JOURNAL OF EXPERIMENTAL ZOOLOGY 301A:701–706 (2004)

Townes and Holtfreter (1955):

Directed Movements and Selective Adhesion of
Embryonic Amphibian Cells
MALCOLM S. STEINBERG1 and SCOTT F. GILBERTn2
1
Department of Molecular Biology, Princeton University, Princeton, NJ 08544
2
Department of Biology, Swarthmore College, Swarthmore, PA 19081

Townes and Holtfreter’s 1955 JEZ classic, towards the wound), Larry Ruben’s doctoral thesis
‘‘Directed Movements and Selective Adhesion of on the implantation of tumors into regenerating
Embryonic Amphibian Cells,’’ was a watershed in limb fields, and a study of dye infusion into
the history of our understanding of the self- regeneration blastemas.
assembly processes in embryonic morphogenesis. Johannes (Hans) Holtfreter was by then a
Like the Spemann and Mangold paper quoted in famous embryologist who was best known for the
this study, the Townes and Holtfreter experiments work he had done in Otto Mangold’s laboratory.
did not so much answer questions as change the While there, he had demonstrated the chemical
questions that needed to be answered. In philoso- nature of the amphibian organizer and had shown
phical terms, it changed the explanandum (‘‘that the critical importance of the mesoderm for
which has to be explained’’) for morphogenesis. inducing the nervous system (see Hamburger,
The revolutionary nature of the JEZ paper can ’88; Holtfreter, ’91). But in addition to showing
be appreciated by looking at the contexts in which induction, these studies had also demonstrated the
it was published. The larger context of experi- need to explain the self-organization of tissues,
mental embryology is represented by the state-of- wherein there was ‘‘the transformation of a
the-art volume, Analysis of Development, which spherical egg into a body of about equal size in
was published the same year by Benjamin Willier, which groups of cells have shifted into specific
Paul Weiss, and Viktor Hamburger. Here, in the arrangements which foreshadow the tissue pat-
chapter on amphibians, one can read Holtfreter tern of the adult organism.’’ (Townes and Holtfr-
and Hamburger’s ironic summation of self-assem- eter, ’55). Holtfreter had come to the conclusion
bly research (p. 281): that although we remained that this transformation was due to ‘‘tissue
ignorant of the principles involved, some research affinity’’ (Gewebeaffinität), about which he had
"is encouraging enough to support the belief that previously written (translated), ‘‘Further experi-
the crucial and universal problem of self-organiza- ments have indicated that a whole system of y
tion is not entirely refractory to further analysis." attraction and repulsion phenomena is operating
One can also see the immediate context of between various cell types during development
experimental morphogenesis by looking at the and that information on this system will yield
other articles in the JEZ for that issue. By 1955, valuable information concerning the shiftings and
the Journal of Experimental Zoology, despite its segregations of tissues during organogenesis.’’
more inclusive name, had become a journal for the Moreover, he wrote, ‘‘It seems advantageous to
experimental analysis of morphogenesis. One us to introduce a more fitting term for the forces
paper in this issue (Nelson Spratt’s fate mapping that are instrumental in these processes of
of the epiblast to show the prospective notochord attraction and repulsion. Henceforth we shall
and somite cells) was strict developmental anat- apply the term affinity, ywhich may serve as a
omy. There were two papers that looked at reminder for the existence of analogous phenom-
enzyme changes during development. And there ena in chemistry.’’ (Holtfreter, ’39).
were three papers that dealt with morphological n
Correspondence to: Scott Gilbert, Department of Biology, Swarth-
observations on regeneration and wound healing: more College, Swarthmore, PA 19081.
Jay Lash’s thesis on wound healing in amphibians E-mail: [email protected]
Received 28 June 2004; Accepted 28 June 2004
(where he showed that the epidermal cells dis- Published online in Wiley InterScience (www.interscience.wiley.
sociated after wound formation and migrated com). DOI: 10.1002/jez.a.114

r 2004 WILEY-LISS, INC.

