Cell, Vol. 91, 639–648, November 28, 1997, Copyright ©1997 by Cell Press
Avian hairy Gene Expression Identifies
a Molecular Clock Linked to Vertebrate
Segmentation and Somitogenesis
are laid down sequentially from a terminal growth zone
during the course of development.
In vertebrate embryos, the most obvious metameric
structures are the somites. They constitute the basis of
the segmental pattern of the body and give rise to the
Isabel Palmeirim,* Domingos Henrique,†§
David Ish-Horowicz,† and Olivier Pourquie´‡
*Institut d’Embryologie Cellulaire et Mole´culaire
du Centre National de la Recherche Scientifique
et du Colle` ge de France
axial skeleton, the dermis of the back, and all striated49 bis avenue de la Belle Gabrielle
muscles of the adult body (Christ and Ordahl, 1995).94736 Nogent sur Marne Cedex
Individual pairs of somites, located symmetrically onFrance
either side of the neural tube, emerge from the rostral†Imperial Cancer Research Fund
end of the presomitic mesoderm (PSM), while new mes-PO Box 123, 44 Lincolns Inn Fields
enchymal cells enter the caudal paraxial mesoderm, asLondon WC2A 3PX
a consequence of gastrulation. In the chick embryo, aUnited Kingdom
somite pair is laid down every 90 min in a rostro-caudal‡Developmental Biology Institute of Marseille (IBDM)
progression, and a total of 50 somite pairs are formedLGPD-UMR CNRS 6545 Campus
during embryogenesis. The presomitic mesoderm ap-de Luminy - case 907
pears as a long strip of mesenchymal tissue, and surgi-13288 Marseille cedex 9
cal experiments have shown that approximately 10–12France
prospective somites are contained within the 2-day-old
chick PSM (Packard, 1976; I. P. et al., unpublished data).
Its length becomes progressively reduced during later
Summary
development. Also, it has been suggested that the PSM
includes up to 12 “somitomeres,” segmented arrangeWe
have identified and characterized c-hairy1, an ments of cells that can be visualized using the electron
avian homolog of the Drosophila segmentation gene, microscope and that may correspond to prospective
hairy. c-hairy1 is strongly expressed in the presomitic somites (Meier, 1984).
mesoderm, where its mRNA exhibits cyclic waves of Although various models have been proposed to acexpression
whose temporal periodicity corresponds count for segmentation in vertebrates, little is currently
to the formation time of one somite (90 min). The ap- known about underlying molecular mechanisms (see
parent movement of these waves is due to coordinated Keynes and Stern, 1988; Tam and Trainor, 1994, and
pulses of c-hairy1 expression, not to cell displacement references therein; Discussion below). Numerous vertealong
the anteroposterior axis, nor to propagation of brate homologs of the Drosophila segmentation genes
an activating signal. Rather, the rhythmic c-hairy have been identified but are not expressed during somimRNA
expression is an autonomous property of the togenesis. However, homologs of the neurogenic genes,
paraxial mesoderm. These results provide molecular Notch, Delta (Delta like-1:Delta1) and RBPjk genes,
evidence for a developmental clock linked to segmen- which are not involved in segmentation in the fly, have
been implicated in vertebrate somitogenesis. Targetedtation and somitogenesis of the paraxial mesoderm,
inactivation of these genes in mice leads to a disruptionand support the possibility that segmentation mechaof
somitogenesis (Conlon et al., 1995; Oka et al., 1995;nisms used by invertebrates and vertebrates have
Hrabe de Angelis et al., 1997). Nevertheless, althoughbeen conserved.
somitogenesis is disrupted in Delta1Ϫ/Ϫ
mice, paraxial
mesoderm derivatives such as muscles or skeleton re-Introduction
tain a segmented pattern (Hrabe de Angelis et al., 1997).
These results, therefore, support the view that segmen-Identification and characterization of the Drosophila
tation occurs independently of somitogenesis, and weremelanogaster segmentation genes has led to a recent
also taken as confirmation of the widely held view thatrevival of interest in mechanisms underlying vertesegmentation
arose independently in vertebrates andbrate segmentation (Lewis, 1978; Nu¨ sslein-Volhard and
invertebrates.Wieschaus, 1980; Tautz and Sommer, 1995; Kimmel,
In this paper, we report the identification and analysis1996; De Robertis, 1997). However, the process of segof
the chick c-hairy1 gene, an avian homolog of thementation in Drosophila differs significantly from that of
Drosophila hairy segmentation gene. In Drosophila,
more primitive insects or vertebrates. In long germband
hairy is a member of the pair-rule genes, which are the
insects such as the fly, segments are determined essenfirst
to reveal the prospective metameric body plan of
tially simultaneously in a syncytial unicellular embryo,
the fly (Nu¨sslein-Volhard and Wieschaus, 1980; Ishprior
to gastrulation. In more primitive short germband
Horowicz et al., 1985). Here, we show that c-hairy1
insects like orthopterans, in other arthropods such as
mRNA is expressed in a highly dynamic manner in the
crustaceans, and in vertebrates, segment determination
chick PSM, appearing as a caudo-rostral wave, which
occurs in a cellularized embryo, and posterior segments
is reiterated during the formation of every somite. We
demonstrate that this wavefront is not due to cell movements
within the PSM, nor to the periodic production§Present address: Instituto de Histologia e Embriologia, Faculdade
of an anterior-to-posterior diffusing signal, but is an au-de Medicina, Av. Prof. Egas Moniz, 1699 Lisboa Codex, Portugal.
To whom correspondence should be addressed. tonomous property of the cells in this tissue. We show
Cell
640
Figure 1. Sequence Analysis of c-hairy1
Comparison of the c-hairy1 protein sequence with that of other vertebrate homologs belonging to the Hairy/Enhancer-of-split (HES) family
and the insect hairy proteins using the ClustalX programme (Higgins et al., 1996). c-hairy1 shows highest homology with the Xenopus x-hairy1,
the zebrafish Her6, and the mammalian HES genes. The bHLH domain (between black arrowheads) and the orange domain (Fisher et al.,
1996) (between white arrowheads) are well conserved between all the HES proteins, as well as the tetrapeptide WRPW at the carboxyl
terminus, essential to recruit the corepressor Groucho and exert its negative effect on the transcriptional apparatus (Paroush et al., 1994;
Fisher et al., 1996).
