Dynamics of zebrafish somitogenesis

RESEARCH ARTICLE

Dynamics of Zebrafish SomitogenesisChristian Schroter,1 Leah Herrgen,1 Albert Cardona,2 Gary J. Brouhard,1 Benjamin Feldman,3 andAndrew C. Oates1*

Vertebrate somitogenesis is a rhythmically repeated morphogenetic process. The dependence ofsomitogenesis dynamics on axial position and temperature has not been investigated systematically in anyspecies. Here we use multiple embryo time-lapse imaging to precisely estimate somitogenesis period andsomite length under various conditions in the zebrafish embryo. Somites form at a constant period along thetrunk, but the period gradually increases in the tail. Somite length varies along the axis in a stereotypicalmanner, with tail somites decreasing in size. Therefore, our measurements prompt important modificationsto the steady-state Clock and Wavefront model: somitogenesis period, somite length, and wavefront velocityall change with axial position. Finally, we show that somitogenesis period changes more than threefoldacross the standard developmental temperature range, whereas the axial somite length distribution istemperature invariant. This finding indicates that the temperature-induced change in somitogenesis periodexactly compensates for altered axial growth. Developmental Dynamics 237:545–553, 2008.Published 2008 Wiley-Liss, Inc.†

Key words: zebrafish; somitogenesis; oscillator; period; frequency; rate; time-lapse; temperature; segmentation; Clockand Wavefront model

Accepted 13 November 2007

INTRODUCTION

Vertebrate somitogenesis is a re-peated patterning process that occurssequentially along the axis of the de-veloping embryo. Current hypothesesof somitogenesis posit a genetic oscil-lator, termed the segmentation clock,in the presomitic mesoderm (PSM)and tail bud that interacts with a pos-terior-moving determination wave-front to set the periodic spacing ofsomite boundaries: this is termed theClock and Wavefront model (Cookeand Zeeman, 1976; Dubrulle andPourquie, 2004a; Aulehla and Pour-quie, 2006). In this model, somitogen-esis period, i.e., the time interval be-

tween formation of each successiveboundary, is determined by the periodof the genetic oscillator, and somitelength is the distance traveled by thewavefront in one period of oscillation.Despite increased interest in the un-derlying molecular processes, themorphological outputs of the clock andthe wavefront, somitogenesis periodand somite length, have not yet beenquantified simultaneously in any spe-cies. To understand somitogenesis, aswell as other iterative processes dur-ing development, a precise quantita-tive description is necessary.

Estimates of somitogenesis periodin several species indicate that there

are distinct species-specific dynamics(Pearson and Elsdale, 1979; Tam,1981; Hanneman and Westerfield,1989; Brooks and Johnson, 1994;Wood and Thorogood, 1994; Johnstonet al., 1995; Schubert et al., 2001).However, the influence of genetic het-erogeneity on the period within a sin-gle species has not been explored. Fur-thermore, although most studiesreport some axial variation of somito-genesis period, conflicting data existon the axial extent of such changes(Cooke and Zeeman, 1976; Pearsonand Elsdale, 1979; Hanneman andWesterfield, 1989; Kimmel et al.,1995; Schmidt and Starck, 2004).

The Supplementary Material referred to in this article can be found at http://www.interscience.wiley.com/jpages/1058-8388/suppmat1Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany2Department of Molecular, Cell, and Developmental Biology, University of California, Los Angeles, California3Medical Genetics Branch, National Institutes of Health, Bethesda, MarylandGrant sponsor: the Max Planck Society; Grant sponsor: National Institutes of Health.*Correspondence to: Andrew C. Oates, Max Planck Institute of Molecular Cell Biology and Genetics, Pfotenhauerstr. 108,01307 Dresden, Germany. E-mail: [email protected]

DOI 10.1002/dvdy.21458Published online 7 February 2008 in Wiley InterScience (www.interscience.wiley.com).

DEVELOPMENTAL DYNAMICS 237:545–553, 2008

Published 2008 Wiley-Liss, Inc.†This article is a US Government work and, as such, isin the public domain in the United States of America.

Kimmel et al. (1995) reported a sud-den change in zebrafish trunk somito-genesis period, which has been takenas evidence that different genetic sys-tems pattern anterior and posteriortrunk (Holley, 2006), but the periodmeasurements of Hanneman andWesterfield (1989) and Schmidt andStarck (2004) do not support this.Somite length changes along the bodyaxis in the mouse, chick and Xenopusembryo (Herrmann et al., 1951; Pear-son and Elsdale, 1979; Tam, 1981),but how this correlates with changingperiod has not been examined. Inpoikilothermic organisms such as thezebrafish, developmental dynamics,including somitogenesis, dependstrongly on temperature (Kimmel etal., 1995). However, the precise tem-perature dependence of somitogenesisperiod or somite length has not beeninvestigated systematically.

Time-lapse microscopy has beenused to study somitogenesis in varioussystems, e.g., to observe cell behaviorat the forming somite boundary inBarbus conchonius (Wood and Thoro-good, 1994), zebrafish (Henry et al.,2000), and chick (Kulesa and Fraser,2002) and to investigate dynamic geneexpression patterns during mousesomitogenesis (Masamizu et al.,2006). In each of these cases, only oneembryo was imaged at a time, andobservations were limited to a singletemperature and a small region of theaxis. To efficiently gather precise dataon global somitogenesis dynamics, anassay system that can document mul-tiple embryos at closely spaced timeintervals and defined temperaturesthroughout the entirety of somitogen-esis is required.

