J. St"'ittl Biul. Stfllt:l.

1987 10, 73-83

The organ of form: towards a theory of

biological shape
Francisco J. Varela
CREA. Ecole Polyteclmique, Paris, France
and
Samy Frenk
Rolf Institute, Boulder. Colorado. USA
Cells and the extracellular matrix (ECM) immediately surrounding them engage in
reciprocal d..:terminations. But the ECM is also a global structure because it is continuous
throughout the body. We argue that this local-global articularion is a central element in
the determination of an animal's form, and we show how it participates in all the other
dimensions of animal life. Specific experimental implications and further consequences
of this view are discussed.

What is shape?
Living organisms have shape. This is so obvious a statement that we take it for granted
and become oblivious of the fact that there is no adequate theory of living fonn and
shape in contemporary biology. fn this paper, when we employ a theory of shape we shall
be concerned with not only the principles which determine the spatial pauerns of bodies,
but also how such patterns participate in all the dimensions of animal life such as
movement, cognition. disease, and communication. Thus, we are concerned with fonn
not only in tenns of geometric relations and proportions, as developed in the tradition
of D' Arcy Thompson ( 1961 ), but beyond that with shape as an integral component in
the dynamics of a living system. In this paper. we shall present an outline for such a
theory of biological shape.
Our approach is based on current biological research. However, it is not a mere
aggregate of current facts. but rather a conceptual scaffolding from a very specific
vantage point. It defines a living system as a totality, and how the phenomena proper
to life unfold from this peculiar organization. This framework has been presented
extensively elsewhere (Maltlrana & Varela, 1980; Varela, 1979). Rather than recapitulating on these ideas here. we shall use them in the context of shape; the reader will
thus understand these ideas through examples of their usage.
The knife's distinctions
As a first approximation we can define a living form as a collection of spatial distinctions
i':' an organism. A distinction is the act of defining what constitutes the components of
0110-1750{87/010073+ II SOJ.00/0

(D 1987 Academic Press Inc. (london) limiccd

33

F. J. Varela ami S. Frenk

a given unity. The shape of a table, for instance, is such a collection of spatial relations
between the components (table top and legs) of the unity table. Hence. a discussion of
shape must start by making explicit the distinctions we make in an organism as a
composite unity and their spatial relationships.
Traditionally, biological shape has been the province of anatomy (literally: separating
the parts). Anatomical studies began in earnest with Vesalius and his monumental De
Fabrica Corporis Humana in 1543. Since then, his observations have been refined
sub$tantially to constitute a data base which most scientists would consider 'stable' or
achieved, in same basic sense, although a few minor details are continuously being
added. This, of course, applies to human ana10my. Animal anatomy is a more open field
because of the immense diversity of species.
The spirit of Vesalius' work is present almost unchanged in the human anatomy a
medical student must learn today. What are the fundamental distinctions implicit in this
venerable tradition? They can easily be described: the parts of an organism which can
be distinguished (and related in space) are those which result from the actions of a knife.
Vesalius started his own studies in a cultural context where hunting and butchery were
widespread. Evidently, he drew from that context and, more importantly, from the
distinguishing instruments used then. The knife has now been refined to become a scalpel
but the principle remains the same. The knife edge separates that which falls on both
sides as the distinguished components. It separates bone from muscle and muscle from
viscera. Thus, we end up with the separation between soft parts, muscles, and skeleton
which seems so familiar to our Western minds.
What we have said so far concerns what most biologists would call classical human
anatomy. Modern biology has developed additional tools for dissections and instruments
whose implicit distinctions are radically different. These instruments penetrate into the
cellular and molecular level and belong to microscopic anatomy and cellular and
molecular biology.
The most important of these new tools used to make distinctions is the microscope.
It revealed in the eighteenth century a fundamentally different distinction relative to
bodies: cells. The microscope, and tools for molecular separation developed later, can
distinguish 11nits bounded by membranes which are fundamental components of every
living organism.
What might not be so apparent is the fact that delimiting cells reveals, by contradis. tinction, what is not bounded by cells in the body. This aggretate of non-cellular
substance is the so-called connective tissue (see fig. I). It includes the space under a
covering epithelium, the gaps between muscle bundles, the spacing between viscera. as
well as ligaments and fascia. The remarkable thing about this non-cellular component
is that it is a contimmm.
To make this point apparent. let us consider a cross-section through the neck of a
human body (Fig. 2). Let us move through the tissue from the outside to the inside. At
the otuer surface we find the skin which appears as a layer surrounding the entire
cross-section. Immediately under it we find connective tissue, first in the form of a basal
membrane under the epithelial cells and th<:n as a subcutanous layer, Notice that,
:tlthough there are some cellular elements in this connective tissue such as fibroblasts and
blood cells, typically this is a non-cellular matrix of fibrous and viscous material. We
shall return to th.: constitution of the connective tissue below. Moving a bit deeper into
the cross-section. we come into contact with muscle bundles where cellular eiements
clearly predominate. all hough we are able to see connective tissue in the form of fascia
which surround the muscle. The degree of condensation of the connccti\'e tissue associated

