<<

. 49
( 66 .)



>>

from the more standard accounts, the different emphases of the two
perspectives could not be clearer.
Developmental biology has advanced to its current high level of
sophistication with little explicit analysis of the physical dimension
of the questions it treats. This is changing, however. The DNA sequenc-
ing initiatives of the last decade of the twentieth century con¬rmed
that genes number in the range 10 000--50 000 in all the species tradi-
tionally studied by developmental biologists. Analysis with microar-
rays (computer-interfaced devices for quantifying the abundance of
mRNAs in tissue samples) has shown that many genes are simulta-
neously expressed during any signi¬cant developmental event (Bard,
1999; Montalta-He and Reichert, 2003). The sheer number of gene--
gene interactions that are now known to occur during embryogenesis
has generated a need for new methods to handle such complexity as
well as new concepts that could help investigators ˜˜see the forest for
the trees.”
One response to this has been an increase in computational and
complex systems-based approaches to the analysis of gene expression
data sets (Bard, 1999; Davidson et al., 2002; Oliveri and Davidson;
2004). But while concepts of dynamical systems of the sort discussed
in Chapter 3 sometimes enter into these analyses, such approaches
continue to be largely gene-centered. That is, they are concerned pri-
marily with the control of gene activities by other gene activities.
The physics of soft condensed materials -- diffusion, viscoelasticity,
10 EVOLUTION OF DEVELOPMENTAL MECHANISMS 249


network formation, phase separation -- are the concepts we have used
to explain epithelial and mesenchymal morphogenesis, pattern for-
mation, and organogenesis in earlier chapters. This physics has been
slower to enter into mainstream developmental biology. It is one of
the purposes of this chapter to explore what it is about the orga-
nization of development that has permitted the genetic aspects to
stand in, until now, for the whole. It has taken a reconsideration of
the connection between the development of biological form and its
evolution (Gould, 1977; Raff, 1996; Salthe, 1993), a subject that was
out of fashion for most of the twentieth century, to help bring the
physics of materials back into the analysis of development, where it
once occupied a more prominent place. (See Newman, 2003a, for a
brief history of ˜˜systems” approaches to development.) In what fol-
lows, we address the connections between development, physics, and
evolution.


The physical origins of developmental systems
Since the ˜˜physical” and ˜˜gene-centered” approaches deal with iden-
tical subject matter -- the developing embryo -- it makes sense to try
to understand the relationship between them. Speci¬cally, what is
it about the construction of organisms and embryos that has per-
mitted a relatively coherent (but, we would suggest, mechanistically
incomplete) body of knowledge to be generated that focuses on causal
chains acting at the level of gene interactions?
On the one hand, viscoelastic materials and chemically and me-
chanically excitable systems -- the objects of the major explanatory
framework used in this book -- existed prior to the world of organ-
isms. The ˜˜generic” physical forces and mechanisms used to account
for developmental phenomena in the preceding chapters are capable
of acting on living tissues and nonliving materials alike to produce
similar shapes, patterns, and dynamical activities (a view closely as-
sociated with the biologist D™Arcy W. Thompson (Thompson, 1917)). In
principle, a rough sort of ˜˜development,” that is, a sequence of form-
changes in an aggregate of cells, could be driven solely by physics. On
the other hand, it is also the case that developmental steps in any
real embryo are typically associated with the appearance of one or
more speci¬c gene products at a particular time and place.
Many key gene products that participate in and regulate multicel-
lular development emerged over many millions of years of evolution
in a world containing only single-celled organisms. These gene prod-
ucts (organized with other molecules into cells and their ECMs) consti-
tute the material acted upon by physical mechanisms during embry-
onic development. Physical mechanisms acting on materials by vritue
of their inherent physical properties, not the genes or their products
on their own, are what produces biological forms. But as an embryo
develops, the operation of physical mechanisms is constrained and fo-
cused by hierarchical systems of coordinated gene activities -- so-called
250 BIOLOGICAL PHYSICS OF THE DEVELOPING EMBRYO


˜˜developmental programs.” This implies (since developmental pro-
grams evolved gradually, after the origination of multicellularity)
changes in form and cell type in ancient cell aggregates were not
subject to the precise control seen in present-day multicellular or-
ganisms.
For these reasons, the relationship between the physical and gen-
etic aspects of development can be understood only in an evolutionary
context. In the remainder of this chapter we will explore scenarios
for the origin and evolution of developmental systems, with emphasis
on the physical aspects.