 1552499x, 2004, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jez.a.114, Wiley Online Library on [31/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
702 M.S. STEINBERG AND S.F. GILBERT

In that 1939 paper, Holtfreter had shown that dissociated cells were described in detail. During
bits of amphibian embryonic tissues would round their rearrangements, some of these tissues were
up in vitro, adhere, and spread over one another’s undergoing differentiation, including epithelio-
surfaces in specific configurations. When tissues mesenchymal transformations. They therefore
that normally formed a particular structure were presented moving targets for morphogenetic ana-
combined, they could rearrange to form an lysis, making it difficult to tease out the basic
anatomically correct semblance of that structure. principles involved. Many hypotheses competed
Holtfreter noted that some tissue pairs seem to here. Because cells and tissues of different kinds
associate preferentially, such as mesoderm with cross-adhere but nevertheless then segregate into
either ecto- or endoderm, but that others, such as tissue-specific domains, it was postulated that cells
ectoderm with endoderm, seem to avoid associa- first exhibit a non-specific mutual adhesiveness
tion. that is either replaced by or coexists with cell-type-
In 1943–1944, Holtfreter found that this specific specific adhesiveness. Because cells and tissues
affinity even applied to individual cells. He moved either inwards or outwards within com-
reported that ‘‘By exposing amphibian explants pound aggregates, it was postulated that they
to a pH of around 10.0, the tissues fall apart and are chemotactically attracted or repelled along
form a suspension of free cells. This cell heap can ‘‘an inside-outside gradient of some kind existing
be further disorganized by stirring it with a glass within the aggregate.’’ Noting a resemblance
needle. When returned to normal pH conditions between the spreading of one tissue over another
the cells reaggregate into firm bodies which and the similar behavior of oil and water, they
continue differentiating. Instead of retaining their suggested that ‘‘invagination may be due to a kind
chaotic cell pattern, the aggregates become orga- of cytotactic reaction of the proximal cell surfaces
nized into well separated tissues and organs of a to a gradient of interfacial tension between inside
topographic pattern hardly less perfect than the and outside of the embryo.’’
one developed in a corresponding untreated Gathering together the main generalizations,
explant’’ (Holtfreter, ’44). Between 1941 and encapsulated in the paper’s title, it was concluded
1947, Paul Weiss and Albert Tyler separately (p. 77) that ‘‘in the process of sorting out, the
offered the first molecular hypothesis to explain different cell types exhibit a cell-specific tendency
‘‘selective cell adhesion’’ in terms of the interlock- to arrange themselves in a definite tissue pattern
ing of sterically complementary cell surface which corresponds to that in normal develop-
macromolecules, in the manner of antigen with ment.’’ Moreover, ‘‘The mixed-up individual cells
antibody (see Gilbert and Greenberg, ’84). perform, according to their cell type, the same
The 1955 paper represented the culmination of kinds of y cell-inherent inward and outward y
Holtfreter’s studies on ‘‘tissue affinity’’. Philip directed movements as do the corresponding
(‘‘Lennie’’) Townes was Johannes Holtfreter’s tissue fragments.’’ ‘‘In consequence, y the dif-
graduate student, and this paper was derived from ferent cell types y are sorted out into distinct
his Ph.D. thesis. Townes’ experiments ‘‘were homogeneous layers, the stratification of which
performed with the aim of studying the kinetic corresponds to the normal germ layer arrange-
and morphogenetic phenomena subsequent to the ment.’’ ‘‘The tissue segregation becomes complete
combination of two or more well defined cell because of the emergence of a selectivity of cell
types,’’ either as intact tissues or as single cells adhesion: homologous cells when they meet
after alkali dissociation. The techniques were remain permanently united to form functional
fairly simple. Areas of embryonic tissue (mostly tissues, whereas a cleft develops between certain
cell layers from amphibian neurulae) were teased non-homologous tissues.’’ Beyond this, the
from the embryos with glass needles. These could authors caution that ‘‘At present, it would be
be recombined with one another. When single cell futile to speculate further upon the possible
suspensions were recombined from these different subcellular factors that are engaged in cellular
tissues, the cells were dissociated in an alkaline adhesiveness. It should be pointed out however
solution and then brought together. Unlike tissue that this principle is of universal significance in
culture conditions, where one looks for out- morphogenesis, and that, in connection with
growths, such outgrowths were discouraged by directed cell movements, it is deserving of more
coating the petri dishes with agar. attention than it has received.’’ Figures 15–17 in
The sometimes complex behaviors of many their paper, reproduced here as Figure 1, present a
combinations of early embryonic tissues and their good representation of their results.
 1552499x, 2004, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jez.a.114, Wiley Online Library on [31/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TOWNES AND HOLTFRETER 703

