PATTERNS & PHENOTYPES

Genome Duplication, Subfunction Partitioning, andLineage Divergence: Sox9 in Stickleback andZebrafishWilliam A. Cresko, Yi-Lin Yan, David A. Baltrus, Angel Amores, Amy Singer, Adriana Rodrıguez-Marı, andJohn H. Postlethwait*

Teleosts are the most species-rich group of vertebrates, and a genome duplication (tetraploidization) event in ray-finfish appears to have preceded this remarkable explosion of biodiversity. What is the relationship of the ray-fingenome duplication to the teleost radiation? Genome duplication may have facilitated lineage divergence bypartitioning different ancestral gene subfunctions among co-orthologs of tetrapod genes in different teleost lineages.To test this hypothesis, we investigated gene expression patterns for Sox9 gene duplicates in stickleback andzebrafish, teleosts whose lineages diverged early in Euteleost evolution. Most expression domains appear to havebeen partitioned between Sox9a and Sox9b before the divergence of stickleback and zebrafish lineages, but someancestral expression domains were distributed differentially in each lineage. We conclude that some genesubfunctions, as represented by lineage-specific expression domains, may have assorted differently in separatelineages and that these may have contributed to lineage diversification during teleost evolution. DevelopmentalDynamics 228:480–489, 2003. © 2003 Wiley-Liss, Inc.

Key words: genome duplication; subfunctionalization; transcription factor; chondrogenesis; sex determination;macroevolution; tetraploidization

Received 28 July 2003; Accepted 4 August 2003

INTRODUCTIONHalf of all vertebrate species are te-leost fish (Nelson, 1994), the mostspeciose and diverse group of verte-brates (Fig. 1). What evolutionarymechanisms contributed to this re-markably successful explosion ofbiodiversity? A clue comes from theobservation that chromosomallydiploid teleosts often have severalparalogous copies of single copytetrapod genes (Morizot et al., 1991;Ekker et al., 1992, 1997; Akimenko etal., 1995; Smith et al., 2000, 2001; Ad-amska et al., 2001; Robinson-Re-

chavi et al., 2001). Genetic mappingstudies showed that duplicated ze-brafish genes map in duplicatedchromosome segments co-ortholo-gous to portions of individual humanchromosomes (Amores et al., 1998;Postlethwait et al., 1998; Gates et al.,1999; Woods et al., 2000). The mostparsimonious conclusion was thatthere had been a genome duplica-tion in the zebrafish lineage, andsubsequent analysis shows that thisgenome amplification in the ray-finfish lineage occurred before the te-leost radiation (Amores et al., 1998;

Meyer and Schartl, 1999; Taylor etal., 2001a, b, 2003; Van de Peer etal., 2002; Amores et al., 2003). Didgenome duplication play a role inthe teleost radiation, and if so, howdid it spur lineage diversification andmorphologic variation?

To clarify the relationship betweengenomic and phenotypic complex-ity, we must first understand the pro-cesses by which duplicate genesevolve. Initial models of duplicategene evolution assumed that eachgene performs one or few functions(Ohno, 1970; Nei and Roychoudhury,

Institute of Neuroscience, University of Oregon, Eugene, OregonGrant sponsor: National Institutes of Health; Grant numbers: R01RR10715; 5 F32 GM020892; Grant sponsor: National Science Foundation;Grant numbers: IBN 0236239; IBN 9728587.*Correspondence to: John H. Postlethwait, Institute of Neuroscience, University of Oregon, Eugene, Oregon, 97403-1254.E-mail: jpostle@uoneuro.uoregon.edu

DOI 10.1002/dvdy.10424

DEVELOPMENTAL DYNAMICS 228:480–489, 2003

© 2003 Wiley-Liss, Inc.

1973; Bailey et al., 1978; Takahata andMaruyama, 1979; Li, 1980; Watterson,1983). On the basis of this assumption,retention of both copies of dupli-cated genes was hypothesized to berare and would occur only if onecopy acquired a novel, positively se-lected function (neofunctionalization;Walsh, 1995; Sidow, 1996; Cooke etal., 1997; Nadeau and Sankoff, 1997).

Empirical data, however, show thatduplicate pairs are retained at ahigher rate than the classic models ofduplicate gene evolution predict (Al-lendorf et al., 1975; Bisbee et al., 1977;Ferris and Whitt, 1979; Graf and Kobel,1991; Hughes and Hughes, 1993). Thismay be due to the modular andcomplex nature of many genes,which comprise numerous subfunc-

tions that perform a variety of tasks indifferent tissues and at different devel-opmental times. This view of genescomprising multiple subfunctions hasimportant implications for under-standing both the evolution of dupli-cate gene pairs and the evolution ofphenotypic complexity (Force et al.,2003). Instead of the acquisition ofnovel functions, the partitioning of an-cestral subfunctions among descen-dant gene duplicates by the recipro-cal, neutral fixation of degenerativeregulatory mutations can contributeto permanent preservation of bothcopies (Hughes, 1994, 1999; Force etal., 1999; Hughes, 1999; Stoltzfus, 1999).The partitioning process for duplicategene retention was formalized in theDuplication, Degeneration, Comple-mentation (DDC) model (Force et al.,1999), and the partitioning of ances-tral subfunctions has since been dem-onstrated in many duplicated teleostgene pairs (co-orthologs) in singlelineages (De Martino et al., 2000;Bingham et al., 2001; Bruce et al.,2001; Chiang et al., 2001; Lister et al.,2001; McClintock et al., 2001, 2002;Altschmied et al., 2002; Yu et al.,2003).

A key question for understandingthe evolution of gene duplicatesand the role of gene duplication inthe teleost radiation is whether an-cestral subfunctions assorted beforeor after the divergence of variousteleost lineages. Ancestral subfunc-tions that have not partitioned be-tween duplicates before lineage di-vergence remain available forsubsequent differential partitioningin different lineages, a process thatcan contribute to reproductive iso-lation (Lynch and Force, 2000). Sub-functions of paralogous duplicatesin the different lineages will be largelyidentical if partitioning happens soonafter duplication. If partitioning is slow,however, it may frequently occur in-dependently in different lineages,and as a consequence, duplicateparalogs will retain different combina-tions of ancestral subfunctions in dif-ferent teleost lineages. A centralquestion is the following: how oftendoes subfunction partitioning occurindependently in different lineages?To date, there is no generalizable an-swer, because only a single case ofduplicate gene pairs has been ana-

Fig. 1. A family tree for bony fish (Osteichthys) after (Nelson, 1994).

