme c-type cytochrome essential for optimal current Pettersen, E.F., Goddard, T.D., Huang, C.C., Couch, G.S., Greenblatt, D.M.,
production by Geobacter sulfurreducens. Appl. Environ. Microbiol. 76, Meng, E.C., and Ferrin, T.E. (2004). UCSF Chimera–a visualization system
3999–4007. for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612.
Janiak, C. (2000). A critical account on p-p stacking in metal complexes Qian, X., Mester, T., Morgado, L., Arakawa, T., Sharma, M.L., Inoue, K., Jo-
with aromatic nitrogen-containing ligands. J. Chem. Soc. Dalton Trans. 0, seph, C., Salgueiro, C.A., Maroney, M.J., and Lovley, D.R. (2011). Biochemical
3885–3896. characterization of purified OmcS, a c-type cytochrome required for insoluble
Jiang, X., Futera, Z., Ali, M.E., Gajdos, F., von Rudorff, G.F., Carof, A., Breuer, Fe(III) reduction in Geobacter sulfurreducens. Biochim. Biophys. Acta 1807,
M., and Blumberger, J. (2017). Cysteine linkages accelerate electron flow 404–412.
through tetra-heme protein STC. J. Am. Chem. Soc. 139, 17237–17240. Reardon, P.N., and Mueller, K.T. (2013). Structure of the type IVa major pilin
Klimes, A., Franks, A.E., Glaven, R.H., Tran, H., Barrett, C.L., Qiu, Y., Zengler, from the electrically conductive bacterial nanowires of Geobacter sulfurredu-
K., and Lovley, D.R. (2010). Production of pilus-like filaments in Geobacter sul- cens. J. Biol. Chem. 288, 29260–29266.
furreducens in the absence of the type IV pilin protein PilA. FEMS Microbiol. Reguera, G., McCarthy, K.D., Mehta, T., Nicoll, J.S., Tuominen, M.T., and Lov-
Lett. 310, 62–68. ley, D.R. (2005). Extracellular electron transfer via microbial nanowires. Nature
Leang, C., Malvankar, N.S., Franks, A.E., Nevin, K.P., and Lovley, D.R. (2013). 435, 1098–1101.
Engineering Geobacter sulfurreducens to produce a highly cohesive conduc- Reguera, G., Nevin, K.P., Nicoll, J.S., Covalla, S.F., Woodard, T.L., and Lovley,
tive matrix with enhanced capacity for current production. Energy Environ. Sci. D.R. (2006). Biofilm and nanowire production leads to increased current in
6, 1901–1908. Geobacter sulfurreducens fuel cells. Appl. Environ. Microbiol. 72, 7345–7348.
Leang, C., Qian, X., Mester, T., and Lovley, D.R. (2010). Alignment of the c-type Reguera, G., Pollina, R.B., Nicoll, J.S., and Lovley, D.R. (2007). Possible
cytochrome OmcS along pili of Geobacter sulfurreducens. Appl. Environ. nonconductive role of Geobacter sulfurreducens pilus nanowires in biofilm for-
Microbiol. 76, 4080–4084. mation. J. Bacteriol. 189, 2125–2127.
Li, X., Mooney, P., Zheng, S., Booth, C.R., Braunfeld, M.B., Gubbens, S., Richter, L.V., Sandler, S.J., and Weis, R.M. (2012). Two isoforms of Geobacter
Agard, D.A., and Cheng, Y. (2013). Electron counting and beam-induced mo- sulfurreducens PilA have distinct roles in pilus biogenesis, cytochrome locali-
tion correction enable near-atomic-resolution single-particle cryo-EM. Nat. zation, extracellular electron transfer, and biofilm formation. J. Bacteriol. 194,
Methods 10, 584–590. 2551–2563.
Liu, X., Zhuo, S., Rensing, C., and Zhou, S. (2018). Syntrophic growth with Scheres, S.H. (2012). RELION: implementation of a Bayesian approach to
direct interspecies electron transfer between pili-free Geobacter species. cryo-EM structure determination. J. Struct. Biol. 180, 519–530.
ISME J. 12, 2142–2151. Subramaniam, S., Earl, L.A., Falconieri, V., Milne, J.L., and Egelman, E.H.
Malvankar, N.S., and Lovley, D.R. (2014). Microbial nanowires for bioenergy (2016). Resolution advances in cryo-EM enable application to drug discovery.
applications. Curr. Opin. Biotechnol. 27, 88–95. Curr. Opin. Struct. Biol. 41, 194–202.
Malvankar, N.S., Tuominen, M.T., and Lovley, D.R. (2012). Lack of cytochrome Summers, Z.M., Fogarty, H.E., Leang, C., Franks, A.E., Malvankar, N.S., and
involvement in long-range electron transport through conductive biofilms and Lovley, D.R. (2010). Direct exchange of electrons within aggregates of an
nanowires of Geobacter sulfurreducens. Energy Environ. Sci. 5, 8651–8659. evolved syntrophic coculture of anaerobic bacteria. Science 330, 1413–1415.

