Dev Genes Evol (2003) 213:541–553DOI 10.1007/s00427-003-0360-6

O R I G I N A L A R T I C L E

Christoph Winkler · Harun Elmasri · Barbara Klamt ·Jean-Nicolas Volff · Manfred Gessler

Characterization of hey bHLH genes in teleost fish

Received: 26 June 2003 / Accepted: 23 September 2003 / Published online: 31 October 2003� Springer-Verlag 2003

Abstract Hairy-related basic helix-loop-helix (bHLH)transcription factors are targets of Delta-Notch signalingand represent essential components for a number of cellfate decisions during vertebrate embryogenesis. Heygenes encode a subfamily of hairy-related proteins thathave been implicated in processes like somitogenesis,blood vessel and heart development. We have identifiedand characterized hey genes in three teleost fish lineagesusing degenerate PCR and database searches. Phyloge-netic analysis of Hey proteins suggests a complex patternof evolution with high divergence of hey2 in Takifugurubripes (Fugu, Japanese pufferfish) and possibly loss inthe related Tetraodon nigroviridis (the freshwater puffer-fish). In addition, duplication of hey1 in both pufferfishes,Fugu and Tetraodon, was observed. Conversely, zebrafish(Danio rerio) has the same complement of three heygenes as known from mammals. All three hey genes showmuch more restricted gene expression profiles in ze-brafish when compared to mouse. Importantly, while allthree murine Hey genes are expressed in overlappingpatterns in the presomitic mesoderm (PSM) and somites,in zebrafish only hey1 shows PSM and somite expressionin a highly dynamic fashion. Therefore, while overlappingexpression might account for redundancy of hey functionin higher vertebrates, this is unlikely to be the case inzebrafish. In deltaD (dlD) deficient after-eight zebrafishmutants, the dynamic expression of hey1 in the PSM isimpaired and completely lost in newly formed somito-meres. Overexpression of dlD on the other hand results inthe ectopic expression of hey1 in the axial mesoderm.Hence, hey1 represents a target of Delta-Notch signalingdynamically expressed during somite formation in ze-brafish.

Keywords hey genes · bHLH · Delta-Notch targets ·Somitogenesis · Heart development

Introduction

Hey genes, also known as Hesr, HRT, CHF or HERPgenes, encode “Hairy and Enhancer-of-split”(Hes)-relatedbasic helix-loop-helix (bHLH) transcription factors. Theyhave characteristic C-terminal YRPW motifs and, com-pared to Hairy, a proline to glycine substitution in theirbasic domain (Kokubo et al. 1999; Leimeister et al. 1999;Nakagawa et al. 1999; Chin et al. 2000; Iso et al. 2001a).There are three members in the human and mousegenomes and a single Drosophila counterpart. Hey genesrepresent primary targets of the Delta-Notch signalingpathway (Maier and Gessler 2000). In transgenic mice,overexpression of an activated Notch variant in hairfollicles resulted in the upregulation of HeyL (Lin et al.2000), while expression of all three Hey genes wasdownregulated in mice mutant for the Notch ligand Dll1(Delta-like1; Leimeister et al. 2000a, 2000b). In Xenopus,Xhey-1 expression was altered in response to bothectopically activated or suppressed Notch signaling(Pichon et al. 2002; Rones et al. 2002). Hey proteinsform homodimers, as well as heterodimers with otherHairy-related Hes proteins, like Hes1 in mouse and c-Hairy1 in chicken (Leimeister et al. 2000b; Fischer et al.2002). They bind to so-called E-box motifs on DNA andact as transcriptional repressors (Iso et al. 2001a, 2001b;Fischer et al. 2002).

Hey genes generally show complex and broad patternsof expression in mouse and they have been implicated aseffectors of Delta-Notch-dependent cell fate decisions(Kokubo et al. 1999; Leimeister et al. 1999, 2000a,2000b; Nakagawa et al. 1999). In zebrafish, the ortho-logue of Hey2 has been identified as the gridlock (grl)gene, whose mutation leads to defective aortic develop-ment (Zhong et al. 2000). More recent work has suggestedthat grl participates in the specification of arterial vsvenous cell fate in preangioblasts (Zhong et al. 2001). On

Edited by D. Tautz

C. Winkler ()) · H. Elmasri · B. Klamt · J.-N. Volff · M. GesslerDepartment of Physiological Chemistry I, Biocenter,University of Wuerzburg,Am Hubland, 97074 Wuerzburg, Germanye-mail: winkler@biozentrum.uni-wuerzburg.deTel.: +49-931-8884142Fax: +49-931-8884150

the other hand, knock-out mice deficient for Hey2developed severe cardiac hypertrophy and ventricleseptum defects, whereas the dorsal aorta was not affected(Donovan et al. 2002; Fischer et al. 2002; Gessler et al.2002; Sakata et al. 2002). These differences in phenotypesbetween fish and mouse provided first indications that heyorthologues, although structurally conserved, might havedistinct roles in different organisms. Furthermore, bothmouse and chicken hey2 orthologues exhibit oscillatingexpression in the presomitic mesoderm that has beenlinked to the somitogenesis clock (Leimeister et al.2000b). Thus, it was surprising that the description of thegrl mutant did not include any alteration of somitogenesisor somite patterning. This could be due to either geneticredundancy caused by other hairy-related genes, or elsedifferential evolution of hey gene function after geneduplication in fish.

In order to initiate a functional comparison of verte-brate hey orthologues, we identified hey genes in thezebrafish Danio rerio, the seawater Fugu Takifugurubripes and the freshwater pufferfish Tetraodon ni-groviridis. Through PCR screening and mining of genomeproject databases we identified three hey genes inzebrafish that represent the orthologues of Hey1, Hey2and HeyL from higher vertebrates. Surprisingly, four heygenes exist in the genome of Takifugu rubripes, while theclosely related Tetraodon nigroviridis apparently losthey2 and therefore retained only three hey genes. In thisreport, we characterize the phylogenetic relations betweenall identified hey genes that indicate complex scenariosduring vertebrate evolution. We describe the dynamicexpression of the zebrafish hey genes and show that theyexhibit much more restricted expression patterns com-pared to higher vertebrates. Most importantly, hey1 is theonly hey gene in zebrafish that shows an expressionpattern which is consistent with a function duringsomitogenesis.

Materials and methods

Wild-type and mutant fish

Fish were reared and embryos obtained as described (Westerfield1993) from wild-type stocks T�-AB and K-WT kept as inbredstrains in our fish facility for many generations. Mutant fish werekindly provided by Artemis. The mutant allele used was after-eight(aei, deltaD; allel tr233). Embryos were raised at 28–30�C andstaged according to Kimmel et al. (1995) using standard morpho-logical criteria.