Sox9 EXPRESSION IN STICKLEBACK AND ZEBRAFISH 481

lyzed in detail in two different teleostlineages (Lister et al., 2001; Altschmiedet al., 2002). In this case, subfunctionpartitioning appears to have oc-curred before the divergence of ze-brafish, swordtail, and pufferfish lin-eages at the base of the teleostradiation (see Fig. 1).

To test the hypothesis that dupli-cate genes can be partitioned inde-pendently, we have examined ex-pression patterns for two duplicatesof Sox9 in the threespine sticklebackGasterosteus aculeatus and the ze-brafish Danio rerio lineages that di-verged early in the teleost radiation(Fig. 1). Stickleback is an emergingmodel system for the study of theevolution of development, particu-larly for rapid morphologic changesin bony armor plates and defensivespines over short periods of time (Bellet al., 1993; Bell and Foster, 1994; Belland Orti, 1994; Ahn and Gibson,1999; Bell, 2001; Peichel et al., 2001;Gibson, 2002). Thus, these data arenot only useful for macro-evolution-ary studies across divergent lineagesbut also lay the groundwork for ex-amining how microevolutionarychange in cartilage and bone reg-ulatory genes might contribute tothe origin of phenotypic variationamong stickleback populations.

Mutations in human, mouse, andzebrafish have shown that Sox9 is animportant gene for the regulation ofcartilage formation and the subse-quent development of cartilage-re-placement bones (Wagner et al.,1994; Bi et al., 2001; Kist et al., 2002;Yan et al., 2002). Sox9 is a member ofthe Sox protein family of transcrip-tion factors, which contain a SRY-likeHMG-box that binds and bends DNA(Ng et al., 1997; Wegner, 1999). Inmammals, Sox9 is involved in testisdetermination (Wagner et al., 1994;Vidal et al., 2001) and regulates car-tilage formation by binding to a cis-regulatory sequence in COL2A1, thehuman type II collagen gene (Bell etal., 1997; Lefebvre et al., 1997). Sox9expression occurs in mesenchymalcondensations before and duringchondrogenesis, and the expressionpattern mirrors that of COL2A1 in tet-rapods and teleosts (Zhao et al.,1997; Healy et al., 1999; Chiang etal., 2001; Yan et al., 2002). In additionto the critical role played in chon-

drogenesis, studies on Xenopus em-bryos indicate that reducing theproduction of Sox9 protein with mor-pholino antisense-oligonucleotidesinhibits the formation of neural crest,the progenitors of craniofacial carti-lage (Spokony et al., 2002). Thus,Sox9 is a multifunctional gene, play-ing important roles in testis determi-nation, the formation of neural crest,and chondrogenesis.

Zebrafish has two Sox9 genes:Sox9a and Sox9b (Chiang et al.,2001). Gene phylogenies and ge-netic mapping show that these areco-orthologs of the tetrapod Sox9gene and likely arose in the ray-fingenome duplication event (Chianget al., 2001). Importantly, expressionpatterns of these duplicates in ze-brafish exhibit overlapping subsets ofthe tetrapod expression domains(Chiang et al., 2001; Yan et al., 2002)and together approximate the ex-pression pattern of the Sox9 gene inmouse, suggesting the partitioningof ancestral subfunctions. To deter-mine whether subfunction partition-ing occurred before or after the te-leost radiation, we have cloned twoSox9 genes from stickleback, estab-lished their phylogenetic relation-ships to tetrapod and zebrafish Sox9genes, and compared their expres-sion patterns with those of zebrafishand tetrapods. We found that theevent that produced duplicated te-leost Sox9 genes occurred beforethe divergence of stickleback andzebrafish lineages, and so did thepartitioning of most subfunctions as-sayed by embryonic expression anal-ysis. These results have important im-plications for the generalization ofexperimental results among teleostmodel systems, for the ways in whichduplicated genes evolve, and for themechanisms that generated themagnificent teleost radiation.

RESULTS AND DISCUSSIONSox9 Duplication OccurredBefore Divergence ofStickleback and ZebrafishLineagesBy using redundant primers de-signed to amplify the first exon ofSox9, we screened Fosmid genomiclibraries for stickleback and found

seven positive clones. Restriction en-zyme analysis divided these clonesinto two distinct classes, and wechose two clones from each class toanalyze in depth. A combination ofdirected and shotgun sequencingof the subclones produced two dis-tinct genomic contigs 9908 and 6495base pairs long that each contain aSox gene with DNA sequence simi-larity to Sox9. Gene prediction anal-ysis using GENSCAN (Burge and Kar-lin, 1998) showed that each genecomprises three exons and two in-trons and is predicted to encodetranslated products of 464 and 477amino acids that have high se-quence similarity to Sox9. The pre-sumptive HMG domain at the N-ter-minus of the protein is wellconserved between these pre-dicted peptides and Sox9 co-or-thologs from teleosts (pufferfish, ze-brafish, rice eel; Bagheri-Fam et al.,2001; Chiang et al., 2001; Zhou et al.,2002) and tetrapods (chick, mouseand human; Wright et al., 1993;Wagner et al., 1994; Healy et al.,1999). The C-termini of the predictedproteins, however, are much lessconserved both between the stick-leback paralogs and across theother orthologs.