The orga11 of form

34

Fig. 1. A microscopic view of IIU! network of elemems within a small section of the loose
guinea~ pig stainecl witiJ Bi:zozero 's method; x 800. A: bundieJ of
collagen bundles. B. C: fibroblasts. F: elastic fibers. V: blood esse/. Taken from Ramon
y Caja/ & Tello y M11iioz ( /950: Fig. /91)
connecrive tissue in tlte

with muscular elements varies from extremely lax to a thick packet as is found in a
tendon. Moving even deeper inside, we encounter bone, whicll is also in continuity wilh
the rest of connective tissue. It differs from it by deposition of mineral elements,
especially calcium, and the arrangement of precise geometrical pa!lems produced by a
sparse bur active population of cells.
In short, from this brief journey through a cross-section of a neck. we see that the usual
anatomical descriptions implicit in the knife's actions produce separations in connective
tissue which amount to arbitrary distinctions of degrees of density rather than qualitat
ively different constituents of the neck's shape. A more adequate distinction is between
cellular aggregates (epithelium. muscle bundles, and so on) and the surrounding space
which is tilled with an extracellular mmrix (ECM).
let us now extend this point of view of the continuity of the space between cellular
elements beyond the two dimensions of the cross-sections described above to the entire
three dimensions of the body. Thus, let us consider the shoulder beyond the forearm,
then the trunk, until we encompass the en(ire body. !I is perhaps easiest to evoke what we
mean by yet another thought-experiment. Imagine we take the dead body of an animal,
say a cat, and we drop the entire thing in a detergent which dissolves only cellular
elements, leaving the ECM unmuched. We leave it in the detergent long enough to
extract every piece of the cellular components, and then we pull the cat-minus-cells out
of the detergent bath. What we would see is still a eat's shape, only in negative as it were,
where only the space around the cells remains visible. The eat's shape is a continuum:

F. J. Varela and S. Frenk

35

Fascia pratoWCIU:lt,.:ah

f'orpu'
Ytlhtbr;;u
~en

ic<.tht \"I(

J~nnuutu

,-e ..... u,o.l..:

____ )IUII:tln det

N~lle""

""

:..I

...

Fig. 1. A cross-section thrDIIgh the neck ofa human at the level of the trachea, as seen from
above. Taken from !:i'patellolz ( 1902; fig. 307 ). Feupolster = layer of fat; Muskeln des
Nacketu = neck muscles

there is no dear transition between the basement membrane of the skin, the muscle's
fascia, the bones, or the connective tissue between the viscera.
Our basic intention here is to argue why the continuity (or global interconnectedness)
of the ECM should be brought into the foreground as an essential key to the understanding of biological form. In fact, we believe it constitutes an organ of form. The
st>ctions that follow unfold the arguments to support the adequacy of this designation
and its consequences.
The biology of the extracellular matrix
Studies concerning the ECM have been carried out mostly in the last 20 years, and they
are still at a stage where they are not that familiar to biologists and non-biologistsalike,
although this is changing rapidly. Over the last 20 years, techniques from biochemistry,
ultrastructure. and immunohistochemistry, have revealed the fundamental universality
of the ECM components (Hay. 1981a).

