The Endothelin System: Evolution of Vertebrate-Specific Ligand

The Endothelin System: Evolution of Vertebrate-Specific Ligand–ReceptorInteractions by Three Rounds of Genome Duplication

Ingo Braasch,* Jean-Nicolas Volff,�� and Manfred Schartl**University of Wurzburg, Biozentrum, Physiological Chemistry I, Germany; �Institut de Genomique Fonctionelle de Lyon,Universite de Lyon, France; and �Institut de Genomique Fonctionelle de Lyon, Universite de Lyon, Institut Federatif BiosciencesGerland Lyon Sud, Universite Lyon 1, CNRS, INRA, Ecole Normale Superieure de Lyon, France

Morphological innovations like the acquisition of the neural crest as well as gene family expansions by genomeduplication are considered as major leaps in the evolution of the vertebrate lineage. Using comparative genomic analyses,we have reconstructed the evolutionary history of the endothelin system, a signaling pathway consisting of endothelinligands and their G protein–coupled receptors. The endothelin system plays a key role in cardiovascular regulation aswell as in the development of diverse neural crest derivatives like pigment cells and craniofacial bone structures, whichare hot spots of diversity in vertebrates. However, little is known about the origin and evolution of the endothelin systemin the vertebrate lineage.

We show that the endothelin core system, that is, endothelin ligands (Edn) and their receptors (Ednr), isa vertebrate-specific innovation. The components of the endothelin core system in modern vertebrate genomes date backto single genes that have been duplicated during whole-genome duplication events. After two rounds of genomeduplication during early vertebrate evolution, the endothelin system of an ancestral gnathostome consisted of four ligandand four receptor genes. The previously unknown fourth endothelin ligand Edn4 has been kept in teleost fish but lost intetrapods. Bony vertebrates generally possess three receptor genes, EdnrA, EdnrB1, and EdnrB2. EdnrB2 has been lostsecondarily in the mammalian lineage from a chromosome that gave rise to the sex chromosomes in therians (marsupialsand placentals). The endothelin system of fishes was further expanded by a fish-specific genome duplication andduplicated edn2, edn3, ednrA, and ednrB1 genes have been retained in teleost fishes.

Functional divergence analyses suppose that following each round of genome duplication, coevolution of ligandsand their binding regions in the receptors has occurred, adjusting the endothelin signaling system to the increase ofpossible ligand–receptor interactions. Furthermore, duplications of genes involved in the endothelin system areassociated with functional specialization for the development of particular neural crest derivatives. Our results support animportant role for newly emerging ligands and receptors as components of signaling pathways and their expansionthrough genome duplications in the evolution of the vertebrate neural crest.

Introduction

The evolution of vertebrates from an invertebrate pro-tochordate is one of the major transitions in the animal king-dom. This transition was accompanied with fundamentalchanges in both anatomy and genome structure. At the mor-phological level, neural crest cells, placodes, a complexbrain, and the endoskeleton are key innovations of the ver-tebrate bauplan (Gans and Northcutt 1983; Hall 1999a;Shimeld and Holland 2000). Furthermore, early vertebrateshave passed through a period of miRNA innovation(Heimberg et al. 2008) and massive genome rearrange-ments (Hufton et al. 2008) followed by several rounds ofwhole-genome duplications (WGDs) resulting in a substan-tial expansion of their gene repertoire (Ohno 1970; Dehaland Boore 2005; Panopoulou and Poustka 2005; Putnamet al. 2008).

The first round of WGD (1R) occurred concomitantlywith the rise of vertebrates (Panopoulou and Poustka 2005).The timing of the second round (2R) remains controversialand might have occurred in a common ancestor of agna-thans (lamprey and hagfish) and gnathostomes (Kurakuet al. 2009) or later in the gnathostome lineage after the splitfrom agnathans (Panopoulou and Poustka 2005). A thirdround, the fish-specific genome duplication (FSGD), hasoccurred within ray-finned fishes at the base of the teleost

lineage (Meyer and Van de Peer 2005). WGDs become ev-ident by the presence of so-called paralogons, that is, chro-mosomal blocks of duplicated genes showing conservedsynteny within a genome. Paralogons have been used to re-construct ancestral vertebrate karyotypes such as the pre-1R/2R (Nakatani et al. 2007; Putnam et al. 2008) andpre-FSGD protochromosomes (Jaillon et al. 2004; Woodset al. 2005; Kasahara et al. 2007).

Of particular importance for the evolutionary success ofthe vertebrate lineage was the acquisition of the neural crestand its derivates (Gans and Northcutt 1983; Hall 1999a;Donoghue et al. 2008). The neural crest is a multipotentstem-cell population that originates in the ectoderm at theneural plate border, the region at the junction of the neuralplate and the prospective epidermis. After closure of the neu-ral tube, neural crest cells delaminate and migrate to diverseregions of the embryo, where they give rise to around 50different cell types as diverse as pigment cells, craniofacialskeleton, or enteric neurons (LaBonne and Bronner-Fraser1998; Le Douarin and Kalcheim 1999; Vickaryous and Hall2006). Neural crest–derived structures are important formany vertebrate-specific features like jaw-based predation(Gans and Northcutt 1983; Donoghue et al. 2008).

The neural crest is absent from the most basal livinggroup of chordates, the cephalochordates (Putnam et al.2008; Yu et al. 2008). In contrast, migratory ‘‘neuralcrest-like’’ cells that develop into pigment cells have beenfound recently in urochordates, the sister clade of verte-brates (Jeffery et al. 2004; Jeffery 2006). A bona fide neuralcrest, however, is a synapomorphy of vertebrates includ-ing lampreys (Donoghue et al. 2008; Sauka-Spengler andBronner-Fraser 2008).

Key words: neural crest, endothelin receptor, G protein–coupledreceptor, pigment cell, evo-devo, pseudoautosomal region.

E-mail: [email protected].

Mol. Biol. Evol. 26(4):783–799. 2009doi:10.1093/molbev/msp015Advance Access publication January 27, 2009

� The Author 2009. Published by Oxford University Press on behalf ofthe Society for Molecular Biology and Evolution. All rights reserved.For permissions, please e-mail: [email protected]

Dow

nloaded from https://academ

ic.oup.com/m

be/article/26/4/783/1023825 by guest on 23 January 2022

Several models have been proposed to explain theemergence of the neural crest in vertebrates and its evolu-tionary origin is matter of an ongoing debate (reviewed inDonoghue et al. 2008). Importantly, these models are notmutually exclusive but put emphasis on different aspectsof neural crest evolution at the molecular level. Accordingto the ‘‘gene regulatory co-option model,’’ the origin of theneural crest is based on the recruitment of neural crest spec-ifier genes into a pre-existing gene regulatory network at theneural plate border (Meulemans and Bronner-Fraser 2004,2005; Yu et al. 2008). The ‘‘genome duplication model’’underlines the importance of the two WGDs early in thevertebrate lineage (1R, 2R) for providing the genetic rawmaterial necessary for the evolution of the neural crestand its derivatives (Ohno 1970; Holland et al. 1994;Shimeld and Holland 2000; Wada 2001; Wada and Makabe2006). Finally, the ‘‘new genes model’’ proposes thatthe evolution of the neural crest has been relying on theemergence of genes de novo in the vertebrate lineage,particularly signaling molecules (Martinez-Morales et al.2007).

To further dissect the molecular basis of neural crestevolution and to differentiate between ancestral states andevolutionary novelties, it is necessary to reconstruct theemergence and evolution of key components in the neuralcrest regulatory network. Using comparative genomic anal-yses, we have reconstructed the evolutionary history of theendothelin system.

The endothelin system consists of G protein–coupledendothelin receptors (Ednr) that are activated by endothelin(Edn) signaling peptides. Edn peptides consist of 21 aminoacids cleaved progressively from larger precursor proteinsby furin and endothelin-converting enzymes (Masaki2004). A large body of research and literature is availableon the involvement of the endothelin in blood-pressure reg-ulation and cancer development (reviewed in Masaki 2004;Bagnato and Rosano 2008). Furthermore, the endothelinsystem plays a major role in the determination, migration,proliferation, survival, and differentiation of neural crestcells and their derivatives (reviewed in Pla and Larue2003). Disruption of endothelin signaling at the level of en-dothelin ligands and their converting enzymes, the endothe-lin receptors, their associated G proteins, or furtherdownstream targets leads to malformations of neural crestderivatives such as craniofacial cartilage, enteric neurons,and pigment cells (Yanagisawa et al. 1998; Miller et al.2000; Pla and Larue 2003; Van Raamsdonk et al. 2004;Dettlaff-Swiercz et al. 2005; Walker et al. 2006; Walkeret al. 2007). However, despite its important functions invertebrate physiology and development, relatively little isknown about the evolution of the endothelin system inchordates. Particularly, it remains unclear 1) when and fromwhich pre-existing system the endothelin system emerged,2) whether the expansion of the endothelin system is basedon vertebrate genome duplications or more local duplica-tion events, 3) how it has evolved in the different vertebratelineages, and 4) whether expanding ligand and receptor rep-ertoires have coevolved. We show that the endothelin sys-tem newly emerged in vertebrates before the divergence ofgnathostomes and lampreys and that it was expanded bythree rounds of WGD, each followed by functional diver-

gence that has contributed to the diversity of neural crestderivatives found in vertebrates.

