ggacSaosratsdepscHdtctcb
bceppbtaiosc
Molecular Phylogenetics and EvolutionVol. 14, No. 3, March, pp. 375–388, 2000doi:10.1006/mpev.1999.0707, available online at http://www.idealibrary.com on
PCR Survey of Hox Genes in the Crinoid and Ophiuroid: Evidencefor Anterior Conservation and Posterior Expansion in the
Echinoderm Hox Gene ClusterTaro Mito and Kazuyoshi Endo
Geological Institute, University of Tokyo, Tokyo, 113-0033, Japan
Received March 30, 1999; revised July 1, 1999
bm
accaEsdgooHhc1aodieec
cdmhericmthpijs
To help elucidate the cluster organization of Hoxenes in echinoderms, we amplified a homeobox re-ion by polymerase chain reaction (PCR) and clonednd sequenced the PCR products for the comatulidrinoid Oxycomanthus japonicus and the ophiuroidtegophiura sladeni. The crinoid had at least threenterior, four medial, and four posterior genes, and thephiuroid had at least one anterior, three medial, andix (one of which being a possible trans-paralog) poste-ior genes. The survey of the crinoid detected all threenterior complements (PG1–3 genes). It was inferredhat the Hox genes of each species are organized into aingle cluster and that a novel cognate group of echino-erm posterior genes (tentatively termed HboxP9)xists among echinoderms in addition to the knownosterior genes Hbox4, Hbox7, and Hbox10. The re-ults, combined with the data of other echinodermlasses, strongly suggest that the presence of a singleox gene cluster is a common feature among echino-erms and that the cluster has the general features ofhe deuterostome Hox gene cluster, i.e., the anterioronservation and posterior expansion. The results ofhe ophiuroid imply that the posterior genes in thislass diversified after the phylum Echinodermata hadeen established. r 2000 Academic Press
INTRODUCTION
Living echinoderms generally share a distinct adultody plan of pentaradial symmetry with no apparentephalic structures. The features and functions ofchinoderm Hox genes have been of interest as aossible clue to understanding how such a unique bodylan was formed in the course of echinoderm evolution,ecause Hox genes have been shown to be involved inhe pattern formation along the anteroposterior (AP)xis of various animals. The Hox genes are organizednto a cluster in a chromosome and their arrangementn the cluster is related collinearly with their expres-ion patterns along the AP axis. Such a feature is
onserved among metazoans, including insects, verte- s375
rates, and nematodes; this is thought to have beenaintained from the common ancestor of metazoans.Echinoids have a single Hox gene cluster (Popodi et
l., 1996), and the gene organization has recently beenharacterized through physical mapping in Strongylo-entrotus purpuratus (Martinez et al., 1997, 1999). Thesteroid also has a single Hox gene cluster (Mito andndo, 1997). These results strongly suggest that posses-ion of a single cluster is a common feature of echino-erms. It might be, however, too early to make thiseneralization, because information is lacking on thether major groups of echinoderms, i.e., holothuroids,phiuroids, and crinoids. Concerning the echinodermox gene cluster, complements of its anterior regionave been of special interest. This is because genesorresponding to the vertebrate paralogous group (PG)and 2 were not found in earlier studies, and absence ofnterior genes tended to be linked with the uniquenessf echinoderm morphology (Ruddle et al., 1994). Weemonstrated that a PG1-type gene at least is presentn an asteroid (Mito and Endo, 1997). Recently, pres-nce of a PG1 and a PG2 has been reported in anchinoid (Martinez et al., 1999). However, the extent ofonservation of the anterior region is still unclear.Both morphological and molecular data show that
rinoids diverged earliest from other extant echino-erms (Littlewood, et al., 1997), retaining primitiveorphological features since the early Paleozoic and
aving remarkable differences from other groups ofchinoderms, such as possession of a stalk. Genesegulating crinoid morphogenesis should be importantn elucidating the evolution of echinoderms because therinoids might well have retained the primitive ancientode of echinoderm development. Therefore, in view of
he role of Hox genes in patterning development, weave begun a study of these genes in crinoids. Here, weresent the results of a PCR-based survey of Hox genesn the crinoid species (feather star), Oxycomanthusaponicus. Additional preliminary results for the brittletar, Stegophiura sladeni, are also described. The pre-
ent survey detected crinoid Hox genes including all1055-7903/00 $35.00Copyright r 2000 by Academic PressAll rights of reproduction in any form reserved.
tPGcast
A
(SM(PchbEaacsw(a
b8Bwv1Kstpp
P
swusAHTwapt(
p
tttwTm5aip39iofr37
C
ePmlgadtnrqsct
S
qqFLp(
twadn(Pawar
376 MITO AND ENDO
he complements of the anterior group (PG1, PG2, andG3) as well as some ParaHox genes (Xlox, Cdx, andsh class genes), which are known to constitute a sister
luster of the Hox gene cluster in amphioxus (Brooke etl., 1998). Based on these results, the features andtructure of echinoderm Hox and ParaHox gene clus-ers are discussed.
MATERIALS AND METHODS
nimals and DNA Isolation
The feather star, Oxycomanthus japonicus (Muller)5 Comanthus japonica Muller) was obtained byCUBA from a subtidal rocky substratum near theisaki Marine Biological Station, University of Tokyo
MMBS) in the mouth of Koajiro cove, Sagami Bay.innules of a single male containing sperm were ex-ised and homogenized in 0.54 M KCl solution. Theomogenate was incubated at 55°C for 3 h in 10 vol of auffer containing 100 mM Tris–HCl, pH 7.4, 10 mMDTA, 10 mM NaCl, 0.5% SDS, and 100 µg/ml protein-se K. An additional 100 µg/ml of proteinase K wasdded to the solution and further incubation wasarried out overnight at room temperature with gentlehaking. The resulting solution was extracted twiceith phenol and once with chloroform/isoamyl alcohol
24:1, v/v) and precipitated with 0.1 vol of 3 M sodiumcetate and 2 vol of ethanol.The brittle star, Stegophiura sladeni, was collected
y dredging from a sandy shelf bottom, approximately0 m deep, off the western coast of Jogashima, Sagamiay. A single small individual (disk diameter 5 5 mm)as cut into pieces and homogenized in an appropriateolume of STE (10 mM Tris–HCl, pH 7.4, 10 mM EDTA,0 mM NaCl). Then, 0.1 vol of 20% SDS and proteinase
(final concentration: 70 ng/µl) was added and theolution was incubated with gentle shaking at roomemperature overnight. The resulting solution wasurified as described for the feather star except threehenol extractions rather than two were employed.
