SPECIAL ISSUE REVIEWS–A PEER REVIEWED FORUM

Deciphering the Somite Segmentation Clock:Beyond Mutants and MorphantsJulian Lewis* and Ertugrul M. Ozbudak†

The regular pattern of somite segmentation depends on a clock, the somite segmentation clock, in the formof a gene expression oscillator, operating in the presomitic mesoderm (the PSM) at the tail end of thevertebrate embryo. Genetic screens and other approaches have identified a variety of genes, includingcomponents and targets of the Notch signalling pathway, that show transcriptional oscillations in thisregion and appear to be necessary for correct segmentation. Mathematical modelling shows that theoscillations could plausibly be generated by a simple mechanism of delayed negative feedback, based onautoinhibition of Notch target genes of the Hes/her family by their own protein products. To move beyondplausible models to an experimentally validated theory, however, it is necessary to measure the parameterson which the proposed model is based and to devise ways of probing the dynamics of the system by meansof timed disturbances so as to compare with the model’s predictions. Some progress is being made in thesedirections. Developmental Dynamics 236:1410–1415, 2007. © 2007 Wiley-Liss, Inc.

Key words: segmentation clock; oscillation; gene expression dynamics; mathematical modeling; synchronization; Notchpathway; Hes genes; her genes

Accepted 12 March 2007

INTRODUCTION

The regular pattern of somite segmen-tation, it is generally agreed, dependson a clock—the somite segmentationclock—in the form of a gene expres-sion oscillator, operating in the pre-somitic mesoderm (the PSM) at thetail end of the vertebrate embryo.Each somite consists of the cells thatemerge from the PSM in the course ofone oscillator cycle; and when the os-cillations of gene expression are dis-rupted, so is the pattern of somites(Pourquie, 2003). We would like to un-derstand how the clock works, first ofall on account of its importance in de-fining the vertebrate body plan. Butthere is also a larger reason for focus-

ing on this problem. While develop-mental biologists have made goodprogress in explaining how spatialpatterns arise, the mechanisms thatcontrol the timing of developmentalevents are largely unexplored. In mostcases, we have little more than avague appreciation that one thing fol-lows another, and that biochemical re-actions and cell growth and divisiontake time. The quantitative dynamicsof development—the mechanisms andthe “laws of motion” that govern thetiming of each developmental step—are almost completely unknown. Thesomite segmentation clock exemplifiesin a particularly pure and clear formthe general problem of how the tempo

of events is controlled. In studyingthis clock, we may learn how to tacklethe dynamics of other developmentalprocesses also, where time is likewiseof the essence, even if not marked outin regular repetitive cycles.

THE SEGMENTATIONCLOCK IS A COMPOSITEOF MANY LITTLE CLOCKS

The operation of the clock is manifestin the oscillating expression of a num-ber of different genes, and most obvi-ously of genes coding for componentsof the Notch pathway such as Hes1,Hes7, and Lfng in chick and mouse,and her1, her7, and deltaC in the ze-

Vertebrate Development Laboratory, Cancer Research UK London Research Institute, London, United KingdomGrant sponsor: Cancer Research UK; Grant sponsor: EMBO Fellowship; Grant sponsor: Marie Curie Intra-European Fellowship.†Ertugrul M. Ozbudak’s present address is Stowers Institute for Medical Research, 1000 E. 50th Street, Kansas City, MO 64110.*Correspondence to: Julian Lewis, Vertebrate Development Laboratory, Cancer Research UK London Research Institute,44 Lincoln’s Inn Fields, London WC2A 3PX, UK. E-mail: julian.lewis@cancer.org.uk

DOI 10.1002/dvdy.21154Published online 13 April 2007 in Wiley InterScience (www.interscience.wiley.com).

DEVELOPMENTAL DYNAMICS 236:1410–1415, 2007

© 2007 Wiley-Liss, Inc.

brafish (Pourquie, 2003). Each cell inthe PSM periodically makes and de-grades the products of these genes,and it does so in synchrony with itsneighbours. A basic question, there-fore, is whether each cell has its ownclock or depends instead on a timesignal emitted by some central coordi-nator. And if each cell does have itsown clock, how is synchrony main-tained?

