The caspase family in urochordates: distinct evolutionary fates in ascidians and larvaceans

Biol. Cell (2005) 97, 857–866 (Printed in Great Britain) doi:10.1042/BC20050018 Research article

The caspase family in urochordates:distinct evolutionary fates inascidians and larvaceansMylene Weill*, Alexandre Philips†, Daniel Chourrout‡ and Philippe Fort†1

*Institut des Sciences de l’Evolution (UMR 5554), C.C. 065, Univ. Montpellier II, 34095 Montpellier cedex 5, France, †Centre de

Recherche en Biochimie des Macromolecules (FRE2593), 1919 route de Mende, 34293 Montpellier cedex 5, France, and ‡Sars Centre

for Marine Molecular Biology, Bergen High Technology Centre, Thormøhelnsgaten 55, 5008 Bergen, Norway

Background information. Caspases are cysteine proteases that mediate apoptosis (programmed cell death) in-itiation and execution. Apoptosis is a conserved mechanism shared by all metazoans, although its physiologicalfunction and complexity show considerable taxon-dependent variations. To gain insight into the caspase re-pertoire of putative ancestors to vertebrates, we performed exhaustive genomic searches in urochordates, a sistertaxon to vertebrates in which ascidians and appendicularians display chordate characters at early stages of theirdevelopment.

Results. We identified the complete caspase families of two ascidians (Ciona intestinalis and C. savignyi) and onelarvacean (Oikopleura dioica). We found in ascidian species an extremely high number of caspase genes (17 forC. intestinalis and 22 for C. savignyi), deriving from five founder gene orthologues to human pro-inflammatory,initiator and executioner caspases. Although considered to be sibling species, C. intestinalis and C. savignyi onlyshare 11 orthologues, most of the additional genes resulting from recent mass duplications. A sharply contrastedpicture was found in O. dioica, which displayed only three caspase genes deriving from a single founder genedistantly related to caspase 3/7. The difference between ascidian and larvacean caspase repertoires is discussedin the light of their developmental patterns and life cycles.

Conclusions. The identification of caspase members in two ascidian species delineates five founder genes thatbridge the gap between vertebrates and Ecdysozoa (arthropods and nematodes). Given the amazing diversityamong urochordates, determination and comparison of the caspase repertoires in species from orders additionalto Enterogona (ascidians) and Oikopleuridae might be highly informative on the evolution of caspase-dependentphysiological processes.

IntroductionApoptosis is a major active process leading to celldeath and is critical for many aspects of animal devel-opment. The best-known examples of apoptosis prob-ably are the removal of interdigital webs in verteb-rates (Glucksmann, 1951) and the loss of the tail inanuran tadpoles (Tata, 1996). Apoptotic events have

1To whom correspondence should be addressed (emailphilippe.fort@crbm.cnrs.fr).Key words: apoptosis, cysteine protease, developmental pattern,metamorphosis, tunicate.Abbreviations used: CARD, caspase recruitment domain; CiCSP,C. intestinalis caspase; DED, death effector domain; EST, expressed sequencetag; IAP, inhibitor of apoptosis protein; ICE, interleukin-1β-converting enzyme;NF-κB, nuclear factor κB; RT, reverse transcriptase.

also been reported in lower eukaryotes, in particularCaenorhabditis elegans in which core components ofthe cell-death machinery were identified first (Yuanet al., 1993). Apoptosis results either from proapop-totic signals emanating from the cellular neighbour-hood or simply from withdrawal of survival factors orcell attachment (Strasser et al., 2000). Both types ofevents converge on intracellular activation of caspases,a group of cysteine proteases that regulate cell-deathinitiation and degrade critical cellular components atlater stages.

Caspases are synthesized as inactive precursorscomposed of an N-terminal prodomain, followed byp20 and p10 catalytic subunits. Activation is trig-gered by proteolytic processing that cleaves apart

www.biolcell.org | Volume 97 (11) | Pages 857–866 857

M. Weill and others

the prodomain, the p20 and the p10 subunits, al-lowing the formation of the active complex made oftwo p20 and two p10 subunits (Salvesen and Dixit,1997). Thirteen different caspases have been identi-fied in the human and murine genomes. One class in-cludes caspases 1, 4, 5, 11 and 12 also known as ICEs(interleukin-1β-converting enzymes), which matur-ate pro-inflammatory cytokines. The other caspaseshave pivotal roles in apoptosis, and some of them haveroles in control of proliferation and differentiationof specific cell types (reviewed in Abraham andShaham, 2004). Apoptotic caspases are grouped asinitiator or executioner caspases: initiator caspasesdisplay long prodomains containing either a CARD(caspase recruitment domain), found in caspases 2 and9, or a DED (death effector domain), found in cas-pases 8 and 10. Both domains mediate caspase auto-activation through aggregation with cellular adaptermolecules. Executioner caspases (caspases 3, 6, 7 and14) display short prodomains and are usually activ-ated by initiator caspases.