PCR amplification of zebrafish hey sequences

Zebrafish hey gene sequences were amplified from cDNA andgenomic DNA by PCR using various combinations of a series ofdegenerate primers located in the highly conserved basic, helix-loop-helix and orange domains or the carboxyterminus. Degenerateprimers were designed using conserved mammalian, chicken andDrosophila Hey protein sequence blocks and the CODEHOPprogram as described before (Leimeister et al. 2000b). Thefollowing primers were used as 50 or 30 primers in differentcombinations: allhey5a (50-CGAAAGAAGCGACGAGGART-

NATHGARAA-30), allhey5i (50-ACTGGTGCCTACCGCCTTY-GARAARCARG-30), allhey3a (50-CCACCTCGGTCAGGCAYT-CNCKRAA-30), allhey3i (50-CACCTCGGTCAGGCACTCNCK-RAANCC-30), codeblik2 (50-AAGGCCGAGATCCTGCARATG-ACNGT-30) and flik-code-30 (50-GCGTTAGAAGGCGCCGATC-TCNGTDATYTC-30).

Cloned PCR products and available EST clones were sequencedusing vector or custom primers. Sequence analysis was done usingNCBI BLAST servers, the GCG package (Version 10.0, Accelrys)and ClustalX (Thompson et al. 1997). More than 100 cloned PCRproducts were analyzed which resulted in the repeated isolation ofonly three classes of sequences. The deduced protein sequencesshowed high similarity to human Hey1, Hey2 or HeyL. For hey1,the complete cDNA sequence of 3.3 kb was later determined fromzebrafish kidney EST clone fo76e12 (GenBank BG727748). Ahey2 (grl) cDNA sequence of 1.9 kb has been published before(Zhong et al. 2000). The zebrafish heyL cDNA sequence wascomplemented by a recent 50 EST sequence (fk74e12, GenBankBE017032) and 30RACE products that encompass 1.6 kb. For30RACE, the primer RXGT12 (50-CGGAATTCTCGAGAT-CTTTTTTTTTTTT-30) was used for cDNA synthesis and ampli-fication was performed with RXGT12 and zf-heyx-50Bam (50-GC-GGGATCCGGACTGAGGAACCTGTGGAG-30). The obtainedzebrafish hey sequences were submitted to GenBank (accessionnumbers AJ510221 for hey1 and AJ510222 for heyL).

In situ hybridization

For preparation of a zebrafish hey1 riboprobe, the part of the hey1coding region corresponding to nt 391–959 of AJ510221 wascloned into pDK101 (clone zf-hey1-e6). This vector was used togenerate the antisense riboprobe by NotI digestion and T7transcription.

Zebrafish hey2 (gridlock) was amplified from cDNA preparedfrom 24 hpf embryos using the primers zfgrd-50 (50-CGAGGATC-CATCATGAAGCGGCCCTGTG-30) containing a terminal BamHIsite and zfgrd-30 (50-GCGGAATTCTGCTGACCGAAGCAGGCA-CAA-30) containing a terminal EcoRI site. The amplified fragmentwas sequenced, digested with BamHI and EcoRI and then clonedinto the corresponding sites of the expression vector pCS2P+ toresult in pCS2-zf-hey2. For preparation of RNA antisense probes,this plasmid was linearized with HindIII, and the T7 RNApolymerase was used for in vitro transcription.

A zebrafish heyL riboprobe construct was prepared as aboveusing the primers zfheyx-50 (50-GCGAGATCTGGACTGAGGAA-CCTGTGGAG-30) containing a terminal BglII site and zfheyX-30

(50-GCGGAATTCCCAAACTCCAAACTTGAGG-30) containing aterminal EcoRI site. For preparation of RNA antisense probes, theplasmid pCS2-zf-heyL was linearized with HindIII, and T7polymerase was used for in vitro transcription. One- and two-colorin situ hybridizations were performed as described previously(Winkler and Moon 2001).

Phylogenetic analysis

Multiple sequence alignments were generated using “PileUp” of theGCG Wisconsin package or ClustalX. Phylogenies were deter-mined with PAUP* (D.L. Swofford, Smithsonian Institution,Washington, D.C.) by bootstrap analysis using maximum parsimo-ny (100 replicates) and neighbor-joining (1,000 replicates; Saitouand Nei 1987). Maximum likelihood analysis was performed byquartet puzzling (10,000 puzzling steps; Strimmer and von Haeseler1996) using TREE-PUZZLE 5.0 (Schmidt et al. 2002). Thepairwise number of synonymous vs non-synonymous substitutionsper site between two aligned sequences was estimated with“Diverge” of GCG using an unambiguous alignment of nucleotidesequences (495 nt in length) encoding the N-terminal, basic, helix-loop-helix and orange domains. Gene structure was analyzed usingprograms available on the NIX server (http://menu.hgmp.mrc.a-

542

c.uk/menu-bin/Nix). Deduced sequences from Takifugu rubripesand Tetraodon nigroviridis are available on request.

Microinjection of RNA into zebrafish embryos

Capped RNA encoding DeltaD was obtained from the expressionvector pCS2-DeltaD kindly provided by J. Campos-Ortega (Dorn-seifer et al. 1997). In vitro transcription of the NotI linearizedvector using the Sp6 mMessage mMachine kit from Ambion wasperformed as described previously (Winkler and Moon 2001). TheRNA concentration was determined photometrically and 350 pgRNA were injected into the yolk of embryos just underneath singleblastomeres of 2- to 4-cell-stage zebrafish embryos. Injectedembryos were raised at 28�C until they reached the desired stageand then processed for in situ hybridization.

Results

Isolation and structural characterization of fish hey genes

Zebrafish hey gene sequences were amplified by degen-erate and RACE PCR, as well as database searches (fordetails see Materials and methods). Three classes ofsequences were obtained, the deduced protein sequencesof which showed high similarity to either human Hey1,Hey2 or HeyL (Fig. 1). Extensive searches of zebrafishEST and genomic trace sequence databases as well as thenewly released second genomic assembly did not revealthe presence of additional hey genes. In contrast, four heygenes were identified in the draft of the genome of theJapanese pufferfish Takifugu rubripes (Aparicio et al.2002; http://fugu.hgmp.mrc.ac.uk/blast/), one correspond-ing to each mammalian Hey2 and HeyL prototype, andtwo representing different hey1 gene copies (Figs. 1, 2).The Takifugu hey1a, hey1b, hey2 and heyL genes havealso been described in an earlier report in which they werenamed Frhey1.1, Frhey1.2, Frhey2 and Frhey3, respec-tively (Gajewski and Voolstra 2002). In a similarapproach the Tetraodon nigroviridis trace file archives(Roest Crollius et al. 2000;http://www.genoscope.cns.fr;http://www-genome.wi.mit.edu/annotation/tetraodon/)were searched yielding three different hey genes. Basedon gene structure prediction analysis as well as thededuced exon-intron structure that fits to the mammaliangene organization, the putative complete coding regionswere assembled for each Takifugu rubripes and Tetra-odon nigroviridis hey gene.

The protein sequences for all known human, mouse,chicken, zebrafish and pufferfish Hey proteins werealigned to build a phylogenetic tree. A high similaritywithin the orange domain, a region with so far unknownfunction (Dawson et al. 1995), the bHLH and to a lesserextent in the carboxyterminus was observed (Fig. 1). Theimmediate aminoterminus and most of the carboxytermi-nal half of the proteins were less conserved and proved tobe rather specific for each family member regardless ofspecies.