An answer to these questions wassuggested by study of a set of ze-brafish mutants (van Eeden et al.,1996) in which somite segmentationwas disrupted by mutations that, aswe now know, lie in genes coding forthe receptor Notch1a (Holley et al.,2002), for its ligands DeltaD andDeltaC (Holley et al., 2000; Julich etal., 2005), or for the Mind-bomb pro-tein, required to enable the ligands toactivate Notch (Jiang et al., 1996; Itohet al., 2003). In all these mutants, insitu hybridisation revealed that thePSM cells were in a variety of differ-ent states, some expressing the oscil-lator gene deltaC strongly, others ex-pressing it weakly, in a higgledy-piggledy mixture that was preciselywhat one would expect if each cell wascontinuing to oscillate, but doing soindependently and out of synchronywith its neighbours (Jiang et al.,2000). Moreover, the first few (four toeight) somites in all these mutantsshowed correct segmentation, corre-lated with synchronized expression ofdeltaC in the PSM cells in the periodpreceding their formation, suggestingthat oscillation was triggered synchro-nously in all the PSM cells at the be-ginning of somitogenesis and that themutant cells then took several cyclesto drift out of synchrony (Jiang et al.,2000). Given the identity of the mu-tant genes, this carried two furtherimplications: first, that cell–cell com-munication via the Notch pathwaywas required to maintain synchronybetween adjacent cells; and, second,that somites (or at least anteriorsomites) could still form normally inthe zebrafish in the absence of Notchsignalling so long as neighbouringcells remained synchronized.

Subsequent experiments by severalgroups have directly examined theability of PSM cells to oscillate auton-omously. Thus, in the chick, Hes1genes (c-hairy1 and c-hairy2) continue

to show regular oscillation in smallfragments of PSM tissue, but less reg-ular behaviour in isolated cells (Ma-roto et al., 2005). Similarly in themouse, Hes1 expression oscillates (asshown by a luciferase reporter) evenin isolated PSM cells, but less regu-larly than in the intact tissue(Masamizu et al., 2006). Individualcells, it seems, do indeed contain theirown oscillators; but these are erratic,and require cell–cell communicationto keep them all going regularly andat the same rate. Further support forthis view has come from experimentsin the zebrafish, in which cells weregrafted between wild-type embryosand embryos injected with morpholi-nos to knock down components of theNotch pathway and the results wereanalysed with the help of some math-ematical modelling. The conclusionwas that Notch signalling coordinatesthe oscillations of adjacent cells,which, in the absence of such commu-nication, oscillate somewhat irregu-larly (Horikawa et al., 2006).

The somite segmentation clock, asobserved at the tissue level, is there-fore a collective phenomenon, reflect-ing the operation of the many littleclocks in the individual cells, all en-trained to the same rhythm by Notch-mediated cell–cell communication. Agene-expression clock in a single iso-lated cell is inevitably imprecise, ifonly because of the stochastic natureof biochemical reactions (Lewis, 2003);but by coupling many such clocks inparallel, a high precision can be at-tained.

As we discuss below, it is possiblethat the somite segmentation clock ismade of many little clocks in anothersense also: it is conceivable that theindividual cell may contain severalparallel systems of molecules that arecapable of generating oscillations andhave evolved to oscillate with a simi-lar period. These could be held in syn-chrony by some coupling betweenthem, combining to create a sophisti-cated timepiece with all the requiredproperties of temperature depen-dence, buffering against changes ofenvironmental chemistry, responsive-ness to levels of Fgf8, and so on. Theway the system adjusts to changes intemperature is particularly impres-sive. The rate of growth in the PSM,and presumably the kinetics of Fgf8

signalling, vary steeply with tempera-ture, but the period of the clockchanges in a precisely coordinatedway, so that, regardless of the temper-ature, somites of the same standardsize are produced.

GENETIC STUDIESIDENTIFY ESSENTIALCLOCK COMPONENTS

From the foregoing, it seems reason-able to break down the problem of theclock into two parts: How are oscilla-tions generated in the individual cells,and how are the oscillations in neigh-bouring cells coupled?

