. 50
( 66 .)


processes (see Fig. 8.15), if they come to regulate adhesive differentials between cells,
can lead to segmentation or other periodic structures. (After Newman, 1994.)

into account) reactor-diffusion system (see Chapters 7 and 8). Here
is where the chemically excitable nature of cells comes into play.
In our discussion of the Keller transcription-factor network model
(Chapter 3, Figs. 3.5 and 3.6) we saw how dynamical multistability
could lead to the generation of alternative cell types. Single-cell or-
ganisms can use this capacity to function in environmental niches
with different biosynthetic demands, without undergoing genomic
evolution. But if any genetic change does occur that ˜˜locks” the cell
into one of the alternative states, the original uniform population
would simply break into subpopulations of distinct types of single-
cell organism. Once multicellularity arose, however, there was the
possibility that the alternative cell types would be present simultane-
ously in the same organism. This would have created a premium on
retaining the plasticity (i.e., the condition-dependent reversibility) of
cell-type switching and of utilizing physical mechanisms such as dif-
fusion for reliably determining where in the organism the different
cell types arise (see Chapter 7, Fig. 7.7). Experiments with ˜˜synthetic”
multicellular systems consisting of colonies of bacteria with genet-
ically engineered biosynthetic and signalling networks demonstrate
that providing cells with the simple circuitry described above indeed
enables them to reliably produce spatiotemporal patterns of cell states
(Basu et al., 2005).
The use of a diffusible signaling molecule with an externally de-
termined gradient as the sole cue causing switching between alter-
native cell types is a simple, hierarchical, mechanism of pattern for-
mation; ˜˜hierarchical” here implies a unidirectional, ˜˜feed-forward”,
determination in the way that the individual cells acquire their iden-
tity, without any feedback on the cue or each other. An even simpler
mechanism is the formation of an intracellular gradient by the sedi-
mentation of dense cytoplasmic materials. Once division occurs, the
progeny will be biochemically distinct (Fig. 10.1). If the responding
cells or nuclei in a syncytium (see below) themselves participate in the
formation of the gradient, pattern formation can be ˜˜emergent,” that
is, the pattern will arise from self-organizing dynamics (e.g., reaction--
diffusion, or isologous diversi¬cation).
Oscillation in chemical composition is another physical process
that took on novel functions in the multicellular context. The cell
division cycle is a temporally periodic process driven by chemical
oscillation (see Chapter 3). In a world of free-living cells it leads to
generation of more of the same; it has no special morphological conse-
quence. Even in a multicellular aggregate the division cycle typically
acts only to increase the number of cells. But, as we saw in Chapter 7,
in a multicellular aggregate that contains a ˜˜gating” chemical gradi-
ent (Dubrulle et al., 2001) or an additional biochemical oscillation
with a period different from the cell cycle (Newman, 1993; Holtzen-
dorff et al., 2004; Fig. 7.1), spatially periodic populations of cells with
distinct recurrent states can be generated. The developmental conse-
quences of such oscillatory mechanisms must have appeared early in
the history of multicellularity (Fig. 10.1).

