. 32
( 66 .)


ily (TGF-βs, activin, Nodal, BMPs, GDFs), or the ¬broblast growth factor
FGF family (FGF-2, -3, -4, etc.). In Chapter 1 we discussed the reasons
why free diffusion inside cells can only occur on scales that are small
compared with the cell™s linear dimensions. The situation is quite dif-
ferent, however, in multicellular embryos and in the cell aggregates

that form organs. Morphogens can diffuse in the extracellular space
and form gradients by interacting with cell surface receptors. The
best-documented case of the use of this mechanism is the induction
of mesoderm by Xenopus Xnr factors, mentioned in the introduction
to this chapter. Morphogens can also spread by cell relay mechanisms.
Here the effect of the morphogen spreads as if by diffusion, but what
is actually propagating across the domain is an activity passed from
cell to cell by signaling rather than by any substance (Reilly and
Melton, 1996).
In addition, morphogens can act on cells in an autocrine fashion
(i.e., on the same cells that produce them). If the morphogen happens
also to stimulate its own production (as is the case with the TGF-
βs; Van Obberghen-Schilling et al., 1988), a positively autoregulatory
chemical reaction loop will result. Such positive feedback situations
become ˜˜explosive” unless something is present to curtail them. In
any realistic biological system, therefore, an ˜˜inhibitory” morphogen,
i.e., one that suppresses the production of the ˜˜activating” or pos-
itively autoregulatory morphogen, will also be present. In his 1952
paper in which he introduced the term ˜˜morphogen,” Turing consid-
ered the situation when an activating morphogen also induces the
production of the inhibiting morphogen. This implies that the acti-
vator and inhibitor both emanate from the same sites. Turing then
assumed that the two molecules diffused away from their (common)
sites of production at different rates. Such ˜˜reaction--diffusion” sys-
tems (which could be purely chemical, as well as biological) were
demonstrated mathematically to have inherent pattern-forming ca-
pability (Turing, 1952).
With certain parameter choices reaction--diffusion systems can
thus produce morphogen patterns that are signi¬cantly more com-
plex than the monotonic gradients that would result from simple
morphogen diffusion (Turing, 1952; Harrison, 1993; Meinhardt and
Gierer, 2000). The cell patterns induced by these morphogen pro¬les
will then be correspondingly complex. It has been suggested by vari-
ous workers that reaction--diffusion mechanisms underlie the gen-
eration of planar cell polarity (discussed in Chapter 5 in relation
to convergent extension; Amonlirdviman et al., 2005), axis forma-
tion and the generation of left--right asymmetry in the early embryo
(Meinhardt, 2001; see below), the formation of pigment stripes in the
skin of ¬sh (Kondo and Asai, 1995), feather bud formation in the skin
of birds ( Jiang, T., et al., 1999), and skeletal pattern formation in the
vertebrate limb (Newman and Frisch, 1979; Miura and Shiota, 2000a;
Hentschel et al., 2004; see Chapter 8).
Inductive pattern-forming mechanisms can themselves be subdi-
vided into two broad categories, hierarchical and emergent (Salazar-
Ciudad et al., 2000, 2001a, b) (Fig. 7.1B). In hierarchical patterning
mechanisms, transmission of the inductive signal, whether juxtacrine
or paracrine, is unidirectional. In emergent patterning mechanisms,
there are reciprocal inductive interactions between cells. Both types of
mechanism are prevalent during development; indeed, most episodes
of pattern formation during embryogenesis contain hierarchical and

emergent steps. In Chapter 10 we discuss the implication of these
different mechanisms for the evolution of embryonic pattern.
In the remainder of this chapter we will describe speci¬c biologi-
cal examples of, and physical models for, developmental patterning by
cell-autonomous and juxtacrine mechanisms and by autocrine and/or
paracrine gradient and reaction--diffusion mechanisms. The develop-
mental events we will consider are segmentation, the formation of
periodically arranged, clustered, epithelial-cell subpopulations, spa-
tially regulated mesoderm induction, and axis formation and the as-
sociated generation of left--right asymmetry. These events all occur
during early development but not in every species and not always
in the same order. Mesoderm induction occurs prior to gastrulation
in all triploblasts (organisms with three germ layers, see Chapter 5)
whereas segments or epithelial clusters may form both before and
after gastrulation, depending on the species. Axis formation and the
generation of left--right asymmetry are postgastrulation determinants
of form, but the sequence of these events differs in different species.
Here we will follow an order of presentation that is motivated by the
logic of the physical mechanisms involved.