Comprehensive searches in genomes showed thatcaspases belong to a larger family that also includesparacaspases and metacaspases, found in all kingdoms(Uren et al., 2000). True caspase homologues werefound only in Metazoa, from cnidarians (Cikala et al.,1999) and sponges (Wiens et al., 2003). Caspase fam-ilies show a great taxon-dependent diversity in size,ranging from three members in Caenorhabditis elegansto 10–13 in vertebrates. Urochordates (or tunicates)represent a sister taxon to vertebrates in which ascid-ians and appendicularians display chordate charac-ters at early stages of their development (Jeffery andSwalla, 1997). As such, urochordates have become abiological model that bridges the gap between ver-tebrates and Ecdysozoa (arthropods/nematodes). Inthis study, we undertook a search for caspase genesin three recently sequenced tunicate genomes [the as-cidians C. intestinalis (Dehal et al., 2002) and C. sav-ignyi, and the appendicularian Oikopleura dioica].

ResultsCaspase diversity in C. intestinalisWe reported previously the identification of 15 dis-tinct short conserved caspase domains in the rawgenomic sequences of C. intestinalis (Chambon et al.,2002). We have completed here the analysis of thecaspase family by exhaustive searches in genomic and

EST (expressed sequence tag) databases. We recon-structed a total of 17 mRNAs encoding caspases(CSP), 14 of which encoded complete p20/p10 con-served domains (Table 1). Six caspases are additionalto those described in a previous survey of componentsof the C. intestinalis cell-death machinery (Terajimaet al., 2003). mRNA expression of each potentialcaspase gene was estimated both from RT (reversetranscriptase)–PCR analysis and searches in EST data-bases (Table 2). Except for CSP4d and CSP5, forwhich no EST was found and RT–PCR was negative,caspase genes were all expressed as mRNAs althoughat very low levels, as illustrated by CSP2a, only de-tected from RT–PCR. Occurrences in EST databasesranged from 1 (CSP1b2) to 21 (CSP1b1) with an av-erage of 5.6 on the 239086 sequenced ESTs. Thesewere values too low to allow significant compari-sons of the spatio-temporal expression of each caspasemRNA. However, global comparison showed signi-ficant excess of CSP1b1, CSP2b, CSP3c and CSP8/9ESTs (P < 0.003), as well as higher overall incidenceof caspase EST in blood cells (P < 0.003) and theneural complex (P < 0.03).

C. intestinalis thus expresses an unexpectedly highnumber of different caspases compared with Caen-orhabditis elegans (three), Drosophila (seven) and evenvertebrates (10–13). To identify putative pseudo-genes, we compared the catalytic p20/p10 domains ofall the 17 proteins (Figure 1A). Alignment showed anoverall conservation of the residues involved in cata-lytic activity and in formation of the substrate pocket.A particular feature of CiCSP1a (C. intestinalis CSP1a)is the presence of a Glu residue instead of the canon-ical Gln involved in the co-ordination of the P1 as-partate of the substrate (position −2 with respect tothe catalytic Cys residue). Only CiCSP5 appearedto be totally deficient for these features. CiCSP5 lacksin particular the H-C catalytic dyad, whereas it isconserved in the distantly related paracaspase andmetacaspase families (Uren et al., 2000). In addition,CiCSP5 was not found in ESTs nor amplified by RT–PCR (Table 1), suggesting that it probably is a silentpseudogene.

The complexity of the C. intestinalis caspase familywas next compared with that of the human family bymultiple alignment and neighbour-joining tree clus-tering. As shown in Figure 1(B), p20/p10 domainsclustering delineated six distinct C. intestinalisgroups: CSP1 (a, b1 and b2), CSP2 (a and b), CSP3

858 C© Portland Press 2005 | www.biolcell.org

Evolution of the caspase family in urochordates Research article

Table 1 Members of the caspase family in C. intestinalisAnnotated sequences are identified using their accession number from Ensembl or from Grail or JGI (Joint Genome Institute) models. Twounannotated genes were detected from BLAST searches of the EST database and are identified accordingly (*). The sizes of complete andpartial (>) putative proteins as well as the presence of CARD, DED, p20 and p10 structural regions are indicated.

