. 51
( 66 .)


segment and the anterior half of the next segment posterior to it;
see Lawrence, 1992, and Fig. 10.3). The stripe patterns of the pair-
rule genes are generated by a complex set of interactions among

Fig. 10.3 (A) Schematic representation of a portion of the even-skipped gene,
including the promoter and transcription start site (cf. Fig. 3.9). Contained within the
promoter is a subregion (the stripe-2 enhancer) responsible for the expression of the
second eve stripe (see Fig. 10.2). This sequence contains binding sites for gap-gene-class
transcription factors that positively (Hunchback, red) or negatively (Giant, blue; Kr¨ ppel,
yellow) regulate eve expression. (B) Lower part: the Drosophila syncytial blastula. The
upper part shows, schematically the distribution of Giant, Hunchback, and Kr¨ ppel
proteins in the region containing the ¬rst six prospective parasegments. At the position
of prospective parasegment 3 (which comprises the posterior half of the last head
segment plus the anterior half of the ¬rst thoracic “ upper body “ segment) the levels of
the activator Hunchback is high and those of the inhibitors Giant and Kr¨ ppel are low.
This induces eve stripe 2. To either side of this position Giant and Kr¨ ppel are high,
restricting the expression of eve to a narrow band. (Panel A is based on Small et al., 1992;
panel B is after Wolpert, 2002.)

transcription factors in the syncytium that encompasses the entire
embryo. A simple reaction--diffusion model would suggest that all
the stripes (which are quite similar-looking, Fig. 10.2) are produced by
chemically identical conditions emerging at evenly spaced locations.
Instead, individual stripes are generated by stripe-speci¬c, rather than
generic, mechanisms (Akam, 1989). The formation of eve stripe num-
ber 2, for example, requires the existence of sequences in the eve
promotor that switch on the eve gene in response to a set of spatially
distributed morphogens that under normal circumstances have the
requisite values only at the stripe 2 position (Small et al., 1991, 1992,

1996) (Fig. 10.3). In particular, these promoter sequences respond to
speci¬c combinations of ˜˜gap” gene products (e.g., Giant, Kr¨ ppel, em-
bryonic Hunchback), a set of nonuniformly distributed transcription
factors that act as activators and competitive repressors of the pair-
rule gene promoters (Clyde et al., 2003; see also the discussion of the
Keller model in Chapter 3). The patterned expression of the gap genes,
in turn, is controlled by particular combinations of ˜˜maternal” gene
products (e.g., the transcription factor Bicoid and the RNA-binding
protein Staufen), which are distributed as gradients along the em-
bryo at even earlier stages (Fig. 10.2). As the category name suggests,
the maternal gene products are deposited in the egg during oogeneis.
Bicoid, interestingly, is in evolutionary terms a recently acquired
gene in ¬‚ies (Stauber et al., 1999). Its graded product acts in a
concentration-dependent fashion to activate the embryos™s hunchback
gene. But Bicoid is not essential for performing this function. In its
absence, an ancient regulatory circuit involving maternal Hunchback
protein can properly activate embryonic hunchback (Wimmer et al., 2000).
The expression of pair-rule genes and engrailed during the devel-
opment of arthropods other than Drosophila has also been explored.
In the grasshopper, Schistocerca, a short-germ-band insect, Engrailed is
localized in stripes marking the borders of segments (Patel et al., 1989,
1992; Fig. 10.2), although no pair-rule genes have been found to be
expressed in stripes (Dawes et al., 1994). In Tribolium, an ˜˜intermediate-
germ-band” beetle (having both a growth zone and a syncytium in
different regions of the embryo), the pair-rule genes eve and ftz are
expressed in a striped pattern in both syncytial and growth zone
regions and this is followed by the expression of en along the pos-
terior margin of each segment (Brown et al., 1994a, b; see also Patel
et al., 1994). While the modes of segmentation may have changed
over the course of evolution, the expression patterns of the segment-
polarity gene engrailed, and to a lesser extent, the pair-rule genes,
appear to be conserved. Based on the conservation of these gene ex-
pression patterns throughout the insects, and their similarities with
those found in other groups, it seems reasonable to assume that the
last common ancestor of Drosophila and modern intermediate-germ-
band insects was itself intermediate germ-band and probably had a
pattern of pair-rule and en expression in the syncytial region of its
embryo similar to that found in present-day Drosophila.