How, then, could patterns of cell state or type resulting from dif-
fusion gradients and the excitable properties of cell aggregates have
led to the evolution of new organismal shapes and body plans? It is
likely that they did so by mobilizing another simple generic property
of cell aggregates -- cell adhesion. This, of course, is the de¬ning con-
dition of multicellularity. That is to say, organisms became multicel-
lular by evolving cell-surface molecules that either prevented them
from separating after division or caused individual cells to aggregate
(Bonner, 1998). Modern organisms have multiple highly evolved gene-
network-based regulatory mechanisms devoted to controlling the pre-
cise strength of intercellular adhesion (Buckley et al., 1998; Braga,
2002). In contrast, the earliest cell aggregates would have been ama-
teurs in the regulation of cell--cell interactions: cell stickiness is likely
to have been less stringently regulated in the earliest metazoa than
it is at present.
Just as organizer cells secreting a proto-morphogen could have
arisen randomly in primitive cell aggregates, it is reasonable to envi-
sion that some cells with distinct adhesive properties -- either more
or less adhesive than the parental cells -- could have emerged ran-
domly in these ancient multicellular masses. We have just seen how
the existence of a proto-morphogen-secreting cell has developmen-
tal consequences in a tissue mass of a certain scale, as a result of
the physical process of diffusion. In a similar fashion, differential ad-
hesion within cell mixtures, even if randomly arising, mobilizes its
own physically based morphogenetic effects. When cell types with
different amounts of adhesion molecules on their surfaces are mixed
together they will sort out into islands of more cohesive cells in a
sea composed of their less cohesive neighbors (as discussed in de-
tail in Chapter 4). Eventually, by random cell movement, the islands
coalesce and an interface is established across which cells do not in-
termix (Steinberg, 1998; Fig. 10.1). In other words, the presence of
two or more differentially adhesive cell populations within the same
tissue mass immediately establishes the conditions for the formation
of multiple layers or ˜˜compartments,” i.e., distinct spatial domains
in which no interchange or mixing of cells occurs across common
boundaries (as shown in Chapter 5).
Although compartment boundaries in the developing embryos of
modern organisms are typically allocated with precision by spatially
distributed cues based on juxtacrine signaling (e.g., the Notch--Delta
couple) or paracrine-type pattern-forming mechanisms (Chapter 7),
even a random assignment of cells to distinct adhesive states can
result in a compartmentalized tissue. The reason is that the sorting-
out process will eventually bring the cells of similar adhesive state
to one side or the other of a common boundary. In this fashion, our
hypothetical ancient cell cluster could have undergone a primitive
form of gastrulation, achieved by physical means (Fig. 10.1).
Now recall the discussion above regarding the proto-organizer and
various physically instigated pattern-forming processes. If the alter-
native cell states in that example were adhesive states, this would

provide a way to regulate the generation, number, and position of
differentially adhesive cells in a reliable, rather than a random, fash-
ion. The linking, as an outcome of gene evolution, of distinct physical
mechanisms -- differential adhesion, chemical gradient formation (by
sedimentation, diffusion, or reaction--diffusion), and chemical oscil-
lation -- would thus produce authentic developmental mechanisms
for body-plan generation (Fig. 10.1).
The previous discussion implies that long before complex, modern-
type, body morphologies emerged in the course of evolution, simpler
forms resembling gastrulae and budding and segmented tubes (i.e.,
the forms that can be generated by relatively simple physical mech-
anisms acting on cell aggregates) should have arisen. Despite their
appearance, these would not have been embryonic stages of more
complex organisms, but rather the morphologically most advanced
organisms of their time. It is signi¬cant, therefore, that such struc-
tures have been found in Ediacaran-period sediments dating from
∼ 600 Ma unaccompanied by the ˜˜adult” metazoan forms character-
istic of the later Cambrian period (Xiao and Knoll, 2000; Xiao et al.,
2000; Chen, J. -Y., et al., 2004).

Analyzing an evolutionary transition using physical
concepts: segmentation in insects
Segmentation is a major puzzle for those who study the evolution of
developmental mechanisms (Minelli and Fusco, 2004). For example,
evolutionarily related organisms such as beetles (˜˜short-germ-band”
insects) and fruit ¬‚ies (˜˜long-germ-band” insects) have apparently dif-
ferent modes of segment formation. Similarly to somitogenesis in
vertebrates, in short-germ-band insects (Patel et al., 1989) as well as in
other arthropods such as the horseshoe crab (Itow, 1986), segmental
primordia are added in sequence from a zone of cell proliferation
(a ˜˜growth zone;” Fig. 10.2). In contrast, in long-germ-band insects,
such as the fruit ¬‚y Drosophila, a series of chemical stripes (i.e., paral-
lel bands of high concentration of a molecule) forms in the embryo,
which at this stage is a syncytium, a large cell with a single cytoplas-
mic compartment containing about 6000 nuclei arranged in a single
layer on the inner surface of the plasma membrane.
The stripes of chemical concentration that form in the Drosophila
syncytial embryo are actually alternating evenly spaced bands of tran-
scription factors of the ˜˜pair-rule” class. When cellularization (the en-
closure of each nucleus and nearby cytoplasm in their own complete
plasma membrane) takes place shortly thereafter, the cells of the re-
sulting blastoderm will have periodically distributed identities that
are determined by the particular mix of transcription factors they
have incorporated. The different cell states, de¬ned in part by the
subsequent expression of genes of the ˜˜segment polarity” class, are
later transformed into states of differential adhesion (Irvine and Wi-
eschaus, 1994), and morphological segments form as a consequence