The organ ofform

36

As a first approximation. the ECM is a matrix of fibrous materials secreted by cells

of various kinds and bound together in intricate tangles. The most conspicuous, and first
to be described, of these fibrous components is collagen, an ubiquitous protein which can
exist in various degrees of aggregation. Next to collagen in abundance are the polysaccharides and a combination of polysaccharides and proteins, or glycoproteins. There
is also a rich variety of mucopolysaccharides, including hyluronate, chondroitin sulfate,
collectively called glycosaminoglycans (GAG) (Fig. 3).
Dy and large. these biochemical characterizations have remained separate from
cellular biology until recently. Interest in this area has increased because of the steady
accumulation of observations pointing to the precise and extensive relationship between
the ECM and the surfaces of all the cells in the body. According to these observations,
there are multiple ways in which collagen, glycoproteins, and GAG can be arranged to
form highly specific links to receptors located on cells' membranes. Thus, the ECM is in
a position to exert specific and dramatic changes on the cellular dynamics, just as much
as, say, a hormone or a neurotransmitter (Fig. 4).
Test lube

Collagen
(640 A)

Tissue

?
:.

~!

__....

Native
collagen
(640 A)

Staggered

TC

TC
Acid

Test tube

(%6QOA)
:'\..,v'

~~A~ :=f==:J:::=r
~

lln

-TC

~~!i.=f

FLS
Fig. ]. Colllrgm is a protein whicl1 can take many different fo~ms ~ependin'! on the
comiitions. Troporollagen ( TC) is the building block. which polymer~:es mthe vartous wa.vs
shown in the tliagram. Taken from Ha. ( /98/a)

F. J. Varela am/ S. Frenk

37

Act in complex

ACIIIl complex

co
( 0)

(bl

I I I I

Cell

(c )

(d)

Fig. 4. Diagrams of several models depicting the possible relationship of ECM molecules
to the cell surface. All models ass11me specific cell receptors .for various fibrillar componems
of tile ECM. Taken from Hay ( J98Ja)

In brief, the imimate milieu of every cell in our bodies is not a bland and homogeneous
soup of nutrients. Instead, this intimate milieu has a precise achitecture provided by all
of the intricacies of the ECM components, with an ongoing dynamic exchange with the
cell surface.

Morphocyclcs and the orgnn of shape

Having introduced the basic questions about shape and the key qualities of the ECM,
we can now turn to the centr;1l idea we wish to introduce here. h consists of considering
simultaneously 1he local and global qualities of the ECM. The link between global and
local is given by the cyclic (or self-referential) nature oft he interactions between cells and
their surrounding space containing the ECM. Let us clarify this.
At every location. the ECM is produced by cellular elements of that particular region.
But the local ECM can also influence cell dynamics, thus constituting a cycle of reciprocal
interaction between cellular and non-cellular constituents. Dut this local reciprocity is
not lhe entire story, for whatever local action occurs is necessarily conditioned by the