Here, we investigate somitogenesisdynamics using the zebrafish model.Because of their transparency and ex-ternal development, zebrafish em-bryos allow for easy visualization athigh sampling rates without adverseperturbation. In addition, a largenumber of closely staged embryos caneasily be obtained, allowing measure-ment precision to be estimated. To si-multaneously document somitogen-esis period and somite length, we useda semiautomatic microscopy setupwith a motorized stage that can docu-ment approximately 40 zebrafish em-bryos in parallel throughout somito-genesis. We show that somitogenesis

period is constant in the trunk butincreases gradually in the tail. Somitelength decreases in the posterior body,indicating that wavefront velocity alsochanges during somitogenesis in thewild-type embryo. Finally, we mea-sure a strong linear dependence ofsomitogenesis frequency on tempera-ture, whereas somite length distribu-tion is temperature invariant, indicat-ing that the period of the clockcompensates for altered growth.

RESULTS

To precisely measure somitogenesisperiod and somite length, we set out todocument multiple embryos develop-ing under the same conditions usingbrightfield time-lapse microscopy. Amotorized stage was used to repeat-edly screen up to 41 animals at de-fined time intervals and at specified,constant temperatures. Applying anImageJ plugin (for details, see the Ex-perimental Procedures section andFig. 1A), we automatically generatedtime-lapse movies of these embryos inparallel (Supplementary Movie S1,which can be viewed at http://www.interscience.wiley.com/jpages/1058-8388/suppmat).

Precise Determination ofSomitogenesis Period

The increase in somite number overtime can be read from consecutiveframes of the movies (Fig. 1B). To de-termine somitogenesis period, we an-alyzed the movies, sequentially notingthe time at which the boundary poste-rior to each somite formed. For em-bryos developing inside the chorion,this approach allowed most trunksomites to be scored. We first analyzedsomitogenesis period between somites6 to 18; the period at other axial posi-tions is described below. Analysis of asingle movie gave the somitogenesisperiod for an individual (Fig. 1C). Toexamine somitogenesis period as apopulation property, we analyzedmultiple movies that had been re-corded in parallel, normalized thedata, and calculated the mean forma-tion time, as well as the standard de-viation (SD), for each boundary. Theuncertainty of the data points, com-prising both sampling error and bio-logical variation, remained constantwith recording time (Fig. 1D). Linear

fits to the data from somites 6 to 18yielded R2 values between 0.998 and0.999, indicating a linear increase insomite number over time. Therefore, aprecise estimate of the somitogenesisperiod (P) for the population could bederived from the slope of the linearcurve, which was 24.9 � 0.7 min at atemperature of 27.1 � 0.1°C for theexperiment shown (Fig. 1D). Confi-dence intervals for fits of comparableexperiments were consistently lessthan 10% of the measured value (Ta-ble 1). Therefore, observation of mul-tiple embryos and several boundariesalong the axis enabled us to preciselydetermine the somitogenesis periodbetween somites 6 through 18.

Somitogenesis Period IsIndependent of Wild-TypeGenetic Background

We next tested the influence of geneticheterogeneity on somitogenesis pe-riod. To determine whether the com-monly used zebrafish strains (Guryevet al., 2006) have strain-specific perioddifferences, we measured trunk somi-togenesis period for the AB strain indirect pairwise comparison to threeothers: Golden, Wik, and TL (2 exper-iments per strain, 10 � n � 20 perstrain and experiment). We found nosignificant difference in somitogenesisperiod between these strains and theperiod established for AB (data notshown), suggesting that somitogen-esis period is not modified by strain-specific genetic differences.

Somitogenesis Period IsConstant Along the Trunkbut Increases in the Tail

While somitogenesis period is con-stant from somites 6 to 18, the periodin the anterior trunk or in the tailmight deviate from this behavior. Toinvestigate the extent of axial varia-tion of somitogenesis period, we ex-tended our measurements to the en-tire embryonic axis.

First, we measured the somitogen-esis period of the first 6 somites incomparison to the rest of the trunk,which ends at somite 17. By orientingdechorionated embryos dorsally (Fig.1A, upper left), we were able to followsomitogenesis in the anterior trunk

546 SCHROTER ET AL.

(Supplementary Movie S2). In paral-lel, we documented embryos from thesame clutch inside their chorion and

followed somitogenesis from a lateralview for the rest of the trunk. We ex-cluded the possibility that dechorion-

ation has adverse effects on the onsetor period of somitogenesis by compar-ing the absolute time when a refer-ence boundary formed in dechorion-ated vs. undechorionated embryos.When fitting linear curves to the datapoints obtained from embryos ori-ented dorsally, we found good linear-ity (Fig. 2A, red line; R2 � 0.9996).Therefore, somitogenesis period in theanterior trunk can once again be esti-mated as the slope of the linear fit tothe data. The somitogenesis period forsomites 1 to 6 (Fig. 2A, red line; P �24.7 � 5.0 min) and the period be-

Fig. 1.