that blocking protein synthesis in embryo explants leads screen a random-primed cDNA library prepared from
chick embryonic mRNA, and several positive clonesto an arrest of somitogenesis but that the oscillations
of c-hairy1 expression persist. This provides evidence were isolated. Sequence analysis of a fraction of these
cDNAs revealed that they arise from a new gene, namedagainst the cyclic c-hairy1 expression beingunder negative
autoregulatory control.Together, these results dem- c-hairy1. Comparison with other vertebrate Hairy-like
genes reveals that c-hairy1 is most similar to the Xeno-onstrate that cells of the PSM undergo a defined and
constant number of c-hairy1 expression cycles between pus laevis hairy1, the mammalian HES, and the zebrafish
Her6 genes (Figure 1).emergence from the primitive streak and incorporation
into a somite. The rhythmic oscillations of the c-hairy1 The putative c-hairy1 protein is 291 amino acids long,
including a bHLH domain and the tetrapeptide WRPWmessenger RNA in prospective somitic cells provide the
first molecular evidence in favor of a developmental at the carboxyl terminus, which are characteristic features
of the hairy-related class of bHLH transcriptionclock involved in vertebrate segmentation.
factorsin flies and vertebrates (Figures1 and 2). Analysis
of the c-hairy1 sequence suggests that it belongs to aResults
subgroup of the WRPW-containing bHLH proteins,
which includes mammalian HES1 and HES2, XenopusIdentification of an Avian hairy Homolog (c-hairy1)
Expressed in the Paraxial Mesoderm X-hairy1, zebrafish Her6, and the fly and tribolium hairy
(Figure 1). The Enhancer-of-split and the zebrafish Her1To identify chick homologs of the fly pair-rule gene hairy,
we used a PCR-based approach with degenerate oligo- genes are only distantly related to these hairy-like genes
(Figure 2). In Drosophila, these proteins act as transcrip-nucleotides that correspond to sequences conserved
between the two hairy-like genes in Drosophila (hairy tional repressors in a variety of developmental contexts
(Ohsako et al., 1994; Paroush et al., 1994; Van Doren etand deadpan). An initial PCR fragment was used to
Molecular Clock Linked to Vertebrate Segmentation
641
This dynamic expression sequence is reiterated during
the formation of every somite and can be represented
as a cycle of three successive stages (Figure 3,
bottom). In stage I, c-hairy1 transcripts are detected in
a broad domain comprising the posterior 70% of the
PSM (corresponding to at least eight prospective somites)
and in a narrow band in the prospective caudal
part of the forming somite (somite 0). In stage II, the
posterior band of c-hairy1 expression has narrowed to
about 3 somite-equivalents in length and has moved
anteriorly, so it now lies in the rostral half of the PSM.
In stage III, the c-hairy1 expression domain becomes
narrower than a somite-equivalent and moves further
anteriorly, forming a stripe coincident with the caudal
part of prospective somite 0.
Transitions between these stages are observed, indicating
that c-hairy1 is expressed as a continuous and
dynamic sequence rather than abrupt switches from one
stage to the other. For example, the broad caudal stripe
observed in stage I begins to appear during stage III
(Figure 3C), indicating that stage III is indeed a precursor
Figure 2. Phylogenetic Tree Analysis of b-HLH Proteins of the Hairy to the next stage I and that the anterior c-hairy1 stripe
and Enhancer-of-Split Families Indicate that c-Hairy1 Belongs to in stage III is a precursor to the stripe in somite 0 of
the Hairy Family of Proteins
stage I. In addition, the intensity of c-hairy1 expression
The tree was generated using the Distances and Growtree progam- increases between stage I and stage III. Out of 71 emmes
of the GCG package Version 9.0 (Madison, Wisconsin). Another
bryos analyzed, 24 were found in stage I, 22 in stage II,
bootstrapped tree calculated on a Power Macintosh using the Clusand
25 in stage III. Based on a cycle time of 90 min, wetalX programme (Higgins et al., 1996) generated a substantially simiestimate
that each stage lasts about 30 min.lar tree (not shown). All compared sequences are accessible in
The reiterated patterns of c-hairy1 expression at theGenbank. Espl, Enhancer-of-split; HES, Hairy and Enhancer-of-split
genes; Her, Hairy and Enhancer-of-split related genes; X, Xenopus; different stages examined suggest that the wavefront
M, mouse; Zf, zebrafish; Hu, human; Dm, Drosophila Melanogaster; of c-hairy1 in the unsegmented mesoderm occurs in
Trib, Tribolium Castaneum; Hes1, Hes2, Hes3, and Hes5 are the rat a cyclic fashion correlated with somite formation. To
genes.
investigate this further, we cultured bilaterally divided
avian embryos in vitro, under conditions where the PSM
yields at least three new somites according to in vivo
al., 1994; Jime´ nez et al., 1996; Fisher et al., 1996). Given kinetics (one somite per 90 min). The caudal parts of
the structural conservation, it is likely that c-hairy1 func- 2-day-old embryos including the PSM were removed
tions similarly during chick development. Expression of and separated surgically along the midline into two
the c-hairy1 gene was analyzed during chick embryonic halves. One embryonic half was fixed immediately, and
the other half cultured on a filter for 30–270 min priordevelopment and was detected in several tissues (data
to fixation. Both halves were then hybridized with thenot shown). In this paper, we focus our attention on
c-hairy1 probe, and the expression pattern on the twomesoderm expression, in particular on the presomitic
sides was compared (Figure 4).mesoderm, where c-hairy1 revealed a very dynamic
After culturing for 30 to60 min, the patterns of c-hairy1mRNA expression pattern.
expression in the cultured and uncultured presomitic
mesoderm always differ (Figure 4A, n ϭ 18), demonstratCyclic
c-hairy1 mRNA Expression in the Paraxial ing the extremely dynamic nature of the expression of
Mesoderm Is Correlated with Somite Formation this mRNA. However, when half embryos are cultured
Analysis of the c-hairy1 mRNA expression pattern in the for 90 min, the time required to form one somite, the
paraxial mesoderm was carried out by whole mount in c-hairy1 expression patterns in the PSMs of cultured
situ hybridization of embryos containing between 1 and and uncultured halves are identical, reflecting the cyclic
25somites. c-hairy1 expressionis detectedin thecaudal property of this expression pattern (Figure 4B; n ϭ 25).
part of somites at all stages examined, where it persists The same rhythmicity of c-hairy1 expression profile is
in the caudal sclerotome for at least 15 hr (Figure 3 and also observed when half embryos are cultured for 270
data not shown). By contrast, c-hairy1 expression in the min, corresponding to the time required to form three
presomitic mesoderm is highly dynamic, as reflected somites in vivo (n ϭ 3; data not shown). Therefore, the
by the variety of expression patterns that are seen in wavefront of c-hairy1 expression in the PSM occurs
embryos with an identical number of somites. The do- in a cyclic fashion, with a periodicity that correlates
main of c-hairy1 expression has the appearance of a precisely with somite formation.
wavefront beginning in the broad, caudal PSM, progressing
anteriorly and intensifying into the narrow ante- The Wave of c-hairy1 mRNA Expression in
rior PSM (Figure3, top). Finally,in each cycle,expression the Presomitic Mesoderm Is Independent
decays sharply throughout the PSM except for a thin of Cell Movement
stripe corresponding to the posterior part of the forming Several mechanisms could account for the kinetics of
c-hairy1 expression in the PSM. One simple possibilitysomite (somite 0).