Tlte organ of form

38

continuity of c:ach local ECM with the adjacent ECM. and, through them. the entire
body. As in the notion of a field and its corrc:sponding particles. there is in living shape
a dynamic complementarity: the entire global shape of the body affects the local conditions for ECM/cell relationship, but at the same time the local dynamics conditions
how the entire body is actually built.
We call this reciprocal determination between cellular elements of a multicellular
anin1al and the continuous extracellular matrix a morplwcycle. Thus. a morphocycle is a
process. that is an ongoing bootstrapping, whereby a shape is produced by the body's cells.
Gut this sha'pe in turn conditions (through the continuity of the ECM) what the cells do.
Or, in other words, a morphocycle is the process whereby a local action between the ECM
and cell surfaces produces the global effect of shape and is in turn constrained by it.
(n this kind of dynamics of mutual and complementary reciprocity, it is tempting to
take one side of the process as dominant (Goguen & Varela, 1979). However, it is clear
that at any given time, a body is the result of a very prolonged history of uninterrupted
morphocycles, and what is due to celts and what is due to shape is inseparable. In fact,
even if we retrace the steps of a body's shape back in time, the problem is not solved,
for even the zygote did not exist in a vacuum, but already inside another shape.
From what we have said, it seems appropriate to refer to the whole continuous ECM
as an organ of shape, since it is through it that existence and form become inseparable.
Cases
Let us consider some examples which illustrate the above ideas in action.
Development morphology of organs
The way in which a specific (i.e.' local) kind of ECM can condition the differentiation of
cells, and further, be an integral part of specifying the characteristic morphology of an
organ, is a recent and much debated possiblility (Hay, 198tb; Lewis. 1984). Forinstance,
for many years researchers have tried to induce normal differentiation of mammary
glands in vitro with the aid of inducing hormones. Such attempts met with little or no
success. However, when mammary cells are cultivated in the presence of the ECM of the
mammary gland, adequate differentiation takes place and produces functional mammary
glands. Furthermore, this differentiaton is possible with just the mammary stroma (i.e.
the isolated fibrous components of the local ECM) and in the absence of any inducing
hormone. If the llexibility of the ECM is inhibited by various means (such as accessibility
to oxygen), the capacity for differentiation is correspondingly lost (Shannon & Pitelka,
1981 ). Thus, the ECM is capable of acting back onto cells by mechanisms which involve
genetic repression and derepression. giving rise to changes in the celts they enclose which,
in turn, produce an ECM peculiar to their configuration. Also, the role of fibroblasls.
the cell class found sparsely in the ECM, is beginning to be clearer. They have been found
to have remarkable traction properties through their secretion of a collageneous matrix,
capable of dictating much of the structure of the skeleton. location of the muscles. routes
of m:rves, and pauerning of the skin (Chevallier & Kinney, 1982; Lewis et al. 1981).
Thus. morphocycles are centrally involved in the differentiation of the function and
shape of organs.

Cmu:er am/ cotJIIt!Clile tiss11e

One of the most devastating aspects of cancer is the capacity of tumors to spread from
their primary site to other organs. Purl of the difficulty of understanding these processes.

39

F. J. Varela and S. Frenk

which is a key for the preventive treatment of cancer, is the diversity of cells in the
primary tumor and the way in which metastatic tumors are selected accordingly. Various
different locations in the body select different cellular classes upon which to stan the
growth of a malignant tumor. Predictably. this selection occurs with the concourse of the
cells of the target organs. Further, such selection also occurs with the participation of
the specific kind of ECM because it mediates between the invading malignant cells and
the future site of tumor growth (Fidler & Hart, 1982; Auerbach & Auerbach, 1981).
Thus, a malignant tumor might not grow in the lymph nodes of the neck, but il does so
actively in the nodes under the ann pit. Thus, one key to the mechanism of metastasis is
the presence of ongoing morphocycles at each location.

Do muscles act by pulling on the tendons?