To answer such questions at a mo-lecular level, one has to begin by iden-tifying the components of the underly-ing control system. We have alreadymentioned several of them. Even be-fore the discovery of the clock, mu-tagenesis studies revealed a criticalrole in somite segmentation for theNotch pathway. Knockout mutationsof Notch1 and Delta1 in the mousecaused disruption of somite segmenta-tion (Conlon et al., 1995; Hrabe deAngelis et al., 1997), while a large-scale mutagenesis screen in the ze-brafish (van Eeden et al., 1996) iden-tified a rather small set of genes thathad similar effects; and most of thesesubsequently turned out also to codefor components of the Notch signallingpathway. But while these findingssuggested that the Notch pathwaywas somehow central to the mecha-nism of segmentation, they gave nohint of how this might be so. More-over, since the zebrafish screens fellshort of saturation, and in any casewere liable to disregard mutationscausing drastic disruption of the bodyplan before the onset of somite forma-tion, it remained possible that otherimportant players had been missed.

The discovery that Hes1 (c-hairy-1)shows oscillating expression in thePSM, matching the rhythm of somito-genesis, transformed the situation(Palmeirim et al., 1997). Hes genes areprimary targets of regulation byNotch, mediate most of the effects ofNotch signalling, and can regulate theexpression of Notch ligands. Their os-cillating expression not only revealedthe existence of the somite segmenta-tion clock, but also hinted at a way inwhich the two parts of the clock prob-

DECIPHERING THE SOMITE SEGMENTATION CLOCK 1411

lem might be reduced to molecularterms. An intracellular oscillatorbased on the dynamics of Hes geneexpression would be susceptible toregulation by Notch signalling, andwould be capable of regulating expres-sion of Notch ligands so as to influenceHes gene expression in neighbouringcells; oscillations in neighbouring cellswould thus be coupled.

The discovery that Hes1 expressionoscillates in chick and mouse (Jouve etal., 2000) was soon followed by thefinding that certain other Hes familygenes—her1 and her7 in the zebrafish(Holley et al., 2000; Sawada et al.,2000; Henry et al., 2002; Oates andHo, 2002; Gajewski et al., 2003), Hes7in the mouse (Bessho et al., 2001,2003)—not only oscillated in the PSM,but were required to oscillate to giverise to regular somite segmentation.These clearly were essential elementsof the clock—either components of thebasic pacemaker (like the pendulumof a pendulum clock) or componentsneeded, like the hands of a clock, forreadout of the clock state (Giudicelliand Lewis, 2004).

SYSTEMATIC MICROARRAYSCREENING REVEALSADDITIONAL GENES THATOSCILLATE IN THE PSM

Other studies, in a variety of verte-brates, implicated other Notch path-way genes, such as Lfng (Forsberg etal., 1998; McGrew et al., 1998; Dale etal., 2003), in the somite clock mecha-nism, adding to the evidence that theNotch pathway had a central role. Butthe picture began to seem less simplewhen it was found that, in the mouseat least, cells in the PSM also showedoscillating expression of the Wnt path-way components Axin2, Nkd1, andDact1 (Aulehla et al., 2003; Ishikawaet al., 2004; Suriben et al., 2006).

If components of the Notch and Wntpathway oscillate in the PSM, are therestill other classes of molecules that alsodo so? Recent work in the Pourquie labhas tackled this question directly andfound that the answer is yes, through amicro-array screen for genes whose ex-pression level oscillates in the PSMwith a periodicity matching that of thesegmentation clock (Dequeant et al.,2006). Meanwhile, a separate study hasdemonstrated that Snail genes (Snail1

in the mouse, Snail2 in the chick, cod-ing for transcription factors not gener-ally thought to lie in the Notch path-way) also show oscillating expression inthe PSM (Dale et al., 2006). Moreover,we cannot exclude the possibility that,underlying the observed transcriptionaloscillations, there is some other type ofoscillator, based for example on cycles ofprotein phosphorylation and dephos-phorylation, as has been beautifullydemonstrated for the circadian clock ofthe cyanobacterium Synechococcus(Kageyama et al., 2006).

In this situation, how can we workout which genes are essential compo-nents of the ultimate pacemaker ofthe segmentation clock?