368 Cell 177, 361–369, April 4, 2019

Tan, Y., Adhikari, R.Y., Malvankar, N.S., Pi, S., Ward, J.E., Woodard, T.L., acids required for pili conductivity and long-range extracellular electron trans-
Nevin, K.P., Xia, Q., Tuominen, M.T., and Lovley, D.R. (2016a). Synthetic bio- port in Geobacter sulfurreducens. mBio 4, e00105–e00113.
logical protein nanowires with high conductivity. Small 12, 4481–4485. Vignon, G., Köhler, R., Larquet, E., Giroux, S., Prévost, M.-C., Roux, P., and
Tan, Y., Adhikari, R.Y., Malvankar, N.S., Ward, J.E., Nevin, K.P., Woodard, Pugsley, A.P. (2003). Type IV-like pili formed by the type II secreton: specificity,
T.L., Smith, J.A., Snoeyenbos-West, O.L., Franks, A.E., Tuominen, M.T., and composition, bundling, polar localization, and surface presentation of pep-
Lovley, D.R. (2016b). The low conductivity of Geobacter uraniireducens pili tides. J. Bacteriol. 185, 3416–3428.
suggests a diversity of extracellular electron transfer mechanisms in the genus Wang, F., Coureuil, M., Osinski, T., Orlova, A., Altindal, T., Gesbert, G., Nassif,
Geobacter. Front. Microbiol. 7, 980. X., Egelman, E.H., and Craig, L. (2017). Cryoelectron microscopy reconstruc-
Tan, Y., Adhikari, R.Y., Malvankar, N.S., Ward, J.E., Woodard, T.L., Nevin, tions of the Pseudomonas aeruginosa and Neisseria gonorrhoeae type IV pili at
K.P., and Lovley, D.R. (2017). Expressing the Geobacter metallireducens sub-nanometer resolution. Structure 25, 1423–1435.e4.
PilA in Geobacter sulfurreducens yields pili with exceptional conductivity. Wang, R.Y.-R., Kudryashev, M., Li, X., Egelman, E.H., Basler, M., Cheng, Y.,
mBio 8, e02203–16. Baker, D., and DiMaio, F. (2015). De novo protein structure determination
Tang, G., Peng, L., Baldwin, P.R., Mann, D.S., Jiang, W., Rees, I., and Ludtke, from near-atomic-resolution cryo-EM maps. Nat. Methods 12, 335–338.
S.J. (2007). EMAN2: an extensible image processing suite for electron micro- Williams, C.J., Headd, J.J., Moriarty, N.W., Prisant, M.G., Videau, L.L., Deis,
scopy. J. Struct. Biol. 157, 38–46. L.N., Verma, V., Keedy, D.A., Hintze, B.J., Chen, V.B., et al. (2018). MolProbity:
Vargas, M., Malvankar, N.S., Tremblay, P.L., Leang, C., Smith, J.A., Patel, P., More and better reference data for improved all-atom structure validation.
Snoeyenbos-West, O., Nevin, K.P., and Lovley, D.R. (2013). Aromatic amino Protein Sci. 27, 293–315.

