DOI: 10.1126/science.1085887
, 328 (2003);301Science
Olivier Pourquié
Pattern
The Segmentation Clock: Converting Embryonic Time into Spatial
This copy is for your personal, non-commercial use only.
clicking here.colleagues, clients, or customers by
, you can order high-quality copies for yourIf you wish to distribute this article to others
here.following the guidelines
can be obtained byPermission to republish or repurpose articles or portions of articles
):December 30, 2010www.sciencemag.org (this infomation is current as of
The following resources related to this article are available online at
http://www.sciencemag.org/content/301/5631/328.full.html
version of this article at:
including high-resolution figures, can be found in the onlineUpdated information and services,
http://www.sciencemag.org/content/301/5631/328.full.html#related
found at:
can berelated to this articleA list of selected additional articles on the Science Web sites
220 article(s) on the ISI Web of Sciencecited byThis article has been
http://www.sciencemag.org/content/301/5631/328.full.html#related-urls
30 articles hosted by HighWire Press; see:cited byThis article has been
http://www.sciencemag.org/cgi/collection/development
Development
subject collections:This article appears in the following
registered trademark of AAAS.
is aScience2003 by the American Association for the Advancement of Science; all rights reserved. The title
CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.
(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience
onDecember30,2010www.sciencemag.orgDownloadedfrom
tal stage 26. By stage 41, the eyes release melatonin
rhythmically, indicating that Xenopus embryos
develop functional photoresponsive circadian
clocks within the first few days of life. In
embryonic zebrafish, a circadian oscillator that
regulates melatonin synthesis becomes functional
and light-responsive between 20 and 26 hours
postfertilization (27). However, there is little
evidence yet that suggests these embryonic
clocks are gating key patterning events within
the spatial axes. Clocks that play that role are
more likely to have periods similar to those of
the somite clocks that exhibit a Hes1 gene–
dependent oscillation with a period of about 2
hours (28, 29). Recent studies showing that a
mouse strain that bears mutations at the
PER2 locus exhibits enhanced susceptibility to
cancer (30) may implicate a role for circadian
rhythmicity in cell proliferation, although it has
not been ruled out that PER2 may have a role
independent from circadian control that is yet to
be elucidated.
Circadian clocks have a well-defined role in
regulating physiological and behavioral events
on a 24-hour basis and have extended that role
into seasonal timing and photoperiodism. It
remains to be elucidated what early developmental
patterning events might be gated by
circadian clocks, although a major role here
seems unlikely given the relatively normal
morphology and lack of heterochronic events
exhibited by organisms that bear strong mutations
in their circadian systems.
References and Notes
1. M. J. Dowson-Day, A. J. Millar, Plant J. 17, 63 (1999).
2. D. Alabadi et al., Science 293, 880 (2001).
3. Z. Y. Wang, E. M. Tobin, Cell 93, 12077 (1998).
4. R. Schaffer et al., Cell 93, 1219 (1998).
5. P. Mas, D. Alabadi, M. J. Yanovsky, T. Oyama, S. A.
Kay, Plant Cell 15, 223 (2003).
6. D. E. Somers, A. A. Webb, M. Pearson, S. A. Kay,
Development 125, 485 (1998).
7. A. Millar, S. Kay, Proc. Natl. Acad. Sci. U.S.A. 93,
15491 (1996).
8. H. G. McWatters, R. M. Bastow, A. Hall, A. J. Millar,
Nature 408, 716 (2000).
9. X. L. Liu, M. F. Covington, C. Fankhauser, J. Chory,
D. R. Wagner, Plant Cell 13, 1293 (2001).
10. A. Mouradov, F. Cremer, G. Coupland, Plant Cell 14,
S111 (2002).
11. I. Kardailsky et al., Science 286, 1962 (1999).
12. A. Samach et al., Science 288, 1613 (2000).
13. P. Suarez-Lopez et al., Nature 410, 1116 (2001).
14. M. J. Yanovsky, S. A. Kay, Nature 419, 308 (2002).
15. R. S. Poethig, Science 301, 334 (2003).
16. M. Yano et al., Plant Cell 12, 2473 (2000).
17. S. Kojima et al., Plant Cell Physiol. 43, 1096 (2002).
18. Y. Takahashi, A. Shomura, T. Sasaki, M. Yano, Proc.
Natl. Acad. Sci. U.S.A. 98, 7922 (2001).
19. S. Sugano, C. Andronis, M. S. Ong, R. M. Green, E. M.
Tobin, Proc. Natl. Acad. Sci. U.S.A. 96, 12362 (1999).