GENETIC ANALYSIS CANNARROW THE FIELD OFCANDIDATE PACEMAKERCOMPONENTS

In principle, we can narrow the field ofcandidates through a simple genetictest: if component X still oscillateswhen component Y is missing, but thereverse is not true, then Y cannot bean essential component of the masteroscillator, but X might be. In reality,such tests are difficult to interpret be-cause of problems of genetic redun-dancy, and they are dubious at a moreprofound level, because the concept ofa master oscillator may be miscon-ceived. Instead of a single master os-cillator, there may be several weakly-coupled oscillation mechanisms thatnormally operate in parallel but canalso function in isolation. In themouse, the Axin2 oscillations were re-ported to persist, with coordinationbetween cells, even when the Notchpathway was blocked (Aulehla et al.,2003; Hirata et al., 2004), and thesame was found for the Snail oscilla-tions, which did, however, depend onthe Wnt oscillator (Dale et al., 2006).These findings suggested that themaster oscillator in the mouse andchick was after all not based on Notchsignalling, but merely required theNotch pathway for proper readout soas to define the pattern of somiteboundaries. An alternative interpre-tation would be that in the mouse andchick, a Notch-related oscillator and aWnt-related oscillator are present inparallel, with a loose coupling be-

tween them to maintain coordinationunder normal circumstances.

In the zebrafish, in any case, thesituation is different. No evidence ofoscillation in homologs of Axin2 orSnail has so far been reported, andNotch pathway components remainstrong candidates to be the key ele-ments of a master oscillator.

Identifying the molecules that formthe clock is, however, only a first step.To understand the mechanism, we needto know how each of the relevant com-ponents interacts with the others, aslinked elements of a control system.And even that is not enough. We need toexplain how these interactions betweenthe components cause the control sys-tem to behave in the way we observe.

HES/HER AUTOINHIBITIONCOULD BE THE SOURCEOF OSCILLATIONS

Hes genes (in mouse and chick) andtheir counterparts the her genes (infish) code for inhibitory transcriptionfactors. Negative feedback is the gen-eral mechanism underlying oscilla-tions in biological systems. An obvioussuggestion, therefore, is that the in-tracellular oscillations of the segmen-tation clock might arise simply as aconsequence of a direct autoinhibitoryfeedback loop, in which the proteinproducts of Hes or her genes would actback as negative regulators of theirown transcription. Bessho et al. sup-ported this idea with experiments inthe mouse showing that Hes7 proteinbinds to the Hes7 promoter region anddoes indeed inhibit Hes7 transcription(Bessho et al., 2003). Morpholinoknock-down experiments in the ze-brafish to test whether the products ofher1 and her7 inhibited their own ex-pression led to results that were not soclear. Different laboratories drew dif-ferent conclusions (Henry et al., 2002;Holley et al., 2002; Oates and Ho,2002; Gajewski et al., 2003), but auto-inhibition remained a possibility forthese genes too. Other negative feed-back loops, involving Lfng or Axin2,were also suggested as a possible basisfor the oscillations (Aulehla et al.,2003; Dale et al., 2003), but autoinhi-bition of Hes7 (in the mouse) or her1/7(in the fish) was the simplest proposal.

Whichever of these hypotheses onemight espouse, two questions have to be

1412 LEWIS AND OZBUDAK

answered. Is the hypothesized mecha-nism capable in principle of giving riseto the observed oscillations? If so, is thathypothetical mechanism the actualmechanism at work in the embryo?

EXPLANATION OF THECLOCK MECHANISMREQUIRES SOMEMATHEMATICS

Traditionally, developmental biologistshave been content to summarise theirmechanistic findings and speculationsin the form of little diagrams, in whichthe molecules of interest are linked byarrows that show which componentsregulate which other components, andin each case whether the effect is posi-tive or negative. For simple linear path-ways without feedback, this can indeedbe a useful kind of description; but evenin those cases, such cartoons convey noinformation about the dynamics of thesystem. For control mechanisms withfeedback, which is to say for almost allbiological control systems, and certainlyfor all those that generate oscillations,simple circuit diagrams by themselvesdo not explain the behaviour. One can-not look at such a diagram and predict,by unaided intuition and without quan-titative information, what the systemwill do. Newton did not explain the or-bits of the planets simply by saying thatthe Sun exerts an attraction, or even bysaying that the attraction obeys an in-verse square law: he had to show thatthese postulates lead to the ellipticalorbits observed. To understand themechanism of the segmentation clock,or of any other control system with feed-back, we have to do some mathematicalanalysis, or at least some computermodelling. Given a hypothesis as to themechanism, we need to examine itmathematically to discover whether itcould in principle work.