CR Amplification
A pair of degenerate primers matching highly con-erved regions in the homeoboxes of Hox genes, eachith a restriction site for BamHI or PstI added, wassed to amplify a Hox homeobox region. The nucleotideequences of the primers are HoxME1, 58 AAA AGGTC CGA RCT NGA RAA RGA RTT 38 (27mer); andoxME2, 58 AAA ACT GCA GYT TCA TNC GNC GRTYT GRA ACC ADA TYT T 38 (40mer). This primer setas designed and used in our previous survey on thesteroid (Mito and Endo, 1997) and it generates PCRroducts of 143 bp in length including 76 bp of informa-ive sequence spanning homeobox positions 60–135amino acid positions 21–45).
Prior to PCRs each solution containing 200 ng tem-
late DNA was treated with GeneReleaser (Bioven- Tures) following the manufacture’s instructions in ordero suppress possible polymerase inhibitors contained inhe DNA solution. Then, the mixture of PCR reagentsas added to the GeneReleaser-treated DNA solution.he final reaction contained 200 µM each of dNTP, 2.5M MgCl2, 50 mM KCl, 10 mM Tris–HCl, pH 8.3, and
0 pmol each of the primers. PCR was performed usingPerkin–Elmer DNA thermal cycler under the follow-
ng conditions: 94°C for 3 min and then 2.5 U rTaq DNAolymerase (TOYOBO) was added; 5 cycles of 94°C for0 s, 50°C for 1 min, and 72°C for 1 min; 30 cycles of4°C for 30 s, 55°C for 1 min, and 72°C for 1 min;ncubation at 72°C for 5 min. A second PCR was carriedut using 1% of the primary PCR products under theollowing conditions: 94°C for 3 min and then 2.5 UTaq DNA polymerase was added; 30 cycles of 94°C for0 s, 55°C for 1 min, and 72°C for 1 min; incubation at2°C for 5 min.
loning and Sequencing of PCR Products
The products of the second PCR were purified bythanol precipitation and digested with BamHI andstI (Boehringer Mannheim). The purified DNA frag-ents were ligated into the pUC19 vector using a DNA
igation kit (Takara). In the survey of crinoid Hoxenes, inserts were derived from four independentlymplified PCR products to reduce the effect of PCRrift. The recombinant plasmids were then used toransform competent Escherichia coli NM522. Recombi-ant clones were selected by blue/white screening andestriction enzyme digestion, and the nucleotide se-uences were determined using an ABI prism 377 DNAequencer (Applied Biosystems) with Dye terminatorycle sequencing FS ready reaction kit (Applied Biosys-ems).
equence Analyses
For initial estimation of similarities to known se-uences, the informative 76 bp of the obtained se-uences were searched to DDBJ on the Web site usingASTA 3.0 (Lipman and Pearson, 1985; Pearson andipman, 1988) for nucleotide sequences (using fastarogram) and for the deduced amino acid sequencesusing tfasta program).
For the fragments showing definite relationships tohe Hox class genes, further analyses of relationshipsith typical Hox sequences of mouse and Drosophiland with previously reported Hox sequences of echino-erms were done by maximum likelihood (ML) andeighbor-joining (NJ) methods using MOLPHY 2.3
Adachi and Hasegawa, 1996). The MOLPHY programsrotML and NJdist were used for the ML and NJnalyses using the Dayhoff model (Dayhoff et al., 1978)ith the ‘‘2F’’ (with data frequencies) option. MLnalyses were carried out with the ‘‘2R’’ (local rear-angements) option using the NJ tree as the initial tree.
he local bootstrap probability (LBP) (Kishino et al.,
1u
HHgti
apD((HHHHHHHHHHHHS(h(Ah((HT(aA((((
hfifiTtClvco
c(wes
tftcnXaCC
Hdsr
asH(at
ms3sr
E
pgcscTacltheowSwmTa
377CRINOID AND OPHIUROID Hox GENES
990; Hasegawa and Kishino, 1994) was also calculatedsing the same programs.From the results of these sequence analyses, eachox-type fragment was assigned to a specific class ofox genes or to a group of the known echinoderm Hox
enes. In some cases, we also utilized information onhe characteristic positions of amino acid replacementsn the homeodomain.