Cell
642
Figure 3. c-hairy1 mRNA Expression in the Presomitic Mesoderm Defines a Highly Dynamic Caudal-to-Rostral Expression Sequence Reiterated
during Formation of Each Somite
(Top) In situ hybridization with c-hairy1 probe showing the different categories of c-hairy1 expression patterns in embryos aged of 15 (A, B,
and C), 16 (D, E, and F), and 17 (G, H, and I) somites. Rostral to the top. Bar ϭ 200 ␮m.
(Bottom) Schematic representation of the correlation between c-hairy1 expression in the PSM with the progression of somite formation. While
a new somite is forming from the rostral-most PSM (somite 0:S0), a narrow stripe of c-hairy1 is observed in its caudal aspect, and a large
caudal expression domain extends rostrally from the tail bud region (stage I; A, D, and G). As somite formation proceeds, as evidenced by
the visualization of the appearing caudal fissure, the c-hairy1 expression expands anteriorly, the caudal-most domain disappears, and c-hairy1
appears as a broad stripe in the rostral PSM (stage II; B, E, and H).
When somite 0 is almost formed, the stripe has considerably narrowed, and c-hairy1 is detected in the caudal part of the prospective somite
(stage III; C, F, and I) while a new caudal expression domain arises from the tail bud region (in C can be seen the beginning of stage I of the
next cycle). This highly dynamic sequence of c-hairy1 expression in the PSM was observed at all stages of somitogenesis examined (from 1
to 25 somites), suggesting a cyclic expression of the c-hairy1 mRNA correlated with somite formation. Arrowheads point to the most recently
completely formed somite (somite I:SI).
is that the wavefront reflects extensive caudo-rostral patterns differ (Figure 5). This experiment clearly indicates
that the progression of the c-hairy1 wavefrontmovement of c-hairy1 expressing cells during somite
formation. This appears unlikely because previous work occurs independently of cell movement. It also confirms
and extends the results of cell grafting experiments inhas indicated that cell movement within the PSM is restricted
(Tam and Beddington, 1986; Stern et al., 1988). the mouse and of tracer injection into single PSM cells,
which demonstrated that their progeny never encom-If the c-hairy1-expressing cells in stage II were to derive
from cells in stage I, they would have to move across pass more than two consecutive segments (Tam and
Beddington, 1986; Stern et al., 1988).about 50% of the PSM, a distance greater than 450 ␮m,
in less than 30 min.
To exclude the possibility that cell migration contrib- Rhythmic Expression of c-hairy1 Is an Autonomous
Property of the Presomitic Mesodermutes to the dynamics of c-hairy1 expression, we have
marked small clusters of cells at the same anteroposte- What might drive the caudal-to-rostral wavefront of
c-hairy1 expression in the PSM? One possibility is thatrior level in both the left and right PSM with DiI. The
caudal part of these embryos was then separated into it results from a periodic signal originating at the posterior
end of the PSM, which spreads and activatesits two halves as described previously, and one half was
immediately fixed while the other was cultured for 30 c-hairy1 in successively more anterior cells. This relay
hypothesis predicts that a discontinuity within the PSMmin prior to fixation. The DiI was then photoconverted
to an insoluble DAB precipitate, and both halves were would interrupt spreading of thesignal and halt the anterior
progression of c-hairy1 expression. To test this idea,hybridized with the c-hairy1 probe. In all observed cases
(n ϭ 8), DiI labeled cells are found at exactly the same c-hairy1 expression was assayed in half embryos in
which the caudal part of the PSM including the tailbudlevel in the two halves whereas the c-hairy1 expression
Molecular Clock Linked to Vertebrate Segmentation
643
Figure 4. CyclicExpression of c-hairy1RNA in thePresomitic Meso- Figure 5. Cell Movements Do Not Account for c-hairy1 Expression
derm Correlates with Somite Formation Kinetics
The caudal regions of 15- to 20-somite embryos (including the pre- The caudal regions of 15- to 20-somite embryos were sagittally
somitic mesoderm and the last few somites) were sagittally divided divided into two halves after DiI labeling of a small group of cells
into two halves. One half (left side) was immediately fixed, and the at the same anteroposterior level in the left and right PSM. One half
other half (right side) was incubated on top of a millipore filter. Both (left side) was immediately fixed, and the other half (right side) was
halves were hybridized with c-hairy1 probe. incubated on top of a millipore filter for 30 min. Both halves were
(A) Experimental half-embryo cultured for 30 min. A different expres- hybridized with c-hairy1 probe after photoconversion of the DiI. In
sion pattern is observed between the two halves, indicating the (A), the c-hairy1 expression pattern changes from stage I to stage
extremely dynamic nature of c-hairy1 expression. IIϩ while in (B), it progresses from stage II to stage III. Cells labeled
(B) Experimental half-embryo cultured for 90 min (the time required with DiI appear brown owing to the formation of DAB precipitate
for the formation of one somite). The same expression pattern is and are indicated by white arrows. In both examples, the cells are
found in both halves, indicating that c-hairy1 expression pattern located at exactly the same anteroposterior level after a 30 min
cycles over a period exactly corresponding to somite formation. culture periodwhile the c-hairy1expression patternhas progressed,
Open arrowhead, somite 0; arrowheads, segmented somites. Ros- indicating that expression dynamics are not due to cell movements
tral to the top. Bar ϭ 350 ␮m. in the PSM. Asterisk marks the last formed somite. Bar ϭ 150 ␮m.
was surgically ablated (n ϭ 8). The same expression their circuitry involves unstable components that are
pattern is observed in ablated and unoperated halves, subject to negative autoregulation (reviewed in Saseven
after extended culture (Figures 6A–6C). Therefore, sone-Corsi, 1994; Dunlap, 1996). The dynamic pattern
cycling of c-hairy1 expression in the rostral PSM is inde- of c-hairy1 expression and the likelihood that c-hairy1
pendent of the presence of a caudal PSM, and the pro- is a transcriptional repressor led us to ask whether
gression of the c-hairy1-expressing wavefront during c-hairy1 is itself a central component of the clock mechsomite
formation is not related to the spreading of a anism or if its cyclical transcription reflects an output
signal originating in the posterior part of the embryo and from the clock. To address these questions, we examtravelling
anteriorly along the cells in the PSM. ined the effects of blocking protein synthesis on c-hairy1
These experiments suggest that the dynamic c-hairy1 expression.