MATHEMATICALMODELLING CONFIRMSTHAT A SIMPLE HES/HERAUTOINHIBITIONOSCILLATOR COULD WORK,BUT ONLY ON CERTAINCONDITIONS

The idea that Hes or her genes mightgenerate intracellular oscillations

through direct autoinhibition, andthat the link between these genes andNotch signalling might serve to syn-chronize adjacent cells, has been for-mulated in a simple mathematicalmodel based on the case of the ze-brafish, with her1 and her7 jointlyplaying the pacemaker role, and theNotch ligand DeltaC expressed in anoscillatory fashion under their control(Lewis, 2003).

Translating this qualitative ideainto a quantitative mathematicalmodel led to some conclusions thatmight not have been obvious. Itshowed that a simple her1/7 autoinhi-bition circuit could indeed give rise tooscillations, but only if it incorporatedtranscriptional and/or translationaldelays, that is, delays in the relation-ship between the amount of Her1/7protein and the rate of production ofmature her1/7 mRNA molecules, orbetween the amount of her1/7 mRNAmolecules and the rate at which mol-ecules of Her1/7 appear in the cell nu-cleus. Quantitative estimates fromother systems suggested that the mostimportant of these delays would be thetranscriptional delay, the delay fromthe moment when a molecule of inhib-itory protein dissociates from its DNAbinding site in the her1/7 promoterregion, permitting transcription to be-gin, to the moment when the maturemRNA molecule emerges into the cy-toplasm to direct protein synthesis.The modelling showed that for sus-tained oscillation, the sum of the tran-scriptional and translational delaysmust be long compared with the life-times of the oscillating molecules. Theoscillations that then occur are pre-dicted to have a period that is roughlytwice the sum of the total feedbackcircuit delay. Notch signalling can in-deed synchronise the oscillations of in-dividual cells according to this model,but only if the delays in Notch-medi-ated communication between theneighbouring cells lie within the rightrange. Judging from crude estimatesbased on information from other bio-logical systems, it seemed plausiblethat all these quantitative require-ments might be met, and plausible,too, that the result could be oscilla-tions with the observed 30-min periodof the zebrafish somite segmentationclock.

Playing with the model led to some

other insights. It showed that even aquite severe general inhibition of therate of protein synthesis might be ex-pected to leave the mRNA oscillationspractically unaffected (Lewis, 2003),explaining observations of Palmeirimet al. (1997). Likewise, thanks to neg-ative feedback, even a quite severemorpholino knockdown of her1 or her7translation would not necessarily beexpected to have much impact on theoscillatory transcription of thesegenes (explaining some of the puzzlingfindings from such experiments in thezebrafish: Henry et al., 2002; Holley etal., 2002; Oates and Ho, 2002; Gajew-ski et al., 2003). Applied to the mouse,the model showed too that a modestincrease in Hes7 protein lifetimewould be expected to lead to dampingand rapid failure of the clock oscilla-tions, just as observed (Hirata et al.,2004).

ANALYSIS OF THE CLOCKMECHANISM REQUIRES AQUANTITATIVEEXPERIMENTALAPPROACH

The mathematical model survives themost basic test. It shows that theideas that it embodies could in princi-ple give rise to the observed behav-iour, including some experimentalphenomena that might seem at firstsight contradictory. Quantitativelyand qualitatively, the model appearsplausible.

But a plausible hypothesis is not thesame thing as a true theory. To decidewhether the model represents the ac-tual mechanism at work in the em-bryo, we need to measure the param-eters that the modelling identifies ascrucial, and to carry out experimentsthat will test the predictions quantita-tively and so discover whether thepostulates of the model are right orwrong.