Physical mechanisms and the evolution
of insect segmentation
As noted above, the kinetic properties that give rise to a limit-cycle
chemical oscillation can also give rise to standing or travelling spa-
tial periodicities of chemical concentration, when one or more of
the components is diffusible. Whether a system of this sort exhibits
purely temporal, or spatial, or spatiotemporal periodicity depends on
particular ratios of reaction and diffusion coef¬cients. A simple dy-
namical system that exhibits temporal oscillation or standing waves,
depending on whether diffusion is permitted, is shown in Fig. 10.4.


‚g1 ‚ 2g
w13 g3
dg1 w13 g3 = ’ µ1g1 + D1 21
= ’ µ1g1
‚t w13 g3 + kM ‚x
dt w13 g3 + kM
‚g2 w 23 g3 ’ w 24 g4 ‚ 2 g2
w 23 g3 ’ w 24 g4
dg2 = ’ µ2 g2 + D2
= ’ µ 2 g2
‚t w 23 g3 ’ w 24 g4 + kM ‚x 2
dt w 23 g3 ’ w 24 g4 + kM
w 31g1 ’ w 34 g4 ‚g3 w 31g1 ’ w 34 g4 ‚ 2 g3
= ’ µ3 g3 = ’ µ3 g3 + D3
‚t w 31g1 ’ w 34 g4 + kM
dt w 31g1 ’ w 34 g4 + kM ‚x 2
w 42 g2 ’ w 43 g3 ‚g4 w 42 g2 ’ w 43 g3 ‚ 2 g4
dg4 g3
= ’ µ4 g4 = ’ µ4 g4 + D4
w 42 g2 ’ w 43 g3 + kM ‚t w 42 g2 ’ w 43 g3 + kM ‚x 2

0.3 0.25





0 0
1 21 41 61 81
3550 3525
3500 3600


Fig. 10.4 Example of a network (central box) that can produce, for the same
parameter values, sequential stripes when acting as an intracellular biochemical clock in a
one-dimensional cellularized blastoderm with a posterior proliferative zone, or
simultaneously-forming stripes when acting in a one-dimensional diffusion-permissive
syncytium. The arrows in the central box indicate positive regulation and the lines
terminating in circles indicate negative regulation. In the upper parts of the blue boxes
the equations governing each of the two behaviors are shown. The four genes involved in
the central network diagram, as well as their levels of expression in the equations, are
denoted by g1 , g2 , g3 , and g4 . In the reaction“diffusion case, g1 and g2 can diffuse
between nuclei (note that the two sets of equations differ only in the presence of
diffusion terms for the products of genes 1 and 2). The lower boxes indicate the levels of
expression of gene 2 for the two systems. On the left, for the intracellular clock, the
horizontal axis represents time t whereas on the right, in the reaction“diffusion system,
this axis represents the space variable x . The patterns produced by the two different
behaviors are not exactly equivalent because the reaction“diffusion system pattern has a
small dependency on the initial conditions. In the pattern shown, the initial condition
consisted of all gene product levels being set to zero except gene 1 in the central of 81
nuclei, which was assigned a small value (the exact quantity did not affect the pattern).
The patterns shown were found when the following parameter values were used: kM =
0.01; W13 = 0.179; W23 = 0.716; W24 = ’0.704; W31 = 0.551; W34 = ’0.466; W42 =
0.831; W43 = ’0.281; µ1 = 1.339; µ2 = 2.258; µ3 = 2.941; µ4 = 2.248. For the
reaction“diffusion case, the same parameter values were set but in addition the values
D1 = 0.656 and D2 = 0.718 were taken. (Redrawn from Salazar-Ciudad et al., 2001b.)