The following sequences were used in the presentnalysis (database accession numbers or references inarentheses): Drosophila, lab (X13103), pb (X63728),fd (X05136), Scr (X14475), Antp (M20704), Ubx
X76210), abd-A (X54453), abd-B (X51663), enM29285), NK-1 (M27289); mouse, Hoxa-1 (X06023),oxa-2 (M93148), Hoxa-3 (K02591), Hoxa-4 (X13538),oxa-5 (M36604), Hoxa-6 (M11988), Hoxa-7 (M17192),oxa-9 (M28449), Hoxa-10 (L08757), Hoxa-11 (U20370),oxb-1 (X59474), Hoxb-2 (M34004), Hoxb-3 (X66177),oxb-4 (M36654), Hoxb-5 (M26283), Hoxb-6 (X56461),oxb-7 (X06762), Hoxb-8 (X13721), Hoxb-9 (M34857),oxc-4 (S62287), Hoxc-5 (U28071), Hoxc-6 (X16511),oxc-8 (X07646), Hoxc-9 (X55318), Hoxc-10 (D11330),oxc-12 (U04839), Hoxc-13 (U04838), Hoxd-1 (X60034),oxd-3 (X73573), Hoxd-4 (J03770), Hoxd-8 (X56561),oxd-9 (X62669), Hoxd-10 (X62669), Hoxd-11 (X60762),oxd-12 (X58849), Hoxd-13 (S80533), EN1 (L12703),ax-1 (X75384), Sax-2 (U94419), Gsh-1 (U21224), Gsh-2
S79041); Amphioxus, Amphihox-1 (Z35142), Amphi-ox-2 (Z35143), Amphihox-3 (X68045), Amphihox-4
Z35144), Amphihox-5 (Z35145), Amphihox-6 (Z35146),mphihox-7 (Z35147), Amphihox-8 (Z35148), Amphi-ox-9 (Z35149), Amphihox-10 (Z35150), Brahoxy57L14869); echinoids, HeHbox1 (U30935), HeHbox6U31447), HeHbox7 (U31564), HeHbox9 (U31563), He-box10 (U31600), Heen (U58775), TgHbox3 (X13146),gHbox4 (X13147), HpHbox11 (U83423), HpHbox13
U83424), HpHbox14 (U83425), SpHox1 (Martinez etl., 1999), SpHox2 (Martinez et al., 1999); asteroid,M-1 (Mito and Endo, 1997), AM-3 (D86362), AM-4
D86363), AM-5 (D86364), AM-6 (D86365), AM-7D86366), AM-9 (D86367), AM-Xlox (D86368), AM-GbxD86369); Xenopus, XlHbox8 (X16849); human, IPF-1U35632), GBX1 (L11239); chicken, Ovx1 (M76985).
RESULTS
Of the sequenced 107 clones containing a crinoidomeobox fragment, 18 different sequences were identi-ed (Fig. 1). Thirteen different fragments were identi-ed from the sequenced 36 ophiuroid clones (Fig. 2).hese homeobox fragments are named with a prefix ofhe derived species (CJ, Oxycomanthus japonicus 5omanthus japonica; SS, Stegophiura sladeni), fol-
owed either by the name of the inferred orthologousertebrate or echinoid gene or by a laboratory identifi-ation number for CJ-28, SS-11, SS-12, and SS-13, the
rthologies of which have not been clearly established. SVariations by one nucleotide (in 24 clones for therinoid; in 1 clone for the ophiuroid) or two nucleotidesin 3 clones for the crinoid; in 9 clones for the ophiuroid)ere interpreted as either PCR mutations or alleles. Inach of those cases, the nucleotide sequence was repre-ented by the most frequently found fragments.Twelve of the crinoid fragments were clearly related
o the Hox genes by the database search using bothasta and tfasta. Each of the remaining fragments, onhe other hand, indicated a definite relationship with alass of homeobox genes other than Hox. Of thoseon-Hox fragments, two sequences corresponded to thelox class, two to the Gbx class, one to the NK-1 class,nd one to the Gsh class. Therefore, they were namedJ-Xlox1/CJ-Xlox2, CJ-Gbx1/CJ-Gbx2, CJ-NK1, andJ-Gsh, respectively.The 13 ophiuroid fragments contained 11 distinctox-type sequences. The remaining 2 sequences showedefinite similarity to Xlox and Gbx by fasta and tfastaearches, and they are named SS-Xlox and SS-Gbx,espectively.Each pair of SS-Hbox4 and CJ-Hbox4, and SS-Xlox
nd CJ-Xlox1/CJ-Xlox2 has an identical amino acidequence, but pairwise nucleotide comparisons of SS-box4/CJ-Hbox4 (similarity: 75.0%), SS-Xlox/CJ-Xlox1
71.1%), and SS-Xlox/CJ-Xlox2 (72.4%) show that theyre different and did not result from cross-contamina-ion.
The amino acid sequences of all the obtained frag-ents were aligned with mouse, amphioxus, Dro-
ophila, and known echinoderm homeodomains (Fig.), with the typical Hox gene, Antp, as the reference forhowing the characteristic positions of amino acideplacements among related sequences.
stimation of Allelic Variants
The estimation of allelic variants was based onairwise comparisons of the corresponding 76-bp re-ion between paralogous Hox genes of the four-Hoxluster and between two Gsh class genes of mouse. Theequence pairs used were those which are either identi-al or different at one amino acid site in the region.hose were Hoxa-2/b-2, a-3/d-3, b-3/d-3, a-5/b-5,-6/b-6, a-6/c-6, b-6/c-6, b-8/c-8, a-9/c-9, a-9/d-9,-9/d-9, and Gsh1/Gsh2. The range of nucleotide simi-arities in the above pairs are 71.1–90.8%. The nucleo-ide similarity between CJ-Hbox4 and CJ-Hbox4a is asigh as 92.1%, though those fragments code for differ-nt amino acid residues at positions 21 and 24. Becausef their remarkably high nucleotide similarity, theyere interpreted as allelic variants with each other.S-9 and SS-9.2 encode the same amino acid sequencehich shows one amino acid difference in the homeodo-ain position 32 from the sequence encoded by SS-9a.he pairwise comparisons of nucleotide sequencesmong SS-9, SS-9a, and SS-9.2 (SS-9/SS-9a, 94.7%;
S-9/SS-9.2, 81.6%; SS-9a/SS-9.2, 78.9%) suggest that
SpaDswCs
8pa
P
C
f ate
378 MITO AND ENDO
S-9 and SS-9a are allelic variants and that SS-9.2 is aossible trans-paralog (terminology by Cartwright etl., 1993) of those, considering the fact that genomicNA from a single individual was used in the present
urvey. The allelic variants CJ-Hbox-4a and SS-9aere omitted from the phylogenetic analyses. AlthoughJ-Xlox1 and CJ-Xlox2 code for the same amino acid
FIG. 1. Nucleotide sequences (homeobox positions 60–135) and deragments from Oxycomanthus japonicus. Numbers at the right indic
equence, the nucleotide similarity between them is g
1.6%, which is within the range of mouse paralogousairs. We thus could not conclude whether they arelleles or paralogs.
hylogenetic Analysis of Hox Gene Fragments
The NJ and ML trees in Fig. 4 show relationships ofJ and SS fragments to mouse and Drosophila Hox
d amino acid sequences (homeodomain positions 21–45) of homeoboxthe number of inserts sequenced.