Materials and MethodsSequence Database Surveys

Nucleotide sequences of preproendothelin and endothe-lin receptor genes were identified using Blast searchesagainst GenBank (nr and expressed sequence tag databases;www.ncbi.nlm.nih.gov/blast/Blast.cgi), the current Ensemblgenome assemblies (www.ensembl.org; version 50, July2008) of human (Homo sapiens), mouse (Mus musculus),opossum (Monodelphis domestica), platypus (Ornitho-rhynchus anatinus), chicken (Gallus gallus), anole lizard(Anolis carolinensis), frog (Xenopus tropicalis), zebrafish(Danio rerio), medaka (Oryzias latipes), spotted green puf-ferfish (Tetraodon nigroviridis), torafugu (Fugu rubripes),and stickleback (Gasterosteus aculeatus). Usually, humansequences were used as initial queries. For some genes, cod-ing sequences were annotated manually from genome assem-blies based on ESTs and sequence homology to otherspecies. The preliminary genome assemblies of sea lamprey(Petromyzon marinus; www.pre.ensembl.org) and elephantshark (Callorhinchus milii; http://esharkgenome.imcb.a-star.edu.sg/) as well as EST sequences from agna-thans and cartilaginous fishes were included in our study,but were insufficient for in-depth analyses. We also analyzedgenomes from invertebrate species (sea squirt, Ciona intes-tinalis; amphioxus, Branchiostoma floridae; sea urchin,Strongylocentrotus purpuratus; fruitfly, Drosophilamelanogaster; and nematode, Caenorhabditis elegans) atGenBank, Ensembl and JGI (www.jgi.doe.gov/).

Sequence Alignments and Phylogenetic Reconstructions

Nucleotide sequences obtained from Blast searcheswere loaded into BioEdit (Hall 1999b), translated into pro-teins, and aligned using ClustalW (Thompson et al. 1994).Alignments were carefully checked, and ambiguouslyaligned regions were removed before phylogeny analyses,leaving ;30% and ;73% of Edn and Ednr coding regions,respectively, for phylogenetic analyses. Protein and nucle-otide maximum likelihood (ML) phylogenies were com-puted with PHYML (Guindon et al. 2005) with 100bootstrap replicates. Models of protein evolution and param-eter values were determined with ProtTest (Abascal et al.2005). Nucleotide phylogenies were performed using thegeneral time reversible (GTR) þ I þ G model. MEGA4(Tamura et al. 2007) was used to obtain Neighbor-Joiningbootstrap values of 10,000 replicates and to draw phylogenies.

Synteny Analyses

To establish syntenic relationships between vertebrategenomes within the chromosomal regions of interest, weused the Reciprocal Blast Hit method as described inBraasch et al. (2007) and made use of information providedby Ensembl Ortholog Predictions. For gene families thatwere identified to be part of endothelin paralogons (phos-phatase and actin regulator, Phactr; human immunodefi-ciency virus type I enhancer binding protein, Hivep) or

784 Braasch et al.

Dow


ic.oup.com/m


www.ncbi.nlm.nih.gov/blast/Blast.cgi

www.ensembl.org

www.pre.ensembl.org

http://esharkgenome.imcb.a-star.edu.sg/

http://esharkgenome.imcb.a-star.edu.sg/

http://www.jgi.doe.gov/

endothelin receptor paralogons (Spry, Brn3, and Slain), wereconstructed their phylogenies as described above.

Functional Divergence Analyses

Functional divergence of endothelin receptors afterduplication was analyzed using the ML approach (Gu1999, 2001). Aligned full-length Ednr protein sequenceswere loaded into DIVERGE 2.0 (http://xgu.zool.iastate.edu/) and clustered using the NJ Tree-Making option. Co-efficients of type I and of type II functional divergence (hI,hII) and site-specific profiles of posterior ratio for functionaldivergence were determined (Zheng et al. 2007). Cutoffvalues were determined by removing residues with de-scending posterior ratios until h . 0 was no longer signif-icant. Residues above the cutoff were mapped onto the 2Dstructure of the human EDNRB1 protein obtained from thedatabase of G protein–coupled receptors (GPCRDB; http://www.gpcr.org/7tm/).

Results and Discussion

Previous studies have failed to identify components ofthe endothelin core system outside vertebrates (Hyndmanand Evans 2007; Martinez-Morales et al. 2007). In contrast,genes encoding the G proteins associated with endothelinreceptors as well as the endothelin-converting enzymes(Ece) predate the vertebrate lineage (Zhang et al. 2001;Hyndman and Evans 2007; Zheng et al. 2007; Blandet al. 2008). Outside vertebrates, Ece proteins might func-tion as more general peptidases (Hyndman and Evans2007). The present study therefore focuses on the evolu-tionary history of the endothelin core system, particularlyin view of three rounds of genome duplication in the ver-tebrate lineage. Genomic locations of endothelin ligandsand receptors are given in table 1, accession numbers insupplementary table 1, Supplementary Material Online.

Evolution of Endothelin Ligands

Emergence and Expansion of the Endothelin Family inVertebrates

Preproendothelin genes have not been found so far inlineages basal to tetrapods and teleost fishes (Hyndman and

Evans 2007; Martinez-Morales et al. 2007). Consistently, wewere not able to identify any preproendothelin gene in thegenome assemblies of Ciona (urochordates), the cephalo-chordate amphioxus, or more distantly related invertebrates.Moreover, no endothelin domain was found in these ge-nomes using pattern-hit initiated Blast (PHI-Blast) searches(Zhang et al. 1998) with a pattern derived from endothelinpeptides (C-[STA]-C-x(4)-D-x-E-C-x-[YF]-[YF]-C-H-[LI]-[DG]-I-[IV]-W).

In contrast, multiple endothelin domains were foundhere for the first time in lamprey (supplementary table 2,Supplementary Material online). An unambiguous phylo-genetic assignment to the different endothelin groups ofthese short sequences from the preliminary genome assem-bly, however, was not possible. The presence of endothe-lins in lamprey suggests that endothelins are a vertebrateinnovation that arose after the divergence from protochor-dates but before the appearance of gnathostomes. Alterna-tively, endothelin genes may be present in invertebrategenomes but are too divergent to be identified by themethods used here.

The three known preproendothelin genes—Edn1,Edn2, and Edn3 (Masaki 2004)—were found in all tetrapodgenomes. Partial sequences of at least three different endo-thelins were also found in shark (Edn1, Edn2, and Edn3;supplementary table 2, Supplementary Material online)but were too short for further analyses. In teleost fish, anexpansion of the endothelin family was found, with fivegenes in medaka and pufferfishes and six genes in zebrafishand stickleback (table 1).

Sequence conservation among preproendothelin genesis mainly restricted to the ‘‘big endothelin’’ region (contain-ing the endothelin peptide) and the ‘‘endothelin-like’’ do-main (Arinami et al. 1991). Because the remaining partsof the preproendothelin genes are highly divergent, theycould not be aligned unambiguously and were difficultto predict from genomic DNA sequence in the absenceof EST sequence data using orthologous protein sequencesfrom related species. Thus, a nucleotide phylogeny of pre-proendothelin genes was reconstructed using 195 bp of thebig endothelin domain (144 bp) and the endothelin-likedomain (51 bp).

The endothelin phylogeny (fig. 1) provides evidencefor the presence of four gene family members in bony ver-tebrates. In addition to the three known family members,

Table 1Genomic Locations of Vertebrate Preproendothelin Genes

Species Edn1 Edn2 Edn3 Edn4

Human (Hsa) Hsa6p24 Hsa1p34 Hsa20q13 —Mouse (Mmu) Mmu13 Mmu4 Mmu2 —Opossum (Mdo) Mdo3 Mdo4 Mdo1 —Platypus Ultracontig474 Contig19327 Ultracontig516 —Chicken (Gga) Gga2 Gga23 Gga20 —Lizard Scaffold128 Scaffold762 Scaffold177 —Frog Scaffold33 Scaffold478 Scaffold1103 —

edn1 edn2a edn2b edn3a edn3b edn4Zebrafish (Dre) Dre19 Dre19 Dre16 Dre11 Dre23 Dre20Medaka (Ola) Ola11 Ola11 — Ola5 Ola7 Ola24Stickleback (Gac) GacX GacX GacXX GacXVII GacXII GacXVIIITakifugu Scaffold69 Scaffold61 Scaffold22 — Scaffold20 Scaffold120Tetraodon (Tni) Scaffold9996 Tni21 Tni8 — Tni9 Tni14

Evolution of the Endothelin Signaling System 785

Dow


ic.oup.com/m


http://xgu.zool.iastate.edu/

http://xgu.zool.iastate.edu/

http://www.gpcr.org/7tm/

http://www.gpcr.org/7tm/

supplementary table 4



a fourth endothelin gene, designated Edn4, was foundexclusively in teleost genomes. Additionally, two co-orthologous copies of Edn2, termed edn2a and edn2b, aswell as of Edn3, edn3a, and edn3b, were identified in tele-osts. For each teleost endothelin gene, mRNA or ESTsequences from at least one species were found (supple-mentary table 1A, Supplementary Material online), show-ing that all six teleost endothelin genes are actuallytranscribed and most likely functional.

Because of the absence of invertebrate outgroup se-quences, rooting of the endothelin phylogeny was not pos-sible. Thus, relationships of the four vertebrate Edn genesonly became apparent using synteny analyses (figs. 2 and3). Particularly, it was possible that the Edn4 gene is a fish-specific paralog of Edn1 derived from the FSGD. However,the human EDN1 region (Hsa6p24) shows conserved syn-teny to a single edn1 region in teleost genomes. Further-more, no synteny was observed with the teleost edn4region (fig. 2A and B). A second region syntenic to the tel-eost edn4 paralogon was also identified within teleostgenomes (though not containing a second edn4 gene)

(fig. 2B). These regions are derived from a common proto-chromosome (Kasahara et al. 2007) and were duplicatedduring the FSGD. The location of the second paralogonon a different chromosome than teleost edn1 is further ev-idence that teleost edn1 and edn4 are not FSGD duplicates.Furthermore, the tree topology (fig. 1) as well as syntenydata (figs. 2 and 3) exclude that edn4 is a fish-specific du-plicate of either edn2 or edn3.