The standard textbook interpretation of how a muscle acts is that it pulls on the tendon
in which it terminates. The traction produced by the muscle's shortening is directly
transmitted through the tendon to the bone, which is thus mechanically displaced. We
may ask, however, what is the evidence for this accepted view? It it were true. we would
expect some kind of mechanical continuity between muscle cells and the surrounding
collagen. From the ultrastructural point of view, such continuity is not all that clear.
Muscle fibers are surrounded, but not directly linked, to their surrounding ECM. This
raises the possibility of an alternative interpretation of muscle action, one that puts
further emphasis on the continuity and imegrity of the organ of fonn. In fact, when
a muscle contracts, it not only shortens but also thickens. The diameter of the fibers
is correspondingly increased, which causes the connective sheath to be pulled per
pendicular to the line of sarcomere shortening. If there is a strong continuity in the
connective sheath, the increase in the diameter will also result in a pull on the tendon and
bone. Recent experiments show, in fact, that weakening the continuity of the connective
tissue around the muscle belly also weakens its capacity for action (Kirkwood, Mamrana
& Varela, unpublished data). It is possible, of course, that both mechanisms act in
unison. But it is only if we think about the organ of form as a continuum that the second,
and perhaps predominant mode of action of muscle action is properly understood.
The cases mentioned in this section range from the very detailed to the suggestive and
from the molecular to the macroscopic. They are intended as a showcase of how the
present perspective can be projected into specific problems and contribute fresh new
alternatives.
Natural history of the organ of shape
From the point of view presented here, the organ of shape is the specific structure which
makes possible the spatial co-existence of cells in an aggregate which op~rates as a unity.
as a whole organism. Thus, shape is synonymous with the very existence of a metazoan
or multicellular animal.t Furthermore, the biochemistry of the ECM is. surprisingly
universal throughout the entire range of vertebrate and perhaps also invertebrate life
(Hay, 1981). This universality is also present in other fundamental living dynamics such
as the genetic code, membrane transporl, or metabolic pathways. Like these. the mutual
effects between ECM and cell dynamics tend to be very conservative mechanisms
throughout evolution, as fundamental building blocks which are rarely, if ever. subjecl
to modification.
. t Ahhough this discussion does not in principle exclude planl shapn, there is linle comparative material on
vegetal ECM. What follows, therefore:, applies 10 all kingdoms with the possible exception
plants.

or

The organ of form

40

This is a very interesting fact when considered in the light of the universal nature of
multicdlularity. Contrary to traditional views, cellul<lr aggregates constituting an organism
with a distinct shape e~ists not only amongst the macroscopic creatures, vertebrates and
invertebrates. Multicellularity is present in all of the five kingdoms: monera (i.e. bacterialike), protysta (i.e. protozoa-like), fungi, plants, and animals (Margulis & Schwartz,
1982). (n all of these kingdoms. one can find individuals which are multicellular, although
in tht; case of venebrates this is an obligatory feature. In the first three kingdoms, in
contrast, many members lead a life as independent, free-living, single cells (Fig. S).
There is evidence for the presence of multicellular animals dating back to the
Edicarian period at the beginning of the Phanerozoic eon some 3 billion years ago. This
is the period to which the oldest known living fossils have been traced (Cloud &

Lipid vesicle

Ventral epithelium
(columnar)
Cutaway view

Nucleus
Nucleolus

Call membrane
_.,~-r+-.._

~1\doplasmic

Mitochondria "1-llr-:!"'

reuculum

vacuole
Slime motria

Slime matria

Fig. j. Top: a tlrawing of T richopl:llt adhaercns. one of tile simplt.ft oft!/1/Mng multit-C'Iiule~r.
animals. Btlow l~ji: Labyrinlhula c~lls forming a slime ne~. 8~/o11 rtght: the SlrllCtllf~ oj
a single L:1byrinlhula cell. showi11g slime matrix surrormdm.l( 11. Taken from Margulrs &:
Schwart: { 1982)

F. J. Yare/a and S. Frenk

41

Fig. 6. The history of reciprocal coupling between two autopoietic units, symbolized here by
a circular arrow. can have two possible ou1comes: containmem or juxlaposition. In one case,
ont' Jws a histor)' of S)'mbiosis; in the other, the secretion of a common space, i.e. a shape
Gl;~essner,