expression sequence reflects an autonomous property Half-embryo explants were incubated in cyclohexiof
the PSM. We therefore studied the c-hairy1 expres- mide for up to 90 min while the contralateral half was
sion pattern in explant cultures of presomitic mesoderm fixed immediately. When explants are cultured for less
isolated from all the surrounding tissues that might be than 75 min, the fixed and incubated halves show differproviding
extrinsic signals. The presomitic mesoderm ent patterns, indicating that inhibiting protein synthesis
of one-half of 15- to 25-somite embryos was separated does not block c-hairy1 oscillations (n ϭ 4/4; Figure
from ectoderm, endoderm, neural tube, notochord, lat- 7A). We confirmed this result by studying half-embryos
eral plate, and tail bud while the other half remained cultured for equal times in the presence or absence of
intact. The two halves were cultured separately for peri- cycloheximide. For the first 60 min of culture, treated
ods between 30 and 180 min (n ϭ 31). c-hairy1 expres- and untreated halves showthe same patterns of c-hairy1
sion patterns are similar in both types of explant (Figures expression (n ϭ 11/11; Figures 7C and 7D), suggesting
6D–6F), suggesting that the kinetics of c-hairy1 expres- that the periodicity of c-hairy1 pulsing is initially indesion
are independent of surrounding tissues, and derive pendent of de novo protein synthesis.
autonomously from the PSM. Moreover, these cultures Nevertheless, protein synthesis may be required for
cycle normally although Hensens node is absent, show- continued periodicity of c-hairy1 expression. Explants
ing that cycling in the caudal PSM does not depend on cultured in cycloheximide for 90 min (one somite equivaa
signal from the node (Figure 6). lent) usually show a different pattern of expression from
halves fixed immediately (n ϭ 6/9; Figure 7B). Also,
Periodic Oscillations of c-hairy1 Are Independent c-hairy1 expression in half-embryos cultured for 90 min
of Protein Synthesis or more in the presence of cycloheximide often differs
The above results show that c-hairy1 mRNA is ex- from that in the matched half-embryos incubated withpressed
cyclically in cells of the PSM and are consistent out the drug (n ϭ 8/16; Figures 7E and 7F). Thus, a 90
with clock models for somitogenesis (see Discussion). min periodicity is not maintained in such longer term
cultures.Studies of other clock control mechanisms indicate that
Cell
644
Figure 6. The Cyclic Expression of the
c-hairy1 Gene Is an Autonomous Property of
the Presomitic Mesoderm Independent of the
Anterior-Posterior Integrity of This Tissue
Caudal parts of 15- to 20-somite embryos
including the PSM were sagittally divided into
two halves and were cultured in parallel.
(A–C) The caudal part of the right embryonic
half was surgically removed, and the remaining
part was cultured in parallel with its
contralateral half during 90 min (A), 120 min
(B), and 180 min (C). Expression pattern of
c-hairy1 is similar in operated and control
halves independent of the culture period.
(D–F) In the experimental embryonic half,
the presomitic mesoderm was isolated from
the surrounding tissues and cultured with the
contralateral half during 30 min (D), 90 min
(E), and 180 min (F). The expression pattern
of c-hairy1 gene is preserved in the isolated
presomitic mesoderm, showing that the expression
of this gene is an autonomous property
of the presomitic mesoderm. Rostral to
the top. Bar ϭ 150 ␮m.
To verify that protein synthesiswas efficiently blocked mRNA appears as a wavefront travelling along the
anteroposterior axis, and this scheduled expressionduring such short time periods in explant culture, we
measured [35
S]methionine incorporation in half-embryo constitutes an autonomous property of the paraxial
mesoderm. We discuss these results in terms of a devel-explants incubated with or without cycloheximide (n ϭ
36). At concentrations of 5 or 10 ␮M cycloheximide, opmental clock linked to segmentation of the paraxial
mesoderm.progression of the c-hairy1 wavefront was not affected
after 30 min in culture while 71% and 84% of the protein
synthesis was blocked, respectively (data not shown). Rhythmic c-hairy1 mRNA Expression Provides
Molecular Support for a DevelopmentalIncreasing the concentration to 20 ␮M did not increase
the efficiency of the inhibition. Since cycloheximide Clock Driving Segmentation
Prospective somitic cells begin to express pulses ofdoes not block all protein translation (i.e., mitochondrial
protein synthesis), we consider that treatment effi- c-hairy1 mRNA as soon as they leave Hensen’s node
and rostral primitive streak territory to enter the paraxialciently blocked protein synthesis in our explants. Two
other lines of evidence indicate that the persistence of mesoderm (Figure 8). Thus, PSM cells exhibit periodicity
immediately after gastrulation, well before they are in-c-hairy1 wave of expression after cycloheximide treatment
is not due to a failure of the drug to block protein corporated into a somite. This result is in good agreement
with earlier studies which showed that prospective so-synthesis. First, somitogenesis is blocked in the treated
embryos (Figure 7F). Second, treated explants show mites are determined almost concomitantly with paraxial
mesoderm formation (reviewed in Keynes andstrongly increased levels of c-hairy1 transcripts indicating
mRNA stabilization (Figures 7B, 7E, and 7F; note Stern, 1988). However, c-hairy1 mRNA is not expressed
according to the postulated prepattern of the PSM de-that staining times are reduced by at least 5-fold for
treated explants). Together, these results indicate that fined in the somitomere hypothesis (Meier, 1984).
Rather, we propose that the periodic nature ofduring one cycle of expression, the dynamic regulation
of c-hairy1 mRNA is unlikely to involve feedback regula- c-hairy1 mRNA expression in the PSM, which correlates
precisely with the time it takes to form a somite, is drivention by the c-hairy1 protein. Indeed, the failure of the
cycloheximide treatment to stop the clock suggests that by an underlying molecular clock linked to somitogenesis.
Various experiments in amphibian embryos havec-hairy1 is more likely to be an output of the clock than
a component of the clock. led to the idea of such a clock or an oscillator that would
govern the behavior of the cells that are destined to
segment together and form a somite (Cooke and Zee-Discussion
man, 1976; see Davidson, 1988 for a review). In the
“clock-and-wavefront” model, cells oscillate synchro-We report here the identification of c-hairy1, an avian
homolog of the fly segmentation gene hairy. This gene nously according to the clock while they are in the PSM
and then halt their oscillation as they become matureis expressed in a cyclic fashion in the presomitic mesoderm
with a periodicity corresponding to the formation for somite formation. The boundary between oscillating
(immature, presomitic) and arrested (mature, somitic)time of one somite. The periodic expression of c-hairy1
Molecular Clock Linked to Vertebrate Segmentation
645
Figure 8. Rhythmic c-hairy1mRNA Expression Identifies a Developmental
Clock Driving Segmentation and Somite Formation
The PSM is a rod of mesenchymal cells thought to contain about
12 prospective somites. As a new somite is formed every 90 min at
its rostral extremity, PSM length is maintained by a continuous
addition of new cells arising caudally from the gastrulation site. PSM
cells begin to rhythmically express c-hairy1 mRNA as soon as they
exit the gastrulation site (Hensen’s node and rostral primitive streak)
and stop cyclic expression once they are incorporated into a somite.