Ancient duplication of hey1 in the fish lineage

While single copies of hey2 (accession FT:T014948) andheyL (FT:T005657) were detected in Takifugu rubripes,two hey1 genes, now called hey1a (FT:T004546) andhey1b (FT:T000531), were found (Figs. 1, 2). Both hey1aand hey1b were identified in Tetraodon nigroviridis aswell (Figs. 1, 2). The average ratio of synonymous vsnon-synonymous substitutions was 9.8 (from 8.7 to 11.4)between pufferfish hey1 paralogues (i.e. between hey1aand hey1b genes), 34.3 between Tetraodon nigroviridisand Takifugu rubripes hey1a orthologues, and 34.2between T. nigroviridis and T. rubripes hey1b ortho-logues, compared to 43.8 (from 32.6 to 60.4) betweenhey1 orthologues from different higher vertebrate species.This clearly indicates that both fish hey1 duplicatesglobally evolved under purifying negative selection afterduplication, and therefore probably do not correspond topseudogenes. The fact that the ratio of synonymous vsnon-synonymous substitutions tended to be smallerbetween pufferfish hey1 paralogues than for the uniquehey1 gene in higher vertebrates might be due to relaxedselective constraints during a certain period of evolutionafter duplication (Lynch and Conery 2000). Nevertheless,a more stringent selection was restored on each hey1duplicate, since the ratio observed between both hey1aorthologues as well as between both hey1b orthologueswas closer to that obtained for hey1 in higher vertebrates.

Differential evolution of hey2 in fish

Unique hey2 (accession FT:T014948) and heyL(FT:T005657) genes were detected in Takifugu rubripes.The degree of conservation of Hey2 compared to highervertebrates was strikingly much lower in T. rubripes thanin zebrafish (Fig. 3). Zebrafish and T. rubripes Hey2proteins showed on average 81.3% (79% identity) and69.7% similarity (65.8% identity), respectively, withHey2 from higher vertebrates. Zebrafish Hey2 was evenmore similar to Hey2 from mammals and chicken than toHey2 from T. rubripes (73.6% similarity, 67.6% identity).This was reflected by the Hey phylogeny, where T.rubripes Hey2 was pulled down to the root of the Hey2phylogenetic group probably because of its higherdivergence (Fig. 2). Even in the sequence encoding thehighly conserved basic helix-loop-helix and orangedomains, T. rubripes hey2 accumulated on average 3times more non-synonymous (protein-changing) substitu-tions than zebrafish hey2 when compared to mouse andhuman Hey2 (21.8% vs 7.1%, non-synonymous substitu-tions per 100 non-synonymous sites). Zebrafish andTakifugu hey2 were—for non-synonymous changes—even more divergent (18.6%) than zebrafish compared tohigher vertebrates (7.1% on average, 5.8% and 8.4%divergence with human and mouse, respectively). TheHey2 protein, particularly downstream of the orangedomain, was much shorter in T. rubripes (264 aa) than inzebrafish (324 aa) and higher vertebrates (335–339 aa;

543

Fig. 3). All together, these observations indicate that hey2experienced a much more relaxed selection during acertain period of evolution in the lineage that led to T.rubripes after its separation from the zebrafish lineage.

Since Takifugu rubripes and Tetraodon nigroviridisare “only” separated by 20–30 million years, it appearedof great interest to compare the structure of hey2 in bothpufferfishes. A total of about 3.6 million genomic

Fig. 1 Sequence comparison of vertebrate Hey proteins. Identicalresidues are in black, conservative substitutions in gray (drawn withMacBoxshade). Basic, helix-loop-helix and orange domains areshown, the YRPW motif is overlined. Accession numbers: Hey1

Danio rerio AJ510221; Hey1 Homo sapiens CAB75715; Hey2/Gridlock D. rerio AAF44780; Hey2 H. sapiens NP_036391; HeyLD. rerio AJ510222; HeyL H. sapiens NP_055386

544

sequences from Tetraodon nigroviridis kindly providedby L. Bouneau (Genoscope, http://www.genoscope.cns.fr)or made available by the Whitehead Institute/MIT, Centerfor Genome Research (http://www.ncbi.nlm.nih.gov/blast/mmtrace.html) were screened using Takifugu ru-

bripes Hey1a/b, Hey2 and HeyL nucleotide and proteinsequences as queries. Nineteen sequences matching toHey1a, 29 to Hey1b and 28 to HeyL were unambiguouslyidentified in Tetraodon nigroviridis, corresponding tounique hey1a, hey1b and heyL genes (Figs. 1, 2). Incontrast, no hey2 sequence could be found in Tetraodonnigroviridis, even under stringency conditions detectingother non-hey basic helix-loop-helix sequences. Sincethere is no high quality draft from the genome of T.nigroviridis available at the moment, we can notcompletely exclude that hey2 is present but has not beensequenced so far. Nevertheless, our analysis stronglysuggests that hey2 has been lost in T. nigroviridis.

Differential expression of hey genesduring zebrafish embryogenesis

We investigated the expression of zebrafish hey genesduring embryonic development using RNA whole-mountin situ hybridization in order to compare patterns to thosereported earlier for mouse, chick and Xenopus (Leimeisteret al. 1999, 2000a; Pichon et al. 2002; Rones et al. 2002).While hey genes in the mouse are transcribed in a broadrange of tissues derived from all three germ layers, theirexpression in fish seems much more restricted.

Certain aspects of hey2 (gridlock) expression at the 3-somite stage and in the dorsal aorta at later stages havebeen reported earlier (Zhong et al. 2000, 2001). Inaddition to this, we found dynamic hey2 expression in thedeveloping heart field and in neural tissues (Fig. 4). hey2expression can be detected as early as the 1-somite stage(Fig. 4A) making it one of the earliest heart field markersknown so far. Starting at this stage, expression is found intwo bilateral patches of cells located in the anterior region

Fig. 3 Sequence comparison of vertebrate Hey2 proteins. Identicalresidues are in black, conservative substitutions in gray (drawn withMacBoxshade). Basic, helix-loop-helix and orange domains are

shown, the YRPW motif is overlined. Accession numbers are givenin the legends of Figs. 1 and 2

Fig. 2 Phylogeny of vertebrate Hey proteins. The tree (neighbor-joining) is unrooted. In order to estimate the confidence level ofphylogenetic groups, bootstrap values (in percentage of bootstrapreplicates; Felsenstein 1985) using neighbor-joining (first values;1,000 replicates; Saitou and Nei 1987) and maximum parsimonyanalyses (third values; 100 replicates), as well as the reliabilityvalues for maximum likelihood analysis (second values; quartetpuzzling; 10,000 puzzling steps; Strimmer and von Haeseler 1996)are given. Branches with less than 50% support have beencollapsed. Accession numbers: Hey1 Canis familiaris CAB65543;Hey1 Mus musculus AAD38966; Hey1 Xenopus laevis CAB96791;Hey2 M. musculus NP_038932; HeyL M. musculus CAB71347;Hey2 Gallus gallus (Leimeister et al. 2000b). Other accessionnumbers are given in the legend of Fig. 1