20. B. D. Goldman, J. Biol. Rhythms 16, 283 (2001).
21. P. L. Lowrey et al., Science 288, 483 (2000).
22. K. Shimomura, D. E. Nelson, N. L. Ihara, M. Menaker,
J. Biol. Rhythms 12, 423 (1997).
23. D. S. Saunders, Invert. Neurosci. 3, 155 (1997).
24. E. Tauber, B. P. Kyriacou, J. Biol. Rhythms 16, 381
(2001).
25. L. M. Beaver et al., Proc. Natl. Acad. Sci. U.S.A. 99,
2134 (2002).
26. E. M. Myers, J. Yu, A. Sehgal, Curr. Biol. 13, 526
(2003).
27. N. Kazimi, G. M. Cahill, Brain Res. Dev. Brain Res. 117,
47 (1999).
28. O. Pourquie, Science 301, 328 (2003).
29. M. N. Nitabach, J. Blau, T. C. Holmes, Cell 109, 485
(2002).
30. M. N. Nitabach, J. Blau, Nat. Genet. 32, 559 (2002).
31. We thank S. Panda for Fig. 2 and C. Green and T.
Imaizumi for comments.
R E V I E W
The Segmentation Clock: Converting Embryonic
Time into Spatial Pattern
Olivier Pourquie´
In most animal species, the anteroposterior body axis is generated by the formation
of repeated structures called segments. In vertebrate segmentation, a specialized
mesodermal structure called the somite gives rise to skeletal muscles, vertebrae, and
some dermis. Formation of the somites is a rhythmic process that involves an
oscillator—the segmentation clock—driven by Wnt and Notch signaling. The clock
ticks in somite precursors and halts when they reach a specific maturation stage
defined as the wavefront, established by fibroblast growth factor and Wnt signaling.
This process converts the temporal oscillations into the periodic spatial pattern of
somite boundaries. The study of somite development provides insights into the
spatiotemporal integration of signaling systems in the vertebrate embryo.
In most vertebrate species, somites appear as
epithelial blocks of paraxial mesoderm cells
forming synchronously on both sides of the
body axis from the mesenchymal presomitic
mesoderm (PSM). The rhythm of somite production
is characteristic of the species at a given
temperature (90 min in the chick embryo at
37°C and 20 min for the zebrafish embryo at
25°C). The total number of somites is constant
within a given species. It is usually about 50,
although in some animals, such as snakes, it can
reach up to 400. The process of embryonic
segmentation is now well understood at the
molecular level in Drosophila, in which segments
form simultaneously in the syncytial embryo.
In most invertebrate and vertebrate species,
segmentation takes place sequentially from
head to tail in a cellularized embryo and accompanies
the progressive formation of the body
axis. Accordingly, the genetic networks at play
in fly segmentation do not appear to be conserved
in vertebrates, and it remains unclear
whether segmentation arose independently in
invertebrates and vertebrates.
The Segmentation Clock: A Molecular
Oscillator Underlying Vertebrate
Segmentation
In vertebrates, segmentation involves a molecular
oscillator—the segmentation clock—which
acts in the PSM (1). Evidence for this oscillator
was first provided by the observation of the
regular pulses of expression in PSM cells of the
mRNA coding for the basic helix-loop-helix
(b-HLH) transcription factor c-hairy1, a vertebrate
homolog of the protein encoded by the fly
pair-rule gene hairy (1). c-hairy1 mRNA is expressed
as a wave sweeping across the whole
PSM once during each somite formation (Fig.
1A). This rhythmic expression begins during
gastrulation in the paraxial mesoderm precursors
of the primitive streak and their descendants
and is maintained throughout somitogenesis
(2). Existence of such an oscillator or clock,
whose role is to generate a temporal periodicity
that can be translated spatially into the periodic
boundaries of the somites, was originally proposed
in theoretical models such as the “clock
and wavefront” (3). Several additional genes,
referred to as cyclic genes, that exhibit a dynamic
behavior similar to that of c-hairy1 have now
been identified in fish, frog, chick, and mouse
embryos, suggesting that the segmentation clock
is conserved among vertebrates (4–12).
The best characterized set of cyclic genes is
involved in Notch signaling, suggesting that
Notch activation lies at the heart of the oscillator.