In conclusion, the endothelin family emerged mostlikely in vertebrates and is larger than previously thought.Particularly in teleost fishes, a new endothelin family mem-ber Edn4 is found, which seems to have already been pres-ent in the last common ancestor of all gnathostomes.

Evolution of the Vertebrate Endothelin Repertoire byThree Rounds of Genome Duplication

If the four preproendothelin genes in gnathostomeswere the result of the two rounds of genome duplicationin early vertebrates, further gene families within endothelinparalogons should have undergone 2-fold duplication aswell. Indeed, the endothelin paralogons generally containmembers of the Hivep and Phactr gene families (figs. 2and 3). They have single orthologs in invertebrate genomesbut four (Phactr) and three members (Hivep) in human andchicken (supplementary table 3A, Supplementary Materialonline). We used their phylogenies to clarify the relation-ships of the Edn paralogons (supplementary fig. 1, Supple-mentary Material Online). Consistently, both gene familiesshow the same topology ((Edn1, Edn3),(Edn2, Edn4)).Such topology would be expected for gene families dupli-cated during the 1R and 2R genome duplications. Further-more, the chromosomal segments carrying the fourendothelin paralogons in human (Hsa6p24, Hsa20q13,Hsa1p34, and Hsa6q24) were shown to be derived fromthe same ancestral, pre-1R/2R protochromosome (Putnamet al. 2008). Thus, the presence of four endothelin genes ingnathostomes is indeed the result of the two vertebrategenome duplications.

The endothelin phylogeny further suggests that dupli-cated edn2 and edn3 genes in teleosts were the result of theFSGD (fig. 1), which is also supported by synteny data. Thehuman EDN2 region (Hsa1p34) shows double-conservedsynteny to two different chromosomal regions in teleost ge-nomes, each of them harboring an edn2 duplicate (fig. 3A).These teleost chromosomes have been shown to be derivedfrom a common pre-FSGD protochromosome (Kasaharaet al. 2007). A similar observation was made for humanEDN3 (Hsa20q13), which shows double-conserved syn-teny to two teleost chromosomes (fig. 3B), all derived fromanother common protochromosome and duplicated in thecourse of the FSGD (Kasahara et al. 2007).

We propose the following model for the evolution ofendothelin genes in vertebrates (fig. 4): The ancestral endo-thelin gene newly appeared in an ancestral vertebrate as partof a paralogon including the Hivep and Phactr genes. Thisparalogon was then doubled twice during the 1R and 2Rgenome duplications leading to four endothelin genes ingnathostomes. In the lineage leading to tetrapods, Edn4was lost. In teleost fishes, on the other hand, eight

FIG. 1.—ML phylogeny of endothelin ligand nucleotide sequences(195 bp) using the GTR þ I þ G model. Bootstrap values above 60%(ML/Neighbor-Joining method) are indicated. Four clades, Edn 1–4 arerecognized. In teleosts, duplicates of edn2 and edn3 are present.

786 Braasch et al.

Dow


ic.oup.com/m


supplementary table 3A



supplementary fig. 5

FIG. 2.—Conserved synteny of vertebrate Edn regions—part I. (A) The human EDN1 region corresponds to a single edn1 region in teleosts. (B)The teleost edn4 region (A-paralogon) is syntenic to two chromosomal regions in chicken as well as to another region within teleost genomes (B-paralogon). A- and B-paralogons are derived from a common protochromosome that was duplicated in the FSGD. Hivep and Phactr genes (gray)establish conserved synteny among vertebrate Edn regions. Interjacent genes that do not contribute to conserved synteny are not shown. (C) TeleostEdn4 peptide. The conserved sequence from medaka, stickleback, and pufferfishes is shown. Residues 6, 12, and 13 differ in zebrafish. Three Edn4-specific amino acid substitutions are found in the C-terminal tail compared with vertebrate Edn1–3 peptides, which are invariant in this region.


Dow


ic.oup.com/m


788 Braasch et al.

Dow


ic.oup.com/m


endothelin paralogons were generated by the FSGD, fol-lowed by loss of one copy of edn1 and edn4 so that upto six endothelins are present in teleosts. Differential lossof endothelin duplicates has occurred during the teleost ra-diation because medaka has lost edn2b and pufferfishesedn3a. Such divergent resolution can be an importantmechanism leading to speciation (Volff 2005 and referen-ces therein).

In conclusion, the evolution of the endothelin familysupports both, the new genes model as well as the genomeduplication model of neural crest evolution.

Sequence Analysis of the Teleost Edn4 Protein

The genomic organization of the newly identified tel-eost edn4 gene is conserved with that of other vertebrateEdn genes (supplementary fig. 2, Supplementary Materialonline) showing that it is a genuine Edn gene. However, theputative Edn4 protein shows important structural differen-ces to all other known vertebrates endothelins (fig. 2C).Edn4 proteins of acanthomorphs (medaka, pufferfishes,and stickleback) are invariant but have three amino acid(AA) substitutions compared with the zebrafish protein.Among all vertebrate Edn proteins, four AA were foundto be invariable, including four cysteine residues (supple-mentary fig. 2, Supplementary Material online). The C-terminal tail is generally invariant among vertebrateEdn1–3 proteins. Edn4, in contrast, differs at three AA sitesto all other vertebrate Edns in this part (fig. 2C). Biochem-ical studies are needed to determine whether Edn4 is bind-

ing to endothelin receptors. Importantly, divergence fromthe known endothelin C-terminus does not preclude endo-thelin receptor binding: The endothelin-derived snake ven-oms of the sarafotoxin family also differ from Edn1–3 inthe C-terminal tail and still bind to endothelin receptors(Sokolovsky 1992; Ducancel 2005; Fry 2005).

Evolution of Endothelin Receptors

Emergence of Endothelin Receptors in Early Vertebrates

Endothelin receptors have not been identified outsidebony vertebrates so far (Hyndman and Evans 2007). Endo-thelin receptors are distantly related to the G protein–coupledreceptor 37 (Gpr37) family, also present in amphioxus, andthe Brs3 (Bombesin-like receptor 3) GPCR family, alsopresent in amphioxus and more distantly related inverte-brates (fig. 5A) (Parichy et al. 2000; Nordstrom et al. 2008).

Our surveys of invertebrate genomes generally did notreveal any Ednr sequences except for two putative genesfrom amphioxus. These sequences lack orthologs in Cionaor other invertebrates and group with vertebrate Ednr pro-tein with comparatively low bootstrap support (fig. 5A)(Nordstrom et al. 2008). Therefore, it remains unclearwhether these ‘‘Ednr-like’’ genes in amphioxus and the ver-tebrate Ednr genes are descendants of a common Ednr geneof chordates (that then would have been lost secondarily inurochordates) or whether they might be derived from inde-pendent duplications of ancestral GPRC genes. There is sofar no evidence for conserved synteny between the Ednr-related gene regions and the vertebrate Ednr regions

FIG. 4.—Putative evolution of endothelin ligand paralogons. Three genes, Hivep, Ednr, and Phactr, have been linked in the ancestral vertebrate.This paralogon was duplicated twice during the 1R and 2R genome duplications. Edn4 was lost in the tetrapod lineage, but kept in teleost fishes, whichalso have retained two copies of edn2 and edn3 after the FSGD.

FIG. 3.—Conserved synteny of vertebrate Edn regions—part II. (A) Edn2 region. (B) Edn3 region. The Edn2 and Edn3 regions show double-

conserved synteny between a single human chromosome and two teleost-specific paralogons. Members of the Hivep and Phactr gene families (gray)establish conserved synteny among the vertebrate Edn regions. Interjacent genes that do not contribute to conserved synteny are not shown.


Dow


ic.oup.com/m





FIG. 5.—Phylogeny of Ednr proteins. (A) Unrooted ML phylogeny of Ednr-related GPCR protein sequences (322 AA). Important bootstrap valuesabove 50% (ML/NJ) are shown. Besides vertebrates, GPCRs of the Brs3/Nmbr/Grpr family are found in Drosophila (Dme), sea urchin (Spu), andamphioxus (Bfl). The Gpr37 family is found in vertebrates and amphioxus only. In amphioxus, also two closely related sequences of an Ednr-like typeare found but bootstrap support for their grouping with vertebrate Ednr proteins is comparatively low (79/55%). No ortholog of Ciona was identified forany of the three GPCR families. Bootstrap support for relationships between vertebrate Ednr proteins (arrowhead) is ,50%. (B) Unrooted MLphylogeny of endothelin receptor protein sequences (322 AA) using the Jones-Taylor-Thornton (JTT) þ I þ G model. Bootstrap values above 60%(ML/NJ) are indicated. Three distinct clades, EdnrA, EdnrB1, and EdnrB2 are present. EdnrB1 is present in all vertebrate groups (including amphibianETc) except in therian mammals. In teleosts, two copies of EdnrA and EdnrB1 are found.

790 Braasch et al.

Dow


ic.oup.com/m


(supplementary table 3B, Supplementary Material online),but this cannot be definitively excluded due to the fragmen-tary character of the current amphioxus genome draft. How-ever, because the ligand-binding domains of vertebrateEdnr proteins and of the putative amphioxus Ednr-like pro-teins are divergent (supplementary fig. 3, SupplementaryMaterial online), a bona fide receptor for endothelin ligandsseems to be indeed a vertebrate innovation.