An analogy can be drawn with theproblem of understanding the mecha-nism of the nerve action potential.When Hodgkin and Huxley begantheir work, there were many compet-ing theories, invoking different mech-anisms and assigning the key roles todifferent molecules. Hodgkin andHuxley established the true mecha-nism by focusing on a certain subset of

DECIPHERING THE SOMITE SEGMENTATION CLOCK 1413

factors (primarily the concentrationsof Na� and K� ions, and the mem-brane potential), which they sus-pected to be crucial, and making care-ful quantitative measurements of howeach one influenced the others. Fromthese measurements they were able tocompute the predicted behaviour ofthe system as a whole. Precise agree-ment between the predictions and theobservations showed convincinglythat their chosen set of factors, inter-acting according to the dynamicalrules that their experiments had de-fined, were indeed the crucial ingredi-ents of the action potential mecha-nism, and that their model was a truetheory (Hodgkin, 1964).

Making measurements of the dy-namics of molecular oscillations in thecells of the PSM is hard, and we canscarcely hope to achieve the accuracyand completeness that was possiblefor Hodgkin and Huxley in their verydifferent system. But we have to try, ifwe are to understand truly how theclock works. The model itself high-lights the things that we need to checkand to measure: the logic of the controlcircuitry, the lengths of the delays, themolecular lifetimes, the synthesis rateconstants, the steepness of the autoin-hibition, and a handful of other fac-tors.

ANALYSIS OF CLOCKDYNAMICS REQUIRESTEMPORALINTERVENTIONS ANDOBSERVATIONS

In efforts to analyse the genetic con-trol circuity of the segmentation clock,the traditional approach, used in moststudies up to now, has been to usemutants or morphants to knock downor misexpress a gene that is suspectedto be important, and examine the con-sequences for expression of other can-didate oscillator genes. In this type ofexperiment, the genetic perturbationoperates from the beginning of devel-opment, and it is difficult or impossi-ble to know whether a given observedeffect is direct or indirect. The timingof action of one component on anotherremains unclear. Moreover, the samemolecule may have different actionsat different steps in the operation ofthe segmentation machinery, and in

this case we see only the composite,cumulative effect. For example, Notchsignalling in the mouse is suspected tobe important both in the operation ofthe segmentation clock and in theevents in the anterior PSM that con-vert the clock oscillations into a spa-tial pattern; we cannot disentanglethese actions simply by looking atNotch mutants. Thus, although mu-tant and morphant studies havehelped to identify candidate compo-nents of the oscillator and have sug-gested regulatory relationships be-tween them, they have not allowedfirm conclusions about the basic cir-cuitry or dynamics of the clock controlmechanism.

To make better progress and totest our hypotheses properly, weneed above all two kinds of experi-mental tools: ways of perturbing in-dividual components at definedtimes, and ways of following the con-sequences temporally. Such tools al-ready exist, but they are crude. Inthe zebrafish, transgenic lines carry-ing a gene of interest under a heat-shock promoter allow us to switchon expression of that gene at anytime we please, more or less abruptly(within about 7 min). Transgeniclines carrying GFP (or similar) re-porter constructs let us monitor inthe living state the level of expres-sion of a chosen gene, though with atemporal precision that is rather se-verely limited by the lifetime of thefluorescent reporter protein. As wedescribe elsewhere (Giudicelli et al.,2007; E.O. and J.L., unpublisheddata), we have begun to use thesetools to analyse and measure the cir-cuitry and dynamics of the zebrafishsegmentation clock and to test oursimple mathematical model. So far,the model seems to survive the tests.But there is a long way still to go. Ifwe are ever to reach the end of theroad and establish the true mecha-nism of the segmentation clock, weshall need to improve on the tools fortemporal analysis that we currentlyhave. Effort devoted to tackling theproblem of the somite segmentationclock in this way should help equipus for the larger enterprise of under-standing how the tempo of develop-ment in general is controlled.

ACKNOWLEDGMENTSWe thank Michael Stauber for com-ments on the manuscript. Our workwas supported in part by an EMBOFellowship and a Marie Curie Intra-European Fellowship within the 6thEuropean Community FrameworkProgramme for E.O.

REFERENCES

Aulehla A, Wehrle C, Brand-Saberi B,Kemler R, Gossler A, Kanzler B, Herr-mann BG. 2003. Wnt3a plays a majorrole in the segmentation clock control-ling somitogenesis. Dev Cell 4:395–406.

Bessho Y, Miyoshi G, Sakata R, KageyamaR. 2001. Hes7: a bHLH-type repressorgene regulated by Notch and expressedin the presomitic mesoderm. Genes Cells6:175–185.