A wide variety of animal types, ranging across groups as diverse as
insects, annelids (e.g., earthworms), and vertebrates, undergo segmen-
tation early in development, whereby the embryo, or a major portion
of it, becomes subdivided into a series of tissue modules. These mod-
ules typically appear similar to each other when initially formed;
later they may follow distinct developmental fates and the original
segmental organization can be all but obscured in the adult form.
Somite formation (or ˜˜somitogenesis”) is a segmentation process in
vertebrate embryos in which the presomitic ˜˜paraxial” mesoderm (i.e.,
the mesoderm directly to either side of the notochord) becomes or-
ganized into parallel blocks of tissue. The somites are transient struc-
tures that eventually give rise to mature tissues and structures, some
of which retain the segmental or modular organization (the verte-
brae -- hollow cylinders of bone that surround and protect the spinal
cord -- and the ribs), and some of which do not (the dermis of the
dorsal skin, the muscles of the back, body wall, and limbs) (Gossler
and Hrabe de Angelis, 1998; Keller, 2000; Pourqui©, 2001).
Somitogenesis takes place in a sequential fashion. The ¬rst somite
begins forming as a mesenchymal condensation (see Chapter 6) in the
anterior region of the trunk (i.e., the main body region). Each new
somite forms just posterior to the previous one, budding off from
the rostral (towards the nose) portion of the unsegmented paraxial
mesoderm (Fig. 7.2). As they mature, the somites epithelialize, that is,
become epithelioid tissues, the sclerotome, which forms the vertebrae
and the ribs, and the dermamyotome, which forms the dermis and the
muscles. Eventually, 50 (chick), 65 (mouse), or as many as 500 (certain
snakes) of these segments will form.

Neural fold
Cut edge
plate of amnion



10 days
20 days
22 days

Fig. 7.2 Somitogenesis in the human embryo, looking down on the dorsal surface. On
the left, the 19-day embryo. Gastrulation has advanced, but the primitive streak is still
visible at the posterior end of the embryo (see Fig. 6.1 for a different view of a chicken
embryo, which has similar features at this early stage). The neural plate and neural
groove have formed but no somites have emerged by this stage, which is roughly
equivalent to that shown in Fig. 5.13B. In the center, the 20-day embryo. The amnion is a
¬‚uid-¬lled membrane, composed of embryo-derived cells, that forms a protective
covering over a portion of the (mammalian and avian) embryo. Three pairs of somites
have formed, beginning at the anterior end of the future trunk. The neural folds have
begun to converge (see Fig. 5.13C). On the right, the 22-day embryo. Several additional
pairs of somites have formed, each new pair having been added posterior to the previous
one. The neural folds have begun to fuse, as have the paired heart tubes (see Fig. 8.1),
which reside within the pericardial bulge. (After Langman, 1981.)

In the late nineteenth century the biologist William Bateson spec-
ulated that the formation of repetitive blocks of tissue, such as
the somites of vertebrates or the segments of earthworms, might
be produced by an oscillatory process inherent to developing tis-
sues (Bateson, 1894). More recently, Pourqui© and coworkers made
the signi¬cant observation that c-hairy1, an avian homologue of the
Drosophila gene hairy, is expressed in the paraxial mesoderm of avian
embryos in cyclic waves whose temporal periodicity corresponds to
the formation time of one somite (Palmeirim et al., 1997; Pourqui©,
2003). The c-hairy1 mRNA and its protein product, a transcription fac-
tor, are expressed in a temporally periodic fashion in individual cells,
but since the phase of the oscillator is different at different points
along the embryo™s axis, the areas of maximal expression sweep along
the axis in a periodic fashion (Fig. 7.3).
Other studies indicate that somite boundaries form when cells
that have left a posterior growth zone move suf¬ciently far from
a source of FGF8 in the tailbud (the posterior tip of the embryo)

0 min 90 min 180 min




Fig. 7.3 The model proposed by Pourqui© for segment formation in vertebrates, based
on mouse and chick data. A gradient of FGF8 (see the main text), shown in black,
regresses posteriorly during somitogenesis. The anterior boundary of the gradient
de¬nes the determination front, which corresponds to the position of the wavefront
(the thick black line). (A coordinately expressed gradient of Wnt3A plays a similar role;
Aulehla et al., 2003). The oscillatory (i.e., waxing and waning with the developmental
stage) expression of c-hairy1 and related genes is shown in red. The expression of genes
of the Mesp family, which encode transcription factors involved in somite boundary
formation, is shown in blue-green. (Reprinted, with permission, from Pourqui©, 2003.)