For lamprey, short partial sequences from both EdnrAand EdnrB genes were found in the preliminary genomeassembly (supplementary table 2, Supplementary Materialonline) showing that the emergence of endothelin core sys-tem (ligand plus receptor) is as old as the vertebrate lineageitself. Therefore, the endothelin core system appearedaround the same time as the bona fide neural crest in ver-tebrate evolution, supporting the ‘‘new genes hypothesis’’of neural crest origin (Martinez-Morales et al. 2007). Sim-ilarly, other signaling pathways involved in neural crest de-velopment with peptide ligands and G protein–coupledreceptors emerged at first in vertebrates such as the mela-nocortin system (Selz et al. 2007).

Two endothelin receptor genes, EdnrA and EdnrB,have been found in mammals so far (Arai et al. 1990;Sakurai et al. 1990). A third gene reported to encode anamphibian-specific receptor, ETc, is present in Xenopus(Karne et al. 1993). EdnrB2 found in birds was reportedto be an avian-specific paralog of EdnrB1 (Lecoin et al.1998). In the present study, we identified two (opossumand placental mammals); three (frog, lizard, chicken, andplatypus); five (zebrafish); or six (medaka, stickleback,and pufferfish) Ednr genes. Genomic locations are givenin table 2, accession numbers in supplementary table 1B,Supplementary Material online.

The phylogeny of vertebrate Ednr proteins in shown infigure 5B. An outgroup was not included because rootingwith the amphioxus Ednr-like sequences as well as Gpr37or Brs3 family members gave conflicting tree topologieswith low bootstrap supports (fig. 5A and data not shown).Three clades of Ednr genes are present in bony vertebrates:EdnrA, EdnrB, and a third clade containing ETc from frogand EdnrB2 from chicken. Partial, nonoverlapping sequen-ces of each of the three Ednr genes were also found in car-tilaginous fishes (supplementary table 2, SupplementaryMaterial online), but were too short for further analyses.

This shows that at least three different Ednr genes werepresent in the ancestor of gnathostomes. In teleost fishes,the Ednr gene family is further expanded by the presenceof two copies of ednrA as well as of ednrB1.

Orthology of Vertebrate EdnrB2 Genes

Our results show that frog ETc and chicken EdnrB2are orthologous. Neither is ETc amphibian- nor is EdnrB2avian specific. The ETc/EdnrB2 gene (which will be termedEdnrB2 in the following) is present in all gnathostomegroups including monotreme mammals (platypus). Exten-sive conserved synteny can be found between the EdnrB2regions in the genomes of Xenopus, chicken (fig. 6A) andlizard (not shown). In teleosts, genes of this ancestral chro-mosomal block have been distributed to two chromosomes(fig. 6A), for example, on Ola10 and Ola14 in medaka andDre14 and Dre21 in zebrafish, which are derived froma common protochromosome by duplication during theFSGD (Kasahara et al. 2007). There is no evidence forthe presence of a second copy of ednrB2 in any of the tel-eost genomes under investigation. This gene duplicate wasmost likely lost before the split of the lineages of zebrafishand acanthomorphs (medaka, stickleback, and puffer-fishes). The remaining ednrB2 gene has been lost subse-quently in the lineage leading to zebrafish or is presentbut missing from the current genome assembly.

In mammals, EdnrB2 is present in monotremes(platypus), but absent from therians (marsupials and placen-tals). The EdnrB2 region in platypus is highly syntenic tothose of other, nontherian vertebrates (fig. 6A). Interestingly,this genomic region also shares conserved synteny with partsof the therian sex chromosomes. In the opossum (marsu-pials), at least six of the EdnrB2-neighboring genes inchicken are found in the same order on the X chromosome.In placentals, as exemplified by the human genome, thesegenes are found in the telomeric region of the X chromosome(Xq28) with two genes, SYBL1 and SPRY3 being part of thepseudoautosomal region 2 (PAR2) on the Y chromosome(Yq12). The sex chromosomes of marsupials and placentalsshare a common evolutionary origin (Veyrunes et al. 2008).Our data suggest that the EdnrB2 gene had been maintainedin the mammalian lineage until the split of monotremes andtherians. EdnrB2 has then been lost in the therian lineage

Table 2Genomic Locations of Vertebrate Endothelin Receptor Genes

Species EdnrA EdnrB1 EdnrB2

Human (Hsa) Hsa4q31.23 Hsa13q22 —Mouse (Mmu) Mmu8 Mmu14 —Opossum (Mdo) Mdo5 Mdo7 —Platypus Ultracontig539 Ultracontig230 Ultracontig519Chicken (Gga) Gga4 Gga1 Gga4Lizard Scaffold122 Scaffold78 Scaffold131Frog Scaffold424 Scaffold13 Scaffold258

ednrAa ednrAb ednrB1a ednrB1b ednrB2Zebrafish (Dre) Dre1 ScaffoldNA1838 Dre1 Dre9 Not detectedMedaka (Ola) Ola1 Ola18 Ola17 Ola21 Ola10Stickleback (Gac) GacIX GacVII GacIII GacXVI GacIVTakifugu Scaffold17 Scaf376 Scaffold263 Scaffold103 Scaffold163Tetraodon (Tni) Tni18 Scaffold147678 Tni15 Tni2 Tni1


Dow


ic.oup.com/m


supplementary table 3B





before the split of marsupials and placentals concomitantlywith the rise of the therian sex chromosomes (Veyrunes et al.2008). Given that the endothelin system is involved in tumorformation (Bagnato and Rosano 2008), the disappearance ofEdnrB2 fits the observed depletion of cancer-related genesfrom the therian X chromosome (Graves et al. 2002).

Our data also shed new light on the evolutionary originof the PAR2 gene content in human, which consists of fourgenes. It has been previously suggested that SPRY3 hasbeen added to the eutherian (placental) sex chromosomesafter the divergence from marsupials (Charchar et al.2003). In contrast, we show here that Spry3 is an ancientcomponent of the EdnrB2 paralogon (fig. 6A) dating backat least to the split of teleosts and tetrapods. Thus, Spry3was already present on the therian sex chromosomes beforethe split of marsupials and placentals, just as Sybl1.

Evolution of the Endothelin Receptor Family byVertebrate Genome Duplications

Establishing the relationships of Ednr paralogons us-ing the Ednr phylogeny was not possible because of the lackof suitable outgroup sequences. However, endothelin re-ceptor genes are linked to members of three other gene fam-ilies: Spry, Brn3, and Slain, for example, in the genomes ofhuman, chicken, and teleosts (fig. 6; supplementary table3B, Supplementary Material online). We reconstructedthe phylogenies of these gene families (supplementaryfig. 4A–C, Supplementary Material online), but their phy-logenies were not consistent with each other. The Spry phy-logeny suggests a ((EdnrA, EdnrB1), EdnrB2) topology,the Brn3 phylogeny favors a (EdnrA, (Edrnb1, EdnrB2))topology, and the Slain tree a ((EdnrA, EdnrB2), EdnrB1)

FIG. 6.—Conserved synteny of vertebrate Ednr regions. (A) Synteny of EdnrB2 paralogons supports orthology among vertebrate EdnrB2 genes.Chicken EdnrB2 and frog ETc regions are syntenic to two regions in the genomes of zebrafish and medaka. EdnrB2 is not detected in zebrafish. EdnrB2is also present in platypus. In therian mammals (opossum and human), the syntenic regions devoid of EdnrB2 orthologs are found on the sexchromosomes including the PAR2 in humans. (B) The EdnrA and (C) the EdnrB1 regions show double-conserved synteny between a single humanchromosome and two teleost-specific paralogons. Spry3, Brn3d, and Slain1b genes (gray) establish conserved synteny among vertebrate Ednr regions.Interjacent genes that do not contribute to conserved synteny are not shown.

792 Braasch et al.

Dow


ic.oup.com/m




supplementary fig. 4A–C

supplementary fig. 4A–C

topology. Conflicting tree topologies of paralogon mem-bers have been found for other genomic regions in verte-brates (e.g., Abbasi and Grzeschik 2007), and may becaused by different rates of molecular evolution.

Importantly, the Ednr paralogons are also linked to theParaHox paralogons consisting of the Parahox gene clusterand adjacent receptor tyrosine kinase genes (RTK) (Ferrieret al. 2005; Siegel et al. 2007): EdnrA is linked to ParaHoxC, EdnrB1 to ParaHox A, and EdnrB2 to ParaHox B. Theevolution of the four vertebrate ParaHox paralogons as re-sult of the two rounds of vertebrate genome duplication iswell established (Ferrier et al. 2005; Siegel et al. 2007). Thisstrongly supports the (EdnrA, (Edrnb1, EdnrB2)) topologyof Ednr paralogons (supplementary fig. 4D, SupplementaryMaterial online) and their duplication during 1R and 2Rgenome duplications.

In the five teleost genomes, ednrA and ednrB are pres-ent in two copies (fig. 5B). The duplication of ednrA hasbeen previously reported for zebrafish and fugu (Nairet al. 2007). In addition, we have found two paralogs alsoin the genomes of medaka, stickleback, and Tetraodon. Thetopology of the Ednr phylogeny (fig. 5B) is consistent withthe duplication of ednrA during the course of the FSGD.Furthermore, we have identified double-conserved syntenybetween Hsa4 containing EDNRA and two ednrA paralo-gons in teleost genomes (fig. 6B), which are again derivedfrom a same protochromosome as result of the FSGD(Kasahara et al. 2007).