Fig. 2. Somitogenesis period at different axial positions. A: Somitogenesis period in the anterior trunk does not differ significantly from the period inthe posterior trunk. Populations of embryos from the same clutch were filmed in parallel, either with or without chorion from lateral or dorsal view,respectively. Normalized mean somite formation times � SD were scored from either set of movies. Linear fits to both data sets (somites 1–6 for thedorsal view, somites 6–17 for the lateral view) yielded slopes that were not significantly different. Note that the limited number of data points for periodestimation for the anteriormost somites results in a large confidence interval. T � 27.6 � 0.1°C. B: Somitogenesis period increases in the tail. Tailoutgrowth of embryos homozygous for the nicb107 allele was filmed, and somite formation times were analyzed from somite 4 onward. A linear fit todata points in the trunk region (somites 4–17) indicates nonlinear behavior as somitogenesis proceeds into the tail. T � 27.4 � 0.1°C.

Fig. 1. Measurement of trunk somitogenesisperiod. A: Automated generation of somitogen-esis time-lapse movies. Embryos in differentorientations are held in defined positions in aga-rose molds. A motorized stage scans throughthem at defined time intervals. Focused imagesare generated from Z-stacks using the ImageJStack Focuser, and time-stamped QuickTimemovies are automatically generated from thefocused images. B: Single focused frames of arepresentative movie. Red arrowheads markthe most recently formed somite boundary ineach frame. C: Period measurement from a sin-gle embryo. A movie produced as in A wasanalyzed from the six-somite stage onward,and formation time of individual somites wasnormalized relative to somite 6. A linear in-crease in somite number over time is evident.D: Somitogenesis period of the population. Theplot shows normalized mean formation times �SD of boundaries posterior to somite 6 in agroup of embryos. Note that the SD does notincrease with the duration of recording. Thesomitogenesis period of the population was es-timated as the slope of the linear regression:P � 24.9 � 0.7 min, T � 27.1 � 0.1°C.

ZEBRAFISH SOMITOGENESIS DYNAMICS 547

tween somites 6 to 17 (Fig. 2A, blackline; P � 23.8 � 0.9 min), were notsignificantly different, indicating thatsomites along the entire trunk formwith a similar period.

To measure somitogenesis period inthe tail, we filmed nicb107 homozygousembryos, which lack functional nico-tinic acetylcholine receptors (Wester-field et al., 1990). Because these em-bryos are nonmotile and developotherwise normally far beyond somito-genesis stages (van der Meulen et al.,2005), we could follow tail somitogen-esis without anesthetizing the embryo(Supplementary Movie S3). Homozy-gous nicb107 embryos did not displaysignificant differences in trunk somi-togenesis period compared with theirmotile littermates, and the differencein somite number by the 28 somitestage between these two groups wasless than one (data not shown). Thus,the nicb107 mutation does not affectsomitogenesis period per se. Analysisof the movies showed that, in contrastto the trunk, tail somites formed atcontinuously longer periods along theaxis (Fig. 2B). Thus, somitogenesis pe-riod in the tail marks a gradual slow-ing rather than a separate processwith a distinct period. When fitting alinear curve to the data points ob-tained for trunk somites (4 to 17), wedetected the first significant devia-tions from this fit around somite 20.Therefore, trunk and tail differ insomitogenesis period in the wild-typezebrafish.

Somitogenesis Period IsLinearly Temperature-Dependent

In addition to axial position, environ-mental temperature is likely to setsomitogenesis period in the zebrafish.During the above experiments, we no-ticed that a minor temperature shiftled to a measurable period change(compare Fig. 1C with Fig. 2A). Thisfinding prompted us to explore alarger range of developmental temper-atures corresponding to typical labo-ratory rearing conditions using ourtime-lapse system. We found thatsomite number retained a linear rela-tionship with time at all measuredtemperatures from 20.0°C to 30.8°C(Fig. 3A), whereas the period changedapproximately threefold across this

range (Table 1). When plotting tem-perature against somitogenesis fre-quency, we obtained a good linear re-lationship (Fig. 3B, R2 � 0.9936).Therefore, the somitogenesis fre-quency f [somites.h�1] at any temper-ature T [°C] between 20.0 and 30.8°Ccan be calculated with the formulaf � aT-b, with a � 0.188 � 0.006somites.h�1°C�1 and b � 2.7 � 0.1somites.h�1. Using this formula wecan calculate a temperature coeffi-cient Q10, which describes the foldchange in the rate of a process across a10°C interval. The coefficient for somi-togenesis in the temperature rangefrom 20°C to 30°C is Q10 � 2.8.

Temperature-IndependentAxial Distribution of SomiteLength

We next examined the dynamics ofsomite length at the time of formation(instantaneous length) along the bodyaxis, using movies recorded from apopulation of nicb107 homozygous em-bryos (Fig. 4A). By measuring somitesat the time of formation, we excludecontributions of convergent extensionor differentiation that mask truesomite length if measured later in de-velopment. Along the axis, somitelength increased slightly in the ante-rior trunk, remained approximatelyconstant around 50 �m betweensomites 6 and 14, and decreasedthereafter (Fig. 4B). By somite 25, theinstantaneous length was approxi-mately 30 �m, almost half that seen inthe central trunk. Notably, somitelength decreased in the posteriortrunk, while somitogenesis period re-

mained constant. Thus, the zebrafishembryo shows a continuous change inthe length of the newly formingsomites along the axis.