Therefore, between the moment a cell enters the PSM (0h, arrowhead)
and the time it is incorporated into a somite (18h), a total of
12 somites willhave formed, andconsequently, 12 pulses of c-hairy1
expression will have occurred in the cell. During this time period,
c-hairy1 mRNA expression levels increases progressively, as cells
pass first through stage I, then stage II, and finally stage III before
being incorporated into a somite. c-hairy1 expression is retained
only by cells that lie in the caudal somitic portion. We propose that
the c-hairy1 pulses identify a molecular clock linked to vertebrate
segmentation andsomitogenesis. Thepurpose of sucha clock could
Figure 7. Blocking Protein Synthesis Using Cycloheximide Treat- be to synchronize cells fated to belong to the same somite aspostument
Does Not Block the c-hairy1 Wave Progression lated in somitogenesis models, such as the clock-and-wavefront,
and also to actas a time counting systemfor PSM cells to coordinate(A–D) The caudal region of stage-12 embryos was separated into
the moment of somite formation. R, rostral; C, caudal.two halves along the midline. The left half was immediately fixed
while theright half was cultured in medium containing cycloheximide
(5 ␮M) for 60 min (A) or 90 min (B). Both halves were then hybridized
with the c-hairy1 probeand their expression pattern was compared. of the vertebrate embryo can in principle be either prop(A)
Progression of the wavefront is not stopped since the control agatory (extrinsic) or “kinematic” (independent of the
explant (left) is in stage III while the explant cultured in presence of
propagation of a signal and not stopped by a cut across
cycloheximide for 60 min is in stage I (right). (B) In the majority of
its path; Cooke and Zeeman, 1976). c-hairy1 expressionexplants (6 of 9), a different expression pattern is found in the fixed
is kinematic because it continues to follow an endoge-(left, stage III) and cultured (right, stage II) halves after a 90 min
nous program, even in parts of the PSM that are isolatedculture period. Similar explants as above were cultured for the same
time period in absence (left) or in presence of cycloheximide (right). from the rest.
Explants cultured for 30 min with or without cycloheximide show Strikingly, the observed pattern of c-hairy1 expression
the same expression pattern, stage I (C) or stage III (D), confirming is mimicked by a simple mathematical simulation based
the wavefront progression observed in the previous experiment (A
on a kinematic clock-and-wavefront model of this type,
and B).
in which the wavefront serves to smoothly slow down(E–F) When explants are cultured for longer periods of time, such
and finally freeze the clock. An appendix describing theas 90 min (E) or 120 min (F), treated and control sides are found in
a different stage in 50% of the cases. In (E), the control side (left) model and a movie generated by this simulation (comis
in stage II, and the treated side (right) is in stage III. Segmentation posed by Dr. Julian Lewis, ICRF, London) are available
is blocked by the cycloheximide treatment in these longer cultured
on the Internet at http://www.cell.com/cgi/content/full/
explants (F). The control (left) and treated (right) explants are in
91/5/639. The latter conveys, more clearly than is possi-stage III but are out of register by one somite, owing to the block
ble with static images, the remarkable spatio-temporalof segmentation.
pattern of the oscillations of c-hairy1 expression thatArrowhead points to somite I. Note that in all explants cultured for
longer periods of time, c-hairy1 transcripts become accumulated in we observe and serves as proof that the observed patthe
neural tube and lateral plate (B, C, E, and F). Staining of control tern can be generated by a clock-and-wavefront mechaexplants
lasted 5 times longer than that of treated ones, indicating nism. Although the nature of both clock and wavefront
transcript stabilization. Rostral to the top. Bar ϭ 300 ␮m.
remains undefined in the model, the spatiotemporal pattern
of c-hairy1 expression providesmolecular evidence
for their existence.cells sweeps slowly back along the embryo in an anteroposterior
direction. In this model, the wavefront corre- More recently, it has been proposed that PSM cells
are synchronized using the cell cycle as an internal clocksponding to the anteroposterior gradient of maturation
Cell
646
(Primmett et al., 1989). This proposal is based on experi- In addition, expression in the posterior of somite 0
and then in the caudal part of the newly formed somitements in avian embryos in which a heat shock induces
repeated segmental defects separated by regular inter- suggests that the posterior boundary of c-hairy1 expression
may mark the site at which a new somite boundaryvals of 6–7 somites, an interval corresponding to the
length of one cell cycle in the PSM (9 hr according to should form.c-hairy1 could also contribute to patterning
within somites. Segments in both long and short germ-Primmett et al., 1989). These experiments led to a model
for somitogenesis in which PSM cells are intrinsically band insects are subdivided into anterior and posterior
compartments, domains of lineage restriction that aresynchronized by a clock based on the cell cycle, so that
when groups of cells enter a similar phase of the cycle, required to establish and maintain metamerism and also
to allow further patterning within segments. Althoughthey increase their adhesive properties and segregate
together to form a somite. Indeed, some synchrony in there is no evidence of such lineage restriction in the
early somite, the intrasomitic anteroposterior differencethe cell cycle is observed in the PSM.
Our observations are in agreement with these models inc-hairy1 expression ismaintained duringsomite maturation
(Figure 3) and may help polarize somitic cells intopostulating the existence of an intrinsic clock responsible
for the coordinated behavior of the PSM cells. How- anterior and posterior populations whose interactions
lead to further pattern refinements such as peripheralever, the period of the cell cycle in the PSM is 9 hr while
that of the c-hairy1 cycle is only 90 min. A direct link nervous system segmentation (Keynes and Stern, 1988).
between the cell cycle and the clock driving c-hairy1
expression in the PSM is therefore not obvious. Together,
Evolutionary Implications for Mechanismsthese considerations tend to argue against a direct role
of Segmentation in Invertebratesfor the cell cycle in defining segmental periodicity.
and VertebratesOur results also argue against models in which a reacThe
striking and intriguing pattern of c-hairy1 expressiontion-diffusion mechanism patterns the rostral PSM into
during somitogenesis suggests that it is likely to playtwo states that lead to the segregation of alternating
an important role in mesoderm segmentation in verte-anterior and posterior somitic compartments (Meinbrates.