A phylogenetic analysis of thesesequences confirmed that stickle-back have at least two differentSox9 genes (Fig. 2). Both of these se-quences fell squarely inside the Sox9clade with a very high bootstrapvalue (1,000 of 1,000), showing thatthey are not orthologs of the closelyrelated Sox8 and Sox10 clades. Fur-thermore, one stickleback se-quence clusters within the Sox9aclade with pufferfish, rice eel, andzebrafish (Chiang et al., 2001; Zhouet al., 2002), whereas the other fallswithin the Sox9b clade alongsidepufferfish and rice eel (Bagheri-Famet al., 2001; Zhou et al., 2002) withhigh bootstrap support (728 of 1000).Because zebrafish sox9b fell as anoutgroup to the other teleost Sox9genes, it was important to be sure itwas not a Sox8, Sox9, or Sox10; there-fore, we identified a zebrafish sox8gene as EST fi23c10 (AW153579) iso-lated in the Washington UniversityZebrafish EST Project, and mapped itto LG3 in a region with conservedsynteny with human chromosome

482 CRESKO ET AL.

Hsa16p13.3, the location of humanSOX8 (data not shown). This ze-brafish sequence clusters stronglywith the other vertebrate Sox8genes, which supports the conclu-sion that zebrafish Sox9b is a Sox9ortholog, despite its somewhat unex-pected position in the tree. Impor-tantly, the branching pattern of theSox9a and Sox9b fish clades with re-spect to the tetrapod Sox9 clade,while not completely unambiguous,shows that the duplication eventthat produced the teleost Sox9aand Sox9b clades occurred beforethe divergence of stickleback andzebrafish lineages. Because the ze-brafish Sox9a and Sox9b genes mapin duplicated chromosome seg-ments that are co-orthologous tothe human chromosome Hsa17

(Chiang et al., 2001), the location ofSOX9 (Wagner et al., 1994), we con-clude that Sox9a and Sox9b arose inthe hypothesized whole genomeduplication event in the ray-fin fishlineage (Amores et al., 1998;Postlethwait et al., 1998; Taylor et al.,2001a, 2003). Thus, stickleback, ze-brafish, pufferfish, and rice eel all ap-pear to have retained both Sox9copies formed through the duplica-tion of an ancestral fish genome ap-proximately 300 million years ago.

Shared and DivergentExpression Patterns of Sox9aBecause stickleback and zebrafishSox9 genes were duplicated beforethe species lineages diverged, ex-pression pattern analysis can distin-

guish between evolutionary changesthat occurred after duplication butbefore lineage divergence andthose that occurred between lin-eage divergence and the present.Stickleback embryos are cultured at20°C and are larger and developmore slowly than zebrafish embryoscultured at 28.5°C, but the embry-onic development of each is similarenough that appropriate compari-sons can be made between em-bryos at morphologically similar de-velopmental stages. Sticklebackembryos at 32 hours postfertilization(h) show Sox9a expression in cellsaround the eye and otic placode(Fig. 3A). A zebrafish embryo at ap-proximately the same stage of de-velopment shows strong sox9a ex-pression in the region of the otic

Fig. 2. A phylogenetic tree for Sox9-related genes. Numbers are bootstrap values for 1,000 trials. Dre, Danio rerio, zebrafish; Gac,Gasterosteus aculeatus, threespine stickleback; Gga, Gallus gallus, chicken; Hsa, Homo sapiens, human; Mal, Monopterus albus, rice eel;Mmu, Mus musculus, mouse; Tru, Takifugu rubripes, pufferfish. Dre Sox8 AW153579; Dre Sox9a AY090034; Dre Sox9b AAG09815; Dre Sox10AF402677; Gga Sox9 U12533; Gga Sox10 AF152356; Gga Sox8 AF228664; Hsa Sox8 NP_055402; Hsa Sox9 NP_000337; Hsa Sox10 NP_008872;Mal Sox9a AF378150; Mal Sox9b AF378151; Mmu Sox8 XP_128601; Mmu Sox9 NP_035578; Mmu Sox10 XP_128139; Tru Sox9a AAL32172(mayfold000587); Tru Sox9b mayfold 000421 (fugu assembly 3 at http://fugu.hgmp.mrc.ac.uk/).

Sox9 EXPRESSION IN STICKLEBACK AND ZEBRAFISH 483

placode and in the developingsomites (Fig. 3B). Zebrafish Sox9a ex-pression is weak around the eye,and reciprocally, stickleback Sox9aexpression is weak in the somites,both expression domains appearing

in this stage only after prolongedstaining (data not shown).

By 70 h, stickleback embryos con-tinue to have staining in the eye re-gion, as well as in the forebrain (Fig.3C). At this time, crest cells populat-

ing the pharyngeal arches, perhapsderived from the Sox9a-positive cellsin the otic region, stain intensely,and expression has begun in thepectoral fin bud. Expression is partic-ularly strong in the first (mandibular)

Fig. 3. Sox9a expression patterns. A,C,E: Sox9a expression in 32 hours postfertilization (h), 70-h, and 92-h stickleback embryos, respec-tively. B,D,F: sox9a expression in 14-h, 24-h, and 48-h zebrafish embryos. a1, mandibular arch; a2, hyoid arch; cb; ceratobranchial arches;e, eye; fb, forebrain; hb, hindbrain; mhb, midbrain–hindbrain border; nc, neural crest; op, otic placode region; pf, pectoral fin; s, somite.

Fig. 4. Sox9b expression patterns. A,C,E: Sox9a expression in 32 hours postfertilization (h), 70-h, and 92-h stickleback embryos. B,D,F:sox9a expression in 14-h, 24-h, and 48-h zebrafish embryos. a1, mandibular arch; a2, hyoid arch; e, eye; fb, forebrain; hb, hindbrain; n.notochord; nc, neural crest; op, otic placode region; pcp, prechordal plate; pf, pectoral fin; s, somite; t, tectum.

484 CRESKO ET AL.

and second (hyoid) arches andweaker in the ceratobranchialarches. Weak expression of Sox9aappears in the tail somites in 70-hembryos (Fig. 3C). Most of these ex-pression domains are similar in ze-brafish embryos of approximatelythe same stage of development,with strong expression in forebrain,crest cells populating the arches, tailsomites (Fig. 3D), and pectoral fin(data not shown). Two major differ-ences occur in the midbrain–hind-brain border and the somites, wherezebrafish has significantly strongerexpression than stickleback.

In 92-h stickleback embryos, Sox9acontinues to be expressed in thepharyngeal arches, and in the outermesenchyme of the pectoral fin,more strongly in the outer portion ofthe mesenchyme than the centralcore, as it is in zebrafish (Fig. 3E,F). Inzebrafish, Sox9a is expressed in astriped pattern in the hindbrain, likelyin glial cells as it is in mouse (Pom-polo and Harley, 2001). These resultsshow that the patterns of Sox9a ex-pression in stickleback and zebrafishare largely the same, but differ indetail, with the stickleback genestronger in the eye region, and thezebrafish gene stronger in thesomites and midbrain–hindbrainborder. Expression domains that aretaxon-specific involve evolutionarychange in developmental regula-tion after the divergence of stickle-back and zebrafish lineages.