But the importance of the Townes and Holtfr- we present the following as a chronology of
eter 1955 JEZ paper lies not with its answers, but significant developments in the investigation of
with changing the level at which morphological ‘‘tissue affinities’’ up to the present time. The
questions had to be addressed. Just as Whitehead number of publications that could be cited here
could claim that all of contemporary philosophy is so vast that the reviewers’ mindsets are
was but a footnote to the works of Plato, so one unavoidably reflected in our citations. We apol-
might conclude that our present studies of the ogize for this in advance. See Grunwald (’91);
mechanisms of morphogenesis are footnotes to the Steinberg, (’96) for further analysis and literature
Townes and Holtfreter paper. The attention citations.
Townes and Holtfreter thought cellular adhesive- Dr. Townes subsequently became Chairman of
ness deserved in connection with directed cell the Division of Genetics and Professor of Pedia-
movements was soon in the process of arriving. So trics at the University of Rochester School of
 1552499x, 2004, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jez.a.114, Wiley Online Library on [31/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
704 M.S. STEINBERG AND S.F. GILBERT

Fig. 2. Hans Holtfreter pronouncing his disagreement with the experimental evidence of Mal Steinberg (shown here
laughing) for the thermodynamic model of cell sorting. This photograph was taken at the embryology course at the Marine
Biology Laboratory at Woods Hole, 1971.

Medicine and Dentistry. He is currently Professor Holtfreter J. 1939. Gewebeaffinitüt, ein Mittel der embryona-
of Pediatrics and Chairman of the Division of len Formbildung. Arch. Exp. Zellforsch. Besonders Gewe-
Genetics Emeritus at the University of Massachu- bezücht 23: 169–209. (Revised and reprinted in English,
1964, In: Willier BH, Oppenheimer JM, editors. Founda-
setts Medical School at Worcester. Hans Holtfr- tions of Experimental Embryology, Englewood Cliffs, NJ:
eter died in November, 1992, shortly before his Prentice-Hall. p. 186–225.
92nd birthday. Having fled Nazi Germany in 1939, Holtfreter J. 1944a. Experimental studies on the development
he lived nearly half his life as a professor at of the pronephros. Rev Canad Biol 3:220–250.
Rochester University (Fig. 2). Holtfreter J. 1991. Reminiscences of the life and work of
Johannes Holtfreter. In: Gilbert SF, editor. A Conceptual
LITERATURE CITED History of Modern Embryology. NYC, NY: Plenum Press,
p. 109–27.
Gilbert SF, Greenberg J. 1984. Intellectual traditions in Holtfreter J, Hamburger V. 1955. Amphibians. In: Willier BH,
the life sciences. II. Stereospecificity. Perspec Biol Med Weiss PA, Hamburger V. editors. Analysis of Development.
28:18–34. Philadelphia, PA: W. B. Saunders, p. 230–296.
Grunwald GB. 1991. The conceptual and experimental Steinberg, MS. 1996. Adhesion in development: An historical
foundations of vertebrate embryonic cell adhesion research. overview. Develop Biol 180:377–388.
In: Gilbert, SF, editor. A Conceptual History of Modern Townes PL, Holtfreter J.1955. Directed movements and
Embryology. NYC, NY: Plenum Press, p. 129–158. selective adhesion of embryonic amphibian cells. J Exp Zool
Hamburger V. 1988. The Heritage of Experimental Embryol- 128:53–120.
ogy: Hans Spemann and the Organizer. NYC, NY: Oxford
University Press.
 1552499x, 2004, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jez.a.114, Wiley Online Library on [31/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TOWNES AND HOLTFRETER 705