Ciona gene models

Name Size (amino acids) Domains Ensembl Grail JGI

CiCSP1a 402 p20/p10 – – ci0100152164

CiCSP1b1 279 p20/p10 – grail.101.16.1 ci0100135423

CiCSP1b2 >386 p20/p10 AAF73848 – ci0100145344

CiCSP2a 436 CARD/p20/p10 P29594 grail.48.81.1 ci0100136652

CiCSP2b 529 CARD/p20/p10 NP 071967 grail.37.27.1 ci0100138194

CiCSP3a >320 p20/p10 – grail.37.65.1 ci0100140801

CiCSP3b 344 p20/p10 NP 071596 grail.65.45.1 ci0100139404

CiCSP3c 732 p20/p10 AAN45849 grail.37.75.1 ci0100136167

CiCSP3d 404 p20/p10 – – ci0100146025

CiCSP3e 321 p20/p10 AAC32602 grail.289.3.1 ci0100150284

CiCSP4a >397 p20/p10 – – cicl055o23*

CiCSP4b >308 p20/? AAD41595 grail.1766.1.1 ci0100138028

CiCSP4c >299 p20/? – – cieg20n24*

CiCSP4d >297 p20/? – – ci0100141626

CiCSP5 >315 p20 CAA04196 – ci0100138986

CiCSP6 >329 p20/p10 BAA94747 grail.574.10.1 ci0100142190

CiCSP8/9 697 (DED)2/p20/p10 AAL23700 grail.1.20.1 ci0100134691

Table 2 Expression of caspase mRNA in C. intestinalisFor each C. intestinalis caspase gene, EST occurrences in cDNA libraries from different developmental stages or adult tissues were compiledfrom the Ghost database (http://ghost.zool.kyoto-u.ac.jp/). ‘+’ and ‘−’ indicate positive or negative amplification using a 35 cycle RT–PCRperformed on total mRNA from a mixed developmental stage population. n.d., not determined. nbESTs, total number of sequenced cDNAper library; p, EST occurrence, expressed as ‰. Values statistically different from Poisson or normal distributions are in boldface.

Stage nb ESTs p (‰) 1a 1b1 1b2 2a 2b 3a 3b 3c 3d 3e 4a 4b 4c 4d 5 6 8/9

Eggs 29444 0.24 0 0 0 0 2 0 0 3 0 0 0 0 1 0 0 1 0

Cleaving embryos 26796 0.26 0 0 0 0 0 0 0 2 0 0 2 1 0 0 0 2 0

Gastrulae and neurulae 23475 0.55 0 4 0 0 6 0 0 1 1 0 0 0 0 0 0 1 0

Tailbud embryos 31209 0.32 0 2 0 0 3 3 0 0 0 0 1 1 0 0 0 0 0

Larvae 24532 0.33 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 1 5

Young adults 29138 0.48 1 4 1 0 0 0 4 2 0 0 0 0 1 0 0 0 1

Gonad 16239 0.18 1 0 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0

Testis 4717 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Endostyle 2497 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

The neural complex 10029 0.80 1 2 0 0 1 0 0 0 2 1 0 0 0 0 0 0 1

Heart 12414 0.40 0 1 0 0 0 0 0 1 1 0 0 0 0 0 0 0 2

Blood cell 28596 0.73 0 6 0 0 2 0 1 2 1 1 2 2 0 0 0 0 4

Total 96 3 21 1 0 16 3 5 11 5 2 5 4 2 0 0 5 13

RT–PCR + + n.d. + n.d. + + + + + n.d. n.d. n.d. – – n.d. +

(a–e), CSP4 (a–d), CSP6 and CSP8/9. Except for theCSP1 cluster, each group is more or less closely relatedto one human caspase subfamily. C. intestinalis cas-

pases distribute as apoptosis-related executionercaspases (CSP3 and CSP6), initiator apoptotic caspase(CSP2 and CSP8/10) and inflammatory and immune

www.biolcell.org | Volume 97 (11) | Pages 857–866 859

M. Weill and others

Figure 1 Comparison of the p20/p10 domains of the C. intestinalis caspases(A) C. intestinalis p20/p10 domains were aligned using CLUSTAL X. Included in the alignment are the p20/p10 domains of

human caspase 3 (HsCASP3) as well as the consensus CASc (CASpase consensus) domain [CDD database (Marchler-Bauer

et al., 2003)]. Residues shared by at least 70% of the sequences are shaded. Residues involved in the catalytic activity (H-C

dyad) are indicated by ‘@’ and those lining the substrate pocket by ‘*’. Numbers delineate the position of the p20/p10 domains.

(B) Sequence proximity between C. intestinalis (Ci, green background) and human (Hs) caspases. p20/p10 domains were aligned

using the CLUSTAL X algorithm and a bootstrapped neighbour-joining unrooted tree was inferred. �, nodes with bootstrap values

above 80%; �, nodes with 50–80% values. Shaded are nodes of bootstrap values lower than 15%. The presence of CARD or

DED is indicated.

response caspases (CSP4). Like their human coun-terparts, CiCSP2 members display in their prodo-mains a CARD, which strengthens the relationshipdeduced from the p20/p10 sequence similarity. Simi-larly, CiCSP8/9 displays two DED motifs, like humancaspases 8 and 10. This leaves open the possibilitythat, although CiCSP8/9 shows a p20/p10 domainmore similar to that of human caspase 9, it might befunctionally related to human caspase 8 or 10. To theopposite, no CARD motif was found in sequences ofthe CiCSP4 cluster, whereas CARDs are present inprodomains of human caspases 1, 4 and 5.