Bessho Y, Hirata H, Masamizu Y,Kageyama R. 2003. Periodic repressionby the bHLH factor Hes7 is an essentialmechanism for the somite segmentationclock. Genes Dev 17:1451–1456.

Conlon RA, Reaume AG, Rossant J. 1995.Notch1 is required for the coordinatesegmentation of somites. Development121:1533–1545.

Dale JK, Maroto M, Dequeant ML, Malap-ert P, McGrew M, Pourquie O. 2003. Pe-riodic notch inhibition by lunatic fringeunderlies the chick segmentation clock.Nature 421:275–278.

Dale JK, Malapert P, Chal J, Vilhais-NetoG, Maroto M, Johnson T, Jayasinghe S,Trainor P, Herrmann B, Pourquie O.2006. Oscillations of the snail genes inthe presomitic mesoderm coordinate seg-mental patterning and morphogenesis invertebrate somitogenesis. Dev Cell 10:355–366.

Dequeant ML, Glynn E, Gaudenz K, WahlM, Chen J, Mushegian A, Pourquie O.2006. A complex oscillating network ofsignaling genes underlies the mouse seg-mentation clock. Science 314:1595–1598.

Forsberg H, Crozet F, Brown NA. 1998.Waves of mouse Lunatic fringe expres-sion, in four-hour cycles at two-hour in-tervals, precede somite boundary forma-tion. Curr Biol 8:1027–1030.

Gajewski M, Sieger D, Alt B, Leve C, HansS, Wolff C, Rohr KB, Tautz D. 2003. An-terior and posterior waves of cyclic her1gene expression are differentially regu-lated in the presomitic mesoderm of ze-brafish. Development 130:4269–4278.

Giudicelli F, Lewis J. 2004. The vertebratesegmentation clock. Curr Opin GenetDev 14:407–414.

Giudicelli F, Ozbudak EM, Wright GJ,Lewis J. 2007. Setting the tempo in de-velopment: an investigation of the ze-brafish somite clock mechanism. PLOSBiology (in press).

Henry CA, Urban MK, Dill KK, Merlie JP,Page MF, Kimmel CB, Amacher SL.2002. Two linked hairy/Enhancer ofsplit-related zebrafish genes, her1 and

1414 LEWIS AND OZBUDAK

her7, function together to refine alter-nating somite boundaries. Development129:3693–3704.

Hirata H, Bessho Y, Kokubu H, MasamizuY, Yamada S, Lewis J, Kageyama R.2004. Instability of Hes7 protein is cru-cial for the somite segmentation clock.Nat Genet 36:750–754.

Hodgkin AL. 1964. The conduction of thenervous impulse. Liverpool: LiverpoolUniversity Press.

Holley SA, Geisler R, Nusslein-Volhard C.2000. Control of her1 expression duringzebrafish somitogenesis by a Delta-de-pendent oscillator and an independentwave-front activity. Genes Dev 14:1678–1690.

Holley SA, Julich D, Rauch GJ, Geisler R,Nusslein-Volhard C. 2002. her1 and thenotch pathway function within the oscil-lator mechanism that regulates ze-brafish somitogenesis. Development 129:1175–1183.

Horikawa K, Ishimatsu K, Yoshimoto E,Kondo S, Takeda H. 2006. Noise-resis-tant and synchronized oscillation of thesegmentation clock. Nature 441:719–723.

Hrabe de Angelis M, McIntyre J 2nd, Gos-sler A. 1997. Maintenance of somite bor-ders in mice requires the Delta homo-logue DII1. Nature 386:717–721.

Ishikawa A, Kitajima S, Takahashi Y,Kokubo H, Kanno J, Inoue T, Saga Y.2004. Mouse Nkd1, a Wnt antagonist,exhibits oscillatory gene expression inthe PSM under the control of Notch sig-naling. Mech Dev 121:1443–1453.

Itoh M, Kim CH, Palardy G, Oda T, JiangYJ, Maust D, Yeo SY, Lorick K, WrightGJ, Ariza-McNaughton L, WeissmanAM, Lewis J, Chandrasekharappa SC,Chitnis AB. 2003. Mind bomb is a ubiq-uitin ligase that is essential for efficientactivation of Notch signaling by Delta.Dev Cell 4:67–82.