Double-conserved synteny can also be found betweenthe human EDNRB1 region on Hsa13 and Dre1 and Dre9(fig. 6C), which have been previously shown to contain sev-eral paralogous gene pairs (Woods et al. 2005). In medakaand pufferfish, the ednrB1b paralogon has been well con-served, whereas the members of the ednrB1a paralogonhave been distributed among three different chromosomes(fig. 6C). Thus, although data are less obvious, we considerthe FSGD to be the most likely explanation for the presenceof two ednrB1b genes in teleosts.

A model for the evolution of the endothelin paralogonsis shown in figure 7A. In the Ciona genome, orthologs ofSpry and Brn3 as well as one part of the ParaHox cluster arefound on chromosome 2q, whereas Slain and another partof the ParaHox cluster are found on chromosome 14 (sup-plementary table 3, Supplementary Material online). In am-phioxus, the ParaHox cluster is intact (Ferrier et al. 2005).Thus, it appears that in the last common ancestor of Cionaand vertebrates, a chromosomal block containing Spry,Brn3, Slain, and the ParaHox cluster was present(fig. 7A). In the Ciona lineage, the Slain gene as well asa part of the ParaHox cluster were then translocated to an-other chromosome. In the early vertebrate, in contrast, thechromosomal block had stayed intact and the newly arisingEdnr gene was added to it. Furthermore, the ParaHox pa-ralogon was expanded with RTK genes. Importantly,among these was the precursor of the Kit/Csf1r and Pdgfrtype III RTK genes, which are important regulators for themigration and survival of neural crest derivatives like pig-ment cells, craniofacial cartilage, and others (reviewed inHoch and Soriano 2003; Braasch et al. 2008). Type IIIRTKs as well as some of their ligands (e.g., the Kit ligand)have been shown to be vertebrate-specific as well (Grassot

et al. 2006; Martinez-Morales et al. 2007; D’Aniello et al.2008; Yu et al. 2008). Thus, it appears that a new chromo-somal block was built in the very early vertebrate, fromwhich many important neural crest genes are derived.Presumably, these singleton genes were already involvedin the development of a basal type of neural crest in an earlyvertebrate ancestor.

The Ednr-ParaHox paralogon was then duplicated dur-ing 1R giving rise to EdnrA and EdnrB. After 2R, both cop-ies of EdnrB were kept, whereas one of the two EdnrAgenes was lost. In teleosts, the ednr paralogons were furtherdoubled by the FSGD. The Ednr family therefore furthersupports the importance of genome duplications for theevolution of the neural crest. Only the second copy ofednrB2 was lost before the teleost radiation, so that upto five ednr genes are found in teleosts. Therefore, the Ednrfamily supports the earlier notion that GPCRs are preferen-tially retained after WGD (Semyonov et al. 2008). In ther-ian mammals, finally, EdnrB2 was lost in the course of sexchromosome evolution.

Functional Divergence of Endothelin Receptors byChanges in Ligand-Binding Domains

Generally, each endothelin receptor can bind all endo-thelin ligands but with different affinities. MammalianEdnrA has higher affinity to Edn1 and Edn2 than to Edn3(Masaki 2004). Mammalian EdnrB1 and avian EdnrB2have similar affinities to all three ligands (Lecoin et al.1998; Masaki 2004), whereas frog EdnrB2 (ETc) binds pref-erentially Edn3 (Karne et al. 1993). However, these knownaffinities describe only a subset of possible interactions be-tween endothelins and their receptors in vertebrates. Thepresent study suggests that in tetrapods (except therians),nine (three ligands, three receptors) and in teleost fishes even30 (six ligands, five receptors) different binding interactionsare theoretically possible. Biochemical studies will be nec-essary to characterize the entirety of ligand–receptor interac-tions in the vertebrate endothelin system.

Given that the vertebrate endothelin receptor reper-toire and that of its ligands has expanded concertedly,we were interested to pinpoint important functional changesin the receptors after each round of whole-genome duplica-tion (1R, 2R, FSGD). To this end, functional divergence oftype I (FDI), that is, shift of evolutionary rates, and of type II(FDII), that is, radical change in amino acid property, weredetermined using DIVERGE (Gu and Vander Velden2002). We compared clades originating in the three roundof genome duplication, that is, EdnrA versus EdnrB (1R),EdnrB1 versus EdnrB2 (2R), as well as EdnrAa versusEdnrAb and EdnrB1a versus EdnrB1b (FSGD). A coeffi-cient of functional divergence (h) significantly larger than 0indicates functional divergence between two clades. For eachgenome duplication, significant values for type I functionaldivergence were obtained: hI(EdnrA–B) 5 0.207 ± 0.042,hI(EdnrB1–B2) 5 0.203 ± 0.067, hI(EdnrB1a–B1b) 50.466 ± 0.140. The comparison of EdnrAa versus EdnrAbwas not significant for type I functional divergence, as wellas all comparisons for type II functional divergence (sup-plementary table 4, Supplementary Material online). Impor-tant residues contributing to type I functional divergence


Dow


ic.oup.com/m


supplementary fig. 4D





FIG. 7.—Evolution of Ednr genes by three WGDs. (A) Putative evolution of endothelin receptor paralogons. A chromosomal block containingSpry, Brn3, Slain, and the ParaHox cluster present in an ancestral chordate was expanded in an ancestral vertebrate with the newly arising Ednr andtype III RTK genes. After two rounds of genome duplication (1R, 2R), four Ednr paralogons were present in gnathostomes. The Ednr repertoire ofa typical bony vertebrate after the loss of the second EdnrA gene is best illustrated by the chicken genome. The presence of two Ednr paralogons onchicken chr4 is due to a chromosome fusion in Galliformes (Guttenbach et al. 2003). In the teleost lineage, after FSGD and subsequent gene losses,eight Ednr paralogons with six ednr genes are found. In therian mammals, degeneration of the fourth Ednr paralogon including the loss of EdnrB2occurred during the evolution of sex chromosomes (XY). (B) Structural distribution of functional divergence among endothelin receptors. Importantsites for type I functional divergence (FDI sites) after genome duplications mapped onto the human EDNRB1 receptor. Functional divergence after 1R(EdnrA vs. EdnrB; green) are mainly found from transmembrane domain (TM) II to III including the first extracellular loop, in the vicinity of Lys-161(red), which is highly important for ligand binding. The first extracellular loop in EdnrA has an insertion of five amino acids (AA). Functionaldivergence after 2R (EdnrB1 vs. EdnrB2; yellow) and FSGD (ednrB1a vs. ednrB1b; blue) accumulate in the second extracellular loop and the adjacentTMs IV and V. The N-terminal 26 AAs constitute the cleaved peptide signal.

794 Braasch et al.

Dow


ic.oup.com/m


(FDI sites) were determined from site-specific profiles (sup-plementary fig. 5, Supplementary Material online) andmapped onto the structure of human EDNRB1 (fig. 7B).

Ligand binding of Ednr receptors mainly occurs at theextracellular sides of transmembrane domains (TM) and theextracellular loops, whereas interaction with G proteintakes place at the intracellular loops (Masaki et al. 1999;Orry and Wallace 2000). FDI sites for all three genome du-plications are mainly found in ligand-binding domains andnot in G protein interaction domains consistent with pre-dominant changes in ligand–receptor interactions.

Endothelin receptors consist of two distinct domains forendothelin-ligand binding, namely, the ‘‘message’’ and ‘‘ad-dress’’ domains (Sakamoto, Yanagisawa, Sawamara, et al.1993; Masaki et al. 1999). Functional divergence after eachgenome duplication seems to be differentially associatedwith these domains. For post-1R divergence between EdnrAand EdnrB, four of the eight FDI sites are found in the mes-sage domain. This domain, consisting of TMs I–III, VII andthe intervening loops, interacts with the C-terminus of theendothelin ligand and is involved in ligand–receptor bindingand message transmission (Sakamoto, Yanagisawa, Sakurai,et al. 1993; Sakamoto, Yanagisawa, Sawamara, et al. 1993).Particularly, three of the FDI sites are found in TM II and thefirst extracellular loop (fig. 7B). This region also differs by aninsertion of five amino acids into EdnrA compared withEdnrB and is particularly important for ligand–receptor bind-ing (Orry and Wallace 2000). One of the FDI sites is veryclose to a conserved Lys residue (fig. 7B), which has beenshown to be essential for high affinity binding of Edns(Adachi et al. 1994a, 1994b). This suggests an important di-vergence between EdnrA and EdnrB in the message domainafter their generation in 1R.

Post-2R divergence between EdnrB1 and EdnrB1, incontrast, is mainly concentrated in the second extracellularloop (five of six FDI sites), which is part of the address do-main. The address domain consisting of TMs IV–VI andintervening loops binds to the variable amino terminusof endothelins and is important for ligand–receptor selec-tivity (Masaki et al. 1999). Thus, functional divergence be-tween EdnrB1 and EdnrB2 seems to have mainly changedtheir ligand selectivity. Consistently, EdnrB2 has very lowaffinity to sarafotoxins, which are highly selective EdnrB1agonists (Lecoin et al. 1998).

Post-FSGD divergence between EdnrB1a and EdnrB1b,finallyhasoccurredinboththeaddress(fourofsevenFDIsites)as well as in the message domain (three of seven FDI sites).

We conclude that functional divergence after eachround of genome duplication was concentrated on the mod-ulation of different aspects of ligand–receptor interactions.With progressive expansion of the ligand repertoire, chang-ing the ligand selectivity became predominant. The func-tional divergence analyses therefore suggest coevolutionof the expanding endothelin receptor and ligand repertoires.