545

546

of the lateral plate mesoderm (Fig. 4A, B). These regionsrepresent the cardiac precursor fields located left and rightof the midline. At the 6-somite stage, expression alsoemerges in the lateral mesoderm of the head (Fig. 4C) andtrunk (Fig. 4D). Initially, expression appears uniform inthe trunk lateral mesoderm. At the 16-somite stage,however, hey2 is restricted to a subset of segmentallypositioned cell clusters (Fig. 4E). Double-staining withmyoD shows that segmental hey2 expression in the lateralmesoderm is located at the level of the posterior half of

each somite, but does not overlap with myoD expressionin the paraxial mesoderm (Fig. 4F). The hey2-positive cellclusters represent angioblast precursor cells that have notstarted to migrate medially to form the primordia of thelarge axial vessel, the dorsal aorta. At around the 16-somite stage, we find expression of hey2 in the telen-cephalon, dorsal midbrain, neural crest and, as describedearlier (Zhong et al. 2000, 2001), in the dorsal aorta andits vascular progenitors (Fig. 4E–H). Expression is alsofound in the fusing heart tubes that give rise to a ring-likestructure at the midline (arrows in Fig. 4G, H). At the 22-somite stage (20 hpf), the angioblasts have completelyfused to form the early dorsal aorta that will subsequentlyform its lumen by 24 hpf (data not shown).

Both zebrafish hey1 (see later) and hey2 genes arestrongly expressed during zebrafish embryogenesis. Incontrast to this and to the situation in mouse, heyL showsonly low level expression in the embryo in a highlyrestricted fashion. We found expression of heyL in theventral fin fold of the forming and extending tailbud(Fig. 4I, K). Also, weak staining was observed in theventral-most portions of the neural tube (Fig. 4I–L). Crosssections revealed that the staining is located in the lateralfloor plate, as well as other ventral neural tube cell types.It is excluded from the medial floor plate. Interestingly, inmouse and Xenopus Hey1 is expressed in the floor plate(Leimeister et al. 1999; Pichon et al. 2002).

Zebrafish hey1 is dynamically expressedin the anterior presomitic mesoderm

While all three members of the Hey gene family in miceand chick are expressed in the presomitic mesoderm andits derivatives (Leimeister et al. 2000b), in zebrafish thisis only the case for hey1. Zebrafish hey1 expression startsdirectly after gastrulation is completed and precedes theformation of morphologically visible somites by severalhours. At the 1-somite stage, it shows a segmental modeof expression (Fig. 4M, N). Three to four domains of hey1expression can be seen in the paraxial mesoderm thatotherwise is still unsegmented at this early stage. At laterstages during somitogenesis (16 hpf), hey1 is expressed inthe telencephalon, the region of the otic vesicles andstrongly in the somites and the anterior PSM (Fig. 4O).Co-staining with myoD revealed that hey1 is co-expressedwith myoD in the caudal halves of all epithelializedsomites (Fig. 4P), as well as in one or two rhythmicallyemerging stripes in the anterior PSM (Fig. 4Q–U). AsmyoD is expressed in the adaxial mesoderm and in thesomitomere that is next to become an epithelializedsomite, it serves as a reference marker when identifyingthe relative position of hey1-positive cells in relation tothe PSM (Fig. 4P, myoD-expressing cells marked with redarrows). To analyze the dynamic behavior of its expres-sion in more detail, we collected a clutch of embryos atdifferent phases of the 11-somite stage (14 hpf) andstained for hey1. We identified essentially four differentphases of hey1 expression during one cycle of somite

Fig. 4A–U Expression of hey genes during zebrafish embryogen-esis. RNA whole-mount in situ hybridization using hey2 (A–H),heyL (I–L) and hey1 (M–U) antisense riboprobes. A Dorsal view ofan embryo at the 1-somite stage (10.5 hpf), anterior to the top.Arrows indicate hey2 expression in bilateral patches of lateral platemesoderm cells. B Dorsal view of cardiac fields (arrows) in anembryo at the 3-somite stage (11 hpf). C Dorsal view of thedeveloping head region of an embryo at the 6-somite stage (12 hpf).hey2 expression is seen in the precardiac fields (arrows) and theanterior lateral mesoderm (arrowheads). D Dorsal view of the trunkregion of the same embryo as shown in C with hey2 expression inthe posterior lateral mesoderm (arrows). Anterior cells start tomigrate medially. E Dorsal view of flat-mounted trunk region at the16-somite stage (anterior to the top). hey2 expression is seen inclusters of lateral mesoderm cells (arrows) and the dorsal aortaprecursors (da). Asterisks indicate segmentally arranged neuralcrest cells (nc) located in a different focal plane. F Non-overlappingexpression of myoD (red) and hey2 (blue) in paraxial and lateralmesoderm, respectively, of a 16-somite stage embryo (ventral view,anterior to the top). G Lateral view of a 17 hpf embryo at the 17-somite stage. Left arrow indicates hey2 expression in the region ofdeveloping heart (h), right arrow marks the forming dorsal aorta(da) and asterisk the expression in the dorsal midbrain. Segmentalexpression is found in the neural crest (nc). H Dorsal view of thehead region of a flat-mounted embryo at 17 hpf. Expression in thetelencephalon (arrowhead), posterior midbrain (asterisk) andcardiac progenitor cells (arrows) is indicated. I heyL expressionin the ventral fin fold ectoderm (VE) and the ventral neural tube(VN) seen in a lateral view of the tailbud region of a 14-somitestage embryo. J Cross section through an embryo at 20 hpf showingheyL expression in the ventral neural tube; the medial floor plateappears free of transcripts. K, L heyL expression at 24 hpf. Box inK indicates region shown in L. Note weak expression in cells of theventral neural tube (VN) directly adjacent to the notochord (N). M–U Dynamic expression of hey1 during somite formation inzebrafish. Lateral (M) and dorsal (N) views of embryos at the 1-somite stage (10.5 hpf), anterior to the top, showing segmental hey1expression in the anterior paraxial mesoderm (arrow). O Lateralview of embryo at the 11-somite stage showing her1 expression(red) in the PSM (arrowhead) and hey1 (blue) expression in thetelencephalon (asterisk), otic vesicles (large arrow), somites andthe PSM (small arrows). P Expression of hey1 (in blue) and myoD(in red) in a dorsal view of a flat-mounted trunk region at the 10-somite stage (14 hpf), anterior is to the top. The photograph hasbeen oversaturated using Adobe Photoshop to reveal co-expressionof hey1 with myoD in the most recently formed somites (S 1 and S2; labeled by red arrows). Black asterisks indicate expression ofhey1 in early stage somitomeres (S 0, S 1) devoid of myoDexpression. Q–T Dorsal views of flat-mounted tailbuds showingdynamic changes in hey1 expression in embryos at different phases(I–IV) of the 11-somite stage. Boxed areas demarcate the area ofsomitomere S 1, S 0 (marked by asterisk) and the anteriormost PSMregion. Staining in more posterior areas of the tailbud is located inthe ectoderm. U Lateral view of a tailbud at the 11-somite stageshowing hey1 expression in ectodermal cells (arrowheads), and thechordoneural hinge (small arrow), as well as in the anterior PSM(large arrow) and somitomeres (S 0 marked by asterisk)