(Dubrulle et al., 2001). The FGF8 gradient (along with the coordinated
activity of Wnt3A; Aulehla et al., 2003) thus acts as a ˜˜gate,” which,
when its low end coincides with a particular phase of the segment-
ation clock, results in the formation of a boundary (Pourqui©, 2003)
(Fig. 7.3). The general features of this mechanism, called the ˜˜clock
and wavefront” model, were predicted two decades before there was
any direct evidence for a somitic oscillator (Cooke and Zeeman, 1976).

The Lewis model of somitogenesis
In all the vertebrates studied, certain genes are differentially ex-
pressed in the anterior and posterior portions of each somitomere,
the domains of tissue in the presomitic mesoderm (PSM) that will
eventually give rise to the somites. These always include genes of the
Notch--Delta pathway or their modulators ( Jen et al., 1999; Holley et al.,
2000, 2002; Oates and Ho, 2002; Dale et al., 2003). As noted earlier in
this chapter, the Notch--Delta system is the most common form of jux-
tacrine signaling during early development. Notch is a surface-bound
receptor that transmits signals received from outside the cell into the
cell™s interior. Members of the Delta family of proteins are ligands of
Notch. Delta is not secreted, but remains bound to the cells that pro-
duce it. The Notch--Delta interaction is just the ¬rst step in the gener-
ation of a patterned arrangement of cells; the downstream effects of
Notch activation or lack of it include transcriptional alterations that
lead to new cell types (Artavanis-Tsakonas et al., 1999) (see Chapter 3).

In the mouse ( Jouve et al., 2002) and in zebra¬sh ( Jiang et al., 2000),
mutations that inactivate the Notch pathway disrupt somitogenesis.
During somitogenesis in zebra¬sh the Notch ligand DeltaC oscil-
lates in the PSM with the same period as a pair of transcription fac-
tors, encoded by the genes her1 and her7, which are related to chicken
c-hairy1 (see above). Lewis (2003) has devised a model for somitogene-
sis in the zebra¬sh based on a mechanism for biochemical oscillation
proposed independently by himself and Monk (2003). Lewis hypoth-
esized that her1 and her7 constitute an autoregulatory transcription
factor of the sort discussed in Chapter 3 and, furthermore, that DeltaC
is a downstream effector of this oscillation. The two genes negatively
regulate their own expression (Holley et al., 2002; Oates and Ho, 2002)
and are positively regulated by Notch signaling (Takke and Campos-
Ortega, 1999; Oates and Ho, 2002). Even though, in mutants in which
Notch signaling is disrupted, somitogenesis breaks down (Holley et al.,
2002), the resulting pattern of deltaC expression in the PSM is not what
would be expected if the oscillator had halted in all the cells. Rather,
there is a salt and pepper distribution of DeltaC, suggesting that the
oscillator has become asynchronous. This has led Lewis to propose
that juxtacrine signaling by the Notch pathway, usually considered
to act in the determination of cell fate (see below), in this case acts
to keep cells in the segment-generating growth zone in synchrony
(Lewis, 2003). Such synchrony is an experimental ¬nding (Stern and
Bellairs, 1984; Primmett et al., 1989) that enters into most models of
Lewis provided a simple mechanism for the oscillatory expression
of the her1 and her7 genes, which we brie¬‚y summarize here. The
model is based on the assumption that there exists a feedback loop in
which the Her1 and Her7 proteins bind directly to the regulatory DNA
of their own genes to inhibit transcription. Also incorporated into the
model is the recognition that there is always a delay T m between the
initiation of transcription and the initiation of translation (since it
takes time for the mRNA molecule to translocate into the cytoplasm),
as well as a delay T p between the initiation of translation and the
emergence of a complete functional protein molecule.
These ingredients are put into mathematical language in the fol-


. 32
( 66 .)