Shared and DivergentExpression Patterns of Sox9bLike Sox9a, Sox9b is expressed ingenerally similar patterns in stickle-back and zebrafish. At 32 h, stickle-back embryos express Sox9b aroundthe eye and otic vesicle, and in theneural crest in the head and trunk(Fig. 4A). This stickleback expressionpattern is largely the same as thepattern for the orthologous gene inzebrafish (Fig. 4B), with the main dif-ference being that crest expressionin the trunk is comparatively strongerin zebrafish.

In 70-h stickleback embryos, theexpression pattern of Sox9b is morecomplex and extensive than Sox9ain the cranial region. At this stage,Sox9b transcripts accumulate in the

retina of the eye and in the forebrainand tectum (Fig. 4C). In the hind-brain, Sox9b is expressed in both te-leosts in a striped pattern as forSox9a in zebrafish (Fig. 3F). Sox9b isexpressed in the pharyngeal archesmore strongly in the ceratobranchi-als than in the mandibular and hyoidarches in both teleost embryos. Atthis stage, Sox9b is expressed weaklyin the somites of stickleback but notzebrafish embryos, and, like ze-brafish, Sox9b is expressed in thecrest at the end of the tail (Fig.4C,D).

By 92 h, expression of Sox9b in thestickleback brain is mainly in the pe-ripheral parts of the tectum and incells lining the ventricles (Fig. 4E), asin zebrafish (Fig. 4F). In the fin bud,Sox9b transcripts accumulate in thecentral core of the mesenchyme inboth stickleback and zebrafish. Ingeneral, the expression patterns ofSox9b are more similar between thetwo teleosts than are the expressionpatterns of Sox9a.

Conclusions and FutureDirectionsStickleback and zebrafish embryoshave largely similar embryonic ex-pression patterns for Sox9a andSox9b, and these combined Sox9aand Sox9b domains are generallypossessed by tetrapod outgroups asa function of a single Sox9 ortholog,suggesting that they are ancestral.Most partitioned expression differ-ences between the duplicate genesare common to both teleost species,such as strong expression in themandibular and hyoid arches forSox9a and strong expression in thetrunk crest for Sox9b. These cases dis-play the behavior expected if an-cestral regulatory elements drivingthese expression domains recipro-cally partitioned between the dupli-cate genes in the time interval be-tween the gene duplication eventand the divergence of sticklebackand zebrafish lineages. This is alsothe case for Mitf gene duplicates inteleosts. In tetrapods, there is a sin-gle Mitf gene with transcripts thatare expressed from different promo-tors and 5! exons (Yasumoto et al.,1998; Udono et al., 2000). In teleosts,in contrast, there are duplicated

genes mitfa and mitfb, which pro-duce these alternative isoforms ex-pressed in general in similar patternsto their tetrapod orthologs (Lister etal., 2001; Altschmied et al., 2002).Thus, for the only two cases studied,the general result is that most ances-tral gene subfunctions appear tohave partitioned before the teleostradiation. Clearly more cases needto be examined in detail before astrong generalization can be made.

For some ancestral expression do-mains, however, each co-ortholog isexpressed in a species-specific man-ner. Examples include the strong ex-pression of Sox9a in the zebrafishmidbrain–hindbrain border andsomites, but undetected or weak ex-pression of Sox9a in these domains instickleback, and reciprocally, thestrong expression of Sox9b aroundthe eye in stickleback but the weakor absent expression of sox9b in thesame region of zebrafish embryos. Ingeneral, for the Sox9 gene pair, thespecies-specific features appear tobe mainly quantitative rather thanqualitative. Such unshared domainshave the features predicted for reg-ulatory subfunctions that had notpartitioned at the time of the stickle-back/zebrafish lineage divergencebut then evolved differently in thetwo lineages.

A third class of expression patternsappears to be shared by Sox9a andSox9b in both teleost embryos. Ex-pression in the hindbrain and por-tions of the pectoral fin mesen-chyme appear to be examples. Ifhigher resolution cell-by-cell analysisshows that individual cells are tran-scribing both duplicates at the sametime, then these would be examplesof ancestral regulatory elementsthat remain unpartitioned to thepresent. However, the possibility re-mains that, even though at a grosslevel of examinations the domains ofexpression appear to be overlap-ping, a more detailed analysis mightreveal fine scale partitioning acrosscells within structures such as the finmesenchyme. In-depth studies ofapparently overlapping domainsshould be completed before con-clusions are drawn about whetherdomains remain unpartitioned inthese tissues.

The results with Sox9 and Mitf

Sox9 EXPRESSION IN STICKLEBACK AND ZEBRAFISH 485

genes suggest that most gene sub-functions may have assorted be-tween duplicated genes before theteleost explosion. But did the parti-tioning happen rapidly after dupli-cation or did it take a long period oftime relative to the duplicationevent and the present? An answerdepends upon knowing the date ofthe genome duplication and thedates at which teleost lineages di-verged. Unfortunately, the dating ofboth events is in question. Molecularclock analysis has suggested thatthe ray-fin fish genome duplicationoccurred more than 300 millionyears ago (MYA) (Taylor et al.,2001a) and that the teleost radiationoccurred approximately 140 MYA(Hedges and Kumar, 2002), but pa-leontologic evidence suggests thatthe radiation may have occurred200 MYA or more (Santini and Tyler,1999). Genomic analysis is neededfor basally diverging teleosts such aseels, and basally diverging ray-finfish, such as bowfin and bicher, tofurther resolve the question. At theextreme, however, the times be-tween genome duplication and theteleost radiation, and the radiationto the present, were about equal—approximately 150 million years.However, we hypothesize that theteleost radiation may have oc-curred sooner after the genome du-plication than previously thought,with perhaps only 50–100 millionyears intervening between the two.In either timing scenario, however,most subfunctions appear to havepartitioned before the teleost radia-tion, according to currently avail-able results. This timing would sup-port the idea that subfunctionpartitioning after duplication maybe a relatively rapid process.