Given our observation of an approx-imately threefold change of periodacross the 20.0 to 30.8°C temperaturerange, we tested whether somitelength was affected by temperature.We performed instantaneous somitelength measurements in three sets ofmovies recorded from nicb107 homozy-gous embryos at 21.4°C, 25.0°C, and31.1°C. These measurements did notreveal any consistent change of somitelength with temperature (Fig. 4C).Therefore, we conclude that the axialsomite length distribution of the ze-brafish is independent of steady statetemperature.

DISCUSSION

Precision of the CurrentTechnique

Here, we make use of a novel multipleembryo time-lapse imaging protocol tosystematically investigate the dynam-ics of zebrafish somitogenesis at dif-ferent positions along the developingaxis, over a range of temperatures,and in different wild-type strains. Bysimultaneously recording multipleembryos and scoring multiple bound-aries, we obtain precise mean periodmeasurements that allow reliable de-tection of period differences of �10%between groups of embryos, despitethe relatively low temporal measure-ment resolution of each single bound-ary in a given embryo. Our periodmeasurements in the trunk are ingood agreement with some previously

TABLE 1. Temperature Dependence of Somitogenesis Perioda

Temperature (°C) Somitogenesis period (min) n

20.0 � 0.4 55.4 � 1.9 1120.2 � 0.4 54.6 � 2.0 1222.8 � 0.5 40.2 � 1.1 1124.3 � 0.1 31.5 � 0.9 1427.1 � 0.1 25.1 � 0.5 1527.6 � 0.1 23.9 � 0.6 1629.6 � 0.2 21.4 � 1.0 1130.8 � 0.3 18.7 � 1.1 14

aPeriod measurements were performed at different temperatures. Period values aregiven with 95% confidence intervals and mean temperature � SD during therecordings. Somitogenesis period was estimated as the slope of the linear regressionof the data.

548 SCHROTER ET AL.

published estimates (Hanneman andWesterfield, 1989; van Eeden et al.,1996; Schmidt and Starck, 2004), al-though they disagree with a com-monly cited study that overestimatesthe value at 28.5°C (Kimmel et al.,1995). Precision is currently limitedby a sampling rate of 5 min, due to the

high number of samples and Z-planesand is compounded by uncertainties inthe visual analysis of boundary forma-tion. With higher temporal resolution,even subtler period changes could bedetected.

We find that genetic polymorphismsbetween the commonly used zebrafish

strains AB, Golden, Wik, and TL (Gu-ryev et al., 2006) do not cause strain-specific differences in mean somito-genesis period. Therefore, we proposethat the period is set by a defined com-plement of genes that may be identi-fied by screening for somitogenesis pe-riod mutants using the multiple

Fig. 3. Temperature dependence of somitogenesis period. A: Trunk somite number increases linearly over time at different temperatures. Groups ofembryos were filmed at different temperatures. Plotting of mean formation times � SD for somites 4–18 reveals a linear increase of somite numberover time at 20.0°C (blue), 24.3°C (green) and 30.8°C (red). B: Somitogenesis frequency depends linearly on temperature. Period values from Table1 were expressed as frequencies and plotted against the mean temperature during the recording. The fit shows a good linear dependence of frequencyand temperature in the range assayed. Error bars show SD of the temperature and 95% confidence interval of the frequency value.

Fig. 4. Axial and temperature dependence ofsomite length. A: Measurement of instanta-neous somite length. Somite length was ana-lyzed in focused frames of time-lapse moviesby measuring the length of a straight line (red),connecting the intersections of the dorsal limitof the notochord with the two most recentlyformed somite boundaries. Scale bar � 200 �min main, 50 �m in inset. B: Somite length alongthe axis is independent of somitogenesis pe-riod. Instantaneous somite length (squares) wasmeasured along the entire axis in a populationof nicb107 homozygous embryos and comparedwith somitogenesis period in the same popula-tion (triangles). Somite length increases in theanterior trunk, is approximately constant in thecentral trunk, and decreases thereafter, inde-pendently from somitogenesis period. Errorbars show SD of formation time and lengthmeasurements. Solid line: Linear fit to estimatesomitogenesis period in the trunk. n � 8; P �41.6 � 1.2 min; T � 22.7 � 0.4°C.C: Somite length is independent of develop-mental temperature. Instantaneous somitelength was measured in populations of nicb107

homozygous embryos developing at 21.4°C(blue), 25.0°C (green), and 31.1°C (red). Differ-ent temperatures do not lead to a consistentchange in instantaneous somite length. Errorbars state SD of length measurement.


embryo time-lapse setup describedhere. The identification of segmenta-tion period genes would be a majorstep toward understanding the bio-chemical basis of rhythmicity in somi-togenesis. In addition, measured dif-ferences in mean period between wild-type and mutants could be useful formathematical modeling approaches.