rive
enes. Those in Fig. 5 show relationships among CJ,
SHvgpshmbdptcmgas
tkdctnc
Sm(d(fctssPw(ttetobhCP
C
f e n
379CRINOID AND OPHIUROID Hox GENES
S, and known echinoderm (the echinoid and asteroid)ox genes. Irvine et al. (1997) compared efficiency of
arious phylogenetic methods for reconstructing phylo-enetic relationships among Drosophila and mouseartial homeoboxes (positions: 61–143) or amino acidequences in the corresponding region (positions in aomeodomain: 21–47). The results showed that the NJethod using amino acid data is extremely efficient,
eing followed by the ML method using amino acidata. Therefore, we use the amino acid NJ tree as therimary guide for inferring orthology of the fragmentshat we obtained. The NJ trees indicated two largelades, one of which included the anterior and theedial genes while the other included the posterior
enes (Figs. 4 and 5). The details of the relationshipsmong each of the anterior, medial, and posterior geneshown by the resulting trees are as follows.Hox anterior group (PG1–3). The relationships of
he CJ-Hox1, CJ-Hox2, CJ-Hox3, and SS-Hox1 withnown anterior genes were clearly shown by the initialatabase searches with fasta and tfasta and by theharacteristic positions of amino acid substitutions ofhe anterior Hox genes as shown in Fig. 3. The phyloge-etic analyses also support these relationships. The
FIG. 2. Nucleotide sequences (homeobox positions 60–135) and deragments from Stegophiura sladeni. Numbers at the right indicate th
rinoid fragment CJ-Hox1 and the ophiuroid fragment o
S-Hox1 constitute a clade with the PG1 genes ofouse and Drosophila, Hoxb-1 and lab, in the NJ tree
Fig. 4A) with 91% LBP. The two fragments showedefinite orthology to the asteroid (AM-1) and echinoidSpHox1) PG1 sequences in the NJ tree (Fig. 5A),orming a clade supported with 91% LBP. CJ-Hox2lustered with the PG2 genes Hoxb-2 and Pb in the NJree (Fig. 4A). The LBP of the clade including theseequences is low, but the ML tree (Fig. 4B) alsoupports the close relationship between CJ-Hox2 andb. In addition, CJ-Hox2 showed clear relationshipith the echinoid PG2 gene, SpHox2 in the NJ tree
Fig. 5A), though they did not form a cluster in the MLree (Fig. 5B). CJ-Hox3 clustered with Hoxb-3 in the NJree, but the LBP is low, and in the ML tree those do notven form a cluster. However, the orthology of CJ-Hox3o the PG3-type sequences of echinoderms, HpHbox11f the echinoid and AM-3 of the asteroid, is clear fromoth the NJ and the ML trees in Fig. 5, supported withigh LBP. Thus, CJ-Hox1 and SS-Hox1, CJ-Hox2, andJ-Hox3 were firmly assigned to the anterior groupG1, PG2, and PG3, respectively.Hox medial group (PG4–8). Four crinoid fragments,J-Hbox1, CJ-28, CJ-Hbox3, and CJ-Hbox6, and three
d amino acid sequences (homeodomain positions 21–45) of homeoboxumber of inserts sequenced.
rive
phiuroid fragments, SS-Hbox1, SS-Hbox6, and SS-11,
tm
380 MITO AND ENDO
FIG. 3. Alignment of the crinoid and ophiuroid partial homeodomain sequences with known sequences. Deduced amino acid sequences ofhe crinoid (CJ) and ophiuroid (SS) are aligned with those of echinoid (Hbox or SpHox), asteroid (AM), amphioxus (Amphihox or Brahox),ouse (M), Drosophila (D), Xenopus (X), human (H), and chicken (C). Amino acid identity to Drosophila Antennapedia is indicated by a dash.
waaaqTHa
raaw(SLNwtofts(aCtAtACawMs1CoaowabIg(bn
4HfiHwt
HaCtsHttSnDHP
Cbiqa7iTsTpCbtaab
N
geTgcd1lnNmiocHtTa
381CRINOID AND OPHIUROID Hox GENES
ere included in the large clade consisting of thenterior and medial genes in both NJ trees of Figs. 4nd 5. It is usually difficult to infer specific orthologymong the medial genes based on homeodomain se-uences because they are very similar to each other.he medial fragments, except CJ-Hbox1 and SS-box1, did not cluster with any specific gene of mousend Drosophila.CJ-Hbox1 and HeHbox1 (PG8), in the homeodomain
egion compared in Fig. 3, encode an identical aminocid sequence (nucleotide similarity: 76.3%). SS-Hbox1nd CJ-Hbox1 formed a clade with Hoxb-8 (Fig. 4A) andith the Hbox1 ortholog of the asteroid, AM-7 (PG8)
Fig. 5A). The orthology among CJ-Hbox1, HeHbox1,S-Hbox1, and AM-7 is strongly supported by the highBP (92%) of the node for those four fragments in theJ tree (Fig. 5A). CJ-Hbox6 and SS-Hbox6 form a cladeith HeHbox6 and AM-6 in the NJ tree (Fig. 5A) and
he clade is supported with 77% LBP, suggesting orthol-gy among those sequences. CJ-Hbox3 and CJ-28ormed a clade with AM-4 (PG6) and TgHbox3 (PG6) inhe NJ tree (Fig. 5A). The LBP of the node of those fourequences is relatively high (85%), and the ML treeFig. 5B) also supports a monophyletic relationshipmong the four sequences with 96% LBP. It is clear thatJ-Hbox3 is an ortholog of the PG6 gene, considering
hat the amino acid similarities of this sequence toM-4 and TgHbox3 are 96 and 80%, respectively. On
he other hand, the affinity of CJ-28 to CJ-Hbox3,M-4, or TgHbox3 is weak. The amino acid similarity ofJ-28 to either AM-4 or TgHbox3 is only 64%. Inddition, CJ-28 did not show significant similaritiesith the medial group in the analyses shown in Fig. 4.oreover, the tfasta search suggested a closer relation-
hip of this fragment with echinoid Hbox1 genes (intin:24) than with Hbox3 genes (intin: 117). Thus, theJ-28 fragment could not be assigned to a specificrthologous gene of any known Hox genes used in thenalysis, implying that this fragment represents an-ther paralogous gene. SS-11 did not form a clusterith any specific sequence in all the phylogeneticnalyses. It should be noted, however, that the data-ase searches showed high similarity with PG4 genes.n both fasta and tfasta searches, the amphioxus PG4ene, Amphihox-4, showed the highest initial scoreintin) among the physically mapped genes (intin: 204y fasta, 142 by tfasta; amino acid similarity: 88.0%;ucleotide similarity: 73.7%).Hox posterior group (PG9–10). In the NJ tree (Fig.
