Introduction

Over the last decade, increasing attention has been paid to the study of basal metazoans to establish their phylogenetic relationship with bilaterians. In particular, the analysis of homeobox genes in early animals is of special interest for the understanding of the evolutionary pathway and the origin of metazoans. In this context, Porifera, at the base of the metazoan phylogenetic tree, represents a key group for developmental and evolutionary studies. In Porifera, a variety of homeobox-containing genes has been identified, all of them having bilaterian counterparts except the Sycox genes, so far identified only in calcareous sponges (Manuel and Le Parco 2000). Published sequences are members of the classes Irx (Iroquois) (Perovic et al. 2003), Lim (Wiens et al. 2003), Pax (Hoshiyama et al. 1998), POU (Seimiya et al. 1997), and the Antp class of non-Hox genes, including the families Bar/Bsh (Hill et al. 2004), Msx (Seimiya et al. 1994), NK-2 (Seimiya et al. 1994; Manuel and Le Parco 2000), and the EmH-3/prox2 demosponge homeobox genes, previously assigned to the Tlx/Hox11 family (Coutinho et al. 1994; Seimiya et al. 1994; Degnan et al. 1995; Richelle-Maurer et al. 1998, 2004; Wiens et al. 2003). No Hox genes or para-Hox genes have yet been reported in Porifera, suggesting that these genes appeared after the Porifera branching and that the non-Hox genes are the most ancestral ones (Galliot 2000; Gauchat et al. 2000).

This paper is dedicated to the study of the group of demosponge homeobox genes comprising EmH-3/EflH-1 (Ephydatia muelleri, E. fluviatilis) and the closely related sponge sequences, prox2 (E. fluviatilis), HOXa1 (Suberites domuncula), and Spox-Ta (Tethya aurantia). All these genes form a well-supported orthology group, which we propose to call the Demox family. In expression analyses, the EmH-3 gene appeared to be involved in cell-fate decisions and in the control of cell proliferation/differentiation. The expression of EmH-3 is time- and cell-specific and is necessary for the differentiation of archaeocytes into choanocytes and hence for the completion of a functional sponge (Richelle-Maurer and Van de Vyver 1999a, bb; Nikko et al. 2001). It is worth noting that the EmH-3 promoter is functionally active in mammalian cells, implying common regulatory programs with higher metazoans (Coutinho et al. 1998, 2003). EmH-3 expression is associated with the undifferentiated cell state in the transfected mammalian hematopoietic cells as it is in the young Ephydatia sponges (Coutinho et al. 2003). The presence of EmH-3 homologues in several freshwater sponges (Richelle-Maurer et al. 2004) and of an EmH-3-like gene in two marine sponges (Degnan et al. 1995; Wiens et al. 2003) suggested that this type of gene could be widespread among Porifera.

To address this issue, and to ascertain the position of the Demox genes within the Antp class, we surveyed different orders and classes of Porifera for the presence of an EmH-3-like gene by means of PCR with different sets of primers. Expression analyses using RT-PCR were performed for several species.

Materials and Methods

Sponge Samples

The sponge samples were collected in Florida (Fort Pierce deep and shallow reef, Indian River lagoon, July 2002) and in the Bahamas during a Harbor Branch Oceanographic Institution (HBOI) Research Expedition aboard the R/V Seward Johnson (November 8–25, 2002). Sponges were collected by scuba diving or by the HBOI Johnson-Sea-Link manned submersible. Immediately after collection, sponges were dissociated into single cells using nylon mesh, Ca2+/Mg2+-free seawater, and centrifugation as described elsewhere (Richelle-Maurer et al. 2001). The external surface of the sponge was avoided to minimize potential contamination. The cell pellets were stored at –80°C. Vouchers in 70% ethanol were deposited at HBOI. A total of 28 sponge samples corresponding to 25 species, 19 families, 9 orders, and 2 classes of Porifera were investigated (Table 1). Taxonomic identification and classification of the samples were based on the Systema Porifera (Hooper and van Soest 2002).