Such genes encode several transcription factors
of the Hairy and Enhancer of Split (HES) family,
acting downstream of Notch signaling (4, 7, 8,
10–12), as well as the glycosyl transferase LuStowers
Institute for Medical Research, 1000 East
50th Street, Kansas City, MO 64110, USA. E-mail:
olp@stowers-institute.org
D E V E L O P M E N T A L T I M I N G
18 JULY 2003 VOL 301 SCIENCE www.sciencemag.org328
SPECIALSECTION
onDecember30,2010www.sciencemag.orgDownloadedfrom
natic Fringe (5, 6) and the Notch ligand deltaC
(9). Their cycling behavior in the PSM is regulated
at the transcriptional level (13, 14). All
these genes oscillate largely in synchrony in the
PSM, suggesting that they are downstream of a
common cycling activator.
The mechanism driving the oscillations of
the cycling genes has been actively studied in
zebrafish, chick, and mouse. In fish, oscillations
of deltaC were proposed to drive the periodic
activation of Notch, leading to the cyclic expression
of the genes coding for the Hairy and Enhancer
of Split–Related (HER) b-HLH repressor
genes Her1 and Her7 (9, 15). HER1 and HER7
further play a central role in the oscillator by
establishing a negative feedback loop regulating
their own expression (15, 16). In the chick embryo,
lunatic fringe is also directly activated by
Notch signaling and is involved in a negative
feedback loop leading to Notch inhibition (17).
Given the unstable nature of the Lunatic Fringe
protein, which also oscillates, this inhibition is
transient and could participate in the establishment
of the cyclic expression pattern (17). In
contrast, the role of Lunatic Fringe in the control
of the oscillations in the mouse is currently
unclear (18). In the mouse, lunatic fringe is
regulated through promoter sequences termed
“clock elements” (13, 14), which contain CBF1
binding sites and E boxes. This suggests that the
cyclic genes are directly regulated by Notch and
by b-HLH proteins. Accordingly, in mouse mutant
embryos, disruption of Notch signaling prevents
the oscillations of the cyclic genes (7, 14).
The repressor HES7 appears to play a critical
role in the control of the oscillations in the
mouse, because its mutation prevents the oscillations
of lunatic fringe (8). This gene was
shown to act downstream of Notch and to bind
and negatively regulate its own promoter, suggesting
that like the her genes in zebrafish, it
could establish a negative feedback loop participating
in the control of the oscillations (8).
Another mouse cyclic gene, hes1, exhibits the
required characteristics to implement such a
loop: It acts downstream of Notch and codes for
an unstable protein able to negatively regulate its
own promoter (7, 19). However, hes1 is not
essential for cycling, because hes1 mutants do
not exhibit a segmentation phenotype (7).
Therefore, it seems that in all species examined
thus far, Notch activation lies at the heart of
the oscillator. Notch plays a critical role in the
control of the oscillations by directly activating
the cyclic genes, thus accounting for their synchronous
expression. Notch activation drives the
expression of transcriptional repressors of the
HES family, which in turn negatively regulate
their own expression and that of the other cyclic
genes (Fig. 1B). This negative feedback loop
model requires that the HER proteins be highly
unstable and that their repressive effect be dominant
over Notch activation. Additional regulatory
loops aimed at establishing periodic Notch
activation, such as the Lunatic Fringe–based
loop identified in the chick, act in concert with
this mechanism (17). A role for Notch signaling
in the control of the spatiotemporal coordination
of the cyclic genes’ expression among PSM cells
was also proposed in zebrafish (9).
A second group of cyclic genes linked to the
Wnt signaling pathway has recently been uncovered.
Thus far, only one cycling gene in this
class has been identified: the inhibitor of Wnt
signaling Axin2 (20). In the mouse, axin2 is
expressed in a dynamic sequence similar to, but
out of phase with, that of the Notch-related
cyclic genes (Fig. 1A). axin2 is directly regulated
by Wnt signaling and could participate
in the establishment of an autoregulatory
negative feedback loop involved in its periodic
expression. axin2 oscillations persist in
Notch pathway mutants, whereas both axin2
and lunatic fringe oscillations are disrupted
in wnt3a mutants, indicating that Wnt signaling
acts upstream of the Notch-regulated
cyclic genes (Fig. 1B). Therefore, in the
mouse, the segmentation clock appears to
be composed of a Wnt-based regulatory
loop entraining a series of Notch-based
loops (Fig. 1B). The details of the interactions
between these different loops are presently
not understood. Also, the conservation
of the Wnt-based loop across vertebrates
remains to be examined.