A), the following four crinoid fragments CJ-9, CJ-box4, CJ-Hbox7, and CJ-Hbox10, and the followingve ophiuroid fragments SS-9(SS-9.2), SS-12, SS-box4, SS-13, and SS-Hbox10 formed a large cladeith the posterior genes of mouse (Hoxb-9 and Hoxd-10
o 13) and Drosophila (AbdB). Within this clade, CJ- c
box4 and SS-Hbox4 formed a subclade with Hoxb-9,nd this subclade is supported by the ML tree (Fig. 4B).J-Hbox10 and SS-Hbox10 clustered with Hoxd-10 in
he NJ tree with 51% LBP, but the relationship is notupported by the ML tree. SS-13 formed a clade withoxd-11, Hoxd-12, Hoxd-13, and AbdB with 70% LBP
hough the relationship is not supported by the MLree. Other fragments, CJ-9, CJ-Hbox7, SS-9, andS-12, formed a clade of their own, therefore showingo specific relationship with any of the mouse orrosophila sequences. Thus, only CJ-Hbox4 and SS-box4 could be assigned to a specific paralogous group,G9 in the NJ and ML analyses.In the NJ tree of Fig. 5A, CJ-Hbox4, CJ-Hbox7, and
J-Hbox10 clustered with echinoderm sequences TgH-ox4, HeHbox7, and HeHbox10, respectively, suggest-ng a specific orthologous relationship between se-uences for each pair. SS-Hbox4 is identical in aminocid sequence with CJ-Hbox4 (nucleotide similarity:5.0%), definitely indicating that it is a Hbox4 orthologn the ophiuroid. SS-12 showed a relationship withgHbox4 in the database search by the amino acidequence (intin: 103; amino acid similarity: 68.0%).his relationship, however, was not supported in thehylogenetic analyses. SS-Hbox10 formed a clade withJ-Hbox10 and HeHbox10 in the NJ tree with a longranch (Fig. 5A). The clade was supported by the MLree. SS-13 did not show a specific relationship withny of the known echinoderm sequences. CJ-9, SS-9,nd AM-9 formed a clade supported with high LBP inoth the NJ and the ML analyses (Fig. 5).
DISCUSSION
umber of Echinoderm Hox Gene Clusters
Based on the estimation of orthology to known Hoxenes, the position in a cluster has been inferred forach echinoderm Hox fragment in our study (Fig. 6).he arrangement of the crinoid Hox fragments sug-ests that the crinoid has a single Hox gene cluster. It isoncordant with the previous expectation that echino-erms generally have a single cluster (Mito and Endo,997). The results for the ophiuroid are based on aimited survey of a small number of clones that wouldot be enough to estimate the number of clusters.evertheless, the survey detected 13 different ho-eoboxes from only 36 clones and provided valuable
nformation as discussed below. The limited data on thephiuroid are consistent with the presence of a singleluster. However, it still remains possible that anotherox gene cluster exists in the ophiuroid, considering
he sequence similarity between SS-9/9a and SS-9.2.he SS-9.2 gene could be a trans-paralog of SS-9, asrgued above. If this is the case, then another Hox gene
luster consisting of a few complements may exist;
pdngsaptdvf
pt
E
naNH
N pli
382 MITO AND ENDO
ossibly, the SS-9 ancestor gene duplicated indepen-ently within the ophiuroid lineage. At present, there iso evidence for the presence of trans-paralogous Hoxenes in other echinoderm species. Moreover, the PCRurvey is subject to risks of errors and contaminations,nd extremely careful interpretations of alleles andaralogs are required. It cannot be completely deniedhat the sequence difference between SS-9 and SS-9aerived from PCR bias or that SS-9.2 is an allelicariant of SS-9/9a. We need more information (such as
FIG. 4. Phylogenetic relationship of the crinoid and ophiuroid parJ tree; (B) ML tree. Each number in a node indicates LBP by 1000 re
rom flanking sequences of the region analyzed in the h
resent survey) to establish the relationship amonghese fragments.
chinoderm Archetype Cluster
Combined with information obtained from other echi-oderms, the present results strongly suggest that thencestral echinoderm had a single Hox gene cluster.onvertebrate deuterostomes might also have a singleox gene cluster. The primitive chordate amphioxus
l homeodomain sequences with mouse and Drosophila sequences. (A)cations.
tia
as a cluster composed of 3 anterior, 5 medial, and at
l1tpTcTm
C
a
Taagtapamc
s
383CRINOID AND OPHIUROID Hox GENES
east 4 posterior genes (Garcia-Fernandez and Holland,994, 1996). The hemichordate acorn worm is inferredo have a cluster containing 3 anterior, 5 medial, and 1osterior genes by a PCR study (Pendleton et al., 1993).he complements of the inferred echinoderm archetypi-al (i.e., ancestral) Hox gene cluster are shown in Fig. 6.he cluster consists at least of 11 genes: 3 anterior, 4edial, and 4 posterior.