The caspase family in the Ciona genusWe next took advantage of the C. savigny genomeproject to examine the evolution of the caspase fam-

ily in the Ciona genus. C. intestinalis and C. savignyiare considered to be sister species but display substan-tial sequence divergence. BLAST searches for homo-logues to all C. intestinalis caspase members in theC. savignyi genome (Ensembl database) identified 62distinct DNA scaffolds encoding p20/p10 domains.A total of 39 hits were already annotated in the data-base as coding for proteins with p20/p10 domains.This considerable amount of caspase-like sequencesmight be explained in part by heterozygosity. Thesequenced C. savignyi genome is indeed known todisplay a high level of polymorphism, resulting innumerous occurrences of independent assemblies ofthe two alleles. We thus reiterated the searches inthe Broad Institute C. savignyi genome assembly,which contains already paired (i.e. allelic) contigs

860 C© Portland Press 2005 | www.biolcell.org

Evolution of the caspase family in urochordates Research article

totalling 160 Mb of an estimated 180 Mb genomesize. A total of 22 caspase genes (corresponding to 44Ensembl hits) were found in independent paired con-tigs (Table 3). The 18 remaining caspase domains,found on unpaired contigs, probably result from un-assembled sequences due to poor read quality or fromgene fragments embedded in repetitive sequencesparticularly abundant in unpaired contigs. For clar-ity, we therefore restrict the C. savignyi caspase genefamily to the 22 loci found in paired contigs. Com-parison with C. intestinalis p20/p10 domains usingCLUSTAL X alignment and neighbour-joining treeclustering delineated only seven groups in C. savignyi(1a, 1b, 2, 3, 5, 6 and 8/9; Figure 2). The conserva-tion in both Cionidae is blatant for the CSP2, CSP6and CSP8/9 subgroups, which contain one to threemembers each. A CSP5 homologue was also foundin C. savignyi, having accumulated substitutions atcritical residues like in C. intestinalis, which agreeswell with a probable pseudogene status. No CSP4homologue was found in C. savignyi, whereas fourmembers probably recently duplicated were foundin C. intestinalis. CSP4 was only distantly related tothe human caspase 4 and its absence from C. savignyisuggests that it might have only a side function inC. intestinalis or be complemented by another mem-ber in C. savignyi. For CSP1 and CSP3/7, only a fewmembers are conserved in both species. Mass duplic-ations in CSP1 and CSP3/7 families account for mostof the additional C. savignyi caspase genes. Whetherthe duplicated genes are functional remains to beworked out. Those which have spread since the re-cent speciation of C. intestinalis and C. savignyi (e.g.CsCSP1b) have already accumulated a large numberof substitutions in their p20/p10 domains (results notshown), which suggests that they are evolving withlittle functional constraints.

In summary, the caspase gene families of both Cionaspecies appear to be highly dynamic in number andin amino acid sequences. Amid all this diversity, nineputative proteins (CSP1a, CSP2a, CSP2b, CSP3a,CSP3b, CSP3c, CSP5, CSP6 and CSP8/9) pertainedsequence conservation in both species.

Evolution of the caspase family in tunicatesTo extend the evolution view of the caspase families intunicates, we searched for p20/p10-like sequencesin the genome of O. dioica (Od). This larvacean tuni-cate displays a lifestyle and morphological traits very

distinct from those found in Cionidae: Oikopleura isa free-living planktonic animal that retains a larvamorphology and tiny size (0.2–0.5 mm) all itslifetime. Interestingly, BLAST searches for caspasesonly produced three hits, termed CSP1a, CSP1b andCSP1c, probably deriving all from duplications of asingle gene after the split with ascidians (Figure 3).The p20/p10 domains of CSP1a and CSP1b show65% similarity, while OdCSP1c is more divergent(50% similarity to either CSP1a or CSP1b). Sequencealignments show that most of the residues critical forthe activity are conserved, except for the replacementwith Asn of the Gln residue (yellow; Figure 3) in-volved in the co-ordination of the P1 aspartate of thesubstrate in CSP1a. This position was also substitutedin CiCSP1a (see Figure 1A).

O. dioica caspases delineate a specific branch, whoserelative position with respect to human and Ciona cas-pases cannot be confidently determined (Figure 3B).At the amino acid level, however, O. dioica caspasesappear to be closely related to the caspase 3 fam-ily [for example, OdCSP 1a shares 53 and 51% simi-larity with human caspase 3 and C. intestinalis CSP3a–CSP3e respectively but only 32–48% (45% average)with any other caspase]. In addition, no exon encod-ing putative CARD or DED could be found upstreamof the p20/p10 domains of either O. dioica caspasegene. This suggests that O. dioica caspases all are bothinitiator and effector caspases, as it is the case inCaeaenorhabditis elegans.

DiscussionIn the present study, we have reported the identific-ation of the caspase families in three tunicates andhave found two extreme situations in ascidiansand larvaceans: C. intestinalis and C. savignyi displaya high number of members probably deriving fromfive subfamilies, whereas O. dioica shows only threemembers of a single subfamily. The higher numberof caspase genes in C. savignyi versus C. intestinalis re-sults from local mass amplifications that occurredafter the split. The caspase complexity in tunicatesthus appears to be either too low to challenge othersimpler models like Caenorhabditis elegans or too highto represent a scale model for vertebrates. Never-theless, such a diversity of caspase families in a singlesubphylum represents an unprecedented situationfrom which scenarios can be proposed.

www.biolcell.org | Volume 97 (11) | Pages 857–866 861

M.W

eillando

thers

Table 3 Members of the caspase family in C. savignyiAnnotated Ensembl sequences are identified by PIR/DDBJ accession numbers. Unannotated (‘–’) genes were detected from BLAST searches. The sizes of partial putative proteins aswell as the presence of CARD, DED, p20 and p10 structural regions is indicated. Alleles were identified from BLAST analysis of paired contigs (Broad Institute). For each allele, the contignumber and the position of the p20 domain border are indicated. Paired contigs containing two caspase genes are in boldface.