547

548

formation. These were arbitrarily combined to result inthe series shown in Fig. 4 Q–U. In a first phase (I inFig. 4Q), we found hey1 transcription in a condensedstripe in the anterior PSM representing the caudal domainof the S 0 somitomere (identified by double-labeling withmyoD; for nomenclature of somitomeres, see Pourquieand Tam 2001). Other transcripts were diffusely dis-tributed across the remaining anterior PSM region. Notranscripts were detected in the posterior PSM, however,weak expression was observed in the ectodermal part ofthe posterior tailbud and in the chordoneural hinge(Fig. 4U). During subsequent phases, transcripts appearedto concentrate in the anteriormost PSM at a location thatwas about to become the caudal half of the next formingsomitomere (somitomere S 1; Fig. 4R). At this phase,hey1 transcript levels in the remaining PSM were reducedand eventually disappeared completely. After hey1 ex-pression reached a maximum in S 1 (Fig. 4S), transcriptsagain became detectable in the anterior half of the PSM(the transition zone) and the cycle apparently starts over.This highly dynamic transcription in the transition zone ofthe anterior PSM (boxed areas in Fig. 4Q–T) suggests acomplex regulation of hey1 possibly at both the tran-scriptional level and at the level of RNA stability.

Zebrafish hey1 transcription is linkedto the somitogenesis clock

Given the rhythmical appearance of hey1 transcription inthe PSM, we wanted to analyze whether this reflects acoupling to the somitogenesis clock and therefore a

linkage to the oscillating expression of the hairy-ortho-logues her1 and her7 in zebrafish (Takke and Campos-Ortega 1999; Holley et al. 2002; Oates and Ho 2002). Wedouble-stained embryos at the 9-somite stage with hey1(blue labeling) and her1 (red labeling) and analyzed theirexpression patterns during the different phases of onecycle of somite formation (Fig. 5A–D). When a new cycleof her1 expression started in the posterior PSM (caudaldomain indicated by red line in Fig. 5A), most hey1transcripts were located anterior to the rostral her1domain and no transcripts were found in the interstripedomain. As the wave of caudal her1 expression movedanteriorly, hey1 transcription was induced in the anteriorPSM including the interstripe region (Fig. 5B). In parallelto the rostrally moving her1 domain, the initially broadhey1 expression condensed into a single stripe locatedbetween the caudal and rostral her1 domains (Fig. 5C).Subsequently, hey1 transcription levels increased in theemerging S 0 somitomere, while no hey1 transcripts werefound in the posterior PSM at the time when a new cycleof her1 expression starts at the posterior end (Fig. 5D).Therefore, the complex and dynamic transcription of hey1in the PSM region correlates with the oscillating andwave-like expression of her1. This opens the possibilitythat hey1 transcription is regulated by the Delta-Notchpathway that also induces the oscillating expression ofher1.

Zebrafish hey1 is a target of DeltaD signaling

To analyze the regulation of hey1, we overexpressedDeltaD, a component of the Delta-Notch pathwayinvolved in the anterior development of the presomiticmesoderm (Oates and Ho 2002) and analyzed its effect onendogenous hey1 transcription at the 10-somite stage.Injection of high doses (1.2 ng) of RNA encodingzebrafish DeltaD (DlD) leads to fusion and malformationof individual somites and irregular arrangement of adaxialcells as shown earlier (Dornseifer et al. 1997). Here, weinjected a considerably lower dose of dlD RNA (350 pg)and found that injected embryos developed with overallnormal morphology (92%; n =78, remaining 8% withconvergence problems). However, in these embryos asignificant increase of hey1 expression, especially in thetailbud (in 66.7% of the embryos), was observed. Inaddition to this, we noticed ectopic activation of hey1 inthe axial mesoderm of a large fraction of embryos(52.8%; compare Fig. 5F and H to non-injected controlsin E and G). Furthermore, a strong up-regulation of hey1transcription was evident in the chordoneural hinge tolevels never observed in control embryos that were treatedunder identical conditions (Fig. 5E, F). This shows thatectopic DeltaD is able to enhance hey1 transcription in thedeveloping embryo, however, without apparent morpho-logical consequences, at least at the injected doses used inthese experiments.

To analyze whether Delta/Notch signaling is alsorequired for the transcription of hey1, we analyzed hey1

Fig. 5A–T Zebrafish hey1 is a target of DeltaD. A–D Dynamictranscription of hey1 (in blue) in relation to oscillating her1expression (in red). Arbitrarily assembled dorsal views of tailbudsfrom flat-mounted embryos during four phases of somite formation(I–IV) at the 10-somite stage. At the beginning of a new her1 wave,no hey1 expression is found in the interstripe region (arrowhead, A;blue staining in this area comes from overlaying ectoderm). Asher1 expression sweeps rostrally, hey1 is switched on in a broadarea of the anterior PSM (B; compare to Fig. 4P). C, D The hey1domain is compressed in the her1 interstripe region and forms onediscrete band representing somitomere S 0 (marked by asterisk;compare to Fig. 4R). E–H Overexpression of deltaD results in theectopic expression of hey1. hey1 expression in a non-injectedembryo (wt; E) and an embryo injected with 350 pg RNA encodingDeltaD (dlD; F). Note upregulated hey1 expression in the tailbud,especially in the area of the chordoneural hinge (lower arrow) andectopic expression in the axial mesoderm (upper arrow) of injectedembryos. G, H Cross-sections through trunk regions of non-injected (G) and deltaD-injected embryo showing ectopic hey1expression in the axial mesoderm (arrowhead in H compared to G).I–T Analysis of hey1 (in blue) and her1 (in red) expression in wild-type (wt) and after-eight mutant (aei) embryos at the 14- (I, J, M,N, Q, R) and 17-somite stages (K, L, O, P, S, T). Lateral views ofwhole embryos at lower magnification (I–L), and of tailbuds athigher magnification (M–P) as well as dorsal views of flat-mountedtailbuds (Q–T) are shown. In aei mutant embryos of both analyzedstages, her1 expression is reduced compared to wild-type siblings.Segmental hey1 expression is disorganized in the anterior trunk andthe anteriormost PSM (arrowhead), and strongly reduced in thearea directly adjacent to the PSM (arrow)