Thus, hairy-like genes may function during meta-hardt, 1986). However, an attractive possibility is that
merization in both invertebrates and vertebrates, whosec-hairy1 plays a role in the specification of the posterior
segmentation mechanisms may have more in commonsomitic identity (see below).
than previously thought.Progression of the c-hairy1 wavefront and operation
Most vertebrate homologs of the fly segmentationof the c-hairy1 clock are insensitive to blocking protein
genes do not exhibit expression patterns or mutant phe-synthesis by cycloheximide. Not only does this tend to
notypes indicative of a role in somitogenesis (Patel etexclude c-hairy1 playing a role in the clock mechanism
al., 1989; Kimmel, 1996; De Robertis, 1997). It is currentlyitself, but also it places significant limits on the oscillator
thought that, whereas some of the major patterning sys-mechanism. Theoretical and experimental studies of cirtems
involved in dorsoventral and anteroposterior pat-cadian clocks indicate that cycling machineries use unterning
have been conserved during evolution betweenstable components and negative feedback. Several cirarthropods
and vertebrates, segmentation arose inde-cadian clocks are clearly dependent on transcription
pendently in these two phyla (see Weisblat et al., 1994factors that regulate their own transcription via delayed
for a discussion). However, there is increasing evidencenegative feedback (Sassone-Corsi, 1994; Dunlap, 1996).
that Urbilateria, the common ancestor of invertebratesOur cycloheximide experiments argue against such a
and vertebrates, was segmented (Kimmel, 1996; Mullermodel and any other in which clock activities are reguet
al., 1996; De Robertis, 1997). Moreover, the recentlated by cyclic regulation of protein levels, either by
identification of vertebrate homologs of Drosophila seg-regulated translation or degradation. More likely, the
mentation genes that are expressed during somitogen-clock mechanism revealed by c-hairy1 expression acts
esis, including the zebrafish Her1 gene (Muller et al.,posttranslationally, using protein modifications such as
1996), the avian c-hairy1 gene (this report), and the am-phosphorylation.
phioxus engrailed gene (Holland et al., 1997), raises the
possibility that part of the machinery involved in thePotential Functions for c-hairy1 During
segmentation process may be conserved between in-Segmentation and Somitogenesis
sects and vertebrates.As somitogenesis occurs autonomously within the anteHer1
is expressed in the paraxial mesoderm in alter-rior PSM, the moment of segmentation must be deternating
segment primordia as expected for a pair-rulemined intrinsically (Deuchar and Burgess, 1967; Packgene,
rather than in every segment as seen with c-hairy1.ard, 1976; Menkes and Sandor, 1977). c-hairy1 could
However, Her1 is only very distantly related to c-hairy1play a role as part of a counting mechanism in which
and is also very different from Drosophila Hairy, so itcells would use time to measure their positions within
may belong to a different family of WRPW-containingthe presomitic plate to determine when they should start
bHLH proteins (Figure 2). Moreover, Her1 expressionsomitogenesis, for example, by regulating expression of
constitutes the only evidence for a pair-rule type ofa more stable component whose accumulation triggers
mechanism in vertebrates and, in fact, outside of thesomitogenesis when a threshold concentration is exmore
evolved insects species such as dipterans or cole-ceeded. Alternatively, c-hairy1 might play a direct role
opterans. Other vertebrate pair-rule homologs such asin counting, whereby successive pulses last longer and
the even-skipped-like evx genes do not exhibit pair-rulelead to higher levels of c-hairy1 accumulation (see results
and Figure 8). expression (Bastian and Gruss, 1990), nor do currently
Molecular Clock Linked to Vertebrate Segmentation
647
the same on both left and right PSM before separation of the twoidentified homologs of pair-rule genes in primitive short
halves. One half was immediately fixed, and the other was culturedgerm band insects like orthopterans (Patel et al., 1992).
for 30 min prior to fixation. DiI was photoconverted in both halves
The universality of the pair-rule phenomenon therefore
prior to in situ hybridization with the c-hairy1 probe. DiI labeling and
remains controversial (Sander, 1988). photoconversion were performed as described in Ispizua-Belmonte
By contrast, the segmental pattern of c-hairy1 expres- et al (1993).
In the second series of experiments, embryos were also dividedsion and, more particularly, its cyclical anticipation of
sagittally, and in one of the halves, the caudal part including thesegmentation in the PSM and its expression in the postailbud
was removed. Both halves, the entire and the truncated one,terior of the newly forming somite, are clear hints that
were incubated for the same period of time.
it plays a role in segmentation and/or somitogenesis.
In the third series of experiments, embryos were separated into
This reactivates the debate as to whether vertebrate two halves,and the PSM was dissected from the surrounding tissues
somitogenesis is closely related to more “primitive” in- from one of the two halves after a brief treatement with 4ϫ pancreatin
(Gibco). The isolated PSM and the contralateral intact half ofsect modes of segmentation in which segments are
the embryo were cultured for the same time period as describedadded successively from a terminal growth zone. More
above.evolved insect groups such as dipterans may have subsequently
acquired pair-rule patterns of expression in
Cycloheximide Treatment of Explant Cultures
order to allow extremely rapid segmentation in a syncy- The caudal part of stage 12–13 embryos was separated into two
tial embryo. Of course, this raises the question of halves. The experimental half was incubated in the presence of
whether hairy and c-hairy1 play conserved roles in seg- cycloheximide (Sigma, 5, 10, or 20 ␮M), and the contralateral half
was either cultured for the same time period in normal medium ormentation and, even more intriguingly, whether aspects
fixed immediately. In all series, explants were processed for wholeof their transcriptional regulation might have been conmount
in situ hybridization as described below with the c-hairy1
served. The latter appears paradoxical because hairy
probe.
is directly regulated in the syncytial embryo, whereas To monitor efficiency of protein synthesis inhibition, explants were
spatial regulation of c-hairy1 in the chick embryo must incubated for 30 min in 100 ␮l DMEM without cysteine and methiodepend
on intercellular signals. Future experiments will nine complemented with 1% glutamine and 14 ␮Ci [35
S]Met and Cys
(Amersham) with orwithout cycloheximide (5, 10,or 20 ␮M). Explantsindicate how the dynamic pattern of c-hairy1 transcripwere
then lysed in hot sample buffer, and proteins were precipitatedtion is achieved.
in 10% trichloroacetic acid (TCA) overnight. The protein pellet was
recovered by centrifugation and washed consecutively with 5%
Experimental Procedures TCA, ethanol, ethanol/ether (1:1), and ether and resuspended in
10% SDS prior to counting. Explants were also incubated in DMEM
Cloning of c-hairy1 with 1% glutamine and hybridized with the c-hairy1 probe to ensure
First-strand random-primed cDNAwas synthesized from mRNA pre- that absence of serum did not affect the cycling expression pattern.