Broad Institute assemblyEnsembl assembly

Allele 1 Allele 2

Name Size (amino acids) Domains Contig Start Contig size PIR/DDBJ Contig Start PIR/DDBJ Contig Start

CsCSP1a1 >154 p20/p10 PSA50 112199 151881 AAN45849 14279 134372 BAA94748 20684 69

CsCSP1a2 >151 p20/p10 PSA377 1151570 3244373 XP 320581 31453 1173524 XP 320581 33580 1840808

CsCSP1b1a >121 p20/p10 PSA278 256937 731525 – 14670 400373 – 31165 389402

CsCSP1b1b >121 p20/p10 PSA51 835773 1224790 – 31165 426460

CsCSP1b1c >121 p20/p10 PSA130 17325 449779 – 31165 504225

CsCSP1b2a >121 p20/p10 PSA387 24560 316619 – 32698 14410 – 14769 17325

CsCSP1b2b >121 p20/p10 PSA387 48628 316619 – 31791 24560

CsCSP1b3 >121 p20/p10 PSA278 270498 731525 – 20614 4472 – 30994 582245

CsCSP2a >433 CARD/p20/p10 PSA12 451511 2025132 NP 071967 33490 448441 NP 071967 27854 74388

CsCSP2b >410 CARD/p20/p10 PSA95 634988 1120919 BAC31153 20851 800960 BAC31153 31403 431445

CsCSP3a >129 p20/p10 PSA124 157867 1298227 NP 524551 31682 654604 - 14695 1538960

CsCSP3b1 >156 p20/p10 PSA2 860361 1291007 AAC32602 31201 418238 AAN45849 14685 202288

CsCSP3b2 >288 p20/p10 PSA116 129224 238176 AAC53068 20561 117671 AAC53068 15674 24150

CsCSP3b3 >200 p20/p10 PSA348 392400 980181 BAA94748 31952 605879 P55214 30590 1799525

CsCSP3c >121 p20/p10 PSA335 443591 542329 AAN45849 31629 109185 AAN45849 31252 484810

CsCSP3d >127 p20/p10 PSA63 640410 1203905 AAN45849 14329 333244 – 14695 651929

CsCSP3e >150 p20/p10 PSA6 227364 969958 NP 524551 32177 194155

CsCSP3f1 >103 p20/p10 PSA347 997800 1103586 AAN45850 32010 161769 – 30590 1239806

CsCSP3f2 >389 p20/p10 PSA63 1183175 1203905 1QX3 30696 906876 – 14695 1234348

CsCSP5 >121 p20/p10 PSA409 348093 794312 NP 071613 32660 421773 – 31598 339423

CsCSP6 >126 p20/p10 PSA92 59606 64494 BAA94747 14510 59840

CsCSP8/9 >462 (DED)2/p20/p10 PSA282 199820 1066918 NP 071613 20786 153890 NP 071613 31218 206682

862C ©

Portland

Press

2005|

ww

w.biolcell.org

Evolution of the caspase family in urochordates Research article

Figure 2 Similarity between the caspase repertoires ofC. intestinalis and C. savignyiThe bootstrapped neighbour-joining tree was inferred from

alignment of all C. intestinalis (C) and C. savignyi (Cs; bold-

face in gold background) amino acid sequences. �, nodes

with bootstrap values above 80%; �, nodes with 50–80% val-

ues. Shaded are nodes of bootstrap values lower than 15%.

For clarity, non-significant nodes were removed from the tree.

The presence of CARD or DED is indicated. Gene expansions

in C. savignyi and C. intestinalis are indicated. Note that the

CiCSP4 subgroup has no orthologue in C. savignyi.

The difference in caspase number and complexitybetween O. dioica and Ciona positively correlates withthe number of apoptotic inhibitors IAP (inhibitor ofapoptosis protein): whereas 15 IAPs have been de-tected so far in the genome of C. intestinalis (Terajimaet al., 2003), we detected only three putative IAPsin Oikopleura (results not shown). This likely reflectsqualitative or quantitative differences in the occur-rence of apoptotic events in the biology of theseorganisms. One striking difference is metamorphosis,which mostly involves a change in the tail position inthe case of Oikopleura, whereas, in Ciona, it involvestail regression followed by a complete rearrangement

of organs (Jeffery and Swalla, 1997). In C. intestinalis,programmed cell death affects trunk lateral cells andmostly the central nervous system, which displayedan anterior–posterior wave of apoptosis from theocellus up to the neural tube extremity, followed bya caspase-dependent tail regression (Chambon et al.,2002; Tarallo and Sordino, 2004). At later stages, in-tricate patterns of cell-death and proliferation affectthe stomach, endostyle, gill slits and siphon edgesin juvenile animals (Tarallo and Sordino, 2004). Thelarger ascidian caspase repertoire might thus be in-volved in such developmental features, which prob-ably require time- and tissue-specific caspase activ-ations.