549

expression in the zebrafish mutant after eight (aei, tr233)that carries a point mutation in the deltaD gene (Holley etal. 2000). We analyzed a total of 83 embryos derivedfrom an intercross of heterozygous carriers and identified28 (33.7%) aei homozygous embryos by double stainingwith either her1 (Fig. 5I–T) or myoD (data not shown). Inaei mutant embryos, oscillation of her1 is impaired andtranscripts accumulate in the anterior PSM at earlierstages (8–12 somites; Holley et al. 2000, 2002 and datanot shown) and disappear at later stages (14–17 somites;Fig. 5). Likewise, myoD expression becomes severelydisorganized in posterior somites, formed after the 8-somite stage (Holley et al. 2000). At the 14- and 17-somite stages, expression of hey1 was also significantlyimpaired in the PSM of aei mutants. Expression wasevident in all somites, both in the correctly formed first 6–8 pairs, as well as in later-formed irregular somites. In thelatter, the segmental distribution of hey1 was lost,although expression levels were maintained. The moststriking effect was observed at the level of the S 1 somite.While a distinct band of hey1 transcription is usually seenin this youngest formed somite, no localized expression ofhey1 is evident in this area in aei mutants. Furthermore,while distinct hey1 expression is seen in S 0 and S 1somitomeres in the anterior PSM of wild-type embryos,expression appears reduced and disorganized in thesedomains in aei mutants. Transcription in the her1interstripe regions is lost and no signs of any dynamichey1 expression in the posterior PSM were detected. Ourdata therefore suggest that apparently factors other thanDeltaD are responsible for the induction of hey1 tran-scription in the PSM and later in formed somites.However, DeltaD seems to be absolutely essential forthe dynamic regulation of hey1 in the anterior PSM. Inthis mesenchymal-epithelial transition zone, AP identityof unsegmented somitomeres is established and somitesbecome epithelialized. This process is impaired in aeimutants that are only able to develop epithelializedsomites for the first eight pairs.

Taken together, we have shown that in contrast to thesituation in mouse and chick, hey1 is the only member ofthe hey family in zebrafish expressed during somitogen-esis. Similar to other hairy-related genes, its transcriptionis dynamically regulated by DeltaD in the PSM. Like inmouse, hey1 therefore is a downstream target of theDelta-Notch pathway.

Discussion

Differential evolution of hey genes in fish

Several members of the hey gene family have beenidentified by PCR and database mining in zebrafish, Fuguand Tetraodon. Our phylogenetic analysis suggests com-plex scenarios that might have taken place during theevolution of teleost genomes.

The hey1 gene from zebrafish analyzed in this work isapparently more related to pufferfish hey1a than hey1b

(Fig. 2). This suggests that the duplication that generateda second hey1 gene was relatively ancient and probablyoccurred before divergence between the zebrafish andpufferfish lineages more than 100 million years ago.While six independently derived sequences from thezebrafish hey1 gene were identified in the public genomedatabases, no second hey1 gene could be detected by datamining. The average nucleotide identity between puffer-fish hey1a and hey1b paralogues (495 nt, see Materialsand methods; uncorrected for multiple substitutions) isonly 75.0% compared to 92.5% between mouse andhuman hey1 (divergence about 110 million years ago,Kumar and Hedges 1998). This suggests again a ratherancient event of duplication. Both neighbor-joining (butwith a rather low bootstrap value) and maximum likeli-hood analyses suggest that Hey1a sequences are morerelated to Hey1b than to Hey1 from higher vertebrates,i.e. supported a fish-specific duplication of hey1. It hasbeen proposed that a whole genome duplication has takenplace during the course of evolution of the (ray-finned)fish lineage (Actinopterygia) after its separation from thesarcopterygian lineage including coelacanth, lungfish andland vertebrates (Meyer and Schartl 1999; Wittbrodt et al.1998; Winkler et al. 2003). Pufferfish hey1 duplicatesmight be remnants of this genome duplication with hey1bsubsequently being lost in zebrafish, but they might alsobe the result of a more local event of duplication. It ispossible that one of the pufferfish hey1 duplicates fulfils afunction similar to that of hey2 in other species, whilehey2 has been highly degenerated or perhaps even lost.This would imply that one of the hey1 paralogues inpufferfish might present an expression pattern related tothat of hey2 in zebrafish, and that no mutation impairingthe function(s) of Hey2 have occurred in this hey1duplicate.

Restricted expression of hey genesin zebrafish compared to higher vertebrates

Members of the hey gene family show temporally andspatially tightly regulated expression patterns with partlyoverlapping, but sometimes also complementary, domainsin chick and mouse. For example, all three hey genes areexpressed in somites and PSM in mice with only slightdifferences (Leimeister et al. 2000b). They are underdifferential transcriptional control, as only Hey2 showsoscillating expression throughout the PSM, very similar tochicken hairy1 and mouse Hes1, while Hey1 and HeyLare restricted to anterior aspects of the PSM. Furthermore,while somite and PSM expression of Hey1 and HeyL isreduced in Dll1 knock-out mice, the partial maintenanceof Hey2 in these mutants suggested a possible indepen-dence from the Delta-Notch pathway (Leimeister et al.2000b). In contrast to this, in zebrafish neither hey2 norheyL are expressed in PSM and somites. Only hey1 isexpressed in these structures and shows a dynamic modeof transcriptional regulation. Most importantly, hey1expression is under the control of deltaD, the Dll1

550

orthologue in zebrafish. This suggests that hey genesacquired differential regulatory elements during verte-brate evolution. Future comparison of promoter and UTRsequences in mouse and zebrafish hey genes may help toreveal the molecular correlates of these evolutionarychanges.

We identified one embryonic domain, the telenceph-alon, where two zebrafish orthologues (hey1 and hey2)are co-expressed, similar to the situation in mouse whereall three hey genes are expressed in this region. Other thanthat, it appears as a general feature that in zebrafish heyorthologues are expressed in strictly non-overlappingpatterns, in contrast to the situation in mouse. It can bespeculated that in zebrafish other hairy-related factorshave taken over the function of Hey genes in these organs.Alternatively, Hey genes could be expressed in theseorgans in mouse with species-specific or possibly evendispensable function.

Prominent domains of Hey gene expression in mouseinclude the developing vascular system and the formingheart. Hey1 and Hey2 are expressed in the heart and allthree Hey genes are expressed in the dorsal aorta. Inzebrafish, hey2 was the first gene from the hey family tobe described by cloning the gridlock (grl) mutation(Zhong et al. 2000). Analysis of the grl mutant phenotypeas well as morpholino knock-down experiments showedthat this gene is required for arterial versus venous cellfate decision in preangioblasts (Zhong et al. 2001). Herewe have shown that hey2 in contrast to the situation inhigher vertebrates is the only hey gene expressed in bloodvessels and their precursors. Furthermore, our expressiondata indicate that hey2 represents one of the earliestmarker genes for the precardiac field in the anterior lateralplate mesoderm, comparable to nkx2.5 (Chen and Fish-man 1996). Cardiac precursor cells originate from themarginal zone during gastrulation as has been shown bycell lineage analysis (Stainier et al. 1993). Molecularmarkers for these early precursors, however, are lacking.While first nkx2.5 transcripts are described as appearingat the 6-somite stage, we find first hey2 expression at least2 h earlier at the 1-somite stage, suggesting that hey2 isthe earliest marker for cardiac precursors known so far.Whether hey2 can induce cardiac differentiation in vitro,as has been shown for nkx2.5 (Chen and Fishman 1996),remains to be tested.