pared from 1.5-day-old chick embryos. The cDNA was used in a
PCR reaction (94ЊC for 30 sec, 50ЊC for 2 min, 72ЊC for 1 min, 40 Whole Mount In Situ Hybridization
cycles)with the following degenerateprimers: 5Ј-CGIGCICGIATNAA c-hairy1 probe was produced from an 850 bp fragment of the coding
CAANTG(C/T)(C/T)T-3Ј and 5Ј-ACIGTCTTCTCNAGNAT(S)TCNGC sequence cloned in Bluescript KS, linearized using Hind III, and
(C/T)TT. These primers correspond, respectively, to the sequences transcribed with T7 polymerase. Embryos and explants were fixed
RAR(I/M)N(K/N)CL and KA(D/E)(I/M)LEKTV, located on helices 1 and
overnight at 4ЊC in 4% formaldehyde-2mM EGTA, rinsed in PBS,
2 of the fly hairy/deadpan proteins. A fragment of 117 bp derived
dehydrated througha methanol series,and stored in 100%methanol
from a hairy-like cDNA was obtained and used to screen a randomat
Ϫ20ЊC. Whole-mount in situ hybridization was performed acprimed
cDNA library in lambda gt10, prepared from stage 10–14
cording to the procedure described by Henrique et al. (1995). Emchick
embryos. Three overlapping cDNA clones, which cover the
bryos were photographed as whole mounts in PBT (PBS, 0.1%
entire coding region of the c-hairy1 gene, were obtained and fully
Tween20) using a Wild stereomicroscope.
sequenced using the Sequenase kit (Amersham).
Acknowledgments
Eggs and Embryos
Fertilized chick (Gallus gallus, JA57, Institut de Se´ lection Animale, The authors would like to express their gratitude to Prof. Nicole Le
Lyon, France) eggs, obtained from commercial sources, were incu- Douarin for the use of her laboratory facilities and to Monique Coltey,
bated for up to 48 hr in a humidified atmosphere at 38ЊC. The Christiane Bre´ ant, and Pascale Malapert for their excellent technical
embryos were staged by the number of somite pairs formed and instruction. We thank Dr. Julian Lewis for helpful criticisms and for
according to the developmental table of Hamburger and Hamilton the modeling of the c-hairy1 expression pattern. We thank Miguel
(1992). Soares and Clara Redondo for their participation in the early stages
of this project. We acknowledge an anonymous referee for interesting
suggestions of experiments and discussion. We are indebtedIn Vitro Culture of Chick Explants and DiI Labeling
to Drs. Jonathan Cooke, Rosa Beddington, Christo Goridis, MikeChick embryos ranging from 15 (HH12Ϫ) to 20 (HH13ϩ) somites
McGrew, Charles Ordahl, and Steve Kerridge for critical reading ofwere used throughout this study. Different types of explants were
this manuscript. We thank John Sgouros for help in generating pro-precisely delimited, excised, and cultured for 30 min–3 hr in 35 mm
tein alignments and phylogenies, and Franc¸ oise Viala and Francisculture dishes on Polycarbonate filters (0.8 ␮m; Millipore) floating
Beaujean for photographic work. Financial support was providedon top of culture medium composed of Medium 199 (Sigma) suppleby
the Centre National de la Recherche Scientifique (CNRS), themented with 5% heat inactivated fetal calf serum, 10% chicken
Association Franc¸ aise contre les myopathies (AFM), the Associationserum, 1% L-glutamine and 1% penicillin 5000 IU/ml, streptomycin
pour la Recherche contre le Cancer (ARC), the Fondation pour la5000 IU/ml. Under these conditions, somitogenesis proceeds norRecherche
Me´dicale (FRM), and theImperial Cancer Research Fund.mally, and up to 3 somites can be formed in 270 min.
I. P. was funded by the Portuguese Gulbenkian PhD program inIn the first series of experiments, embryos were divided into two
Biology and Medicine and the French Embassy in Portugal. D. I.-H.halves by cutting across the three germ layers at the midline level.
is an International Research Scholar of the Howard Hughes MedicalOne half was immediately fixed. The other one was cultured, as
Institute.described above. A similar series of experiments was performed in
which small groups of cells at different A/P levels of the the PSM
were DiI labeled. In each embryo, the injection site was precisely Received August 11, 1997; revised October 21, 1997.
Cell
648
References Meinhardt, H. (1986). Models of segmentation. In Somites in Developing
Embryos. R. Bellairs, D.A. Ede, and J.W. Lash, eds. (New York
and London: Plenum Press), pp. 179–191.Bastian, H., and Gruss, P. (1990). A murine even-skipped homologue,
Evx 1, is expressed during early embryogenesis and neuro- Menkes, B., and Sandor, S. (1977). Somitogenesis, regulation potengenesis
in a biphasic manner. EMBO J. 9, 1839–1852. cies, sequence determination and primordial interactions. In Vertebrate
Limb and Somite Morphogenesis. British Society Develop-Bettenhausen, B., Hrabe de Angelis, M., Simon, D., Guenet, J.L.,
mental Biology Symposium 3. D.A. Ede, D.A. Hinchcliffe, and M.and Gossler, A. (1995). Transient and restricted expression during
Balls, eds. (Cambridge: Cambridge University Press), pp. 405–419.mouse embryogenesis of Dll1, a murine gene closely related to
Drosophila Delta. Development. 121, 2407–2418. Muller, M., Weizsacker, E., and Campos-Ortega, J.A. (1996). Expression
domains of a zebrafish homologue of the Drosophila pair-ruleChrist, B., and Ordahl, C.P. (1995). Early stages of chick somite
gene hairy correspond to primordia of alternating somites. Develop-development. Anat. Embryol. (Berl). 191, 381–396.
ment. 122, 2071–2078.
Conlon, R.A., Reaume, A.G., and Rossant, J. (1995). Notch1 is reNu¨
sslein-Volhard, C., and Wieschaus, E. (1980). Mutations affectingquired for the coordinate segmentation of somites. Development.
segment number and polarity in Drosophila. Nature 287, 795–801.121, 1533–1545.
Ohsako, S., Hyer, J., Panganiban, G., Oliver, I., and Caudy, M. (1994).Cooke, J., and Zeeman, E.C. (1976). A Clock-and-Wavefront model
Hairy function as a DNA-binding helix-loop-helix repressor of Dro-for control of the number of repeated structures during animal morsophila
sensory organ formation. Genes Dev. 8, 2743–2755.
phogenesis. J. Theor. Biol. 58, 455–476.
Oka, C., Nakano, T., Wakeham, A.,de la Pompa, J.L., Mori, C., Sakai,
Davidson, D. (1988). Segmentation in frogs. Development. 104,
T., Okazaki, S., Kawaichi, M., Shiota, K., Mak, T.W., and Honjo,
221–230.