Cell 177, 361–369, April 4, 2019 369

STAR+METHODS

KEY RESOURCES TABLE

REAGENT or RESOURCE SOURCE IDENTIFIER

Antibodies
Rabbit monospecific antibody anti-PilA This paper N/A
Bacterial and Virus Strains
Geobacter sulfurreducens wild-type strain PCA Coppi et al., 2001 ATCC 51573
(designated DL1)
Geobacter sulfurreducens wild-type strain PCA Ing et al., 2017 DSMZ 12127
Geobacter sulfurreducens omcS deletion mutant strain Mehta et al., 2005 N/A
(designated DomcS)
Biological Samples
Purified OmcS filaments from electrode-grown cells This paper N/A
Purified OmcS filaments from fumarate-grown cells This paper N/A
Chemicals, Peptides, and Recombinant Proteins
Common lab reagents N/A N/A
NBAF growth Medium O’Brien and Malvankar, N/A
2017
Trypsin Sigma-Aldrich Cat #: T7409
Ethanolamine Sigma-Aldrich Cat #: E9508
Sodium-dodecyl sulfate Ambion, Inc Cat #: AM9822
Deposited Data
OmcS filament Cryo-EM map This paper EMDB: EMD-9046
OmcS filament structure This paper PDB: 6EF8
Software and Algorithms
Spider package Egelman, 2000 https://spider.wadsworth.org/spider_doc/spider/
docs/spider.html
CTFFIND3 Mindell and Grigorieff, http://grigoriefflab.janelia.org/ctf
2003
EMAN2 Tang et al., 2007 http://blake.bcm.tmc.edu/emanwiki/EMAN2
IHRSR Egelman, 2000 N/A
COOT Emsley and Cowtan, http://www2.mrc-lmb.cam.ac.uk/personal/
2004 pemsley/coot/
CHIMERA Pettersen et al., 2004 http://www.rbvi.ucsf.edu/chimera/
MolProbity Williams et al., 2018 http://molprobity.biochem.duke.edu/
Relion Scheres, 2012 https://www2.mrc-lmb.cam.ac.uk/relion
Gwyddion Open Source http://gwyddion.net/
Igor Pro Wavemetrics https://www.wavemetrics.com/
RosettaCM Wang et al., 2015 https://www.rosettacommons.org/docs/latest/
application_documentation/structure_prediction/
RosettaCM
Other
Cypher ES Atomic Force Microscope Oxford Instrument N/A
Probe station with triaxially shielded dark box MPI Corp. N/A
semiconductor parameter analyzer Keithley 4200A-SCS

e1 Cell 177, 361–369.e1–e5, April 4, 2019

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Nikhil S.
Malvankar ([email protected]).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Bacterial strains and growth conditions

The Geobacter sulfurreducens wild-type (WT) strain PCA (designated DL-1) (ATCC 51573, DSMZ 12127) (Coppi et al., 2001) and the
omcS deletion mutant strain (Mehta et al., 2005) (designated DomcS) were obtained from our laboratory culture collection. The cul-
tures were maintained at 30 C or at 25 C under strictly anaerobic conditions in growth medium supplemented with acetate (10 mM)
as the electron donor and fumarate (40 mM) as the electron acceptor in sterilized and degassed NBAF medium (O’Brien and Mal-
vankar, 2017). 1L NBAF medium contained the following: 0.04 g/L calcium chloride dihydrate, 0.1 g /L, magnesium sulfate heptahy-
drate, 1.8 g/L sodium bicarbonate, 0.5 g/L sodium carbonate, 0.42 g/L potassium phosphate monobasic, 0.22 g/L potassium
phosphate dibasic, 0.2 g/L ammonium chloride, 0.38 g/L potassium chloride, 0.36 g/L sodium chloride, vitamins and minerals as
listed in (O’Brien and Malvankar, 2017). Resazurin was omitted and 1 mM cysteine was added as an electron scavenger. All chem-
icals obtained from Fisher Scientific unless otherwise noted. For filament samples used to build the atomic model, the wild-type strain
was grown on electrodes under electron acceptor-limiting conditions that induce filament expression (O’Brien and Malvankar, 2017)
(Figure 1A).