The exact role of this oscillator in the segmentation
process remains unclear. The clock
might serve to implement periodic activation of
Notch signaling in the anterior PSM, which was
shown to be required for the establishment of
somite boundaries and subcompartments (21,
22). Also, the bilateral desynchronization of
somitic boundaries that accompanies the loss of
oscillations of the cyclic genes in Notch pathway
mutant embryos suggests that the clock
could play a role in coordinating the timing of
boundary production during development.
Whether the clock plays a role in the initial
establishment of the segmental pattern remains
to be demonstrated.
No other examples of oscillations involving
such a period range have been reported in developmental
systems. Recently, however, oscillations
of the mouse hes1 gene expression, with
a periodicity similar to that seen in segmentation,
were triggered in vitro by applying a serum
shock to various cultured mouse cell lines
Fig. 1. (A) Expression sequence of lunatic fringe/c-hairy1 (red) and axin2 (blue) mRNAs during the
formation of two somites. (B) Schematic representation of the segmentation clock oscillator in a
PSM cell, integrating mouse and chick data. The Notch-based loop is shown in red; the Wnt-based
loop is in blue. Cyclic genes are boxed. Mb, membrane.
D E V E L O P M E N T A L T I M I N G
www.sciencemag.org SCIENCE VOL 301 18 JULY 2003 329
SPECIALSECTIONonDecember30,2010www.sciencemag.orgDownloadedfrom
(19). These oscillations appear to be mediated
by a periodic repression of hes1 on its own
promoter. These observations were carried out
in cells that are not related to the paraxial
mesoderm, thus raising the possibility that the
segmentation clock ticks in cells other than the
PSM cells. This oscillatory behavior could
reflect a more generic property of gene expression,
perhaps related to transcription. The oscillations,
which are relatively easy to see in segmentation,
where they are well coordinated in
all cells, might also take place in an uncoordinated
manner in other cell types but would be
undetectable given our current methods of analysis.
However, thus far, no gene oscillations
similar to those seen in PSM segmentation have
been observed in other tissues.
FGF Signaling: Translation of the Clock
Pulsation into Spatial Periodicity
Recent studies have shown that the secreted
growth factor FGF8 (fibroblast growth factor
8) could be implicated in converting the clock
pulsation into the periodic arrangement of
segment boundaries (23, 24). fgf8 mRNA is
strongly expressed in PSM precursors in the
primitive streak and tail bud as well as in the
posteriormost PSM, and its expression progressively
decreases in more anterior cells,
thus establishing a gradient over two-thirds of
the PSM length (Fig. 2, black). Its expression
domain correlates with the region of the PSM
in which the segmental pattern is not yet
irreversibly determined (23). Overexpression
of FGF8 in PSM cells can maintain their
posterior identity and block segmentation,
suggesting that high concentrations of FGF8
are required to actively maintain newly
formed posterior PSM cells in an immature
state. It was proposed that, because of the
progressive decrease of fgf8 expression during
maturation of the PSM, when cells become
located in the anterior PSM, they reach
a threshold of FGF signaling allowing them
to activate their segmentation program. This
threshold level, which was termed the “determination
front” (Fig. 2), marks a transition in
genetic regulation in PSM cells, as shown by the
activation of new sets of genes such as paraxis,
the down-regulation of posterior genes such as
Brachyury, and the slowing down and stopping
of the oscillations of the clock genes (21, 23,
24). Wnt3A was also recently proposed to assume
a role similar to that of FGF8 by establishing
a gradient-controlling segmentation in the
PSM (20). However, because Wnt3A acts upstream
of fgf8 in the PSM, it could act together
with or by way of the fgf8 gradient.
At the determination-front level, gene coding
for the transcription factor Mesp2/c-meso1 becomes
periodically activated in a one-somitewide
domain, providing the earliest evidence for
segmentation in the PSM (Fig. 2, blue) (21). The
transcription factors of the Mesp family were
shown to act upstream of a genetic cascade
involving the Notch pathway, which ultimately
results in boundary positioning and formation of
anterior and posterior somitic compartments (21,
22). Therefore, it is tempting to speculate that the
periodic activation of genes of the Mesp family,
which takes place at the determination-front level,
is controlled by the segmentation clock. This
would provide a link between the segmentation
clock, the determination front, and the boundaryformation
process.
As a result of the constant posterior elongation
of the body axis during early development,
the fgf8 gradient and wnt3a expression
are continuously displaced posteriorly, and
thus the absolute position of the determination
front constantly recedes (Fig. 2). This
movement of the determination front ensures
that boundaries will be separated by a distance
corresponding to the posterior displacement
of the determination front during one
period of the oscillation. Accordingly, altering
the dynamics of the fgf8 gradient affects
the position of somite boundaries (23, 24).