Strong TemperatureDependence ofSomitogenesis Period

In the laboratory, zebrafish embryosare commonly raised between 20°Cand 30°C (Kimmel et al., 1995; ourown unpublished observation), andit is known that temperature has astrong effect on the general speed ofdevelopment (Kimmel et al., 1995).Here, we precisely quantify the ef-fect of temperature on somitogenesisfrequency and find an almost three-fold change in the frequency acrossthe standard range of developmentaltemperatures. Interestingly, in thisrange, somitogenesis frequency de-pends linearly on temperature. Weprovide an equation of the form f �aT-b to estimate somitogenesis fre-quency (f) for any temperature (T) inthis range. At the standard rearingtemperature of 28.5°C, we calculatea frequency of 2.7 � 0.3 somites.h�1,corresponding to a period of 22.6 �2.5 min. Additionally, we calculate atemperature coefficient of Q10 � 2.8for somitogenesis between 20°C and30°C. This value agrees well withdata obtained for herring (Johnstonet al., 1995) and biochemical reac-tions in vitro, but contrasts to circa-dian clocks, the free-running periodsof which are strongly temperaturecompensated with a Q10 of approxi-mately 1 (Pittendrigh, 1954; Bun-ning, 1963). The temperature exper-iments presented in this paperprovide a means to directly compareperiod data collected in differentlabs, provided temperature can beaccurately measured. Finally, ourdata suggests that, when using pe-riod estimates as output variables inmathematical modeling approaches,input variables should be normal-ized to specific temperatures.

Changes in SomitogenesisPeriod Along the DevelopingAxis

For the first time, we have measuredsomitogenesis period with high preci-sion along the entire axis of a vertebratespecies. A rapid start to somitogenesishas been reported in amphioxus andamphibian larvae (Pearson and Els-dale, 1979; Schubert et al., 2001), andthere is conflicting evidence concerningthe somitogenesis period in the trunk inzebrafish (Hanneman and Westerfield,1989; Kimmel et al., 1995; Schmidt andStarck, 2004). We find there is a linearincrease in somite number over timealong the entire zebrafish trunk, indi-cating no significant differences insomitogenesis period for the first six vs.later trunk somites. In contrast to anearlier proposal of more rapid anteriorsomitogenesis by Kimmel et al. (1995),the period data presented here elimi-nates one pillar of the “two genetic sys-tems” hypothesis used to explain poste-rior specific segmentation defects inDelta-Notch mutant zebrafish lines(Holley, 2006). Instead, our period dataare consistent with a single genetic sys-tem regulating the timing of trunksomitogenesis, in line with the “run-down” hypothesis first proposed byJiang et al. (2000), and recently sub-stantiated by Riedel-Kruse et al. (2007).

Somitogenesis in the tail was re-ported to occur with a longer periodthan in the trunk in zebrafish (Kim-mel et al., 1995; Schmidt and Starck,2004). Here, we demonstrate andquantify this slowing by measuringsomitogenesis period in the nicb107

strain at high temporal resolution.The increase of period in the tail isgradual, such that it is not possible toexactly define its onset, although it isevident by somite 20. Furthermore,the increase in somite number overtime is nonlinear, indicating thatthere is no sudden transition or switchto a different period. An increase ofsomitogenesis period in the tail hasalso been noted in plaice (Brooks andJohnson, 1994), mouse (Tam, 1981)and amphioxus (Schubert et al.,2001), but not in the amphibians Xe-nopus or Rana (Pearson and Elsdale,1979). These data indicate that an in-crease in tail somitogenesis period islikely the ancestral condition in verte-brates.

Temperature Compensationof Axial Somite LengthDistribution

The distribution of instantaneoussomite length along the zebrafish axisshows a characteristic signature:Somite length increases in the ante-rior trunk, reaching a plateau of ap-proximately constant length in themid-trunk. In the posterior trunk,there is a striking decrease in somitelength despite a constant somitogen-esis period. In the tail, where periodincreases slightly, there is a furtherlength decrease. This distribution ofsegment size may be adaptive for thespecific hydrodynamic problems en-countered by small larvae, allowingfor a more flexible posterior body andthus a better anguiliform swimmingmode (Webb and Weihs, 1986; Mullerand van Leeuwen, 2004; McHenry andLauder, 2006). However, because am-niote mouse and chick embryos (Herr-mann et al., 1951; Tam, 1981) show astrikingly similar profile of somitelength along the axis, it seems likelythat conserved developmental con-straints also underlie this commonfeature of vertebrate somitogenesis.

In their natural habitats, develop-ing zebrafish are exposed to a widevariety of temperatures (Engeszer etal., 2007). We show here that thedistribution of zebrafish somitelength along the axis is independentof the steady state developmentaltemperature. Changes in the periodof somitogenesis due to temperatureare coordinated with altered growthsuch that the proportions of the seg-ments are maintained: in otherwords, segment length throughoutsomitogenesis is temperature-com-pensated. In the Drosophila embryo,the spatial positioning of segmentborders is also temperature-compen-sated (Houchmandzadeh et al.,2002), and we expect that mecha-nisms that shield the proportions ofpoikilothermic embryos from the in-fluence of developmental tempera-ture will be essential.