METHODS DETAILS

G. sulfurreducens filament preparation and biochemical characterization

Filaments were separated from bacteria and extracted via centrifugation (Tan et al., 2016a) and maintained in 150 mM
ethanolamine buffer at pH 10.5 in a manner similar to structural studies on bacterial filaments (Wang et al., 2017). Cells were gently
scraped from the surface using a plastic spatula and isotonic wash buffer (20.02 3 103 M morpholinepropanesulfonic acid, 4.35 3
103 M NaH2 PO4 $H2O, 1.34 3 103 M KCl, 85.56 3 103 M NaCl, 1.22 3 103 M MgSO4 $7H2O, and 0.07 3 103 M CaCl2 $2H2O),
then collected by centrifugation and re-suspended in 150 3 103 M ethanolamine (pH 10.5). Filaments were mechanically sheared
from the cell surface using a Waring Commercial Blender (Cat. No. 7011S) at low speed for 1 min, and then cells were removed by
centrifugation at 13,000 g before collecting filaments with an overnight 10% ammonium sulfate precipitation and subsequent centri-
fugation at 13,000 g. Collected filaments were re-suspended in ethanolamine buffer then cleaned by centrifugation at 23,000 g to
remove debris and a second 10% ammonium sulfate precipitation with centrifugation at 13,000 g (Tan et al., 2016a). The final filament
preparation was re-suspended in 200 ml ethanolamine buffer. Filament preparations were further passed through 0.2 mm filters to
remove any residual cells and stored at 4 C. Cell-free filament preparations were imaged first with transmission electron microscopy
to ensure sample quality (Figure 1A). Dilute 5 ml solutions containing filaments were placed on gold electrodes to achieve individual
filaments across two gold electrodes (Figure 4F, Inset). Prior to all measurements, filaments were imaged with AFM and height
measurements were performed to confirm the presence of individual filaments. For all conductivity measurements, samples were
maintained under hydrated buffer environments (150 mM ethanolamine) and the pH of the buffer was equilibrated to pH 7 using
HCl (Malvankar et al., 2011). Transmission electron microscopy imaging was used to confirm that filaments maintain their structure
and remain morphologically similar at all pH conditions.
For polyacrylamide gel electrophoretic separation (SDS-PAGE) (Figure S1A), all filament samples were prepared in ethanolamine
buffer and dried in a speedvac. Samples were then resuspended in 500 ml ultrapure deionized water and centrifuged at 18,000 x g at
4 C for 1 hour. The pellets were dried again in the speedvac, resuspended in water and dried repetitively 2 times. The final dried
samples were resuspended in 5-8 ml 2.5% sodium-dodecyl sulfate (SDS) to disassemble filaments into their constituent monomers.
The samples were incubated at room temperature for at least 4 hours, then diluted 3-fold with deionized water. The denatured sam-
ples were boiled in 1X SDS sample buffer that included b-mercaptoethanol for 12 min. The samples were run on a 16% Tricine protein
gel (ThermoFisher Scientific, Carlsbad, CA) initially at constant voltage of 30 V for 18 min before changing to 190 V for 12 min. Pre-
cision Plus Protein Prestained molecular weight standards (BioRad, Hercules, CA) and Low Range Protein Ladder (Thermo Scientific)
were used to compare the molecular weight of PilA and cytochromes in the filament preparations. Gels were immediately washed at
least 3 times with ultra-pure deionized water over a 1-hour period, stained with Coomassie R-250 stain (Thermo Scientific, Rockford,
IL), and destained overnight.
Custom monospecific anti-PilA antibody was synthesized by Pacific Immunology (Ramona, CA) by immunizing two rabbits with
synthetic peptide sequence containing targeted epitope on the native protein, PilA, and then affinity purifying the serum against
that peptide sequence. Specificity of antibody in the serum was confirmed by ELISA after 1st stage of immune response and then
verified again with purified antibody (1:125,000 titer) at the final stage. Synthetic peptide sequence contained following 21 amino
acids from C terminus of PilA, Cys-RNLKTALESAFADDQTYPPES, to which a cysteine was added to N terminus and then was con-
jugated with Keyhole Limpet Hemocyanin prior to immunization. The molecular weight of the synthetic peptide was verified by HPLC
and ESI-MS.

Cell 177, 361–369.e1–e5, April 4, 2019 e2

 For LC-MS/MS analysis of filaments (Figure S1B), the PilA band (
6.5 kDa) and OmcS bands (
50 kDa and 30 kDa) were extracted
from the protein gel and treated with trypsin to digest the protein (Figure S1). Proteomic analysis of the cleaved peptides from fila-
ments of electrode-grown cells) was performed by the Proteomics Mass Spectrometry Facilities at Yale University Results gave
unique amino acid sequence matches with the OmcS and c-terminal domain of PilA (Figure S1) whereas the hydrophobicity of
the N-terminal domain may have interfered with the sequence detection procedure (Ing et al., 2017).