The ¬rst metazoa
Multicellular organisms ¬rst arose more than 1.5 billion years ago
(Knoll, 2003). The earliest of these were ¬lamentous algae, plants
whose rigid cell walls would have prevented them from being suscep-
tible to many of the viscoelasticity-based morphogenetic mechanisms
of animal development discussed in this book. (However, see Nagata
et al., 2003, for application of a number of related processes to the
development of plants.) The ¬rst ˜˜metazoa” (multicellular animals) ap-
pear in the fossil record earlier than 700 million years ago (Ma). The
most extensively studied of these fossils, dating from 580--543 Ma,
have been described as relatively simple, ¬‚at, quilt-like creatures,
probably without body cavities (Seilacher, 1992; Brasier and Antcliffe,
2004). Modular body subdivisions often exhibited a fractal branch-
ing pattern instead of the segmentation seen in modern organisms
(Narbonne, 2004; see Chapters 6 and 8 for a discussion of fractals). By
approximately 540 Ma the ˜˜Cambrian explosion” had occurred, a term
denoting the fact that virtually all the general categories of body or-
ganization seen in modern organisms had burst into existence in the
preceding 25--30 million years, a mere blink of the eye geologically
speaking (Conway Morris, 2003).
Metazoan bodies are characterized variously by axial symmetries
and asymmetries, multiple tissue layers, interior cavities, segmen-
tation, and various combinations of these properties. Each species
can be assigned to one of approximately 35 body plans (Raff, 1996;
Arthur, 1997) or organizational categories (Minelli, 2003). These are
essentially the same as the ˜˜phyla” of the standard taxonomic system
(Valentine, 2004). The organs of an animal are constructed using mor-
phological motifs similar to the body plans (see Chapter 8). While the
early world contained many unoccupied ecological settings (˜˜niches”)
within which new organismal forms could ¬‚ourish, this alone cannot
account for the rapid profusion of body plans once multicellularity
was established, nor for the particular forms assumed by bodies and
organs.
Nonliving materials, such as liquids, clays, polymer melts, soap
bubbles, and taut strings, by virtue of their inherent physical proper-
ties can take on only limited, characteristic, sets of shapes and con¬g-
urations -- sessile drops, vortices, standing and traveling waves, and so
forth. The most ancient cell aggregates, lacking the highly integrated
genetic programs of modern-day organisms, would have been molded
10 EVOLUTION OF DEVELOPMENTAL MECHANISMS 251


in a similar fashion, by physical forces and determinants inherent
to their material properties and scale. Can we reconstruct the forms
likely to have been assumed by these ancient cell aggregates?
Multicellular aggregates differ from single cells in their linear
dimensions (i.e., size) and in their susceptibility to physical determi-
nants. As we saw in Chapter 2, for example, liquid surface tension per
se is not an important determinant of individual cell shape, and free
diffusion (Chapter 1) has only limited scope as a mode of intracellu-
lar transport. And while individual cells may assume distinct states
based on the dynamics of their metabolic or transcriptional circuitry
(Chapter 3), such states obviously cannot cooperate with one another
or assume different tasks in an organism that consists of only one
cell.
The cell aggregates that eventually evolved into the metazoa were
a different story. Their cells, like those in modern embryos, were likely
to have been free to rearrange with respect to one another. Primitive
aggregates, therefore, would have been likely to exhibit liquid-like be-
haviors such as rounding-up and spreading (Chapter 4). These cell clus-
ters would also have provided a medium for the diffusion of secreted
proteins and thus for the generation of global (e.g., aggregate-wide)
diffusion gradients (Chapter 7). The single-cell organisms that existed
before the emergence of multicellularity must have been metabol-
ically active, thermodynamically open, systems, much like present-
day cells (the evolution of eukaryotic cells had been under way for
more than a billion years before metazoans appeared; Knoll, 2003).
Since the premetazoan aggregates would thus have been composed of
chemically excitable cells (Chapter 3), they were not only viscoelastic
and chemically heterogeneous, they were also excitable media. This
means they would have had the potential to elaborate self-organized
spatial patterns of cells with different biochemical states (Chapter 7).
And in cases where these cell types exhibited different amounts of ad-
hesive molecules, adhesion-based sorting would have been inevitable.
The evolution of polarized cells and planar basal laminae would have
led, as a physical side-effect, to aggregates with lumens or to elastic
cellular sheets (Chapter 5).
In our earlier discussions (Chapters 1--8), which dealt entirely
with individual cells and multicellular organisms of the present-day
type, not their ancient counterparts, we saw that viscosity, elasti-
city, dynamical-state transitions, differential adhesion, and reaction-
-diffusion coupling all continue to be factors in embryonic develop-
ment, notwithstanding the constraints of evolved architecture and
multisystem integration. In the earliest arising multicellular forms
these physical determinants must have acted in an even less con-
strained fashion, potentially allowing new forms to arise with little
or no genetic change (Newman and M¨ ller, 2000; Newman, 2003b).
u
In the following section we discuss how some of these physical
processes might have ¬rst become relevant to the world of multi-
cellular organisms. We note that, unlike most of the descriptions
in previous chapters, in which physical concepts were linked, wher-
ever possible, to experimentally determined properties of cells and
252 BIOLOGICAL PHYSICS OF THE DEVELOPING EMBRYO