onservation of the Anterior Region—AnteriorConstraint
It is noteworthy that all the anterior genes, PG1, 2,
FIG. 4—
nd 3, have been detected in the survey on the crinoid. s
his is concordant with the recently reported fact thatn echinoid has all of the anterior genes (Martinez etl., 1999), strongly suggesting that the anterior threeenes are conserved throughout echinoderms. Al-hough the data for the members other than crinoidsnd echinoids are still fragmentary, the extremely goodreservation of the anterior genes through triploblasticnimals supports this view. The ophiuroid PG1 frag-ent SS-Hox1 provides additional evidence for the
onservation of PG1 genes throughout echinoderms.The insects, a beetle and a grasshopper, have been
hown to have three anterior genes, PG1–3, and the
ontinued
Cequence and expression pattern of the PG3 gene
s(c(siTagims
tbdp1rtrtac
a od
384 MITO AND ENDO
uggest that it is an ortholog of zen of DrosophilaFalciani et al., 1996). A PCR survey suggested that thehelicerate Limulus polyphemus has orthologs of PG1–3Cartwright et al., 1993). In addition, it has beenuggested, on the basis of a PCR survey, that planar-ans have Hox genes related to PG1–3 (Balavioine andelford, 1995). The suggestion has also been made thatrthropods and platyhelminthes have the anterior threeenes corresponding to the chordate PG1–3. If so, thenference follows that the common ancestor of acoelo-
ates and coelomates had three anterior genes corre-
FIG. 5. Phylogenetic relationship of the crinoid and ophiuroidsteroid (AM) sequences. (A) NJ tree; (B) ML tree. Each number in a n
ponding to PG1–3. Since the anterior genes regulate b
he formation of a part of the cephalic structures inilateral animals, conservation of these genes in echino-erms, which lack cephalic structures, have been ofarticular concern (Ruddle et al., 1994; Popodi et al.,996; Mito and Endo, 1997). Detection of all the ante-ior genes in echinoderms provides conclusive evidencehat the peculiarity of the echinoderm body plan is notelated to a loss or gain of the anterior complements ofhe Hox gene cluster. The conserved feature of thenterior region in the echinoderm Hox gene clusterould be related to the fact that their larvae are
rtial homeodomain sequences with echinoid (Hbox or SpHox) ande indicates LBP by 1000 replications.
pa
asically bilateral.
N
fnpk
PcdaTh
ftet1aeii
da
385CRINOID AND OPHIUROID Hox GENES
ovel Genes in the Medial Region
Specific orthologous relationships of the medial-typeragments CJ-28 and SS-11 to known sequences couldot be established by our phylogenetic analyses. It isossible that each of these genes represents an un-nown paralog of the echinoderm medial genes.The echinoid has only a single gene corresponding to
G4 and 5, i.e., Hbox9 (SpHox4/5), in its Hox geneluster (Martinez et al., 1999). Also, there are someata suggesting that the asteroid has a Hbox9 orthologs the only PG4/5-type gene (Mito and Endo, 1997).hese data suggest that orthologs of either PG4 or PG5
FIG. 5—
ave been lost in the echinoderm lineage. However, e
asta and tfasta analyses revealed sequence similari-ies of SS-11 with PG4 genes of animals other thanchinoderms. Since the deuterostome ancestor ishought to have had both PG4 and 5 orthologs (Burglin,995), it is possible that the fragment SS-11 is ofnother PG4/5 ortholog, which has been lost, at least inchinoids. For precise assignment of this gene, morenformation about organization of the Hox gene clustern ophiuroids and other echinoderms is required.
CJ-28 was included in the clade of echinoderm me-ial sequences, TgHbox3, AM-4, and CJ-Hbox3, in thenalysis of echinoderm sequences only (Fig. 5). How-
ontinued
Cver, the relationship between CJ-28 was not supported
ismrodoAifmcpdtdttfct
E
mtHtocasr
gdt(tsldsgartawcdusttts
grqatcT
aM
386 MITO AND ENDO
n the analyses of crinoid, ophiuroid, mouse, and Dro-ophila sequences (Fig. 4). If the CJ-28 sequence is of aedial gene, the extent of amino acid differences seems
emarkably high in comparison to other medial genesf echinoderms. The CJ-28 and Antp encoded sequencesiffer by 10 amino acids in the compared region. On thether hand, the average number of differences betweenntp and all the echinoderm medial genes except CJ-28
s only 3.9 6 2.0 (SD). The CJ-28 gene might haveunctions largely different from other medial genes oright be a pseudogene. One possibility is that in the
rinoid an Hbox3 ortholog duplicated to produce twoaralogous genes, CJ-Hbox3 and CJ-28, and after theuplication, amino acid substitutions accumulated inhe CJ-28 encoded sequence due either to functionalifferentiation from the CJ-Hbox3 encoded sequence oro loss of function. Examination of its genomic struc-ure and linkage with other genes might reveal uniqueeatures of the crinoid cluster organization, as in thease of the pufferfish (Aparicio et al., 1997) and that ofhe zebrafish (Prince et al., 1998; Amores et al., 1998).
xpansion of the Posterior Region—Tail Flexibility
The present survey revealed that echinoderms haveore posterior Hox genes than previously known. In
he echinoid, three distinct genes, Hbox4, Hbox10, andbox7, have been identified (Popodi et al., 1996) and
hese are arranged in this order toward the 58 terminaln a chromosome (Martinez et al., 1999). AM-9 wasonsidered the ortholog of Hbox7 in the asteroid (Mitond Endo, 1997). However, the results on the crinoiduggest that AM-9 is not the ortholog of Hbox7 but
FIG. 6. Inferred cluster organization of Hox genes in echinodermsbove the row shows PG (vertebrate paralogous group) and Hbox showito and Endo (1997). The echinoid cluster is based on Martinez et al
epresents another paralogous gene in the posterior o
roup, because an AM-9-type homeobox, CJ-9, wasetected in addition to Hbox4-, Hbox7-, and Hbox10-ype fragments. We designate this new group HboxP9Posterior 9, which does not necessarily mean orthologyo PG9). The results of the ophiuroid are furtherurprising; despite the small number of clones ana-yzed, a total of six posterior genes assigned to fiveistinct types were detected, and none of them corre-pond to Hbox7. As Fig. 6 shows, at least four cognateroups (Hobx4, Hbox7, Hbox10, and HboxP9) and andditional two distinct types, SS-12 and SS-13, could beecognizable among echinoderm posterior genes, thoughhis does not necessarily mean that all of these arerranged on a single cluster. It is an intriguing questionhether a particularly large posterior expansion oc-
urred only in the ophiuroids or whether all the echino-erms share a similar expansion but by other as yetnknown posterior genes. The present results seem toupport the former hypothesis because, in the survey ofhe crinoid, in which four posterior genes were de-ected, the number of analyzed clones (107) is thoughto be sufficient to detect a high proportion of amplifiableequences (Dick and Buss, 1994).In vertebrates, five paralogous groups of posterior
enes, PG9–13, have been known to exist. From theesults of phylogenetic analysis of homeodomain se-uences, Zhang and Nei (1996) suggested that twoncestral posterior genes, the ancestor of PG9/10 andhat of PG11/12/13, had already existed before the lastommon ancestor of nematodes, insects, and chordates.he ancestral two genes are thought to have duplicated
ach box indicates a Hox gene. The 38 terminal is at right. The numbern echinoderm cognate group (see the text). The asteroid cluster is by99).