Filament preparation from fumarate-grown cells (Figure S4A)

G. sulfurreducens cells grown in fumarate were used for filament purification (Ing et al., 2017). Briefly, G. sulfurreducens (DSMZ
strain 12127) was grown anaerobically in 1 L cultures of sterilized and degassed NBAF medium (0.04 g/L calcium chloride dihydrate,
0.1 g/L magnesium sulfate heptahydrate, 1.8 g/L sodium bicarbonate, 0.5 g/L sodium carbonate, 0.42 g/L potassium phosphate
monobasic, 0.22 g/L potassium phosphate dibasic, 0.2 g/L ammonium chloride, 0.38 g/L potassium chloride, 0.36 g/L sodium chlo-
ride, vitamins and minerals, using 20 mM acetate as the electron donor and 40 mM fumarate as the electron acceptor). Resazurin was
omitted and 1mM cysteine was added as an electron scavenger. All chemicals were obtained from Fisher Scientific unless other-
wise noted.
Upon reaching stationary phase, cells were pelleted by centrifugation at 5000 x g for 20 min at 4 C. Cell pellets were removed and
extracellular filaments were purified from the supernatant through dropwise additions of 1.1 M aqueous MgCl2, which was added to a
final ratio of 1:10 for aqueous MgCl2:supernatant. Supernatants were stored at 4 C overnight and centrifuged at 50,000 x g for 1 h at
4 C to precipitate extracellular filaments. Pellets of extracellular filaments were resuspended with ultrapure water adjusted to pH 4.3
with HCl and passed through 0.2 mm filters. Additional contaminants were removed from the resuspended pellets using 50 kDa dial-
ysis tubing. After two 1 L exchanges of pH 4.3 water, the dialyzed sample was removed from dialysis tubing and stored at 4 C.
For analysis of filaments from fumarate-grown cells (Figure S4A), the purified filament samples were analyzed for protein content by
12.5% PhastGel (GE Healthcare) and stained with Coomassie. There were two prominent protein bands around 47 kDa, and these
were cut from the gel, digested by trypsin overnight, and then analyzed by mass spec at the University of Virginia core facility. The LC-
MS system consisted of a Thermo Electron Q Exactive HFX mass spectrometer system with an Easy Spray ion source connected to a
Thermo 75 mm x 15 cm C18 Easy Spray column. 5 mL of the extract was injected and the peptides eluted from the column by an
acetonitrile/0.1 M formic acid gradient at a flow rate of 0.3 mL/min over 2.0 hours. The nanospray ion source was operated at
1.9 kV. The digest was analyzed using the rapid switching capability of the instrument acquiring a full scan mass spectrum to deter-
mine peptide molecular weights followed by product ion spectra (10 HCD) to determine amino acid sequence in sequential scans.
This mode of analysis produces approximately 60,000 MS/MS spectra of ions ranging in abundance over several orders of
magnitude.
Prior to use, purified filaments were characterized by MALDI-TOF. For MALDI-TOF analysis, filaments were denatured with
10% OG overnight and then diluted with ultrapure water to a final OG concentration of 2% prior to the addition of matrix solution
(a-Cyano-4-hydroxycinnamic acid dissolved in a 2:1 solution of ultrapure water: acetonitrile and 0.2% trifluoroacetic acid). The final
sample for MALDI was a 12:7:5 ratio of matrix solution to ultrapure water to diluted and denatured protein. Mass spectra were
collected in positive ion mode.

Cryo-EM sample preparation conditions

For samples used to build the atomic model, holey carbon coated Quantifoil grids (R 2mm/2mm) were used. Prior to use, the grids were
floated in 0.05% Triton 100X solution (Cheung et al., 2013). Cryo-EM specimens were prepared with a FEI Vitrobot Mark IV at 22 C
with 100% humidity. 3 mL of sample solution containing WT filaments were dropped on the grids and spread gently by pipette tip
before loading to the Vitrobot, blotted for 5.5 s and plunge-frozen in liquid ethane.