The partitioning of ancestral sub-functions among duplicated genesmay contribute to phenotypic evo-lution. Darwin said that it is generallyacknowledged that all organic be-ings have been formed on two greatlaws - Unity of Type and Conditionsof Existence (Darwin, 1859). The re-sults of subfunction partitioning in te-leost Sox9 co-orthologs suggests thehypothesis that subfunction parti-tioning may contribute mechanisti-cally to Darwin’s generalization. Theearly subfunction partitioning that is

shared across lineages could pro-vide the unity of type, whereas laterlineage-specific, largely quantitativepartitioning, could provide the ge-netic fodder that allows lineages toacquire different distributions of traitsappropriate for their conditions ofexistence, and thus their divergence.Because genome duplication pro-vides a genome’s worth of gene du-plicates, even within a largely sharedset of partitioning that happenedsoon after duplication, it would still of-fer enormous opportunities for subse-quent independent subfunction parti-tioning and multiple dimensions alongwhich lineages could diverge.

The above considerations arebased on a very small data set.What is now needed is developmen-tal genetic analysis of a large num-ber of gene duplicates in a variety ofdifferent teleosts, including modelsfor microevolution, such as stickle-back and swordtails (Walter and Ka-zianis, 2001) and models amenableto mutagenesis such as zebrafishand medaka (Grunwald and Streis-inger, 1992; Loosli et al., 2000) to un-derstand the relative frequency ofconserved vs. independent parti-tioning, and what role independentpartitioning plays in divergence. Ona practical note, establishing this fre-quency is very important because itwill provide a sense of how often weshould expect model systems thathave experienced a genome dupli-cation event, such as pufferfish,medaka, and zebrafish, to unambig-uously provide insight into the func-tion of genes in other organisms, par-ticularly humans. At a morefundamental level, the hypothesisthat subfunction partitioning playeda role in the teleost radiation makesthe testable prediction that lineageswith the most independent partition-ing should be the most phenotypi-cally diverse. Ruling out this hypoth-esis would focus attention on othercauses for the most spectacular ver-tebrate example of biodiversity.

EXPERIMENTAL PROCEDURESIsolation of Stickleback Sox9aand Sox9bWe constructed two stickleback Fos-mid genomic libraries by using the

CopyControl Fosmid Library Produc-tion Kit from Epicentre (catalog no.CCFOS110). Each library was madefrom a single individual, both col-lected from the wild in Alaska. Oneindividual, collected from RabbitSlough, was from an anadromouspopulation exhibiting extensivebony lateral plate and pelvic armor.The other individual, collected fromBear Paw Lake, was from a popula-tion of stickleback whose membershave lost almost all bony armor.These fish were transported to theUniversity of Oregon stickleback fa-cility and reared for 2 months beforebeing killed. Stickleback clone in-serts average approximately 50 kbpin length, and the arrayed librariesprovide approximately 12" cover-age of the stickleback genome, as-suming 600 Mbp in the haploid ge-nome (Vinogradov, 1998).

We screened the library for Sox9genes by using polymerase chain re-action (PCR) primers designed toconserved regions of the first exonsof the Fugu rubripes Sox9a andSox9b genes. The forward primer was5!-TGAATCTCCTCGACCCTTACC-3!,and the reverse primer was 5!-TG-CAGCCTGAGCCCACAC-3!. Sevenindependent positive clones wereidentified. We sheared and sub-cloned four fosmid clones that wereboth positive for Sox9, but which hadtwo distinct restriction enzyme diges-tion patterns. Shearing was per-formed by using a nebulizer and theends of the subclones were repairedby using a mixture of T4 polymeraseand Klenow fragment. The bluntedsubclones were then ligated into thePCR4-BLUNT cloning vector (Invitro-gen TOPO Shotgun Subcloning Kit,catalog nos. K7000-01, K7010-01,K7050-01, and K7060-01). The aver-age size of inserted DNA was 2–5 kb.Subclones were arrayed into 96-wellplates as single clones and thenscreened by means of PCR with thesame first exon primers describedabove. Positive clones were se-quenced from both ends of the vec-tor. Additionally, 16 randomly cho-sen subclones from each Fosmidwere sequenced. In all, this methodprovided full-length genomic se-quence for Sox9a (9908 bp) andSox9b (6495 bp), including 5! and 3!untranslated regions (UTRs), introns,

486 CRESKO ET AL.

and perhaps some regulatory ele-ments. Each Sox9-positive contig wassequenced to an average of 6" cov-erage, equally distributed in either di-rection (GenBank accession nos.AY351914 and AY351915). The genestructure of the Sox9 gene in eachcontig was predicted by using theGENSCAN (Burge and Karlin, 1998)Web server (http://genes.mit.edu/GENSCAN.html).

Phylogenetic AnalysisProtein sequence for zebrafish, riceeel, mouse, human, and chicken or-thologs of Sox8, Sox9, and Sox10 wereobtained from Entrez (http://www.ncbi.nlm.nih.gov:80/entrez/query.fcgi?db#Nucleotide). The zebra-fish Sox9a and Sox9b sequenceswere used to BLAST (Altschul et al.,1997) against the Fugu database(http://ensembl.fugu-sg.org/Fugu_rubripes/blastview), and we ob-tained one single strong alignmentfor each paralog. The gene struc-ture for each was determined byanalyzing the genomic sequenceusing the GENSCAN Web server, aswas done for the stickleback genes.The published Sox9 protein sequencesand the inferred pufferfish and stickle-back protein sequences were ana-lyzed by using ClustalX (version 1.82;Thompson et al., 1994) to establish thephylogenetic relationship of these se-quences. Node confidence was as-certained by bootstrapping the dataset 1,000 times (Felsenstein, 1985), andthe resulting tree was visualized byusing NJPlot (Perriere and Gouy,1996; http://pbil.univ-lyon1.fr/software/njplot.html).

Detecting Gene Expression inStickleback Embryos by In SituHybridizationLinearized genomic clones of eachgene were transcribed in vitro tomake riboprobes using digoxigenin-labeled UTP. The Sox9a probe wasgenerated from a subclone in aPCR4-Shotgun/Not1-linearized tem-plate using T3 RNA polymerase,whereas the Sox9b probe was syn-thesized from a PCR4-TOPO/Not1-lin-earized plasmid also using T3 RNApolymerase. The Sox9a probe is 650bp in length, covering the 5!UTR and

most of the first exon. The Sox9bprobe is 1,800 bp in length, coveringmost of exons 1 and 3, and all ofexon 1 and introns 1 and 2. Embryoswere fixed in 4% paraformaldehydeat 20°C for 1 week, after which timethey were dechorionated by handand processed for in situ hybridiza-tion as described (Chiang et al.,2001; Yan et al., 2002). Sticklebackembryos were staged based onmorphologic criteria in analogy withthe zebrafish staging series (Kimmelet al., 1995).