Besides, the preferential expression of several com-ponents of the apoptotic pathways in haemocytessuggests that these cells might undergo apoptosislike their vertebrate homologues (Shida et al., 2003).Caspases might also participate in the induction ofantimicrobial peptide genes in haemocytes. In Droso-phila, this pathway is indeed controlled by the tran-scription factor Relish [NF-κB (nuclear factor κB)],whose activation results from a cleavage mediated bythe caspase DREDD, structurally similar to humancaspase 8 (Rathore et al., 2004). Ciona homologuesto caspase 8 (CiCSP8/9) and NF-κB (Yagi et al.,2003) could thus participate in the secretion of anti-microbial peptides, such as clavanins or styelinsproduced by several tunicates (Azumi et al., 1990).Although no similar peptide could be found encodedin the genome of Oikopleura, further investigationsare needed to determine unambiguously whether an-timicrobial peptide production represents a differ-ential feature between ascidians and larvaceans.

Caspase activation has also been reported to con-trol cell proliferation and differentiation in mammals,independent of apoptosis (Schwerk and Schulze-Osthoff, 2003). One spectacular case is caspase 3,whose knockout in mice produced skeletal muscledefects due to a lack of myotube and myofibre forma-tion (Fernando et al., 2002). The much larger cas-pase repertoire in Ciona might thus well participatein the differentiation of adult tissues. Along the sameline, programmed cell death could be also requiredfor adult tissue homoeostasis since another major dif-ference between ascidians and larvaceans lies in theircell number at the adult stage, fixed in almost all tis-sues in Oikopleura, whereas ascidians are subjected tocontinuous growth.

www.biolcell.org | Volume 97 (11) | Pages 857–866 863

M. Weill and others

Figure 3 The caspase families of ascidians and larvaceans(A) O. dioica p20/p10 domains were aligned using CLUSTAL X. Included in the alignment are the p20/p10 domains of the

Caenorhabditis elegans CED3 (CeCED3; where CED stands for Caenorhabditis elegans death) as well as the consensus CASc

domain [CDD database (Marchler-Bauer et al., 2003)]. Residues shared by at least 70% of the sequences are shaded. Residues

involved in the catalytic activity (H-C dyad) are indicated by ‘@’ and those lining the substrate pocket by ‘*’. Numbers delineate

the position of the p20/p10 domains. (B) Sequence proximity between O. dioica (Od, blue background), C. intestinalis (Ci, green

background), C. savignyi (Cs, gold background), Caenorhabditis elegans Ced-3 (cyan background) and human (Hs) caspases.

p20/p10 domains were aligned using the CLUSTAL X algorithm and a bootstrapped neighbour-joining unrooted tree was inferred.

For clarity, only single members of C. intestinalis and C. savignyi multigene families are shown. �, nodes with bootstrap values

above 80%; and �, nodes with 50–80% values. Grey shading indicates nodes of bootstrap values lower than 15%. The presence

of CARD or DED is indicated.

The identification of the conserved caspase mem-bers in Ciona bridges the gap between vertebratesand Ecdysozoa (arthropods and nematodes). A scen-ario now can be proposed (Figure 4), in which twoof the three ancestral caspase genes (Lamkanfi et al.,2002) have duplicated before the split urochordates/chordates. This produced five caspase genes, encodinga single ICE-type caspase, two initiator and two ef-fector caspases. After divergence, further duplicationstook place in the taxon leading to vertebrates. In Uro-chordates, local amplifications of genes for effectorcaspases occurred in Ciona, whereas Oikopleura under-went a major loss, followed by amplification of theunique remaining gene. The situation of Oikopleurain Urochordates parallels that of Caenorhabditis inEcdysozoa, in which a similar loss occurred. In bothspecies, the reduced complexity of the caspase familyis associated with a low cell number at the adult stageand absence of major metamorphosis.

Urochordates display an amazing diversity ofdevelopmental patterns and life cycles, in which ex-treme situations can be examined. An interesting caseis Botryllida, in which adults in a colony die synchron-ously and in a cyclical fashion by programmed celldeath (Lauzon et al., 1993). A contrasted situation isprovided by salps and pyrosomids, which develop ac-cording to a body plan comparable with that of adultascidians without passing through a tadpole stage.Combined with the present analysis, the determina-tion and comparison of the caspase repertoires of otherurochordate species might be highly informative onthe evolution of caspase-dependent physiological pro-cesses.

Materials and methodsAnimal husbandryAdult C. intestinalis were collected from the Thau laguna(Sete, Herault, France). Oocytes and sperm were obtained by

864 C© Portland Press 2005 | www.biolcell.org

Evolution of the caspase family in urochordates Research article

Figure 4 Evolution of the caspase families fromEcdysozoa to vertebratesAbove each split, putative ancestor genes are identified

using the human caspase nomenclature. Gene duplications

or losses are indicated by positive and negative numbers

respectively. Below each taxon are indicated the number of

founder genes (i.e. complexity) (number shown in circles) and

the number of genes in the caspase families.

dissection of gonoducts and cross-fertilization was performedin plastic Petri dishes. Embryos were cultured at 18◦C in0.2 µm filtered seawater containing 100 units/ml penicillin and100 µg/ml streptomycin.