Jiang YJ, Brand M, Heisenberg CP,Beuchle D, Furutani-Seiki M, Kelsh RN,Warga RM, Granato M, Haffter P, Ham-merschmidt M, Kane DA, Mullins MC,Odenthal J, van Eeden FJ, Nusslein-Vol-hard C. 1996. Mutations affecting neuro-genesis and brain morphology in the ze-brafish, Danio rerio. Development 123:205–216.

Jiang YJ, Aerne BL, Smithers L, HaddonC, Ish-Horowicz D, Lewis J. 2000. Notchsignalling and the synchronization of thesomite segmentation clock. Nature 408:475–479.

Jouve C, Palmeirim I, Henrique D, BeckersJ, Gossler A, Ish-Horowicz D, PourquieO. 2000. Notch signalling is required forcyclic expression of the hairy-like geneHES1 in the presomitic mesoderm. De-velopment 127:1421–1429.

Julich D, Lim CH, Round J, Nicolaije C,Schroeder J, Davies A, Geisler R, LewisJ, Jiang YJ, Holley SA, Tubingen 2000Screen Consortium. 2005. beamter/deltaC and the role of Notch ligands inthe zebrafish somite segmentation, hind-brain neurogenesis and hypochord dif-ferentiation. Developmental Biology 286:391–404.

Kageyama H, Nishiwaki T, Nakajima M,Iwasaki H, Oyama T, Kondo T. 2006.Cyanobacterial circadian pacemaker:Kai protein complex dynamics in theKaiC phosphorylation cycle in vitro. MolCell 23:161–171.

Lewis J. 2003. Autoinhibition with tran-scriptional delay: a simple mechanismfor the zebrafish somitogenesis oscilla-tor. Curr Biol 13:1398–1408.

Maroto M, Dale JK, Dequeant ML, PetitAC, Pourquie O. 2005. Synchronised cy-cling gene oscillations in presomitic me-soderm cells require cell-cell contact. IntJ Dev Biol 49:309–315.

Masamizu Y, Ohtsuka T, Takashima Y,Nagahara H, Takenaka Y, Yoshikawa K,Okamura H, Kageyama R. 2006. Real-

time imaging of the somite segmentationclock: revelation of unstable oscillatorsin the individual presomitic mesodermcells. Proc Natl Acad Sci USA 103:1313–1318.

McGrew MJ, Dale JK, Fraboulet S, Pour-quie O. 1998. The lunatic fringe gene is atarget of the molecular clock linked tosomite segmentation in avian embryos.Curr Biol 8:979–982.

Oates AC, Ho RK. 2002. Hairy/E(spl)-re-lated (Her) genes are central componentsof the segmentation oscillator and dis-play redundancy with the Delta/Notchsignaling pathway in the formation ofanterior segmental boundaries in the ze-brafish. Development 129:2929–2946.

Palmeirim I, Henrique D, Ish-Horowicz D,Pourquie O. 1997. Avian hairy gene ex-pression identifies a molecular clocklinked to vertebrate segmentation andsomitogenesis. Cell 91:639–648.

Pourquie O. 2003. The segmentation clock:converting embryonic time into spatialpattern. Science 301:328–330.

Sawada A, Fritz A, Jiang YJ, Yamamoto A,Yamasu K, Kuroiwa A, Saga Y, TakedaH. 2000. Zebrafish Mesp family genes,mesp-a and mesp-b are segmentally ex-pressed in the presomitic mesoderm, andMesp-b confers the anterior identity tothedevelopingsomites.Development127:1691–1702.

Suriben R, Fisher DA, Cheyette BN. 2006.Dact1 presomitic mesoderm expressionoscillates in phase with Axin2 in thesomitogenesis clock of mice. Dev Dyn 235:3177–3183.

van Eeden FJ, Granato M, Schach U,Brand M, Furutani-Seiki M, Haffter P,Hammerschmidt M, Heisenberg CP,Jiang YJ, Kane DA, Kelsh RN, MullinsMC, Odenthal J, Warga RM, AllendeML, Weinberg ES, Nusslein-Volhard C.1996. Mutations affecting somite forma-tion and patterning in the zebrafish,Danio rerio. Development 123:153–164.

DECIPHERING THE SOMITE SEGMENTATION CLOCK 1415