The Clock and WavefrontModel in the DevelopingEmbryo

The Clock and Wavefront model hasbeen widely and successfully used to

550 SCHROTER ET AL.

interpret specific genetic and embry-ological perturbations of somitogen-esis. In textbook illustrations ofwild-type somitogenesis, oscillatorperiod, somite length, and wavefrontvelocity are assumed to be constantalong the axis—that is, the model’sdynamics are at steady state (e.g.,Wolpert, 2006). Cooke and Zeeman(1976) originally discussed the possi-bility that period, length, and wave-front might change during develop-ment. Here, we quantify the changesin these primary variables through-out somitogenesis of wild-type ze-brafish. Thus, we provide a descrip-tion of the process that can be usedto adjust the steady state Clock andWavefront model to the developingembryo.

The changes in somitogenesis pe-riod and somite length along the axisof the developing zebrafish are moststriking in the tail, where segmentlength decreases as somitogenesisperiod increases. In a strict interpre-tation of the Clock and Wavefrontmodel, an increase in somitogenesisperiod should result in longer ratherthan shorter somites. Our appar-ently paradoxical findings are re-solved by postulating that the wave-front velocity also changes along theembryonic axis, independently fromthe somitogenesis oscillator period,and that its velocity slows strikinglyin the posterior half of the embryo.Only in the central trunk are somi-togenesis period and somite length,and hence wavefront velocity, all ap-proximately constant. Therefore, inthis region, the steady-state Clockand Wavefront model best approxi-mates the situation in the develop-ing embryo.

What are the cellular and molecu-lar underpinnings of the observedchanges in somitogenesis period andwavefront velocity along the axis?Both the clock and wavefront phe-nomena arise in the PSM. Conver-gence and extension movements andcell proliferation supply the PSMwith cells, whereas the wavefront ac-tivity concomitantly diminishes thecell number in the PSM by arrestingoscillations and gating cells anteri-orly for differentiation. In embryoswhere tissue size has been artifi-cially decreased, segment lengthadapts to the total length of tissue

available, preserving the relativeproportions of the embryo (Cooke,1975; Tam, 1981). The initial pur-pose of the Clock and Wavefrontmodel was to account for the appar-ent conservation of segment numberunder these circumstances by cou-pling the wavefront velocity to ini-tial tissue size (Cooke and Zeeman,1976). It is evident from our moviesthat the PSM size decreases signifi-cantly during somitogenesis (Fig. 4and Supplementary Movie S3). Wespeculate that a mechanism contin-uously coupling the wavefront veloc-ity to tissue size might be responsi-ble for the decreasing somite lengthin a single embryo along the devel-oping axis.

Gradients in signaling molecules pro-duced in the tail bud have been invokedto position the wavefront during somi-togenesis (Dubrulle et al., 2001;Sawada et al., 2001; Aulehla et al.,2003; Diez del Corral et al., 2003;Dubrulle and Pourquie, 2004b). Themovement of the wavefront relative tothe most recently formed somite is afunction of the continuous posteriormovement of the tail bud and the profileof the signaling gradients it generates.We expect that, in a PSM of decreasingsize, the flux of signaling moleculesfrom the posterior will diminish, result-ing in altered profiles of signaling gra-dients. Thus, tissue size-dependent al-terations in signaling flux might be onemechanism for coupling tissue size towavefront movement. Similarly, periodmay be controlled by a factor that issynthesized in the tail bud in proportionto the resident cell number. Limitedsupply of the period-controlling factor ina continuously shrinking PSM mightunderlie the gradual increase in somi-togenesis period in the tail.

In summary, our findings lead usto propose a modification of thesteady-state Clock and Wavefrontmodel in the zebrafish. The adjustedmodel does not alter its basic struc-ture, but the specific values ofits variables change in an axial- andtemperature-dependent manner.The incorporation of the molecularand cellular dynamics of the PSMinto the Clock and Wavefront modelto explain these axial and tempera-ture changes is the current chal-lenge.

EXPERIMENTALPROCEDURES

Preparation of Embryos,Molds

Zebrafish embryos were obtained bynatural spawning. Wild-type embryoswere AB strain if not otherwise stated.Tail somitogenesis was recorded fromnicb107 homozygous embryos (Wester-field et al., 1990). Dechorionation wasperformed between shield and budstage if necessary. Embryos wereplaced in depressions molded by sili-cone cones in 2% agarose/E3 in a Petridish. Silicone cones were made as Syl-gard 184 (Dow Corning) casts of cus-tom-milled molds either producedfrom aluminum or Lucite. Two andtwo-tenths-mm-wide � 2.5-mm-deepconical wells were used for embryoswith chorions or for dorsal viewingduring early somitogenesis. Conicaldepressions 0.4 to 0.5 mm deep wereused to analyze tail outgrowth. To ob-serve anterior trunk somite forma-tion, dechorionated embryos were ori-ented with the equator pointingtoward the objective shortly before thestart of recording. When measuringsomitogenesis period in the anteriortrunk, care must be taken to avoidtemperature differences between in-cubator, dissection microscope, andtime-lapse microscope. Embryos ana-lyzed for tail outgrowth were orientedlaterally around bud stage.