Cryo-EM data collection conditions

Micrographs were acquired by FEI Titan Krios electron microscopy performed at 300kV equipped with a Gatan K2 summit camera.
Quantum energy filter with a slit width at 20 eV was applied to remove inelastically scattered electrons. Movies were collected using
super-resolution imaging mode with a physical pixel size of 1.05 Å and an exposure rate of 6.8 electrons per pixel per second. A total
exposure time of 9.75 s was fractioned into 30 frames and 1447 movies were generated using serial EM auto-collection. The data
collection and processing parameters are summarized below:

Magnification 130,000 X
Voltage (kV) 300
Electron exposure (e-/Å2) 63
Energy filter slit width 20 eV
Physical Pixel Size (Å) 1.05
Defocus range (mm) 0.5 – 3.0

e3 Cell 177, 361–369.e1–e5, April 4, 2019

 For analysis of filaments from fumarate-grown cells (Figure S4A), sample (4 mL,
1 mg/ml) was applied to discharged lacey carbon
grids and plunge-frozen using a Vitrobot Mark IV (FEI, Inc.). Grids were imaged in a Titan Krios at 300 keV and recorded with a Falcon
III direct electron detector at 1.4 Å per pixel. Micrographs were collected using a defocus range of 1.25–2.25 mm, with a total
exposure of 2 s (amounting to
60 electrons/Å2) distributed into 24 fractions. All the micrographs were first motion corrected
(ignoring the first fraction) using MotionCorr (Li et al., 2013) version 2 and then used for CTF estimation by the CTFFIND3 program
(Mindell and Grigorieff, 2003).

Cryo-EM image analysis

For filaments of fumarate-grown cells (Figure S4A), images were extracted using the e2helixboxer program within EMAN2 (Tang
et al., 2007) from the dose-weighted fractions 2-10 (amounting to
20 electrons/Å2), after the images were corrected for the CTF
through multiplication by the theoretical CTF (a Wiener filter in the limit of a very poor SNR). A total of 17,800 overlapping 384-px
long segments (with a shift of 55 pixels,
1.5 times the axial rise per subunit) were generated. The determination of the helical sym-
metry was unambiguous given the 1/(47 Å) meridional layer-line and the 1-start helix of
200 Å pitch. A reconstruction was generated
using the IHRSR method implemented in Spider (Egelman, 2000), and this reconstruction was subsequently filtered to 20 Å for the
starting reference in the reconstruction of the second dataset. To build the atomic model of filaments from electrode-grown cells, a
total of 573 images were selected, motion-corrected, dose weighted (amounting to
20 electrons/Å2) and CTF-corrected in the same
manner as for filaments of electrode-grown cells. About 2,000 long filaments were extracted using e2helixboxer and then 384-pixel
long 28,293 overlapping segments with a shift of 60 pixels were generated. The final volume from the IHRSR reconstruction was esti-
mated to have a resolution of
3.7 Å based on the model:map FSC (Figure S3). It was filtered to 3.6 Å and sharpened with a negative
B-factor of 100. Micrographs with all fractions and boxing coordinates were also imported into Relion for 3D reconstruction. A similar
final reconstruction to the one from SPIDER was generated in Relion using the same helical symmetry, starting with the SPIDER vol-
ume filtered to 10 Å. Interestingly this reconstruction was estimated to have a resolution of 3.2 Å resolution based on a ‘‘gold stan-
dard’’ map:map FSC.

Model building of OmcS filaments

First, about 400 protein Ca atoms and six heme cofactors corresponding to a single subunit within the filaments were built manually
(Figure 2B) in Coot (Emsley and Cowtan, 2004). Then the density corresponding to a single subunit was segmented from the exper-
imental filament density using Chimera (Pettersen et al., 2004). Using the proteins identified in the MS/MS results and real space
fitting to the cryo-EM density, the sequence of the filament was unambiguously determined to be cytochrome OmcS (Figures 2
and S2). The full length OmcS protein as well as six heme molecules were then built manually in Coot. Then the OmcS/heme was
rebuilt with the RosettaCM protocol (Wang et al., 2015) to remove bad geometries. A total of 3,000 full-length models were generated
and the top 20 models were selected based on Rosetta’s energy function, Ramachandran plots and overall fit to the map. These 20
models were then combined into one model by manual editing in Coot using the criteria of the local fit to the density map and the
geometry statistics of the model. To better refine heme interacting areas at this resolution, bond/angle restraints for the heme mole-
cule itself, His-Fe, and Cys-heme thioester bonds were created based on high resolution cytochrome c3 crystal structures NrfB (PDB
2P0B) and NrfHA (PDB 2J7A). Then a filament model was generated and further refined using Phenix with additional heme area re-
straints, and MolProbity (Williams et al., 2018) was used to evaluate the quality of the filament model. The refinement statistics are
given in Table 1.