Drosophila melanogaster
Schistocerca gregaria





Fig. 10.2 Sequential and parallel segmentation modes in short and long germ-band
insects. On the left, the embryo of a short-germ-band insect, the grasshopper
Schistocerca gregaria, is shown schematically. In these insects, single segments or groups
of a few segments appear in succession. The brown stripes indicate the expression of a
segment-polarity gene such as engrailed. With further development, additional segments
appear sequentially from a zone of proliferation that remains posterior to the most
recently added segment. On the right, the embryo of a long-germ-band insect, the
fruit-¬‚y Drosophila melanogaster, is shown, again schematically. In these insects, gradients
of maternal gene products (e.g., bicoid and nanos) are present in the egg before
cellularization of the blastoderm (see the main text). As development proceeds, the
maternal gene products induce the expression of gap genes (e.g., hunchback, Kr¨ ppel ),
the products of which in turn induce the expression of pair-rule genes in a total of 14
stripes (e.g., eve, fushi tarazu, hairy). Each pair-rule gene is expressed in a subset of the
stripes; eve, for example, is expressed in seven alternating stripes, shown in green. The
pair-rule gene products provide a prepattern for alternating bands of nuclear expression
of segment polarity genes. Once cellularization has occurred (bottom right panel), these
are distributed in a similar way to that in short-germ-band embryos. (Schistocerca series
after Patel et al., 1994; Drosophila series after Ingham, 1988.)

(Fig. 10.2). Finally, the individual segments develop into recognizably
different structures under the in¬‚uence of combinations of Hox gene
products, a subset of the ˜˜homeobox” class of transcription factors.
All the members of this class of proteins contain the ˜˜homeodomain,”
a highly conserved DNA-binding protein motif (see Wilkins, 2002, for
a detailed discussion of the developmental functions and evolution
of these regulatory pathways).

So far, no cell-autonomous oscillations have been identi¬ed dur-
ing the segmentation of invertebrates, in contrast with the situation
in vertebrates. However, the sequential appearance of gene expres-
sion stripes from the posterior proliferative zone of short-germ-band
insects and other arthropods such as spiders and horseshoe crabs, has
led to the suggestion that these patterns arise from a segmentation
clock like that found to control vertebrate somitogenesis (Stollewerk
et al., 2003; see also the discussion in Chapter 7). This clock would
regulate, directly or indirectly, downstream genes such as engrailed
(en), a segment polarity gene that speci¬es a transcription factor, in
cells as they leave the proliferative zone (Fig. 10.2).
On both theoretical grounds (Boissonade et al., 1994; Muratov, 1997)
and experimental grounds (Lengyel and Epstein, 1992) it has long
been recognized that the kinetic properties that give rise to a chem-
ical oscillation (the Hopf bifurcation; see Chapter 3) can, when one
or more of the components is diffusible, also give rise to standing or
traveling spatial periodicities of chemical concentration (the Turing
bifurcation, see Chapter 7). If the Drosophila embryo were patterned
by a reaction--diffusion system, it would be tempting to hypothesize
that the Drosophila mode of segmentation arose by an evolutionary
change (i.e., delayed cellularization of the blastoderm) that placed
the ancestral oscillator in a context in which diffusion was possible.
Things are not this simple, however. In reality, Drosophila segmen-
tation is controlled by a hierarchical system of genetic interactions
that has little resemblance to the self-organizing pattern-forming sys-
tems associated with reaction--diffusion coupling. Only by examining
the Drosophila segmentation hierarchy in comparison with the means
used to achieve the same morphological outcome in short-germ-band
insects is it possible to formulate a plausible hypothesis for the mech-
anistic and evolutionary relationships between the two systems.

Molecular mechanisms of segmentation in insects
The formation of overt segments in Drosophila (see St Johnson and
Nusslein-Volhard, 1992, and Lawrence, 1992, for reviews) requires the
prior expression of a stripe of Engrailed in the cells of the posterior
border of each of 14 presumptive segments (Karr et al., 1989). By means
of a gene network (the segment polarity module), cells that express en
cause cells that do not acquire a distinct adhesive af¬nity, leading to
a boundary of cell immiscibility (Rodriguez and Basler, 1997).
The positions of the engrailed stripes are largely determined by
the activity of the pair-rule genes even-skipped (eve) and fushi tarazu (ftz),
which exhibit alternating, complementary, seven-stripe patterns prior
to the formation of the blastoderm (Frasch and Levine, 1987; Howard
and Ingham, 1986). Even-skipped, for example, as its name suggests, is
expressed in only odd-numbered ˜˜parasegments.” (The parasegments
are developmental modules consisting of the posterior half of each


. 50
( 66 .)