Generation of Time-LapseMovies

Time-lapse series were recorded on aZeiss Axioskop 200M with a �5 PlanNEOFLUAR objective using standardbrightfield optics. For tail outgrowthimaging, a �0.63 tube lens (Zeiss TV2/3“C) was inserted in front of the cam-era. Images were acquired with a Pho-tometrics Coolsnap HQ Camera. Themicroscope was equipped with a mo-torized stage (Zeiss MCU 28) drivenby MetaMorph software (version6.2r4, Universal Imaging Corp.).MetaMorph’s MultiDimensional Ac-quisition tool was used to scanthrough up to 41 positions per exper-iment. For each position, a Z-stackconsisting of five planes with 50 �mspacing was recorded. The time inter-val was one frame every 5 min, exceptfor experiments at temperatures be-


low 25°C, where the sampling ratewas reduced to 8 min.

ImageJ was used to produce focusedand time-stamped movies. The Im-ageJ Stack Focuser plugin (http://rsb.info.nih.gov/ij/plugins/stack-focuser.html)selects and combines the in-focus areasfrom different Z-planes of one timepoint. All focused images from one posi-tion were combined into a time-stack,time-stamped, and saved both as aTIFF-stack and QuickTime movie. Wewrote a multithreaded Java plugin,combining Stack Focuser, TimeStamper, and Movie Writer plugins tosimultaneously process data from mul-tiple embryos in parallel. This plugin isavailable from the authors upon re-quest.

Determination ofSomitogenesis Period andSomite Length

Somitogenesis movies were analyzedvisually by annotating the time ofsomite boundary formation. A bound-ary was considered formed when itwas clearly visible and spanned thetissue either from axial to lateral (dor-sal view) or from dorsal to ventral (lat-eral view). If embryo angle precludedobserving the somitic furrow as a darkline in lateral view, the appearance ofa notch-like structure in the dorsallimit of the somitic mesoderm wasconsidered indicative of boundary for-mation. To normalize for slightly differ-ent spawning times, we set the time atwhich somite boundary 6 formed to zeroand calculated the times at which pre-vious and following boundaries formedrelative to this reference point.

For period measurements, 8 to 16embryos were analyzed per experi-ment and view, and normalized meanformation time and SD for everyboundary was calculated. We assignedthe highest SD that was obtained inthe experiment as the SD for forma-tion time of somite 6. Somitogenesisperiod was determined by weightedlinear regression using Origin soft-ware (OriginLab). Period values aregiven as the slope of the linear regres-sion, errors represent the 95% confi-dence interval.

Somite length was measured in lat-eral view from focused TIFF-stacks inImageJ. The program was calibratedusing a 1 mm stage micrometer pho-

tographed with the same microscopesettings. The length of each somitealong the axis was measured from theframe in which it was first consideredformed. A straight line was drawnconnecting the intersections of thedorsal limit of the notochord with thetwo most recently formed somiteboundaries. If embryo angle precludedobserving the somitic furrow as a darkline, auxiliary lines connecting ipsilat-eral notch-like structures radiallywith the surface of the yolk ball wereused as indicators of boundary position.The flattened geometry of the paraxialmesoderm in lateral view at the time ofanterior segment formation preventedanalysis of some segments. Between 8and 12 embryos were measured in eachlength experiment. Somite lengths arestated as mean � SD.

Temperature Experiments

The room with the time-lapse equip-ment was heated with an electric ra-diator (AKO-ISMET, type R909TSIII); temperature regulation wasachieved using a built-in air-condi-tioning system (Silent, Axair). Tem-perature inside the dish was mea-sured with a K-type thermocoupledipped into the agarose and connectedto a data logger system (Voltcraft PlusK202). It was re-calibrated using icewater and boiling water as references.Temperature values were recorded ev-ery minute. Mean temperature � SDduring the experiments is stated.

ACKNOWLEDGMENTSWe thank the Oates Lab, Saul Ares,Luis Morelli, Ingmar Riedel-Kruse,Carl-Philipp Heisenberg, and LaurelRohde for helpful discussions on themanuscript, the MPI-CBG fish facilityfor healthy and fertile fish, and BrittaSchroth-Diez and the Howard Lab forhelp with microscopy. We acknowl-edge the mechanical workshop atMPI-CBG, Dresden, and Justin Costaand the NIH Mechanical Instrumen-tation Design and Fabrication branch.Healthy, adult nicb107 fish were pro-vided by ZIRC. This work was sup-ported by the Intramural ResearchProgram of the National Human Ge-nome Research Institute, National In-stitutes of Health.

REFERENCES

Aulehla A, Pourquie O. 2006. On periodic-ity and directionality of somitogenesis.Anat Embryol (Berlin) 211(Suppl)1:3–8.

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.

Brooks S, Johnson IA. 1994. Temperatureand somitogenesis in embryos of the pla-ice. J Fish Biol 45:699–722.

Bunning E. 1963. The physiological clock.Berlin: Springer-Verlag.

Cooke J. 1975. Control of somite numberduring morphogenesis of a vertebrate,Xenopus laevis. Nature 254:196–199.

Cooke J, Zeeman EC. 1976. A clock andwavefront model for control of the num-ber of repeated structures during animalmorphogenesis. J Theor Biol 58:455–476.

Diez del Corral R, Olivera-Martinez I, Go-riely A, Gale E, Maden M, Storey K.2003. Opposing FGF and retinoid path-ways control ventral neural pattern,neuronal differentiation, and segmenta-tion during body axis extension. Neuron40:65–79.