The Endothelin System and the Evolution of the NeuralCrest

As the endothelin core system is a vertebrate-specificinnovation, it supports the ‘‘new genes’’ hypothesis of neu-

ral crest evolution. The origin of the endothelin system withone ligand–receptor pair seems to correlate with the emer-gence of the bona fide neural crest. But what has been thefunction of the endothelin system in the first place? Inter-estingly, even cnidarians and ciliate protozoans show phys-iological responses to endothelin treatment (Kohidai et al.2001; Zhang et al. 2001). Thus, the early vertebrate mostlikely was preadapted to perceive a signal from the newlygenerated endothelin by binding to a G protein–coupled re-ceptor. Neofunctionalization of a duplicated GPCR fromthe Gpr37 or Bsr3 family specializing for Edn bindingmight have generated the first bona fide endothelin receptor.

In hydra, ectopic endothelin signals lead to musclecontractions (Zhang et al. 2001) that can be seen as analogyto smooth muscle contraction during endothelin-stimulatedvasoconstriction in vertebrates (Masaki 2004). Regulationof the cardiovascular system—which predates vertebrates(Schubert et al. 2006)—therefore might have been the an-cestral function of the endothelin system and its involve-ment in neural crest development evolved afterward. Inthis case, the endothelin system would also support the generegulatory co-option model of neural crest evolution.

During the specification and development of the ver-tebrate neural crest and its derivatives, the endothelin sys-tem is required at different time points. In Xenopus, EdnrAis expressed during early neural crest induction in the pro-spective neural crest. Receiving Edn1 signals from the un-derlying mesoderm, EdnrA functions as a neural crestspecifier and is required for neural crest maintenance(Bonano et al. 2008).

In the following steps of development, the fate of neu-ral crest cells is determined by their position along the an-terior–posterior axis. The anterior, cephalic neural crestgives rise to the ectomesenchymal neural crest (ENC;skeletogenic fate), whereas the more posterior, somitic neu-ral crest gives rise to the nonectomesenchymal neuralcrest (NENC; neuroglial fate including pigment cells)(Vickaryous and Hall 2006).

In gnathostomes, Edn1/EdnrA signaling is requiredfor the development of ENC derivatives, particularly forcraniofacial structures such as the lower jaw (Milleret al. 2000; Pla and Larue 2003; Nair et al. 2007). EdnrAis expressed by the migrating and/or differentiating cephalicneural crest cells, whereas Edn1 is produced by the ecto-and endodermal components of the pharyngeal archesand the paraxial mesoderm (Pla and Larue 2003). Edn3/EdnrB signaling, in contrast, is necessary for the develop-ment of the NENC, in particular for enteric glia and pigmentcells. In the somitic neural crest, expression of EdnrB genesis found during premigratory, migratory and differentiationstages and Edn3 is secreted in the ectoderm and in the gutmesenchyme (Pla and Larue 2003).

During vertebrate evolution, the partitioning into ENCand NENC emerged before the origin of lampreys (Donoghueet al. 2008). As the endothelin system is required for thedevelopment of both of these neural crest groups, the firstEdn/Ednr pair seems to have evolved its role in neural crestdevelopment in the very early vertebrate. The ancestralEdnr gene was then duplicated during 1R giving rise toEdnrA and EdnrB. Subsequently, subfunctionalization ofexpression along the anterior–posterior axis and, thus, with


Dow


ic.oup.com/m




respect to the neural crest fate seems to have occurred. Thespecialization of EdnrA for the anterior ENC and of EdnrBfor the more posterior NENC was also associated with func-tional divergence between the two receptors in the messagedomain. In contrast, the 1R duplication of the ligand mighthave separated neural crest functions from other functionsbecause only Edn1/3 but not Edn2/4 seem to have functionsin neural crest development.

The 2R genome duplication then led to the formationof Edn1 and Edn3 and spatial subfunctionalization of themoccurred subsequently. At this point, the spatial separationof the Edn1/EdrnA and the Edn3/EdnrB systems along thebody axis (Pla and Larue 2003), and thus, their functionalspecialization for ENC or NENC fates, respectively, wascompleted. Interestingly, our data suggest that contempora-neously expanding repertoires of ligands and their receptorsdo not necessarily have to follow a similar timing of func-tional specialization (Ednrs: post-1R, Edns: post-2R).

The 2R genome duplication also led to the formation oftwo EdnrB genes. This was associated with further special-ization of posterior NC functions. Trunk neural crest cellsmigrate along two routes, the dorso-ventral and the dorso-lateral pathways (Pla and Larue 2003). In mice and zebrafish,which both have lost the EdnrB2 gene, EdnrB1 is expressedin NC cells of both pathways (Parichy et al. 2000; Pla andLarue 2003). In birds, in contrast, EdnrB1 is expressed in theventral pathway, and these NC cells give rise to the entericneurons of the peripheral nervous system. NC cells of thedorso-lateral pathway giving rise to the pigment cells, how-ever, express EdnrB2 (Lecoin et al. 1998; Pla et al. 2005).Thus, the 2R duplication of EdnrB was followed by furthersubfunctionalization of spatial expression patterns as well asby changes in the address domain of the receptor.

EdnrB genes generally play an important role for thedevelopment of pigment cells (Karne et al. 1993; Lecoinet al. 1998; Parichy et al. 2000; Pla and Larue 2003).The 2R genome duplication correlates with important pig-ment cell innovations such as the emergence of the yellowxanthophores (Mellgren and Johnson 2002; Braasch et al.2008) and the duplication of EdnrB might have been an im-portant contribution to this increase in pigment cell diver-sity in vertebrates.

The FSGD, finally, has resulted in the most diverseendothelin system in vertebrates as seen in teleost fishes.The ednrA paralogs in the zebrafish are expressed in over-lapping domains in the cranial NC and have partially redun-dant roles for lower jaw formation (Nair et al. 2007). Thedouble knockdown of both ednrA genes leads to a pheno-type similar to the mutation in the single gene of their Edn1ligand (Miller et al. 2000; Nair et al. 2007) suggesting thatthe two receptors have partially subdivided the ancestraljaw function without fundamental changes of ligand–receptorinteractions. This is consistent with our analysis because wedid not find evidence for functional divergence of the teleostEdnrA receptor proteins after the FSGD (supplementarytable 4, Supplementary Material online).

The divergent fates of the edn3 and ednrB1 paralogsafter the FSGD remain elusive. In zebrafish, loss of ednr-B1a leads to defects in the adult pigment pattern, but theearly larval pattern is unaffected. Furthermore, enteric neu-rons also do not seem to be affected like in mice (Parichy

et al. 2000). Therefore, Parichy et al. (2000) have proposedthat these roles might be fulfilled by other ednrB paralogs.Our results now suggest that ednrB1b is the most likely can-didate, as no ednrB2 gene seems to be available in zebra-fish. Furthermore, the loss of ednrB2 might be responsiblefor the lack of the white leucophore pigment cells in thelarval pigment pattern of zebrafish.

We have previously suggested that the FSGD hasplayed an important role in the diversification of the pig-mentary system in teleost fishes, which is the most complexamong vertebrates (Braasch et al. 2007, 2008). The presentstudy puts further evidence into this direction as the Edn3/EdnrB1 system has been retained in two copies in teleostsafter the FSGD.

In summary, the three rounds of vertebrate genome du-plications resulted in progressive specializations in the ex-panding endothelin system and might have been associatedwith major innovations within the NC cell lineage.

Conclusions

The neural crest and its derivatives are key innovationsof vertebrates. Many genes involved in neuralcrest spec-ification and development date back to the metazoan ances-tor (Larroux et al. 2008), but others are vertebrate specific(Martinez-Morales et al. 2007). The endothelin system issuch a vertebrate-specific signaling pathway, and its emer-gence in an early ancestor of all vertebrates might have beena key event in neural crest evolution. Around 20% of thevertebrate proteome is vertebrate specific, and genes emerg-ing within vertebrates are predominantly singletons(Prachumwat and Li 2008). In contrast, endothelin ligandsand receptors have been preferentially retained after thethree rounds of genome duplications in the early vertebratesand in ray-finned fishes. Other, more ancient components ofthe endothelin system like the endothelin-converting en-zymes and the G protein–coupled receptors have largergene families in vertebrates compared with invertebrates(Hyndman and Evans 2007; Zheng et al. 2007; Blandet al. 2008) putatively also as result of 1R/2R. This expan-sion has not only increased the complexity of the endothelinsystem but was also associated with further evolution of theneural crest and its derivatives. Our analysis of the endo-thelin system thus provides support for all three major mod-els of neural crest evolution, namely, vertebrate-specificgene origin of major NC regulators, gene co-option forNC functions and gene family expansions during vertebrategenome duplications. Therefore, at the molecular level, theemergence and diversification of vertebrate neural crest hasto be considered as a rewiring of gene regulatory networksthat were supplemented by the integration of new compo-nents and expanded through WGDs. Future studies of neu-ral crest-related gene families will have to integrate at thesame time their specific functions, the phylogenetic timingof their appearance as well as their subsequent amplificationduring vertebrate WGDs.

Developmental and biochemical studies of the Ednr-like genes from amphioxus as well as the different Edn/Ednr genes from agnathans and cartilaginous fisheswill now be necessary to obtain a comprehensive viewof the contribution of the endothelin system to neural crest

796 Braasch et al.

Dow


ic.oup.com/m




evolution in vertebrates. Furthermore, it will be highly in-teresting to investigate putative correlations between the ex-pansion of the endothelin system in the FSGD, differentialreduction after duplication, and the diversity of jaw struc-tures and pigmentation in teleost fishes (Mabuchi et al.2007; Braasch et al. 2008; Salzburger 2008).

Supplementary Material

Supplementary tables 1–4, supplementary figures1–5, and color versions of figures 2–6 are available atMolecular Biology and Evolution online (http://www.mbe.oxfordjournals.org/).