ACKNOWLEDGMENTSWe thank for support NSF grants IBN0236239 for stickleback, NSF grant IBN9728587 for sex gene research, NIHgrant R01RR10715 for zebrafish, andNIH grant 5 F32 GM020892 for post-doctoral training support (W.A.C.).NSF IGERT grant DGE 9972830 sup-ported summer undergraduate re-searchers Diana Bradley, RebeccaLoda, Melanie Robinson, and MarkRothgary, who helped screen libraries.We also thank Mike Bell, Susan Foster,John Baker, and Jeff Walker and theirstudents for developing the system ofAlaskan stickleback populations intothe great resource for studies in micro-evolution and development that it istoday.

REFERENCESAdamska M, Wolff A, Kreusler M, Witt-

brodt J, Braun T, Bober E. 2001. FiveNkx5 genes show differential expres-sion patterns in anlagen of sensory or-gans in medaka: insight into the evolu-tion of the gene family. Dev Genes Evol211:338–349.

Ahn D, Gibson G. 1999. Expression pat-terns of threespine stickleback Hoxgenes and insights into the evolution ofthe vertebrate body axis. Dev GenesEvol 209:482–494.

Akimenko M-A, Johnson SL, WesterfieldM, Ekker M. 1995. Differential inductionof four msx homeobox genes during findevelopment and regeneration in ze-brafish. Development 121:347–357.

Allendorf FW, Utter FM, May BP. 1975.Gene duplication within the family Sal-monidae: II. Detection and determina-tion of the genetic control of duplicateloci through inheritance studies andthe examination of populations. In:Markert CL, editor. Isozymes. New York:Academic Press. p 415–432.

Altschmied J, Delfgaauw J, Wilde B, Dus-chl J, Bouneau L, Volff JN, Schartl M.2002. Subfunctionalization of duplicate

mitf genes associated with differentialdegeneration of alternative exons infish. Genetics 161:259–267.

Altschul SF, Madden TL, Schaffer AA,Zhang J, Zhang Z, Miller W, Lipman DJ.1997. Gapped BLAST and PSI-BLAST: anew generation of protein databasesearch programs. Nucleic Acids Res 25:3389–3402.

Amores A, Force A, Yan Y-L, Joly L,Amemiya C, Fritz A, Ho RK, LangelandJ, Prince V, Wang Y-L, Westerfield M,Ekker M, Postlethwait JH. 1998. Ze-brafish hox clusters and vertebrate ge-nome evolution. Science 282:1711–1714.

Amores A, Suzuki T, Yan Y, Pomeroy J,Singer A, Amemiya C, Postlethwait J.2003. Developmental roles of pufferfishHox clusters and genome evolution inray-fin fish. Genome Research (inpress).

Bagheri-Fam S, Ferraz C, Daemaille J,Scherer G, Pfeifer D. 2001. Compara-tive genomics of the SOX9 region inhuman and Fugu rubripes: conserva-tion of short regulatory sequence ele-ments within large intergenic regions.Genomics 78:73–82.

Bailey GS, Poulter RTM, Stockwell PA.1978. Gene duplication in tetraploidfish: model for gene silencing at un-linked duplicated loci. Proc Natl AcadSci U S A 75:5575–5579.

Bell MA. 2001. Ontogeny and evolution oflateral plate number in low morphthreespine stickleback fish, Gasteros-teus aculeatus. J Morphol 248:205.

Bell MA, Foster SA. 1994. The evolutionarybiology of the threespine stickleback.Oxford: Oxford University Press. 571 p.

Bell MA, Orti G. 1994. Pelvic reduction inthreespine stickleback from Cook Inletlakes: geographical distribution and in-trapopulation variation. Copeia:314–325.

Bell MA, Orti G, Walker JA, Koenings JP.1993. Evolution of pelvic reduction inthreespine stickleback fish: a test ofcompeting hypotheses. Evolution 47:906–914.

Bell DM, Leung KK, Wheatley SC, Ng L-J,Zhou S, Ling KW, Sham MH, Koopman P,Tam PPL, Cheah KSE. 1997. SOX9 di-rectly regulates the type-II collagengene. Nat Genet 16:174–178.

Bi W, Huang W, Whitworth DJ, Deng JM,Zhang Z, Behringer RR, de Crombrug-ghe B. 2001. Haploinsufficiency of Sox9results in defective cartilage primordiaand premature skeletal mineralization.Proc Natl Acad Sci U S A 98:6698–6703.

Bingham S, Nasevicius A, Ekker SC, Chan-drasekhar A. 2001. Sonic hedgehogand tiggy-winkle hedgehog coopera-tively induce zebrafish branchiomotorneurons. Genesis 30:170–174.

Bisbee CA, Baker MA, Wilson AC, Iran-dokht HA, Fischberg M. 1977. Albuminphylogeny for clawed frogs (Xenopus).Science 195:785–787.

Bruce AE, Oates AC, Prince VE, Ho RK.2001. Additional hox clusters in the ze-brafish: divergent expression patterns

Sox9 EXPRESSION IN STICKLEBACK AND ZEBRAFISH 487

belie equivalent activities of duplicatehoxB5 genes. Evol Dev 3:127–144.

Burge CB, Karlin S. 1998. Finding thegenes in genomic DNA. Curr OpinStruct Biol 8:346–354.

Chiang E, Pai C-I, Wyatt M, Yan Y-L,Postlethwait J, Chung B-C. 2001. Twosox9 genes on duplicated zebrafishchromosomes: expression of similartranscription activators in distinct sites.Dev Biol 229:149–163.

Cooke JMA, Nowak MA, Boerlijst M, May-nard Smith J. 1997. Evolutionary originsand maintenance of redundant geneexpression during metazoan develop-ment. Trends Genet 13:360–364.

Darwin C. 1859. The origin of species. Lon-don: John Murray.

De Martino S, Yan Y-L, Jowett T, Postleth-wait JH, Varga ZM, Ashworth A, AustinCA. 2000. Expression of sox11 gene du-plicates in zebrafish suggests the recip-rocal loss of ancestral gene expressionpatterns in development. Dev Dyn 217:279–292.