. Es a
. (19
r triplicated in tandem in the cluster. Such an increase
ogpataTa1Siai‘mHgtpwndb1plctwb
P
cdgc1grGHswe
P
sddmoHtHr
mimossrptte
tareefcbpdmd
aSswS
A
A
A
B
B
B
C
387CRINOID AND OPHIUROID Hox GENES
f the posterior genes is a feature of deuterostome Hoxene clusters. Amphioxus has at least four genes in theosterior region, PG9 and PG10 orthologs and andditional two genes in a more upstream region, butheir relationships to the vertebrate paralogous groupsre unclear (Garcia-Fernandez and Holland, 1996).he ascidian Ciona intestinalis was suggested to havet least three posterior genes, putative PG10, 12, and3, based on a PCR survey (Di Gregorio et al., 1995).ince the posterior genes show high levels of variabil-
ty, the relationships of their homeoboxes among thenimals are unclear, especially in PG . 10. The variabil-ty in the number of posterior genes is known as‘laxitas terminalis’’ (van der Hoeven et al., 1996) or
ore easily ‘‘tail flexibility’’ (Garcia-Fernandez andolland, 1996). Expansion of the number of posterior
enes in echinoderms would also reflect tail flexibility,hough it is not clear whether the gain of the fourosterior genes in the echinoderm common ancestoras independent from gains in other animal phyla orot. The possibility of relationship between Hox geneuplication and evolution of body plan complexity haseen pointed out (Holland, 1992; Pendleton et al.,993). The posterior genes play important roles inattern formation not only in the trunk but also in theimb buds in vertebrates. In echinoderms, the dupli-ated posterior genes may have acquired novel func-ions. If so, the posterior region of the Hox gene clusterill be even more important in the study of echinodermody plan evolution.
araHox Gene Cluster
The Xlox homologues were detected from both therinoid and the ophiuroid, and the Gsh homologue wasetected from the crinoid. The Xlox, Gsh, and Cdx classenes constitute the ParaHox gene cluster and areollectively known as ParaHox genes (Brooke et al.,998). The asteroid and echinoid also have Xlox classenes. Thus, it is now clear that extant echinodermsetain the Xlox class genes. In addition, detection of thesh fragment from the crinoid suggests that the Para-ox gene cluster is conserved in echinoderms. The
equence information from the Xlox and Gsh fragmentsould be useful for further study of the structure andxpression of the echinoderm ParaHox genes.
erspectives on Body Plan Evolution
The echinoderm Hox genes occur in a single con-erved cluster, as is the Hox gene cluster in othereuterostomes, such as hemichordates, cephalochor-ates, and vertebrates. The ParaHox gene clusteright also be conserved in echinoderms. Thus, in the
rigin of the phylum Echinodermata, components of theox and ParaHox systems appear to have been re-
ained. These pieces of evidence on the Hox and Para-ox systems strengthen the idea that changes in
egulatory cascades may have been critical for establish-
ent of different body plans (Raff, 1996). This evidences also concordant with the hypothesis that gene recruit-
ent to a new role could have been a main driving forcef echinoderm evolution (Lowe and Wray, 1997). Thetudy of expression patterns and upstream/down-tream regulation of the Hox genes could clarify theole of these genes in determining the echinoderm bodylans. Comparisons of the Hox and ParaHox systems tohose of other bilaterians could provide insights intohe origin of the echinoderm body plan from that of anxpected bilateral ancestor with cephalic structures.On the other hand, the present survey demonstrated
hat the number of Hox gene complements is differentmong the echinoderm classes, especially in the poste-ior group. The exact extent of gene loss and gain inach class is unclear at present; yet, independentxpansion of posterior genes in ophiuroids is suggestedrom the present results. Such changes in the Hox geneluster could be a factor that generated differences ofody plans among echinoderm classes. Physical map-ing of all of the Hox genes for each class of echino-erms, as well as comparisons of expression patterns,ight well provide insights on the diversity of echino-
erm body plans.
ACKNOWLEDGMENTS
We thank Dr. Tatsuo Oji for critically reading the manuscript. Were also grateful to staff members of the Misaki Marine Biologicaltation for collecting the animals used in our study. T.M. wasupported by the Japan Society for the Promotion of Science. Thisork was funded by grants from the Ministry of Education, Science,ports and Culture.