known molecules, we are not on such solid ground in attempting to
characterize the origination of multicellular forms. While the discus-
sion that follows is based on plausible biological and physical argu-
ments, it is explicitly in a hypothetical mode, as is frequently the case
with issues of causation in evolutionary biology.

Physical mechanisms in the origin of body plans
Let us imagine an ancient aggregate of cells before the emergence of
true metazoa. If a group of cells in one region of this cluster released
a protein product at a higher rate than their neighbors -- either ran-
domly or because some external cue stimulated them to do so -- the
aggregate would thereby have become chemically nonuniform from
one end to the other. Let us further assume that in some of these
cases the cells happen to react differently to different concentrations
of the molecule in question (we will call it a ˜˜proto-morphogen”) and
thus assume different states in a fashion that depends on the con-
centration of the proto-morphogen. Under these circumstances, a spa-
tially non-homogeneous distribution of a chemical (e.g., a gradient)
would have given rise fortuitously to a nonhomogenous distribution
of cell states (Fig. 10.1). We can thus see how ˜˜generic” physics (i.e.,
diffusion) acting on a scaled-up biological system (i.e., a multicellular
aggregate) can give rise to an incipient developmental process.
But how could such a haphazardly determined effect be perpetu-
ated from one generation to the next? In present-day embryos the po-
sition of the embryonic ˜˜organizer” (i.e., a cell or group of cells that
is a unique source of a diffusible morphogen) is often determined
by maternally deposited cues, or some other genetically in¬‚uenced
process, in conjunction with external cues, such as the sperm entry
point. In such cases, hereditary transmission of the relevant genes or
gene variants creates reproducible conditions for the recapitulation
of the event from one generation to the next. In our hypothetical
ancient form, lacking such a genetic program, recurrence of the de-
velopmental event could have been perpetuated by less formal means.
If, for example, the cells in the primitive aggregate had a 1% chance
of randomly producing and secreting the proto-morphogen, then half
of all 50-cell aggregates would have a (proto-)organizer cell. These vari-
ants would ˜˜develop,” that is to say, they would self-organize into a
nonuniform distribution of cell states.
In this scenario, if there were a selective advantage to having a
phenotype containing nonuniformly distributed cell types then cell
clusters whose genotype inclined them to produce proto-organizer
cells at a higher frequency would become more prevalent. This ten-
dency would be balanced by the fact that if all cells became organizers
there would be no gradient. And this, in turn, would put a premium
on genetically variant clusters in which proto-organizer cells limit the
appearance of other proto-organizer cells, that is, produce a lateral
inhibitory factor simultaneously with the proto-morphogen.
In order for such biochemical circuits to generate stable spatial
patterns they must be part of a reaction--diffusion or (taking cells
10 EVOLUTION OF DEVELOPMENTAL MECHANISMS 253




Sedimentation Chemical
gradient oscillation


Cell polarity

Differential adhesion


Cell polarity


Reaction-diffusion Diffusion gradient




Fig. 10.1 Schematic representation of the hypothesized origination of body plans via
the morphogenetic consequences of linking the regulation of cell“cell adhesion to
various physical and chemical pattern-forming mechanisms. Red lettering, cell properties;
black lettering, pattern-forming mechanisms. The central box denotes the effects of
differential adhesion in causing the formation of boundaries within a tissue mass, across
which cells will not mix (see Chapter 4 and Fig. 4.5). The polarized expression of
adhesion molecules leads to cavities and other lumenal structures (see Fig. 4.2). All the
peripherally arranged boxes denote pattern-forming mechanisms which, when deployed
in conjunction with differential adhesion, can lead to standard body-plan organizational
motifs. The sedimentation of a dense cytoplasmic component or the diffusion of a
morphogen (see Fig. 7.7) are ways in which an egg or early blastula can become spatially
nonuniform; if these nonuniformities lead to the generation of the expression of
differentially adhesive cell populations, gastrula-like structures will form (see Fig. 5.6).
Similarly, chemical oscillations (see Fig. 7.3) or pattern-generating reaction“diffusion

<<

. 49
( 66 .)



>>