Acknowledgments

We would like to thank the three anonymous re-viewers for helpful comments. Our work is supported bygrants from the German Science Foundation (to M.S.and J.N.V.) and by the Biofuture program of the Bundes-ministerium fur Bildung und Forschung (BMBF), the As-sociation pour la Recherche contre le Cancer (ARC), theFrench Institute for Agronomy Research (INRA PHASE),the French Research Agency (ANR). and the Foundationpour la Recherche Medicale (FRM) (to J.N.V.).

Literature Cited

Abascal F, Zardoya R, Posada D. 2005. ProtTest: selection of best-fit models of protein evolution. Bioinformatics. 21:2104–2105.

Abbasi AA, Grzeschik KH. 2007. An insight into the phylogenetichistory of HOX linked gene families in vertebrates. BMC EvolBiol. 7:239.

Adachi M, Furuichi Y, Miyamoto C. 1994a. Identification of a ligand-binding site of the human endothelin-A receptor and specificregions required for ligand selectivity. Eur J Biochem. 220:37–43.

Adachi M, Furuichi Y, Miyamoto C. 1994b. Identification ofspecific regions of the human endothelin-B receptor requiredfor high affinity binding with endothelin-3. Biochim BiophysActa. 1223:202–208.

Arai H, Hori S, Aramori I, Ohkubo H, Nakanishi S. 1990.Cloning and expression of a cDNA encoding an endothelinreceptor. Nature. 348:730–732.

Arinami T, Ishikawa M, Inoue A, Yanagisawa M, Masaki T,Yoshida MC, Hamaguchi H. 1991. Chromosomal assign-ments of the human endothelin family genes: the endothelin-1gene (EDN1) to 6p23-p24, the endothelin-2 gene (EDN2) to1p34, and the endothelin-3 gene (EDN3) to 20q13.2-q13.3.Am J Hum Genet. 48:990–996.

Bagnato A, Rosano L. 2008. The endothelin axis in cancer. Int JBiochem Cell Biol. 40:1443–1451.

Bland ND, Pinney JW, Thomas JE, Turner AJ, Isaac RE. 2008.Bioinformatic analysis of the neprilysin (M13) family ofpeptidases reveals complex evolutionary and functionalrelationships. BMC Evol Biol. 8:16.

Bonano M, Tribulo C, De Calisto J, Marchant L, Sanchez SS,Mayor R, Aybar MJ. 2008. A new role for the endothelin-1/endothelin-A receptor signaling during early neural crestspecification. Dev Biol. 323:114–129.

Braasch I, Schartl M, Volff JN. 2007. Evolution of pigmentsynthesis pathways by gene and genome duplication in fish.BMC Evol Biol. 7:74.

Braasch I, Volff JN, Schartl M. 2008. The evolution of teleostpigmentation and the fish-specific genome duplication. J FishBiol. 73:1891–1918.

Charchar FJ, Svartman M, El-Mogharbel N, Ventura M, Kirby P,Matarazzo MR, Ciccodicola A, Rocchi M, D’Esposito M,Graves JA. 2003. Complex events in the evolution of thehuman pseudoautosomal region 2 (PAR2). Genome Res.13:281–286.

D’Aniello S, Irimia M, Maeso I, Pascual-Anaya J, Jimenez-Delgado S, Bertrand S, Garcia-Fernandez J. 2008. Geneexpansion and retention leads to a diverse tyrosine kinasesuperfamily in amphioxus. Mol Biol Evol. 25:1841–1854.

Dehal P, Boore JL. 2005. Two rounds of whole genomeduplication in the ancestral vertebrate. PLoS Biol. 3:e314.

Dettlaff-Swiercz DA, Wettschureck N, Moers A, Huber K,Offermanns S. 2005. Characteristic defects in neural crestcell-specific Galphaq/Galpha11- and Galpha12/Galpha13-deficient mice. Dev Biol. 282:174–182.

Donoghue PC, Graham A, Kelsh RN. 2008. The origin andevolution of the neural crest. BioEssays. 30:530–541.

Ducancel F. 2005. Endothelin-like peptides. Cell Mol Life Sci.62:2828–2839.

Ferrier DE, Dewar K, Cook A, Chang JL, Hill-Force A,Amemiya C. 2005. The chordate ParaHox cluster. Curr Biol.15:R820–R822.

Fry BG. 2005. From genome to ‘‘venome’’: molecular origin andevolution of the snake venom proteome inferred fromphylogenetic analysis of toxin sequences and related bodyproteins. Genome Res. 15:403–420.

Gans C, Northcutt RG. 1983. Neural crest and the origin ofvertebrates: a new head. Science. 220:268–273.

Grassot J, Gouy M, Perriere G, Mouchiroud G. 2006. Originand molecular evolution of receptor tyrosine kinaseswith immunoglobulin-like domains. Mol Biol Evol. 23:1232–1241.

Graves JA, Gecz J, Hameister H. 2002. Evolution of the humanX–a smart and sexy chromosome that controls speciation anddevelopment. Cytogenet Genome Res. 99:141–145.

Gu X. 1999. Statistical methods for testing functional divergenceafter gene duplication. Mol Biol Evol. 16:1664–1674.

Gu X. 2001. Maximum-likelihood approach for gene familyevolution under functional divergence. Mol Biol Evol.18:453–464.

Gu X, Vander Velden K. 2002. DIVERGE: phylogeny-basedanalysis for functional-structural divergence of a proteinfamily. Bioinformatics. 18:500–501.

Guindon S, Lethiec F, Duroux P, Gascuel O. 2005. PHYMLOnline–a web server for fast maximum likelihood-basedphylogenetic inference. Nucleic Acids Res. 33:W557–W559.

Guttenbach M, Nanda I, Feichtinger W, Masabanda JS,Griffin DK, Schmid M. 2003. Comparative chromosomepainting of chicken autosomal paints 1–9 in nine different birdspecies. Cytogenet Genome Res. 103:173–184.

Hall BK. 1999a. The neural crest in development and evolution.New York: Springer-Verlag.

Hall TA. 1999b. BioEdit: a user-friendly biological sequencealignment editor and analysis program for Windows 95/98/NT. Nucl Acids Symp Ser. 41:95–98.

Heimberg AM, Sempere LF, Moy VN, Donoghue PC,Peterson KJ. 2008. MicroRNAs and the advent of vertebratemorphological complexity. Proc Natl Acad Sci USA.105:2946–2950.

Hoch RV, Soriano P. 2003. Roles of PDGF in animaldevelopment. Development. 130:4769–4784.

Holland PW, Garcia-Fernandez J, Williams NA, Sidow A. 1994.Gene duplications and the origins of vertebrate development.Development. (Suppl):125–133.


Dow


ic.oup.com/m


Supplementary tables 1

supplementary figures 1

supplementary figures 1

http://www.mbe.oxfordjournals.org/

http://www.mbe.oxfordjournals.org/

Hufton AL, Groth D, Vingron M, Lehrach H, Poustka AJ,Panopoulou G. 2008. Early vertebrate whole genome dupli-cations were predated by a period of intense genomerearrangement. Genome Res. 18:1582–1591.

Hyndman KA, Evans DH. 2007. Endothelin and endothelinconverting enzyme-1 in the fish gill: evolutionary andphysiological perspectives. J Exp Biol. 210:4286–4297.

Jaillon O, Aury JM, Brunet F, et al. (61 co-workers). 2004. Genomeduplication in the teleost fish Tetraodon nigroviridis reveals theearly vertebrate proto-karyotype. Nature. 431:946–957.

Jeffery WR. 2006. Ascidian neural crest-like cells: phylogeneticdistribution, relationship to larval complexity, and pigmentcell fate. J Exp Zoolog B Mol Dev Evol. 306:470–480.

Jeffery WR, Strickler AG, Yamamoto Y. 2004. Migratory neuralcrest-like cells form body pigmentation in a urochordateembryo. Nature. 431:696–699.

Karne S, Jayawickreme CK, Lerner MR. 1993. Cloning andcharacterization of an endothelin-3 specific receptor (ETCreceptor) from Xenopus laevis dermal melanophores. J BiolChem. 268:19126–19133.

Kasahara M, Naruse K, Sasaki S, et al. (37 co-authors). 2007.The medaka draft genome and insights into vertebrate genomeevolution. Nature. 447:714–719.

Kohidai L, Toth K, Ruskoaho H, Csaba G. 2001. Effect ofvasoactive peptides on Tetrahymena. chemotactic propertiesof endothelins (ET-1, ET-2, ET-3, fragment 11-21 of ET-1and big endothelin-1): a short-term inducible signallingmechanism of chemotaxis. Cell Biol Int. 25:1173–1177.

Kuraku S, Meyer A, Kuratani S. 2009. Timing of genomeduplications relative to the origin of the vertebrates: didcyclostomes diverge before, or after? Mol Biol Evol. 26:47–59.

LaBonne C, Bronner-Fraser M. 1998. Induction and patterning ofthe neural crest, a stem cell-like precursor population.J Neurobiol. 36:175–189.

Larroux C, Luke GN, Koopman P, Rokhsar DS, Shimeld SM,Degnan BM. 2008. Genesis and expansion of metazoantranscription factor gene classes. Mol Biol Evol. 25:980–996.

Le Douarin N, Kalcheim C. 1999. The Neural crest. Cambridge:Cambridge University Press.

Lecoin L, Sakurai T, Ngo MT, Abe Y, Yanagisawa M, LeDouarin NM. 1998. Cloning and characterization of a novelendothelin receptor subtype in the avian class. Proc Natl AcadSci USA. 95:3024–3029.