REFERENCES
dachi, J., and Hasegawa, M. (1996). ‘‘MOLPHY, version 2.3, Pro-grams for Molecular Phylogenetics Based on Maximum Likeli-hood,’’ computer program package distributed by the authors, TheInstitute of Mathematics, Tokyo.mores, A., Force, A., Yan, Y.-L., Joly, L., Amemiya, C., Fritz, A., Ho,R. K., Langeland, J., Prince, V., Wang, Y.-L., Westerfield, M., Ekker,M., and Postlethwait, J. H. (1998). Zebrafish hox clusters andvertebrate genome evolution. Science 282: 1711–1714.
paricio, S., Hawker, K., Cottage, A., Mikawa, Y., Zuo, L., Venkatesh,B., Chen, E., Krumlauf, R., and Brenner, S. (1997). Organization ofthe Fugu rubripes Hox clusters: Evidence for continuing evolutionof vertebrate Hox complexes. Nat. Genet. 16: 79–83.
alavoine, G., and Telford, M. (1995). Identification of planarianhomeobox sequences indicates the antiquity of most Hox/homeoticgene subclasses. Proc. Natl. Acad. Sci. USA 92: 7227–7231.rooke, N. M., Garcia-Fernandez, J., and Holland, P. W. H. (1998).The ParaHoX gene cluster is an evolutionary sister of the Hox genecluster. Nature 392: 920–922.
urglin, T. R. (1995). The evolution of homeobox genes. In ‘‘Biodiver-sity and Evolution’’ (R. Arai, M. Kato, and Y. Doi, Eds.), pp.291–336. The National Science Museum Foundation, Tokyo.
artwright, P., Dick, M., and Buss, L. W. (1993). HOM/Hox typehomeoboxes in the Chelicerate Limulus polyphemus. Mol. Phylo-
genet. Evol. 2: 185–192.
D
D
D
F
G
G
H
H
I
K
L
L
L
M
M
M
P
P
P
P
R
R
v
Z
388 MITO AND ENDO
ayhoff, M. O., Schwartz, R. M., and Orcutt, B. C. (1978). A model ofevolutionary change in proteins. In ‘‘Atlas of Protein Sequence andStructure’’ (M. O. Dayhoff, Ed.), Vol. 5, Suppl. 3, pp. 345–352, Natl.Biomed. Found., Washington, DC.ick, M. H., and Buss, L. W. (1994). A PCR-based survey of homeoboxgenes in Ctenodrilus serratus (Annelida: Polychaeta). Mol. Phylo-genet. Evol. 3: 146–158.i Gregorio, A., Spagnuolo, A., Ristoratore, F., Pischetola, M., Aniello,F., Branno, M., Cariello, L., and Di Lauro, R. (1995). Cloning ofascidian homeobox genes provides evidence for a primordial chor-date cluster. Gene 156: 253–257.
alciani, F., Hausdorf, B., Schroder, R., Akam, M., Tautz, D., Denell,R., and Brown, S. (1996). Class 3 Hox genes in insects and theorigin of zen. Proc. Natl. Acad. Sci. USA 93: 8479–8484.arcia-Fernandez, J., and Holland, P. W. H. (1994). Archetypalorganization of the amphioxus Hox gene cluster. Nature 370:563–566.arcia-Fernandez, J., and Holland, P. W. H. (1996). Amphioxus Hoxgenes: Insights into evolution and development. Int. J. Dev. Biol.Suppl. 1: 71S–72S.asegawa, M., and Kishino, H. (1994). Accuracies of the simplemethods for estimating the bootstrap probability of a maximumlikelihood tree. Mol. Biol. Evol. 11: 142–145.olland, P. W. H. (1992). Homeobox genes in vertebrate evolution.BioEssays 14: 267–273.
rvine, S. Q., Warinner, S. A., Hunter, J. D., and Martindale, M. Q.(1997). A survey of homeobox genes in Chaetopterus variopedatusand analysis of polychaete homeodomains. Mol. Phylogenet. Evol.7: 331–345.ishino, H., Miyata, T., and Hasegawa, M. (1990). Maximum likeli-hood inference of protein phylogeny, and the origin of chloroplasts.J. Mol. Evol. 31: 151–160.
ipman, D. J., and Pearson, W. R. (1985). Rapid and sensitive proteinsimilarity searches. Science 227: 1435–1441.
ittlewood, D. T. J., Smith, A. B., Clough, K. A., and Emson, R. H.
(1997). The interrelationships of the echinoderm classes: Morpho-logical and molecular evidence. Biol. J. Linn. Soc. 61: 409–438.
owe, C. J., and Wray, G. A. (1997). Radical alteration in the roles ofhomeobox genes during echinoderm evolution. Nature 389: 718–721.artinez, P., Lee, J., and Davidson, E. H. (1997). Complete sequenceof SpHox8 and its linkage in the single Hox gene cluster ofStrongylocentrotus purpuratus. J. Mol. Evol. 44: 371–377.artinez, P., Rast, J. P., Arenas-Mena, C., and Davidson, E. H.(1999). Organization of an echinoderm Hox gene cluster. Proc. Natl.Acad. Sci. USA 96: 1469–1474.ito, T., and Endo, K. (1997). A PCR survey of Hox genes in the seastar, Asterina minor. Mol. Phylogenet. Evol. 8: 218–224.
earson, W. R., and Lipman, D. J. (1988). Improved tools forbiological sequence comparison. Proc. Natl. Acad. Sci. USA 85:2444–2448.
endleton, J. W., Nagai, B. K., Murtha, M. T., and Ruddle, F. H.(1993). Expansion of the Hox gene family and the evolution ofchordates. Proc. Natl. Acad. Sci. USA 90: 6300–6304.
opodi, E., Kissinger, J. C., Andrews, M. E., and Raff, R. A. (1996).Sea urchin Hox genes: Insights into the ancestral Hox cluster. Mol.Biol. Evol. 13: 1078–1086.
rince, V. E., Jolly, L., Ekker, M., and Ho, R. K. (1998). Zebrafish hoxgenes: Genomic organization and modified colinear expressionpatterns in the trunk. Development 125: 407–420.
aff, R. A. (1996). ‘‘The Shape of Life—Genes, Development, and theEvolution of Animal Form,’’ Univ. of Chicago Press, Chicago.
uddle, F. H., Bentley, K. L., Murtha, M. T., and Risch, N. (1994).Gene loss and gain in the evolution of the vertebrates. DevelopmentSuppl.: 155–161.
an der Hoeven, F., Sordino, P., Fraudeau, N., Izpisua-Belmonte,J. C., and Duboule, D. (1996). Teleost HoxD and HoxA genes:Comparison with tetrapods and functional evolution of the HoxDcomplex. Mech. Dev. 54: 9–21.
hang, J., and Nei, M. (1996). Evolution of Antennapedia-class
homeobox genes. Genetics 142: 295–303.