Dubrulle J, Pourquie O. 2004a. Couplingsegmentation to axis formation. Develop-ment 131:5783–5793.

Dubrulle J, Pourquie O. 2004b. fgf8 mRNAdecay establishes a gradient that couplesaxial elongation to patterning in the ver-tebrate embryo. Nature 427:419–422.

Dubrulle J, McGrew MJ, Pourquie O. 2001.FGF signaling controls somite boundaryposition and regulates segmentationclock control of spatiotemporal Hox geneactivation. Cell 106:219–232.

Engeszer R, Patterson L, Rao A, Parichy D.2007. Zebrafish in the wild: a review ofnatural history and new notes from thefield. Zebrafish 4:21–40.

Guryev V, Koudijs MJ, Berezikov E, John-son SL, Plasterk RH, van Eeden FJ,Cuppen E. 2006. Genetic variation in thezebrafish. Genome Res 16:491–497.

Hanneman E, Westerfield M. 1989. Earlyexpression of acetylcholinesterase activ-ity in functionally distinct neurons of thezebrafish. J Comp Neurol 284:350–361.

Henry CA, Hall LA, Burr Hille M, Solnica-Krezel L, Cooper MS. 2000. Somites inzebrafish doubly mutant for knypek andtrilobite form without internal mesen-chymal cells or compaction. Curr Biol 10:1063–1066.

Herrmann H, Schneider MJB, Neukom BJ,Moore JA. 1951. Quantitative data onthe growth process of the somites of thechick embryo: linear measurements, vol-ume, protein nitrogen, nucleic acids. JExp Zool 118:243–268.

Holley SA. 2006. Anterior-posterior differ-ences in vertebrate segments: specifica-tion of trunk and tail somites in the ze-brafish blastula. Genes Dev 20:1831–1837.

Houchmandzadeh B, Wieschaus E, LeiblerS. 2002. Establishment of developmentalprecision and proportions in the early

552 SCHROTER ET AL.

Drosophila embryo. Nature 415:798–802.

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.

Johnston IA, Vieira VV, Abercromby M.1995. Temperature and myogenesis inembryos of the Atlantic herring Clupeaharengus. J Exp Biol 198:1389–1403.

Kimmel CB, Ballard WW, Kimmel SR,Ullmann B, Schilling TF. 1995. Stages ofembryonic development of the zebrafish.Dev Dyn 203:253–310.

Kulesa PM, Fraser SE. 2002. Cell dynam-ics during somite boundary formation re-vealedbytime-lapseanalysis.Science298:991–995.

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 U S A 103:1313–1318.

McHenry MJ, Lauder GV. 2006. Ontogenyof form and function: locomotor morphol-ogy and drag in zebrafish (Danio rerio). JMorphol 2006:1099–1109.

Muller UK, van Leeuwen JL. 2004. Swim-ming of larval zebrafish: ontogeny of

body waves and implications for locomo-tory development. J Exp Biol 207:853–868.

Pearson M, Elsdale T. 1979. Somitogenesisin amphibian embryos. I. Experimentalevidence for an interaction between twotemporal factors in the specification ofsomite pattern. J Embryol Exp Morphol51:27–50.

Pittendrigh CS. 1954. On temperature in-dependence in the clock system control-ling emergence time in Drosophila. ProcNatl Acad Sci U S A 40:1018–1029.

Riedel-Kruse I, Muller C, Oates A. 2007.Synchrony dynamics during initiation,failure, and rescue of the segmentationclock. Science 317:1911–1915.

Sawada A, Shinya M, Jiang YJ, KawakamiA, Kuroiwa A, Takeda H. 2001. Fgf/MAPK signalling is a crucial positionalcue in somite boundary formation. De-velopment 128:4873–4880.

Schmidt K, Starck JM. 2004. Developmen-tal variability during early embryonicdevelopment of zebra fish, Danio rerio. JExp Zoolog B Mol Dev Evol 302:446–457.

Schubert M, Holland LZ, Stokes MD, Hol-land ND. 2001. Three amphioxus Wntgenes (AmphiWnt3, AmphiWnt5, andAmphiWnt6) associated with the tailbud: the evolution of somitogenesis inchordates. Dev Biol 240:262–273.

Tam PP. 1981. The control of somitogen-esis in mouse embryos. J Embryol ExpMorphol 65(Suppl):103–128.

van der Meulen T, Schipper H, van Leeu-wen JL, Kranenbarg S. 2005. Effects ofdecreased muscle activity on developingaxial musculature in nicb107 mutant ze-brafish(Daniorerio).JExpBiol208:3675–3687.

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.

Webb PW, Weihs D. 1986. Functional loco-motor morphology of early life historystages of fishes. Trans Am Fish Soc 115:115–127.

Westerfield M, Liu DW, Kimmel CB,Walker C. 1990. Pathfinding and syn-apse formation in a zebrafish mutantlacking functional acetylcholine recep-tors. Neuron 4:867–874.

Wolpert L. 2006. Principles of develop-ment. Oxford: Oxford University Press.

Wood A, Thorogood P. 1994. Patterns ofcell behaviour underlying somitogenesisand notochord formation in intact verte-brate embryos. Dev Dyn 201:151–167.


Documents

Dynamics of zebrafish somitogenesis