Atomic force microscopy

To visualize individual filaments (Figure 4), 5 ml of buffer solution containing filaments were deposited on mica or on a silicon wafer
insulated by a 100 nm silicon dioxide dielectric layer with gold electrodes patterned by e-beam or nanoimprint lithography (Tan et al.,
2016a). For nanoimprinting, the substrate was cleaned with a Piranha solution (H2SO4:H2O2 = 3:1) and a diluted Hydrofluoric acid
solution before patterning. Two layers of resists (50-nm-thick poly methyl methacrylate and 60-nm-thick UV-curable resist) were
then sequentially spin-coated onto the cleaned substrate. Circuit patterns including nanoelectrodes separated by nano-sized
gaps, microscale fan outs, and contact pads were transferred from a quartz mold to the UV resist with nanoimprint lithography in
a homemade imprint chamber. The residual UV-resist layer and the poly methyl methacrylate underlayer were removed by reactive
ion etching with fluorine based (CHF3/O2) and oxygen-based gases respectively. The excess buffer was absorbed with filter paper.
The sample was air-dried and was mounted on a metal puck (Oxford Instrument, Cypher ES). Atomic force microscopy (AFM) ex-
periments were performed using soft cantilevers (ASYELEC-01, Oxford Instrument Co.) with a nominal force constant of 2 N/m
and resonance frequencies of 70 kHz. The free-air amplitude of the tip was calibrated with the Asylum Research software and the
spring constant was captured by the thermal vibration method. The sample was imaged with a Cypher ES scanner using intermittent
tapping (AC-air topography) mode. AFM showed that gold electrodes were bridged with individual filaments to facilitate conductivity
measurements (Figure 4F, inset). To visualize helical features of filaments, AFM was operated in attractive force imaging mode and
both phase and height channel images were reported to visualize axial periodicity (Figure 4A). All lateral and axial height analyses for
filaments (Figure 4) were performed using Gwyddion and IGOR Pro software (WaveMetrics Inc).

Cell 177, 361–369.e1–e5, April 4, 2019 e4

Direct current conductivity measurements
All direct current (DC) conductivity measurements of filaments were performed under fully hydrated buffer conditions in a 2-electrode
configuration inside a triaxially shielded dark box using a probe station (MPI Corp.) connected to a semiconductor parameter analyzer
(Keithley 4200A-SCS) equipped with preamplifiers, allowing 0.1 fA current resolution and 0.5 mV voltage resolution. A DC voltage,
typically in the range of 0.5V to +0.5V, was applied between the two electrodes and the current was measured over a minimum
period of 120 s until the steady state was reached (Figure 4F). The linearity of the I-V characteristics was maintained by applying
an appropriate low voltage and the slope of the I-V curve was used to determine the conductance (G). Measurements were performed
at low voltages (< 0.5 V) and over longer times (> 100 s) to ensure a lack of electrochemical leakage currents or faradic currents as
evidenced by the absence of significant DC conductivity in buffer or filaments of DomcS strain that were maintained under identical
buffer conditions as filaments of the WT strain (Figures 4F and 4G). All analysis was performed using IGOR Pro (WaveMetrics Inc.).

Conductivity calculations
The conductivity (s) of filaments was calculated using the relation (Malvankar et al., 2011) s = G∙(L/A) where G is the conductance,
L is the length of the filament, and A = pr 2 is the area of cross section of filament with 2r as the height of the filament measured by AFM
(Figures 4F and 4G).

QUANTIFICATION AND STATISTICAL ANALYSIS

Quantification and statistical analyses employed in this publication pertain to the analysis on atomic force microscopy and electron
microscopy data and the determination of structures by electron microscopy, which are integral parts of existing algorithms and soft-
ware used.