1982). Since multicellularity is as old as life is, sha_pe, in the sense understood
here, is almost as old as life. Bodies and shapes did not begin with fishes or lizards.
Life must have arisen by the constitution of minimal autopoietic units, self-producing
units capable of generating their own boundaries (Maturana & Varela, 1980; Varela, 1979).
Dut once populations of such autonomous units arose, the possibility of being factors of
reciprocal selective histories arose at the same time for these units. In such histories of
recurrent interactions between two primitive cells, there are two possible logical outcomes:
either their boundaries dissolve by one becoming contained in the other, or else their
boundaries do not dissolve but become juxtaposed in the same space to each other (see
Fig. 6).
The first possibility is a case of symbiosis, where one kind of cell becomes a permanent
host to another. This seems to have been precisely the path taken in the history of
modern, eucaryotic cells (Margulis, 1981 ). However, the logical dual nature of this
symbiosis is that cells become strongly bound by the specification of a common space
produced by the joint dynamics of the participating cells. This is tantamount to saying
that the participating cells secrete their own surrounding space: an extracellular matrix
which delimits precisely what is and what is not part of it. To state this in yet another
way, crossing the boundaries (as in the origin of eucaryotes) could be described us
endo-symbiosis. Preserving the cellular boundaries while sharing a mutually specified
space could be described as exo-symbiosis. which becomes another word for shape.
Endo- and exo-symbiosis have been presem from the very beginning of the natural
history of life, since lhese were options open to the very first populations of autopoietic
systems. Further, the dual option of endo- and cxo-symbiosis can operate not only
between cells, but also between multicellular organisms themselves. A lichen and a
parasite are examples of this principle applied at a higher level of recursion.
Conclusion
The origin of shape and its morphocylcs are token names for an entire context and
research progmm in which to understand biological shape, its material substrate, its

The organ ofform

42

natural history, and the way in which it can participate in various aspects of an animal's
life. This proposed perspective consists basically of bringing into alignment a number of
data from current research with a specific perspective on living systems. The intuition
behind our. framework is that space is a constitutive element in the dynamics of living
organisms just as much as the solidity of their molecular constituents. We are only just
beginning to realize the imparlance of this mulUal partnership between cellular dynamics
and specitledjspecifying space.
In the light of the present perspective, the understanding of biological phenomena has
hopefully been enriched and unified. Beyond such an aesthetic rewiud-, the present
hypothesis does lead to interesting new questions which can be addressed experimentally,
such as those outlined in the section 'Cases', above. It is also interesting to consider the
usefulness of this perspective as a foundation for the whole array of disciplines and
techniques collectively known as 'bodywork', where shape and posture are seen as being
inseparable from consciousness itself and the wholeness of human experience.

.
Acknowledgements
F.V. is Foudation de France Professor ofCoquitive Science. The financial support of the
Prince Trust Fund is gratefully acknowledged.

References
Auerbach. R. & Auerbach, W. (1981). Science 215, 127-134.
Chevallier, A. & Kinney, M. (1982). Arch. dev. Bioi. 191, 277.
Cloud, P. & Glaessner, M. (1982). Science 218, 783-792.
D'Arcy Thompson, W. (1961). On Growth and Form. Cambridge; Cambridge University Press.
Fidler, I. & Hart, I. (1982). Science 217, 998-1003.
Goguen, J. &. Varela, F. (1979). Int. J. gen. Syst. 5, 31-43.
Hay, E. {1981a). J. cell. Bioi. 91. 20Ss-235s.
Hay, E. {l981b). In (E. Hay, Ed.): Cell Biology of tl1e E.'Ctracel/ular Matri.Y:. New York: Plenum
Press.
Lewis, J. (1984). Nawre 307, 413-414.
Lewis, J., Chevallier, A., Kinney, M. & Wolpert, L. (1981). J. embrrol. e:cp. Morpho/. 64, 211.
Margulis, L. ( 1981). Symbiosis and Cell Evolution. San Fransisco: Freeman.
Margulis, L. & Schwartz, K. {l982). Five Kingdoms. San Fransisco: Freeman.
Maturana, H. & Varela, F. {1980). Autopoiesis and Ct~gnition. Boston: D. Reidel.
Ramon y Cajal. S. & Tello y Munoz. J. {1950). Elemtruos tie lliswlogia Nomwl. Madrid.
Shannon, J. & Pitelka, D. (1981). /11 viuo 17,0106-1028.
Spau:holz, \V. ( 190 I). Hamlml11s tier Ant~tt>mie des ;\-leusclrens. leipzig.
Varela. F. (1979). Prilft:iples of Biological A11~tomy. New York: North-Holland/Elsevier.