APPENDIX
1952 A. Moscona expands the useful experimental material to later stage avian and
mammalian embryos with his introduction of trypsinization as a means of tissue
dissociation.
1960 Moscona proposes that ‘‘ECM,’’a viscous extracellular material appearing during tissue
trypsinization, represents ‘‘a cell-bonding framework and an ‘information’ network ...
carrying and transmitting a multiplicity of signals directed to a multiplicity of responding
mechanisms in cells.’’
1961 A. Curtis proposes that cell-type-speci¢c adhesion is governed by a ‘‘timing’’ mechanism
actuated by cell dissociation.
1962 M. S. Steinberg reports that ‘‘ECM’’ is DNA released from broken cells and gelled by the
action of the trypsin used to dissociate them.
1962^1976 Moscona’s lab reports that tissue-speci¢c ‘‘factors,’’ later called ‘‘cognins,’’ responsible for
‘‘tissue-speci¢c cell adhesion’’are released from embryonic tissues in soluble form and
promote cell aggregation tissue-speci¢cally.
1962^1964 Steinberg reports experiments designed to distinguish between ‘‘directed migration,’’
‘‘di¡erential adhesion’’and ‘‘timing’’ models based upon the cell sorting and tissue
spreading behaviors each model predicts. Among these three models, only the ‘‘di¡erential
adhesion’’ model makes the correct predictions.
1963^1964 Steinberg explores mathematically the properties of a hypothetical model system (the
‘‘Stochastic’’or ‘‘Site-Frequency Model’’) consisting of two kinds of cells expressing
identical adhesion molecules at di¡erent levels. The results predict that two such cell
populations should rearrange to produce a con¢guration in which the lower expression
cell line completely envelops the higher-expression line.
1967^1968 S. Roth and J.Weston report the ¢rst e¡orts to measure cell-cell ‘‘adhesiveness’’and its
speci¢city by measuring the rate of capture of labeled chick embryonic cells by unlabeled
aggregates of like and unlike kind in shaker cultures.They report that homotypic cells are
captured faster than heterotypic cells.
1969 H. Phillips and Steinberg identify cell aggregates’ relative speci¢c interfacial free energies
(surface tensions) as the physical parameter of ‘‘cell-cell adhesiveness’’capable of
specifying cell sorting and tissue spreading behavior and con¢gurations. Their
measurements of chick embryonic limb bud, heart and liver aggregate relative surface
tensions correctly predict these three tissues’ mutual envelopment behaviors.
1970 Steinberg reports combining the tissues and the dissociated cells of six embryonic tissues
in all 15 possible binary combinations. Their mutual spreading preferences rank the
tissues in a hierarchy from the presumably most cohesive (highest surface tension;
enveloped by all the others) to the presumably least cohesive (lowest surface tension;
envelops all the others. This implies that ‘‘cohesivity’’or surface tension ^ a quantitative,
physical parameter ^ speci¢es tissue strati¢cation rather than the particular molecular
interactions, however diverse, which underlie it.
1976 A. Harris presents several alternative hypotheses, among which he favors the ‘‘di¡erential
surface contraction hypothesis’’ that ‘‘The more strongly contractile a given cell type is
over its exposed surface, the more internally it should sort out relative to other, less
contractile cell types.’’
1976 W. Moyer and Steinberg report that the rates of cell-aggregate adhesion measured by Roth
and Weston do not comport with the surface and interfacial tensions required to explain
the con¢gurations adopted by mutual tissue spreading and cell sorting.
1977 N-CAM, the ¢rst true vertebrate cell-cell adhesion molecule, is identi¢ed by J.-P.Thiery, R.
Brackenbury, U. Rutishauser, and G. Edelman.
1977 M. Takeichi reports the co-existence of two classes of adhesion molecules in vertebrate
cells: Ca++- independent (CI) and Ca++- dependent (CD), and a E150kDa cell surface
protein associated with the latter activity.
1979 Takeichi and colleagues show that cells expressing CD vs. CI adhesion systems do not
 1552499x, 2004, 9, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jez.a.114, Wiley Online Library on [31/03/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
706 M.S. STEINBERG AND S.F. GILBERT