Importantly, the loss-of-function phenotypes of Hey2/hey2 in mouse and fish differ remarkably. While themutation in hey2 in fish results in a maturation defect ofthe aorta, resembling human aortic coarctation (Zhong etal. 2000), the Hey2 knock-out in mouse leads to fatalcardiomyopathy and cardiac septation defects with oth-erwise normal aortic development (Donovan et al. 2002;Fischer et al. 2002; Gessler et al. 2002; Sakata et al.2002). Thus, the divergence or possible lack of hey2genes in different fish species, the variation in expressionpatterns of hey genes during vessel and heart formationand the distinct loss-of-function phenotypes in mouse andfish indicate that hey genes may have acquired distinct

functions as key players in cell fate decisions during theseprocesses in vertebrates.

A possible lack of redundancyof hey gene function during somitogenesis in zebrafish

Epithelialized somites emerge from the PSM in atemporally and spatially tightly regulated manner. Asomitogenesis clockwork has been postulated to accountfor oscillating waves of gene transcription in the PSM thatare integrated into the periodic appearance of epithelial-ized somites at the rostral end of the PSM growth zone.The formation of each pair of somites can be subdividedinto three steps: first, precursor cells in the PSM exit theirundetermined state and are prepatterned after havingencountered a certain number of transcriptional waves.Subsequently, these cells form somitomeres in theanterior part of the PSM that acquire an anterior-posteriorpolarity. Finally, they become epithelialized and bud offfrom the PSM.

According to a recently proposed model, the hairy-related genes her1 and her7 are central componentsresponsible for the oscillation of cycling genes, ratherthan a mere output of the somitogenesis clockwork(Holley et al. 2002; Oates and Ho 2002). Autoregulatoryloops involving Delta ligands, Notch receptors and hairy-related targets induce the periodic transcription of genesthat are propagated to the neighboring cells and therebycause periodic waves of gene transcription that travelthrough the PSM. While cyclic deltaC expression seemsresponsible for the onset of the oscillator, deltaD isthought to provide a basal activation of Notch signalingand seems essential for processes at the anterior end of thePSM (Oates and Ho 2002). We have shown that zebrafishhey1 transcripts periodically accumulate in the anteriorPSM, form stripes in the early somitomeres and laterreside in the posterior domains of epithelialized somites.

The dynamic mode of transcription and the regulationby deltaD opens the possibility that hey1 is an additionalhairy-related gene that either functions as an output of theoscillator or even as a central component of the clock-work. However, its mode of expression is significantlydifferent when compared to other cyclic hairy genes.Her1 transcription for example starts in the posterior PSMand then migrates rostrally, while hey1 transcripts areinitially distributed across the anterior PSM and latercondense into a narrow stripe. This indicates differentmodes of regulation and makes it rather improbable thathey1 is part of the clockwork. Future functional analysesincluding a characterization of the regulatory elements inthe hey1 promoter and UTR regions have to clarify therole of zebrafish hey1 as output or component of theoscillator. Important for this, the promoter sequences ofzebrafish hey genes have recently been reported (Gajew-ski and Voolstra 2002).

In mouse and chick, all three hey gene family membersare strongly expressed in the PSM with mostly overlap-ping patterns indicating a possible redundancy in activity

551

and function. Consistently, single mutant mice carryingdeletions for either Hey1, Hey2 or HeyL show no somiticphenotype (Gessler et al. 2002; M. Gessler, unpublished).In contrast, hey1, but not hey2 or heyL, is the onlymember identified in zebrafish so far that is involved inearly somitogenesis. The lack of functional overlap withother hey genes in the zebrafish PSM will allow a detailedloss-of-function approach of hey1 during somitogenesis,e.g. by introducing morpholino antisense oligonucleotidesinto early zebrafish embryos (see Nasevicius and Ekker2000). This will help to understand at which step ofsomitogenesis this Hairy-related transcription factor isneeded and how it interacts with the components of themolecular oscillator.

Acknowledgements We thank Jose Campos-Ortega for kindlyproviding the DeltaD expression construct, Artemis for providingthe after-eight zebrafish mutants, Cordula Neuner for perfecttechnical assistance and Cornelia Leimeister and Matthias Sch�ferfor critical comments on the manuscript. We are also extremelygrateful to Laurence Bouneau and the other members of theTetraodon Genome project (Genoscope, Evry, France) for identi-fying and providing T. nigroviridis hey sequences. (Taki)Fugu datahave been provided freely by the Fugu Genome Consortium for usein this publication only. We especially thank Manfred Schartl forhis critical comments and support. This work was supported by theDeutsche Forschungsgemeinschaft (Ge539/9 and SFB 465). J.N.V.is supported by the BioFuture program of the German Bundesmin-isterium f�r Bildung und Forschung (BMBF).

References

Aparicio S, Chapman J, Stupka E, Putnam N, Chia JM, Dehal P,Christoffels A, et al (2002) Whole-genome shotgun assemblyand analysis of the genome of Fugu rubripes. Science297:1301–1310

Chen JN, Fishman MC (1996) Zebrafish tinman homolog demar-cates the heart field and initiates myocardial differentiation.Development 122:3809–3816

Chin MT, Maemura K, Fukumoto S, Jain MK, Layne MD,Watanabe M, Hsieh CM, Lee ME (2000) Cardiovascular basichelix loop helix factor 1, a novel transcriptional repressorexpressed preferentially in the developing and adult cardiovas-cular system. J Biol Chem 275:6381–6387

Dawson SR, Turner DL, Weintraub H, Parkhurst SM (1995)Specificity for the hairy/enhancer of split basic helix-loop-helix(bHLH) proteins maps outside the bHLH domain and suggeststwo separable modes of transcriptional repression. Mol CellBiol 15:6923–6931

Donovan J, Kordylewska A, Jan YN, Utset MF (2002) Tetralogy offallot and other congenital heart defects in Hey2 mutant mice.Curr Biol 12:1605–1610

Dornseifer P, Takke C, Campos-Ortega JA (1997) Overexpressionof a zebrafish homologue of the Drosophila neurogenic geneDelta perturbs differentiation of primary neurons and somitedevelopment. Mech Dev 63:159–171

Felsenstein J (1985) Confidence limits on phylogenies: an approachusing the bootstrap. Evolution 39:783–791

Fischer A, Leimeister C, Winkler C, Schumacher N, Klamt B,Elmasri H, Steidl C, Maier M, Knobeloch KP, Amann K,Helisch A, Sendtner M, Gessler M (2002) Hey bHLH factors incardiovascular development. Cold Spring Harb Symp QuantBiol 67:63–70

Gajewski M, Voolstra C (2002) Comparative analysis of somito-genesis related genes of the hairy/Enhancer of split class inFugu and zebrafish. Genomics 3:21