In silico identificationSearches in C. intestinalis and C. savignyi genomes were originallydone using TBLASTN and BLAST (Altschul et al., 1990) on rawshotgun sequences. Downloaded genomic sequences were as-sembled using ABI Prism Auto-Assembler (v 2.1; PerkinElmer,Courtaboeuf, France). This led to the identification of 15 partialcaspase sequences in C. intestinalis (Chambon et al., 2002).Identification of the caspase families was since then completedby searches in partial genome assemblies (C. intestinalis: http://genome.jgi-psf.org/ciona4/ciona4.home.html; C. savignyi:http://www2.bioinformatics.tll.org.sg/Ciona savignyi/) and inESTs from the C. intestinalis genome project website (http://ghost.zool.kyoto-u.ac.jp/indexr1.html). Allelic-paired loci weredirectly analysed using the Broad Institute web resources(http://www.broad.mit.edu/annotation/ciona/). The O. dioicagenome was sequenced at a high level of coverage (14 times)using the shotgun method. Alignment of 624 EST sequenceswith the shotgun data indeed shows a representation of99.7% of all coding nucleotides (R.B. Edvardsen, M.F. Jensen,A. Hansen, J. Weissenbach, P. Wincker, H. Lehrach, E.M.Thomson, R. Reinhardt and D. Chourrout, unpublished work).Searches in O. dioica databases were performed on raw sequencedata sets using a local standalone BLAST package. TheOikopleura trace archive database is available for BLAST searchesat the NCBI website (http://www.ncbi.nlm.nih.gov/BLAST/mmtrace.shtml).

Cloning proceduresmRNA was isolated from C. intestinalis larvae lysed in TRIzol®

(Invitrogen) according to the manufacturer’s instructions. Afterreverse transcription, PCR was performed on cDNA by usingprimers specific for each caspase. Amplified products were sub-cloned in pCR4-TOPO (Invitrogen) and sequenced. All con-structs and clone sequences are available upon request.

Sequence analysisSequences were aligned using CLUSTAL X (Thompson et al.,1997) with BLOSUM alignment matrix. Alignments were im-proved by adjusting the parameters and refined by manual edit-ing. Trees were derived from alignments using neighbour joining(Saitou and Nei, 1987) or maximum parsimony (PHYLIP 3.6;Felsenstein, 1996) and displayed using TreeView (Page, 1996).

AcknowledgementsWe thank A. Berthomieu (ISEM, CNRS-UMR5554,Montpellier, France), for technical help and A.Blangy, S. Ory and S. Faure (CRBM, CNRS-FRE2593, Montpellier, France) for constant supportand a critical reading of this paper. We are gratefulto the Station de Biologie (Sete, France) for ascid-ian supply. We thank the Genoscope staff (CentreNational de Sequencage, Evry, France), in particularJ. Weissenbach and P. Wincker. We are also indebtedto the C. intestinalis and C. savignyi ascidian genomeconsortium and to N. Satoh for ascidian databases.This work was supported by institutional grants[CNRS (Centre National de la Recherche Scienti-fique)] and from the Association pour la Recherchecontre le Cancer. Contribution 2005–012 of the Insti-tut des Sciences de l’Evolution de Montpellier (UMRCNRS 5554).

ReferencesAbraham, M.C. and Shaham, S. (2004) Death without caspases,

caspases without death. Trends Cell Biol. 14, 184–193Altschul, S.F., Gish, W., Miller, W., Myers, E.W. and Lipman, D.J.

(1990) Basic local alignment search tool. J. Mol. Biol. 215, 403–410Azumi, K., Yokosawa, H. and Ishii, S. (1990) Halocyamines: novel

antimicrobial tetrapeptide-like substances isolated from thehemocytes of the solitary ascidian Halocynthia roretzi.Biochemistry 29, 159–165

Chambon, J.P., Soule, J., Pomies, P., Fort, P., Sahuquet, A.,Alexandre, D., Mangeat, P.H. and Baghdiguian, S. (2002) Tailregression in Ciona intestinalis (Prochordate) involves aCaspase-dependent apoptosis event associated with ERKactivation. Development 129, 3105–3114

Cikala, M., Wilm, B., Hobmayer, E., Bottger, A. and David, C.N. (1999)Identification of caspases and apoptosis in the simple metazoanHydra. Curr. Biol. 9, 959–962

Dehal, P., Satou, Y., Campbell, R.K., Chapman, J., Degnan, B.,De Tomaso, A., Davidson, B., Di Gregorio, A., Gelpke, M.,Goodstein, D.M. et al. (2002) The draft genome of Cionaintestinalis: insights into chordate and vertebrate origins. Science298, 2157–2167

www.biolcell.org | Volume 97 (11) | Pages 857–866 865

M. Weill and others

Felsenstein, J. (1996) Inferring phylogenies from protein sequencesby parsimony, distance, and likelihood methods.Methods Enzymol. 266, 418–427

Fernando, P., Kelly, J.F., Balazsi, K., Slack, R.S. and Megeney, L.A.(2002) Caspase 3 activity is required for skeletal muscledifferentiation. Proc. Natl. Acad. Sci. U.S.A. 99, 11025–11030

Glucksmann, A. (1951) Cell deaths in normal vertebrate ontogeny.Biol. Rev. 26, 58–86

Jeffery, W.R. and Swalla, B.J. (1997) In Embryology of the tunicates.Embryology: Constructing the Organisms (Gilbert, S., ed.),pp. 331–364, Sinauer, Sunderland, U.K.