An important part of both these kinetic schemes is the presence of a
direct or indirect positive autoregulatory circuit. Indeed, modern-day
Drosophila contains the ingredients for this type of mechanism: several
of the pair-rule proteins (e.g., Eve, Ftz) diffuse within the syncytium
over short distances among the cell nuclei that synthesize their
mRNAs and positively regulate their own synthesis (Harding et al.,
1989; Ish-Horowicz et al., 1989; Schier and Gehring, 1993).
Based on the experimental ¬ndings, theoretical considerations,
and evolutionary inferences described above, the following set of
hypotheses has been suggested for the evolution of the mechanism
of Drosophila segmentation (Newman, 1993; Salazar-Ciudad et al.,
(i) Evolutionarily ancient insects generated segments by a mech-
anism that involved the regulation of pair-rule gene spa-
tiotemporal expression by a biochemical oscillator.
(ii) The evolutionary innovation that converted cellular embryos
to syncytial embryos in the ancestor of Drosophila led to the
biochemical oscillator becoming a reaction--diffusion system.
(iii) Over time, the reaction--diffusion system was superseded by a
mostly hierarchical control system in which gradients of ma-
ternal and gap-gene products came to specify the locations
of pair-rule gene expression. This hierarchical control is me-
diated by stripe-speci¬c promoters of the pair-rule genes that
arose by gene duplication in the course of Drosophila evolu-
To summarize this evolutionary hypothesis: the appearance of the
syncytial mode of embryogenesis converted the growth-dependent,
temporal, mechanism found in the more ancient short-germ-band in-
sects into the growth-independent simultaneous stripe-forming mech-
anism seen in the more recently evolved long-germ-band forms. While
suggesting an underlying physical connection between short-germ-
band segmentation and segmentation in the presumed ancestor of
long-germ-band forms, this hypothesis introduces a new puzzle of
its own: why does modern-day Drosophila not use a reaction--diffusion
mechanism to produce its segments?

The evolution of developmental robustness
Genes are always undergoing random mutation. Notions of the evolu-
tionary process that neglect the physical aspect might easily conclude
that morphological change should closely track genetic change, and
indeed this is the default assumption of neo-Darwinism, the most
widely held model of organismal evolution (see Mayr, 1982). If, how-
ever, as the previous chapters have proposed, the generation of biolo-
gical form and pattern is largely dependent on physical mechanisms
acting on tissues, it follows that genetic mutation might produce a

large, a small, or no change in form, depending on the case, suggest-
ing that the standard model might need to be revised (Newman and
M¨ ller, 2000).
Particularly interesting are those cases in which the outward form
of a body plan or organ does not change, but its genetic ˜˜under-
pinning” does. A useful analogy can be found in the phenomenon
of ˜˜pseudomorphism,” described by geologists (Klein and Hurlbut,
2002). This is a process whereby a mineral is formed according to the
inherent crystallization properties of its molecular constituents, and
then, over time, these constituents are gradually replaced by differ-
ent ones, the structure being maintained. Pseudomorphs are there-
fore ˜˜evolved” minerals exhibiting forms that their ¬nal constituents
could not have assumed by themselves. We saw in the earlier portions
of this chapter how the physical determination of form and pattern
may have been even more important at the earliest stages of meta-
zoan evolution than it is in present-day organisms. If true, this would
suggest that modern organisms, whose developmental mechanisms
have undoubtedly evolved but whose forms may re¬‚ect their origi-
nal physically based morphogenetic and patterning mechanisms, may
be considered as biological pseudomorphs (see Newman and M¨ ller, u
2000; Newman, 2003b).
A particular kind of natural selection, termed ˜˜canalizing selec-
tion” (Waddington, 1942; see also Schmalhausen, 1949), will preserve
those randomly acquired genetic alterations that happen to enhance
the reliability of a developmental process. For example, once a suc-
cessful morphological motif (e.g., gastrulation, neurulation, segmen-
tation) is established in a particular group of organisms, the develop-
mental mechanisms by which it is achieved could undergo selection
for canalizing mechanisms, i.e., molecular--cellular interactions that
reinforce a particular developmental pathway. Its development would
thereby become more complex at the molecular level but correspond-
ingly more resistant (˜˜robust”) to external perturbations or internal
noise that could disrupt non-reinforced physical mechanisms of de-
termination. Indeed, the patterns formed by reaction--diffusion sys-
tems are notoriously sensitive to temperature and domain size (Mein-
hardt, 1982; Ouyang and Swinney, 1991; Harrison, 1993; Boissonade
et al., 1994) and any developmental process that was solely dependent


. 51
( 66 .)