initiate cross-adhesions when shaken together but that hamster V79 and chick neural
retina cells expressing the same system (either CI or CD) do so.
1981 D. McClay, G.Wessel and R. Marchase describe a quantitative cell binding assay with
which they demonstrate that weak initial cell-cell binding is followed by a powerful
adhesion-strengthening process requiring active metabolism.
1981 Three laboratories (Edelman’s, Steinberg’s and Takeichi’s) report CD adhesion systems in
di¡erent cell types that cross-adhere but are nevertheless immunologically distinct.
1981 et Takeichi’s laboratory pioneers discovery that CD adhesion systems (‘‘cadherins’’) comprise
seq. a large superfamily of related cell surface proteins.
1981 First report (Takeichi et al.) of failure of cadherin-expressing cells of di¡erent kinds (later
shown to express di¡erent cadherin subtypes) to initiate adhesions in shaken cultures.
1986 et Takeichi’s laboratory demonstrates correlation between embryonic tissue segregations
seq. and di¡erentiation-related changes in tissues’cadherin subtypes.
1987 A. Nagafuchi et al. (Takeichi’s lab) demonstrate transfer of CD adhesion to non-adhesive L
cells by transfection with E-cadherin cDNA.
1987 S. Hirano et al. (Takeichi’s lab) report association of cadherins with cortical actin bundles.
1988 Takeichi’s lab demonstrates that cadherins interact homophilically: one cadherin to
another on apposed cells.
1988 A. Nose, Nagafuchi, and Takeichi report that L cells transfected to express P-cadherin vs.
E-cadherin and combined in shaken cultures aggregate separately, although the separate
aggregates subsequently cross-adhere. They consider various possibilities but conclude
that ‘‘the homotypic interaction of cadherins must overcome the heterotypic interaction in
mixtures of cells with di¡erent cadherin subclasses, thus inducing the segregation of these
cells.’’
1989 D. Friedlander, R. Mege, B. Cunningham and Edelman report that greater vs. lesser
expression of N-cadherin in two otherwise identical cell populations su⁄ces to cause them
to segregate to a certain extent after co-aggregation.
1994 Steinberg and Takeichi report that transfected L cell populations expressing higher vs.
lower levels of P-cadherin display both sorting-out and tissue spreading behavior leading
to complete envelopment of the former by the latter, as predicted in 1963^64.
1994^1996 R. Foty, G. Forgacs, C. P£eger and Steinberg describe a tissue surface tensiometer and use
it to demonstrate that the inside-outside hierarchy of mutual tissue spreading preferences
described in 1970 ranks the tissues, as predicted, in the sequence of their surface tensions.
1997 G. Davis, Phillips and Steinberg report that the surface tensions of amphibian embryonic
germ layers, whose sorting-out and spreading behaviors were documented by Townes and
Holtfreter,‘‘lie in precisely the sequence necessary to account for germ layer strati¢cation.’’
2003 D. Duguay, Foty and Steinberg demonstrate that the separate aggregation in shaken
cultures of cells expressing di¡erent cadherins results from the magni¢cation by shear
forces of the slightly slower rate of initiation of adhesions between di¡erent cadherin
subclasses, re£ected earlier in the results of Roth and Weston.With lower shear forces, all
cadherin subclasses tested cross-adhered. L cells expressing E- and P-cadherin at the same
level both co-aggregated and failed to segregate, indicating that E-E, P-P and E-P cadherin
adhesions are all of equal intensity. Evidence of weaker ‘‘heterocadherin’’adhesion was
found for B- vs. R-cadherin.They report that as little as ‘‘a 26% di¡erence in mean cadherin
expression level has been su⁄cient to produce a degree of cell sorting in mixed
aggregates.’’
2004 Foty, Cho, and Steinberg report that the surface tensions of aggregates of L cells
transfected to express N-, P- and E-cadherin are all identical on a per cadherin basis and
are a linear function of cadherin expression level, verifying the di¡erential adhesion
hypothesis. They also present evidence that the great ‘‘adhesion strengthening’’ ¢rst
described by McClay et al. is largely abolished by drugs that depolymerize f-actin, for
which they suggest an explanation.