Lamkanfi, M., Declercq, W., Kalai, M., Saelens, X. andVandenabeele, P. (2002) Alice in caspase land. A phylogeneticanalysis of caspases from worm to man. Cell Death Differ. 9,358–361

Lauzon, R.J., Patton, C.W. and Weissman, I.L. (1993)A morphological and immunohistochemical study of programmedcell death in Botryllus schlosseri (Tunicata, Ascidiacea).Cell Tissue Res. 272, 115–127

Marchler-Bauer, A., Anderson, J.B., DeWeese-Scott, C., Fedorova,N.D., Geer, L.Y., He, S., Hurwitz, D.I., Jackson, J.D., Jacobs, A.R.,Lanczycki, C.J. et al. (2003) CDD: a curated Entrez databaseof conserved domain alignments. Nucleic Acids Res. 31,383–387

Page, R.D. (1996) TreeView: an application to display phylogenetictrees on personal computers. Comput. Appl. Biosci. 12,357–358

Rathore, N., Matta, H. and Chaudhary, P.M. (2004) Anevolutionary conserved pathway of nuclear factor-kappaBactivation involving caspase-mediated cleavage and N-end rulepathway-mediated degradation of IkappaBalpha. J. Biol. Chem.279, 39358–39365

Saitou, N. and Nei, M. (1987) The neighbor-joining method: a newmethod for reconstructing phylogenetic trees. Mol. Biol. Evol. 4,406–425

Salvesen, G.S. and Dixit, V.M. (1997) Caspases: intracellularsignaling by proteolysis. Cell (Cambridge, Mass.) 91, 443–446

Schwerk, C. and Schulze-Osthoff, K. (2003) Non-apoptotic functionsof caspases in cellular proliferation and differentiation.Biochem. Pharmacol. 66, 1453–1458

Shida, K., Terajima, D., Uchino, R., Ikawa, S., Ikeda, M., Asano, K.,Watanabe, T., Azumi, K., Nonaka, M., Satou, Y. et al. (2003)Hemocytes of Ciona intestinalis express multiple genes involved ininnate immune host defense. Biochem. Biophys. Res. Commun.302, 207–218

Strasser, A., O’Connor, L. and Dixit, V.M. (2000) Apoptosis signaling.Annu. Rev. Biochem. 69, 217–245

Tarallo, R. and Sordino, P. (2004) Time course of programmed celldeath in Ciona intestinalis in relation to mitotic activity and MAPKsignaling. Dev. Dyn. 230, 251–262

Tata, J.R. (1996) Metamorphosis: an exquisite model for hormonalregulation of post-embryonic development. Biochem. Soc. Symp.62, 123–136

Terajima, D., Shida, K., Takada, N., Kasuya, A., Rokhsar, D.,Satoh, N., Satake, M. and Wang, H.G. (2003) Identification ofcandidate genes encoding the core components of the cell deathmachinery in the Ciona intestinalis genome. Cell Death Differ. 10,749–753

Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F. andHiggins, D.G. (1997) The CLUSTAL X windows interface: flexiblestrategies for multiple sequence alignment aided by qualityanalysis tools. Nucleic Acids Res. 25, 4876–4882

Uren, A.G., O’Rourke, K., Aravind, L.A., Pisabarro, M.T.,Seshagiri, S., Koonin, E.V. and Dixit, V.M. (2000) Identification ofparacaspases and metacaspases: two ancient families ofcaspase-like proteins, one of which plays a key role in MALTlymphoma. Mol. Cell 6, 961–967

Wiens, M., Krasko, A., Perovic, S. and Muller, W.E. (2003)Caspase-mediated apoptosis in sponges: cloning and function ofthe phylogenetic oldest apoptotic proteases from Metazoa.Biochim. Biophys. Acta 1593, 179–189

Yagi, K., Satou, Y., Mazet, F., Shimeld, S.M., Degnan, B., Rokhsar, D.,Levine, M., Kohara, Y. and Satoh, N. (2003) A genomewide surveyof developmentally relevant genes in Ciona intestinalis. III. Genesfor Fox, ETS, nuclear receptors and NFkappaB. Dev. Genes Evol.213, 235–244

Yuan, J., Shaham, S., Ledoux, S., Ellis, H.M. and Horvitz, H.R. (1993)The C. elegans cell death gene ced-3 encodes a protein similar tomammalian interleukin-1 beta-converting enzyme.Cell (Cambridge, Mass.) 75, 641–652

Received 8 February 2005/24 March 2005; accepted 7 April 2005

Published as Immediate Publication 7 April 2005, doi:10.1042/BC20050018

866 C© Portland Press 2005 | www.biolcell.org