Ekker M, Wegner J, Akimenko M-A,Westerfield M. 1992. Coordinate em-bryonic expression of three zebrafishengrailed genes. Development 116:1001–1010.

Ekker M, Akimenko M, Allende M, Smith R,Drouin G, Langille R, Weinberg E,Westerfield M. 1997. Relationshipsamong msx gene structure and func-tion in zebrafish and other vertebrates.Mol Biol Evol 14:1008–1022.

Felsenstein J. 1985. Confidence limits onphylogenies: an approach using thebootstrap. Evolution 39:783–791.

Ferris SD, Whitt GS. 1979. Evolution of thedifferential regulation of duplicategenes after polyploidization. J Mol Evol12:267–317.

Force A, Lynch M, Pickett FB, Amores A,Yan Y-L, Postlethwait J. 1999. Preserva-tion of duplicate genes by comple-mentary, degenerative mutations. Ge-netics 151:1531–1545.

Force AG, Cresko WA, Pickett FB. 2003.Informational accretion, gene duplica-tion, and the mechanisms of geneticmodule parcellation. In: Schlosser G,Wagner G, editors. Modularity in devel-opment and evolution. Chicago: Uni-versity of Chicago Press.

Gates MA, Kim L, Egan ES, Cardozo T,Sirotkin HI, Dougan ST, Lashkari D, Aba-gyan R, Schier AF, Talbot WS. 1999. Agenetic linkage map for zebrafish:comparative analysis and localizationof genes and expressed sequences.Genome Res 9:334–347.

Gibson G. 2002. A genetic attack on thedefense complex. Bioessays 24:487–489.

Graf J-D, Kobel HR. 1991. Xenopus laevis:practical uses in cell and molecular bi-ology: genetics. In: Kay BK, Peng HB,editors. Methods in cell biology. NewYork: Academic Press. p 19–34.

Grunwald DJ, Streisinger G. 1992. Induc-tion of recessive lethal and specific lo-cus mutations in the zebrafish with ethylnitrosourea. Genet Res 59:103–116.

Healy C, Uwanogho D, Sharpe PT. 1999.Regulation and role of Sox9 in cartilageformation. Dev Dyn 215:69–78.

Hedges SB, Kumar S. 2002. Genomics.Vertebrate genomes compared. Sci-ence 297:1283–1285.

Hughes AL. 1994. The evolution of func-tionally novel proteins after gene dupli-cation. Proc R Soc Lond B 256:119–124.

Hughes AL. 1999. Adaptive evolution ofgenes and genomes. New York: OxfordUniversity Press.

Hughes MK, Hughes AL. 1993. Evolution ofduplicate genes in a tetraploid animal,Xenopus laevis. Mol Biol Evol 10:1360–1369.

Kimmel CB, Ballard WW, Kimmel SR, Ull-mann B, Schilling TF. 1995. Stages ofembryonic development of the ze-brafish. Dev Dyn 203:253–310.

Kist R, Schrewe H, Balling R, Scherer G.2002. Conditional inactivation of Sox9:a mouse model for campomelic dys-plasia. Genesis 32:121–123.

Lefebvre V, Huang W, Harley VR, Good-fellow PN, de Crombrugghe B. 1997.SOX9 is a potent activator of the chon-drocyte-specific enhancer of the proalpha1(II) collagen gene. Mol Cell Biol17:2336–2346.

Li W-H. 1980. Rate of gene silencing atduplicate loci: a theoretical study andinterpretation of data from tetraploidfishes. Genetics 95:237–258.

Lister J, Close J, Raible D. 2001. Duplicatemitf genes in zebrafish: complemen-tary expression and conservation ofmelanogenic potential. Dev Biol 237:333–344.

Loosli F, Koster RW, Carl M, Kuhnlein R,Henrich T, Mucke M, Krone A, WittbrodtJ. 2000. A genetic screen for mutationsaffecting embryonic development inmedaka fish (Oryzias latipes). MechDev 97:133–139.

Lynch M, Force A. 2000. The origin of in-terspecific genomic incompatibility viagene duplication. Am Nat 156:590–605.

McClintock JM, Carlson R, Mann DM,Prince VE. 2001. Consequences of Hoxgene duplication in the vertebrates: aninvestigation of the zebrafish Hox para-loguegroup1genes.Development128:2471–2484.

McClintock JM, Kheirbek MA, Prince VE.2002. Knockdown of duplicated ze-brafish hoxb1 genes reveals distinctroles in hindbrain patterning and anovel mechanism of duplicate generetention. Development 129:2339–2354.

Meyer A, Schartl M. 1999. Gene and ge-nome duplications in vertebrates: theone-to-four (-to-eight in fish) rule andthe evolution of novel gene functions.Curr Opin Cell Biol 11:699–704.

Morizot DC, Slaugenhaupt SA, KallmanKD, Chakravarti A. 1991. Genetic link-age map of fishes of the genus Xi-phophorus (Teleostei: Poeciliidae). Ge-netics 127:399–410.

Nadeau JH, Sankoff D. 1997. Compara-ble rates of gene loss and functional

divergence after genome duplicationsearly in vertebrate evolution. Genetics147:1259–1266.

Nei M, Roychoudhury AK. 1973. Probabil-ity of fixation of nonfunctional genes atduplicate loci. Am Nat 107:362–372.

Nelson JS. 1994. Fishes of the world. NewYork: Wiley-Interscience.

Ng L-J, Wheatley S, Muscat GEO, Con-way-Campbell J, Bowles J, Wright E,Bell DM, Tam PPL, Cheah KSE, Koop-man P. 1997. SOX9 binds DNA, acti-vates transcription, and coexpresseswith type II collagen during chondro-genesis in the mouse . Dev Biol 183:108–121.

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

Peichel CL, Nereng KS, Ohgi KA, Cole BL,Colosimo PF, Buerkle CA, Schluter D,Kingsley DM. 2001. The genetic archi-tecture of divergence betweenthreespine stickleback species. Nature414:901–905.

Perriere G, Gouy M. 1996. WWW-query:an on-line retrieval system for biologi-cal sequence banks. Biochimie 78:364–369.

Pompolo S, Harley VR. 2001. Localisationof the SRY-related HMG box protein,SOX9, in rodent brain. Brain Res 906:143–148.