Therefore, the determination front corresponds
to the wavefront of the clock and
wavefront model. In the model, when cells of
the PSM reach the wavefront level, the phase
of their oscillation is definitively
imprinted on these cells,
and the oscillation stops (3).
These coordinated processes
will result in the production
of a series of repeated
somites, which subsequently
differentiate into vertebrae
exhibiting different
morphologies depending on
their position along the anteroposterior
axis. The developmental
program defining
the particular shape of the
vertebrae is controlled by the
Hox genes in the somites.
Regulation of the Hox genes
is discussed in the review by
Kmita and Duboule in this
issue (25). The segmentation clock also controls
aspects of spatiotemporal Hox gene activation,
thus ensuring a perfect match between segment
boundary position and future regional identity of
the somites (23).
Although a number of components of the
segmentation clock have been identified, their detailed
interactions resulting in the generation of the
oscillations of the cyclic genes remain to be elucidated.
Also, investigation into the existence of
such a mechanism in invertebrate species that
exhibit a progressive mode of segmentation, such
as annelid worms, should tell us whether this
oscillator was already operating in the ancestor of
vertebrates and invertebrates.
References and Notes
1. I. Palmeirim, D. Henrique, D. Ish-Horowicz, O. Pourquie´,
Cell 91, 639 (1997).
2. C. Jouve, T. Iimura, O. Pourquie´, Development 129,
1107 (2002).
3. J. Cooke, E. C. Zeeman, J. Theor. Biol. 58, 455 (1976).
4. Y. Li, U. Fenger, C. Niehrs, N. Pollet, Differentiation
71, 83 (2003).
5. M. J. McGrew, J. K. Dale, S. Fraboulet, O. Pourquie´,
Curr. Biol. 8, 979 (1998).
6. H. Forsberg et al., Curr. Biol. 8, 1027 (1998).
7. C. Jouve et al., Development 127, 1421 (2000).
8. Y. Bessho et al., Genes Dev. 15, 2642 (2001).
9. Y. J. Jiang et al., Nature 408, 475 (2000).
10. A. Sawada et al., Development 127, 1691 (2000).
11. S. A. Holley et al., Genes Dev. 14, 1678 (2000).
12. C. Leimeister et al., Dev. Biol. 227, 91 (2000).
13. S. E. Cole et al., Dev. Cell 3, 75 (2002).
14. A. V. Morales et al., Dev. Cell 3, 63 (2002).
15. A. C. Oates, R. K. Ho, Development 129, 2929 (2002).
16. S. A. Holley, D. Julich, G. J. Rauch, R. Geisler, C.
Nusslein-Volhard, Development 129, 1175 (2002).
17. J. K. Dale et al., Nature 421, 275 (2003).
18. K. Serth, K. Schuster-Gossler, R. Cordes, A. Gossler,
Genes Dev. 17, 912 (2003).
19. H. Hirata et al., Science 298, 840 (2002).
20. A. Aulehla et al., Dev. Cell 4, 395 (2003).
21. Y. Takahashi et al., Nature Genet. 25, 390 (2000).
22. W. C. Jen, V. Gawantka, N. Pollet, C. Niehrs, C.
Kintner, Genes Dev. 13, 1486 (1999).
23. J. Dubrulle et al., Cell 106, 219 (2001).
24. A. Sawada et al., Development 128, 4873 (2001).
25. M. Kmita, D. Duboule, Science 301, 331 (2003).
26. O. Pourquie´, P. P. Tam, Dev. Cell 1, 619 (2001).
27. The author thanks members of the Pourquie´ lab and
P. Baumann, K. Dale, D. Duboule, J. Dubrulle, A.
Gossler, B. Hermann, D. Ish-Horowicz, R. Kageyama,
and P. Kulesa for helpful comments and discussions
and for communicating data before their publication.
Fig. 2. Model for segment formation in vertebrates based on mouse and chick data. The FGF8/Wnt3A gradient, which
regresses posteriorly during somitogenesis, is shown in black. The anterior boundary of the gradient defines the
determination front, which corresponds to the position of the wavefront (thick black line). The phase I expression of
Notch-related cyclic genes is shown in red (26). The expression of Mesp2/c-meso1 is shown in blue.
D E V E L O P M E N T A L T I M I N G
18 JULY 2003 VOL 301 SCIENCE www.sciencemag.org330
SPECIALSECTION
onDecember30,2010www.sciencemag.orgDownloadedfrom