DATA AND SOFTWARE AVAILABILITY

All data for OmcS filament were deposited in EMDB and PDB with the following entry codes: EMDB: EMD-9046 and PDB: 6EF8.

e5 Cell 177, 361–369.e1–e5, April 4, 2019

Supplemental Figures

Figure S1. Biochemical Characterizations of Filament Preparations Show PilA and OmcS Proteins, Related to Figures 1 and 2
(A) SDS-PAGE gel of filaments showing OmcS and PilA at expected molecular weights of
45 kDa and
6.5 kDa. LC-MS/MS analysis of metalloproteins in
purified G. sulfurreducens filaments from (B) the
45 kDa band and (C) the
30 kDa band. EMPAI: Exponentially modified protein abundance index. LC-MS/MS
analysis showing matched peptides in (D) OmcS in
45 kDa band and (E) PilA in the
6.5 kDa band. (F) Western immunoblotting using anti-PilA antibody showing
PilA in purified filament preparations (left) and cell lysate (right).
Figure S2. De Novo Atomic Model Building of OmcS Filaments, Related to Figures 1 and 2
(A) Sequence-based alignment of OmcS and two other c-type cytochromes detected by mass spectrometry, with the same highlighted colors used in Fig 2A.
Regions indicated by red, orange, yellow, green, cyan, blue and purple arrowheads are shown in (B)-(G), respectively, with OmcS atomic model fit into the
cryo-EM map. Both OmcT and GSU2501 have deletions or insertions at those regions and cannot fit into the cryo-EM map.
Figure S3. Illustration of Cryo-EM Map Quality, Related to Figures 1 and 2
(A) The model:map FSC (Fourier Shell Correlation) calculation using a 0.38 criterion, which is sqrt (0.143), estimates the Spider map to have a resolution of 3.7 Å.
(B) A view of amino acids 325-350 fit into the Spider map used in (A) after filtering the map to 3.6 Å. (C) The Relion map:map ‘‘gold standard’’ FSC using the 0.143
criterion estimates the Relion map to have a resolution of 3.2 Å. (D) A view of the same region as (B) but with the Relion map which has been filtered to 3.2 Å, shows
that the Relion estimate of resolution is considerably over-optimistic. (E) Fitting of six heme co-factors into the corresponding cryo-EM densities. (F) The Histidine
51 interaction with heme 2 shown along with surrounding residues fit into the cryo-EM density.
Figure S4. Power Spectra of Purified Filaments from Fumarate-Grown Strain and Cell-Attached Filaments Are Similar to Power Spectra from
Filaments of the Electrode-Grown Strain Used to Build the Atomic Model, Related to Figures 1 and 2
(A) OmcS filaments (black arrow) from the fumarate-grown strain. Some flagellar filaments (white arrow) are present in this strain. Scale bar, 500 Å. (B) The
averaged power spectrum of the OmcS filaments from fumarate-grown cells (left) shows similar layer lines and helical symmetry as that from filaments of
electrode-grown cells (right). (C) Power spectra generated from previously published images (Leang et al., 2013) of intact filaments attached to G. sulfurreducens
cells. Left, the power spectrum of a bundle of filaments showing the meridional layer line (blue arrow) at 1/(47 Å). Because of the bundling, a strong sampling is
seen on the near-equatorial layer line. Right, averaged power spectra from 13 isolated segments to avoid the sampling. These power spectra show that the cell-
attached filaments have symmetry similar to that as observed in cryo-EM images of purified OmcS filaments, with a rotation of 84 degrees and a rise of 47 Å per
subunit.
Figure S5. The Thinner Filament Observed in Cryo-EM Imaging, Related to Figure 1
(A) The cryo-EM images frequently showed a thinner filament (white arrow) compared to OmcS filaments (black arrow). Scale bar, 200 Å. (B) The averaged power
spectrum of the thinner filaments shows layer lines similar to those from the OmcS filament, but with a somewhat different axial rise of
57 Å and rotation of

160 . (C) A helical reconstruction of the thinner filament filtered to 20 Å. (D) The projection of the reconstruction in (C) with a low threshold (left) and a high
threshold (right). The high threshold shows strong peaks separated by approximately 10 Å, labeled by yellow dots, which could be hemes.