Postlethwait J, Yan Y, Gates M, Horne S,Amores A, Brownlie A, Donovan A,Egan E, Force A, Gong Z, Goutel C, FritzA, Kelsh R, Knapik E, Liao E, Paw B, Ran-som D, Singer A, Thomson M, Abduljab-bar T, Yelick P, Beier D, Joly J, Larham-mar D, Talbot W. 1998. Vertebrategenome evolution and the zebrafishgene map. Nat Genet 18:345–349.

Robinson-Rechavi M, Marchand O, Es-criva H, Bardet PL, Zelus D, Hughes S,Laudet V. 2001. Euteleost fish genomesare characterized by expansion ofgene families. Genome Res 11:781–788.

Santini F, Tyler JC. 1999. A new phyloge-netic hypothesis for the order Tetra-odontiformes (Teleostei, Pisces), withplacement of the most fossil basal lin-eages. Am Zool 39:10A.

Sidow A. 1996. Gen(om)e duplications inthe evolution of early vertebrates. CurrOpin Genet Dev 6:715–722.

Smith S, Metcalfe JA, Elgar G. 2000. Iden-tification and analysis of two snailgenes in the pufferfish (Fugu rubripes)and mapping of human SNA to 20q.Gene 247:119–128.

Smith S, Metcalfe JA, Elgar G. 2001. Char-acterization of two topoisomerase 1genes in the pufferfish (Fugu rubripes).Gene 265:195–204.

Spokony R, Aoki Y, Saint-Germain N,Magner-Fink E, Saint-Jeannet J. 2002.The transcription factor Sox9 is requiredfor cranial neural crest development inXenopus. Development 129:421–432.

Stoltzfus A. 1999. On the possibility of con-structive neutral evolution. J Mol Evol49:169–181.

Takahata N, Maruyama T. 1979. Polymor-phism and loss of duplicate gene ex-

488 CRESKO ET AL.

pression: a theoretical study with appli-cation to tetraploid fish. Proc NatlAcad Sci U S A 76:4521–4525.

Taylor JS, Van de Peer Y, Braasch I, MeyerA. 2001a. Comparative genomics pro-vides evidence for an ancient genomeduplication event in fish. Philos Trans RSoc Lond B 356:1661–1679.

Taylor JS, Van de Peer Y, Meyer A. 2001b.Revisiting recent challenges to the an-cient fish-specific genome duplicationhypothesis. Curr Biol 11:R1005–R1008.

Taylor J, Braasch I, Frickey T, Meyer A,Van De Peer Y. 2003. Genome duplica-tion, a trait shared by 22,000 species ofray-finned fish. Genome Res 13:382–390.

Thompson JD, Higgins DG, Gibson TJ.1994. CLUSTAL W: improving the sensi-tivity of progressive and multiple se-quence alignment through sequenceweighting, positions-specific gap pen-alties, and weight matrix choice. Nu-cleic Acids Res 22:4673–4680.

Udono T, Yasumoto K, Takeda K, Amae S,Watanabe K, Saito H, Fuse N, Tachi-bana M, Takahashi K, Tamai M, Shiba-hara S. 2000. Structural organization ofthe human microphthalmia-associ-ated transcription factor gene con-taining four alternative promoters. Bio-chim Biophys Acta 1491:205–219.

Van de Peer Y, Frickey T, Taylor J, MeyerA. 2002. Dealing with saturation at the

amino acid level: a case study basedon anciently duplicated zebrafishgenes. Gene 295:205–211.

Vidal VPI, Chaboissier MC, de Rooij DG,Schedl A. 2001. Sox9 induces testis de-velopment in XX transgenic mice. NatGenet 28:216–217.

Vinogradov A. 1998. Genome size andGC-percent in vertebrates as deter-mined by flow cytometry: the triangu-lar relationship. Cytometry 31:100–109.

Wagner T, Wirth J, Meyer J, Zabel B, HeldM, Zimmer J, Pasantes J, Bricarelli FD,Keutel J, Hustert E. 1994. Autosomal sexreversal and campomelic dysplasiaare caused by mutations in andaround the SRY-related gene SOX9.Cell 79:111–1120.

Walsh JB. 1995. How often do duplicatedgenes evolve new functions? Genetics110:345–364.

Walter RB, Kazianis S. 2001. Xiphophorusinterspecies hybrids as genetic modelsof induced neoplasia. ILAR J 42:299–321.

Watterson GA. 1983. On the time forgene silencing at duplicate loci. Ge-netics 105:745–766.

Wegner M. 1999. From head to toes: themultiple facets of Sox proteins. NucleicAcids Res 27:1409–1420.

Woods IG, Kelly PD, Chu F, Ngo-HazelettP, Yan Y-L, Huang H, Postlethwait JH,Talbot WS. 2000. A comparative map

of the zebrafish genome. Genome Res10:1903–1914.

Wright EM, Snopek B, Koopman P. 1993.Seven new members of the Sox genefamily expressed during mouse devel-opment. Nucleic Acids Res 21:744.

Yan YL, Miller CT, Nissen R, Singer A, Liu D,Kirn A, Draper B, Willoughby J, MorcosPA, Amsterdam A, Chung BC, Wester-field M, Haffter P, Hopkins N, Kimmel C,Postlethwait JH. 2002. A zebrafish sox9gene required for cartilage morpho-genesis. Development 129:5065–5079.

Yasumoto K, Amae S, Udono T, Fuse N,Takeda K, Shibahara S. 1998. A biggene linked to small eyes encodes mul-tiple Mitf isoforms: many promotersmake light work. Pigment Cell Res 11:329–336.

Yu WP, Brenner S, Venkatesh B. 2003. Du-plication, degeneration and subfunc-tionalization of the nested synapsin-Timp genes in Fugu. Trends Genet 19:180–183.

Zhao Q, Eberspaecher H, Lefebvre V, deCrombrugghe B. 1997. Parallel expres-sion of Sox9 and Col2a1 in cells under-going chondrogenesis. Dev Dyn 209:377–386.

Zhou R, Cheng H, Zhang Q, Guo Y, Coo-per R, Tiersch T. 2002. SRY-relatedgenes in the genome of the rice fieldeel (Monopterus albus). Genet Sel Evol34:129–137.

Sox9 EXPRESSION IN STICKLEBACK AND ZEBRAFISH 489