T. (1995). Disruption of the mouse RBP-J ␬ gene results in early
De Robertis, E.M. (1997). Evolutionary biology. The ancestry of seg- embryonic death. Development. 121, 3291–3301.
mentation. Nature 387, 25–26.
Packard, D.S.J. (1976). The influence of axial structures on chick
Dunlap, J.C. (1996). Genetics and molecular analysis of circadian somite formation. Dev. Biol. 53, 36–48.
rhythms. Annu. Rev. Genet. 30, 579–601.
Paroush, Z., Finley, R.L.J., Kidd, T., Wainwright, S.M., Ingham, P.W.,
Dunwoodie, S.L., Henrique, D., Harrison, S.M., and Beddington, Brent, R., and Ish-Horowicz, D. (1994). Groucho is required for DroR.S.P.
(1997). Mouse Dll3: a novel divergent Delta gene which may sophila neurogenesis, segmentation, and sex determination and incomplement
the function of other Delta homologues during early teracts directly with hairy-related bHLH proteins. Cell 79, 805–815.
pattern formation in the mouse embryo. Development. 124, 3065–
Patel, N.H., Martin-Blanco,E., Coleman, K.G., Poole, S.J., Ellis,M.C.,
3076.
Kornberg, T.B., and Goodman, C.S. (1989). Expression of engrailed
Fisher, A.L., Ohsako, S., and Caudy, M. (1996). The WRPW motif of proteins in arthropods, annelids, and chordates. Cell 58, 955–968.
the hairy-related basic helix-loop-helix repressor proteins acts as
Patel, N.H., Ball, E.E., and Goodman, C.S. (1992). Changing role of
a 4-amino-acid transcription repression and protein–protein interaceven-skipped
during the evolution of insect pattern formation. Nation
domain. Mol. Cell Biol. 16, 2670–2677.
ture 357, 339–342.
Hamburger, V., and Hamilton, H.L. (1992). A series of normal stages
Primmett, D.R., Norris, W.E., Carlson, G.J., Keynes, R.J., and Stern,
in the development of the chick embryo. Dev. Dyn. 195, 231–272.
C.D. (1989). Periodic segmental anomalies induced by heat shock
Henrique, D., Adam, J., Myat, A., Chitnis, A., Lewis, J., and Ish- in the chick embryo are associated with the cell cycle. Development.
Horowicz, D. (1995). Expression of a Delta homologue in prospective 105, 119–130.
neurons in the chick. Nature 375, 787–790. Sander, K. (1988). Studies in insect segmentation: from teratology
Higgins, D.G., Thompson, J.D., and Gibson, T.J. (1996). Using to phenogenetics. Dev. Suppl. 104, 112–121.
CLUSTAL for multiple sequence alignments. Methods Enzymol. 266, Sassone-Corsi, P.(1994). Rhythmic transcription and autoregulatory
383–402. loops: winding up the biological clock. Cell 78, 361–364.
Holland, L.Z., Kene, M., Williams, N.A., and Holland, N.D. (1997). Stern, C.D., Fraser, S.E., Keynes, R.J., and Primmett, D.R. (1988). A
Sequence and embryonic expression of the amphioxus engrailed cell lineage analysis of segmentation in the chick embryo. Developgene
(AmphiEn): the metameric pattern of transcription resembles ment. 104, 231–244.
that of its segment-polarity homolog in Drosophila. Development.
Tam, P.P.L., and Beddington, R.S.P. (1986). The metameric organi-
124, 1723–1732.
zation of the presomitic mesoderm and somite specification in the
Hrabe de Angelis, M., McIntyre, J., and Gossler, A. (1997). Mainte- mouse embryo. In Somites in Developing Embryos. R. Bellairs, D.A.
nance of somite borders in mice requires the Delta homologue DII1. Ede, and J.W. Lash, eds. (New York and London: Plenum Press),
Nature 386, 717–721. pp. 17–36.
Ish-Horowicz, D., Howard, K.R., Pinchin, S.M., and Ingham, P.W. Tautz, D., and Sommer, R.J. (1995). Evolution of segmentation genes
(1985). Molecular and genetic analysis of the hairy locus in Drosoph- in insects. Trends. Genet. 11, 23–27.
ila. Cold. Spring. Harb. Symp. Quant. Biol. 50, 135–144.
Tam, P.P., and Trainor, P.A. (1994). Specification and segmentation
Jime´ nez, G., Pinchin, S.M., and Ish-Horowicz, D. (1996). In vivo of the paraxial mesoderm. Anat. Embryol. (Berl). 189, 275–305.
interactions of the Drosophila Hairy and Runt transcriptional represVan
Doren, M., Bailey, A.M., Esnayra, J., Ede, K., and Posakony,
sors with target promoters. EMBO J. 15, 7088–7098.
J.W. (1994). Negative regulation of proneural gene activity: hairy is
Johnston, S.H., Rauskolb, C., Wilson, R., Prabhakaran, B., Irvine, a direct transcriptional repressor of achaete. Genes Dev. 8, 2729–
K.D., Vogt, T.F. (1997). A family of mammalian Fringe genes impli- 2742.
cated in boundary determination and the Notch pathway. Develop- Weisblat, D.A., Wedeen, C.J., and Kostriken, R.G. (1994). Evolution
ment. 124, 2245–2254. of developmental mechanisms: spatial and temporal modes of rosKeynes,
R.J., and Stern, C.D. (1988). Mechanisms of vertebrate seg- trocaudal patterning. Curr. Top. Dev Biol. 29, 101–134.
mentation. Development. 103, 413–429.
GenBank Accession NumbersKimmel, C.B. (1996). Was Urbilateria segmented? Trends Genet. 12,
329–331.
Accession numbers for the sequences used for the phylogenetic
Ispizua-Belmonte, J.C., De Robertis, E.M., Storey, K.G., and Stern,
tree are: c-hairy1, AF032966; Mhes1, D16464; RHes1, L04527;
C.D. (1993). The homeobox gene goosecoid and the origin of orgaHuhes1,
L19314; ZFher6, X97333; X-hairy1, U36194; Hes2, D14029;
nizer cells in the early chick blastoderm. Cell. 74, 645–659.
Dmhairy, X15904; Tribhairy, S29712; Deadpan, S48025; Espl-m5,
Lewis, E.B. (1978). A gene complex controlling segmentation in Dro- X16552; Espl-m8, X16550; Espl-m7, X16553; Hes3, D13418; ZfHer3,
sophila. Nature 276, 565–570. X97331; Hes5, Q03062; Zfher2, X97330; Zfher4, X97332; Zfher5,
Meier, S. (1984). Somite formation and its relationship to metameric X95301; Zfher1, X97329.
patterning of the mesoderm. Cell Differ. 14, 235–243.