Mabuchi K, Miya M, Azuma Y, Nishida M. 2007. Independentevolution of the specialized pharyngeal jaw apparatus incichlid and labrid fishes. BMC Evol Biol. 7:10.

Martinez-Morales JR, Henrich T, Ramialison M, Wittbrodt J.2007. New genes in the evolution of the neural crestdifferentiation program. Genome Biol. 8:R36.

Masaki T. 2004. Historical review: endothelin. Trends PharmacolSci. 25:219–224.

Masaki T, Ninomiya H, Sakamoto A, Okamoto Y. 1999.Structural basis of the function of endothelin receptor. MolCell Biochem. 190:153–156.

Mellgren EM, Johnson SL. 2002. The evolution of morphologicalcomplexity in zebrafish stripes. Trends Genet. 18:128–134.

Meulemans D, Bronner-Fraser M. 2004. Gene-regulatory inter-actions in neural crest evolution and development. Dev Cell.7:291–299.

Meulemans D, Bronner-Fraser M. 2005. Central role of genecooption in neural crest evolution. J Exp Zool B Mol DevEvol. 304:298–303.

Meyer A, Van de Peer Y. 2005. From 2R to 3R: evidence for a fish-specific genome duplication (FSGD). BioEssays. 27:937–945.

Miller CT, Schilling TF, Lee K, Parker J, Kimmel CB. 2000.sucker encodes a zebrafish Endothelin-1 required for ventralpharyngeal arch development. Development. 127:3815–3828.

Nair S, Li W, Cornell R, Schilling TF. 2007. Requirements forEndothelin type-A receptors and Endothelin-1 signaling in thefacial ectoderm for the patterning of skeletogenic neural crestcells in zebrafish. Development. 134:335–345.

Nakatani Y, Takeda H, Kohara Y, Morishita S. 2007. Re-construction of the vertebrate ancestral genome revealsdynamic genome reorganization in early vertebrates. GenomeRes. 17:1254–1265.

Nordstrom KJ, Fredriksson R, Schioth HB. 2008. The amphioxus(Branchiostoma floridae) genome contains a highly diversi-fied set of G protein-coupled receptors. BMC Evol Biol. 8:9.

Ohno S. 1970. Evolution by gene duplication. New York:Springer-Verlag.

Orry AJ, Wallace BA. 2000. Modeling and docking the endothelinG-protein-coupled receptor. Biophys J. 79:3083–3094.

Panopoulou G, Poustka AJ. 2005. Timing and mechanism ofancient vertebrate genome duplications – the adventure ofa hypothesis. Trends Genet. 21:559–567.

Parichy DM, Mellgren EM, Rawls JF, Lopes SS, Kelsh RN,Johnson SL. 2000. Mutational analysis of endothelin receptorb1 (rose) during neural crest and pigment pattern developmentin the zebrafish Danio rerio. Dev Biol. 227:294–306.

Pla P, Alberti C, Solov’eva O, Pasdar M, Kunisada T, Larue L.2005. Ednrb2 orients cell migration towards the dorsolateralneural crest pathway and promotes melanocyte differentiation.Pigment Cell Res. 18:181–187.

Pla P, Larue L. 2003. Involvement of endothelin receptors innormal and pathological development of neural crest cells. IntJ Dev Biol. 47:315–325.

Prachumwat A, Li WH. 2008. Gene number expansion andcontraction in vertebrate genomes with respect to invertebrategenomes. Genome Res. 18:221–232.

Putnam NH, Butts T, Ferrier DE, et al. (36 co-authors). 2008. Theamphioxus genome and the evolution of the chordatekaryotype. Nature. 453:1064–1071.

Sakamoto A, Yanagisawa M, Sakurai T, Nakao K, Toyo-oka T,Yano M, Masaki T. 1993. The ligand–receptor interactions ofthe endothelin systems are mediated by distinct ‘‘message’’and ‘‘address’’ domains. J Cardiovasc Pharmacol. 22(Suppl.8):S113–S116.

Sakamoto A, Yanagisawa M, Sawamura T, Enoki T, Ohtani T,Sakurai T, Nakao K, Toyo-oka T, Masaki T. 1993. Distinctsubdomains of human endothelin receptors determine theirselectivity to endothelinA-selective antagonist and endothe-linB-selective agonists. J Biol Chem. 268:8547–8553.

Sakurai T, Yanagisawa M, Takuwa Y, Miyazaki H, Kimura S,Goto K, Masaki T. 1990. Cloning of a cDNA encoding a non-isopeptide-selective subtype of the endothelin receptor.Nature. 348:732–735.

Salzburger W. 2009. The interaction of sexually and naturallyselected traits in the adaptive radiations of cichlid fishes. MolEcol. 18:169–185.

Sauka-Spengler T, Bronner-Fraser M. 2008. Insights from a sealamprey into the evolution of neural crest gene regulatorynetwork. Biol Bull. 214:303–314.

Schubert M, Escriva H, Xavier-Neto J, Laudet V. 2006.Amphioxus and tunicates as evolutionary model systems.Trends Ecol Evol. 21:269–277.

Selz Y, Braasch I, Hoffmann C, Schmidt C, Schultheis C,Schartl M, Volff JN. 2007. Evolution of melanocortinreceptors in teleost fish: the melanocortin type 1 receptor.Gene. 401:114–122.

798 Braasch et al.

Dow


ic.oup.com/m


Semyonov J, Park JI, Chang CL, Hsu SY. 2008. GPCR genes arepreferentially retained after whole genome duplication. PLoSONE. 3:e1903.

Shimeld SM, Holland PW. 2000. Vertebrate innovations. ProcNatl Acad Sci USA. 97:4449–4452.

Siegel N, Hoegg S, Salzburger W, Braasch I, Meyer A. 2007.Comparative genomics of ParaHox clusters of teleost fishes:gene cluster breakup and the retention of gene sets followingwhole genome duplications. BMC Genomics. 8:312.

Sokolovsky M. 1992. Structure–function relationships of endo-thelins, sarafotoxins, and their receptor subtypes. J Neuro-chem. 59:809–821.

Tamura K, Dudley J, Nei M, Kumar S. 2007. MEGA4: molecularEvolutionary Genetics Analysis (MEGA) software version4.0. Mol Biol Evol. 24:1596–1599.

Thompson JD, Higgins DG, Gibson TJ. 1994. CLUSTAL W:improving the sensitivity of progressive multiple sequencealignment through sequence weighting, position-specific gappenalties and weight matrix choice. Nucleic Acids Res.22:4673–4680.

Van Raamsdonk CD, Fitch KR, Fuchs H, de Angelis MH,Barsh GS. 2004. Effects of G-protein mutations on skin color.Nat Genet. 36:961–968.

Veyrunes F, Waters PD, Miethke P, et al. (13 co-authors). 2008.Bird-like sex chromosomes of platypus imply recent origin ofmammal sex chromosomes. Genome Res. 18:965–973.

Vickaryous MK, Hall BK. 2006. Human cell type diversity,evolution, development, and classification with special refer-ence to cells derived from the neural crest. Biol Rev CambPhilos Soc. 81:425–455.

Volff JN. 2005. Genome evolution and biodiversity in teleostfish. Heredity. 94:280–294.

Wada H. 2001. Origin and evolution of the neural crest:a hypothetical reconstruction of its evolutionary history.Dev Growth Differ. 43:509–520.

Wada H, Makabe K. 2006. Genome duplications of earlyvertebrates as a possible chronicle of the evolutionary historyof the neural crest. Int J Biol Sci. 2:133–141.

Walker MB, Miller CT, Coffin Talbot J, Stock DW, Kimmel CB.

2006. Zebrafish furin mutants reveal intricacies in regulating

Endothelin1 signaling in craniofacial patterning. Dev Biol.

295:194–205.Walker MB, Miller CT, Swartz ME, Eberhart JK, Kimmel CB.

2007. phospholipase C, beta 3 is required for Endothelin1

regulation of pharyngeal arch patterning in zebrafish. Dev

Biol. 304:194–207.Woods IG, Wilson C, Friedlander B, Chang P, Reyes DK, Nix R,

Kelly PD, Chu F, Postlethwait JH, Talbot WS. 2005. The

zebrafish gene map defines ancestral vertebrate chromosomes.

Genome Res. 15:1307–1314.Yanagisawa H, Yanagisawa M, Kapur RP, Richardson JA,

Williams SC, Clouthier DE, de Wit D, Emoto N, Hammer RE.

1998. Dual genetic pathways of endothelin-mediated intercellu-

lar signaling revealed by targeted disruption of endothelin

converting enzyme-1 gene. Development. 125:825–836.Yu JK, Meulemans D, McKeown SJ, Bronner-Fraser M. 2008.

Insights from the amphioxus genome on the origin of

vertebrate neural crest. Genome Res. 18:1127–1132.Zhang J, Leontovich A, Sarras MP Jr. 2001. Molecular and

functional evidence for early divergence of an endothelin-like

system during metazoan evolution: analysis of the Cnidarian,

hydra. Development. 128:1607–1615.Zhang Z, Schaffer AA, Miller W, Madden TL, Lipman DJ,

Koonin EV, Altschul SF. 1998. Protein sequence similarity

searches using patterns as seeds. Nucleic Acids Res.

26:3986–3990.Zheng Y, Xu D, Gu X. 2007. Functional divergence after gene

duplication and sequence-structure relationship: a case study

of G-protein alpha subunits. J Exp Zool B Mol Dev Evol.

308:85–96.

Billie Swalla, Associate Editor

Accepted December 29, 2008


Dow


ic.oup.com/m


Documents

The Endothelin System: Evolution of Vertebrate-Specific Ligand