Gessler M, Knobeloch KP, Helisch A, Amann K, Schumacher N,Fischer A, Leimeister C (2002) Mouse gridlock: No aorticcoarctation or deficiency, but fatal cardiac defects in Hey2-/-mice. Curr Biol 12:1601–1604

Holley SA, Geisler R, N�sslein-Volhard C (2000) Control of her1expression during zebrafish somitogenesis by a Delta-depen-dent oscillator and an independent wave-front activity. GenesDev 14:1678–1690

Holley SA, J�lich D, Rauch GJ, Geisler R, N�sslein-Volhard C(2002) Her1 and the notch pathway function within theoscillator mechanism that regulated zebrafish somitogenesis.Development 129:1175–1183

Iso T, Sartorelli V, Chung G, Shichinonhe T, Kedes L, HamamoriY (2001a) HERP, a new primary target of Notch regulated byligand binding. Mol Cell Biol 21:6071–6079

Iso T, Sartorelli V, Poizat C, Iezzi S, Wu HY, Chung G, Kedes L,Hamamori Y (2001b) HERP, a novel heterodimer partner ofHES/E(spl) in notch signaling. Mol Cell Biol 21:6080–6089

Kimmel CB, Ballard WW, Kimmel SR, Ullmann B, Schilling TF(1995) Stages of embryonic development of the zebrafish. DevDyn 203:253–310

Kokubo H, Lun Y, Johnson RL (1999) Identification and expres-sion of a novel family of bHLH cDNAs related to Drosophilahairy and enhancer of split. Biochem Biophys Res Commun260:459–465

Kumar S, Hedges SB (1998) A molecular timescale for vertebrateevolution. Nature 392:917–920

Leimeister C, Externbrink A, Klamt B, Gessler M (1999) Heygenes: a novel subfamily of hairy- and Enhancer of split relatedgenes specifically expressed during mouse embryogenesis.Mech Dev 85:173–177

Leimeister C, Schumacher N, Steidl C, Gessler M (2000a) Analysisof HeyL expression in wild-type and Notch pathway mutantmouse embryos. Mech Dev 98:195–203

Leimeister C, Dale K, Fischer A, Klamt B, Hrabe de Angelis M,Radtke F, McGrew MJ, Pourquie O, Gessler M (2000b)Oscillating expression of c-hey2 in the presomitic mesodermsuggests that the segmentation clock may use combinatorialsignaling through multiple interacting bHLH factors. Dev Biol227:91–103

Lin MH, Leimeister C, Gessler M, Kopan R (2000) Activation ofthe Notch pathway in the hair cortex leads to aberrantdifferentiation of the adjacent hair-shaft layers. Development127:2421–2432

Lynch M, Conery JS (2000) The evolutionary fate and conse-quences of duplicate genes. Science 290:1151–1155

Maier MM, Gessler M (2000) Comparative analysis of the humanand mouse Hey1 promoter: Hey genes are new Notch targetgenes. Biochem Biophys Res Commun 275:652–660

Meyer A, Schartl M (1999) Gene and genome duplications invertebrates: the one-to-four (-to-eight in fish) rule and theevolution of novel gene functions. Curr Opin Cell Biol 11:699–704

Nakagawa O, Nakagawa M, Richardson JA, Olson EN, SrivastavaD (1999) HRT1, HRT2, and HRT3: a new subclass of bHLHtranscription factors marking specific cardiac, somitic, andpharyngeal arch segments. Dev Biol 216:72–84

Nasevicius A, Ekker SC (2000) Effective targeted gene ‘knock-down’ in zebrafish. Nat Genet 26:216–220

Oates AC, Ho RK (2002) Hairy/E(spl)-related (Her) genes arecentral components of the segmentation oscillator and displayredundancy with the Delta-Notch signaling pathway in theformation of anterior segmental boundaries in the zebrafish.Development 129:2929–2946

Pichon B, Taelman V, Kricha S, Christophe D, Bellefroid EJ (2002)XHRT-1, a hairy and Enhancer of split related gene withexpression in floor plate and hypochord during early Xenopusembryogenesis. Dev Genes Evol 212:491–495

Pourquie O, Tam PP (2001) A nomenclature for prospectivesomites and phases of cyclic gene expression in the presomiticmesoderm. Dev Cell 1:619–620

552

Roest Crollius H, Jaillon O, Bernot A, Dasilva C, Bouneau L,Fischer C, Fizames C, Wincker P, Brottier P, Quetier F, SaurinW, Weissenbach J (2000) Estimate of human gene numberprovided by genome-wide analysis using Tetraodon ni-groviridis DNA sequence. Nat Genet 25:235–238

Rones MS, Woda J, Mercola M, McLaughlin KA (2002) Isolationand characterization of Xenopus Hey-1: a downstream mediatorof Notch signalling. Dev Dyn 255:554–560

Saitou N, Nei M (1987) The neighbor-joining method: a newmethod for reconstructing phylogenetic trees. Mol Biol Evol4:406–425

Sakata Y, Kamei CN, Nakagami H, Bronson R, Liao JK, Chin MT(2002) Ventricular septal defect and cardiomyopathy in micelacking the transcription factor CHF1/Hey2. Proc Natl Acad SciUSA 99:16197–16202

Schmidt HA, Strimmer K, Vingron M, von Haeseler A (2002)TREE-PUZZLE: maximum likelihood phylogenetic analysisusing quartets and parallel computing. Bioinformatics 18:502–504

Stainier DYR, Lee RK, Fishman MC (1993) Cardiovasculardevelopment in the zebrafish. I. Myocardial fate map and hearttube formation. Development 119:31–40

Strimmer K, von Haeseler A (1996) Quartet puzzling: a quartetmaximum likelihood method for reconstructing tree topologies.Mol Biol Evol 13:964–969

Takke C, Campos-Ortega JA (1999) Her1, a zebrafish pair-rule likegene, acts downstream of notch signalling to control somitedevelopment. Development 126:3005–3014

Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG(1997) The ClustalX windows interface: flexible strategies formultiple sequence alignment aided by quality analysis tools.Nucleic Acids Res 24:4876–4882

Westerfield M (1993) The zebrafish book. University of OregonPress, Eugene

Winkler C, Moon RT (2001) Zebrafish mdk2, a novel secretedmidkine, participates in posterior neurogenesis. Dev Biol229:102–118

Winkler C, Sch�fer M, Duschl J, Schartl M, Volff JN (2003)Functional divergence of two zebrafish midkine growth factorsfollowing fish-specific gene duplication. Genome Res 13:1067–1081

Wittbrodt J, Meyer A, Schartl M (1998) More genes in fish?BioEssays 20:511–515

Zhong TP, Rosenberg M, Mohideen MA, Weinstein B, FishmanMC (2000) gridlock, an HLH gene required for assembly of theaorta in zebrafish. Science 287:1820–1824

Zhong TP, Childs S, Leu JP, Fishman MC (2001) Gridlocksignalling pathway fashions the first embryonic artery. Nature414:216–220

553