Table 1 List of species, their current classification, collection site, gene name, and accession number
Full size table

DNA and RNA Extraction and (RT)PCR Amplification

Genomic DNA was isolated from the dissociated sponge cells using DNAzol reagent (Invitrogen) or GTC buffer followed by phenol/chloroform extraction and isopropanol precipitation. PCR amplifications were performed with different sets of degenerate and nondegenerate primers designed against EmH-3 sequence (Table 2). Ev1 and Ev3 corresponded to the highly conserved homeodomain (HD) sequences ELEKKF and WFQNRR. ELEKKF was used instead of ELEKEF, usually utilized to amplify Hox genes (Manuel and Le Parco 2000), as this sequence was found in EmH-3, Demox-Namo, and Demox-Ndig. Furthermore, when ELEKEF/WFQNRR was used in the freshwater sponge Ephydatia fluviatilis, no homeobox encoding ELEKEF was recovered but two homeoboxes encoding ELRRF (prox1) and ELEKKF (prox2) were recovered (Seimiya et al. 1994).

Table 2 Degenerate and nondegenerate primers designed against the EmH-3 sequence comprising three exons (nt 1–192, nt 245–281, nt 539–753), the homeobox lying in the third exon between nt 591 and nt 661 (Richelle-Maurer et al. 1998)
Full size table

For freshwater sponges, all the primer combinations (Table 2) except those including Em5 gave positive results. On the contrary, for marine sponges, only the degenerate primers Ev1 and Ev3 gave positive amplification with the exception of two Haplosclerida species, Niphates amorpha (Demox-Namo) and N. digitalis (Demox-Ndig), which also gave positive results with the primer sets Em9, Ev9/Em6, Ev6. PCR reactions were carried out in a 50-µl total volume containing 1–5 ng DNA, Taq PCR buffer, 10 mM dNTPs, 25 pmol of each primer, and 2.5 units of the Taq DNA polymerase (Amersham). They were run in a Techne Unit Genius Cycler under the following conditions: 2 min at 95°C, then 35–40 cycles at 94°C for 1 min, 45°–51°C for 1 min 30, 72°C for 2 min, and a final extension step at 72°C for 10 min.

Total RNA was extracted from several samples using TRIzol reagent (Invitrogen) and was amplified using the Promega Access RT-PCR System according to the manufacturer’s protocol. RT-PCR conditions were those described by Richelle-Maurer et al. (2004), with an annealing temperature of 50°C. RT-PCR experiments were carried out with the Em9/Em6 set of primers for Niphates amorpha and N. digitalis (Haplosclerida) as well as with primers designed against Demox-Namo and Demox-Ndig sequences (Sp10F and Sp10R). For the other samples, Spheciospongia vesparium (Hadromerida), Mycale sp. (Poecilosclerida), and Haliclona sp. (Haplosclerida), the Ev1/Ev3 set of primers was used. Ten percent of the RT-PCR products were analyzed on a 1.5% agarose gel.

Reactions without template and with DNA or RNA extracted from E. muelleri adult sponges were performed as negative and positive controls, respectively.

(RT)PCR products of the expected size (≈120–250 bp) were either directly purified using the QIAquick PCR Purification Kit (Qiagen) or extracted from the agarose gel and purified on GenElute Agarose Spin Columns (Sigma).

Sequencing and Analysis

The purified DNA fragments were directly sequenced on both strands using the ABI PRISM BigDye Terminator v3.0 Cycle Sequencing Ready Reaction Kit and an ABI PRISM 3100 automated sequencer according to the recommended protocol (Applied Biosystems). The sequences were edited using the program Sequencer 3.0 (Applied Biosystems). Chromatograms were examined and amino acid sequences were deduced using BioEdit v.7.0.4. Searches for related sequences in other metazoans were performed in the GenBank database using the BLAST network service.

The 24 new homeobox sequences were deposited in the NCBI GenBank under accession numbers AY864744 to AY864767. They were designated Demox for Demospongiae and homeobox, followed by the first letter of the genus and the first three letters of the species (Table 1).

Phylogenetic Analyses

Demox HDs were aligned to representatives from all the non-Hox families HD sequences. Each family included at least one representative from Protostome, one representative from Deuterostome, and representatives from Cnidaria, Ctenophora, and Porifera when characterized. The alignment was unambiguous over the whole length of the sequences. The species codes are as follows: Cnidaria—Af, Acropora formosa; Cv, Chlorohydra viridissima; Hv, Hydra vulgaris; Hs, Hydractinia symbiolongicarpus; Ms, Metridium senile; and Nv, Nematostella vectensis; Ctenophora—Pp, Pleurobrachia pileus; Porifera—Ef, Ephydatia fluviatilis; Efr, Eunapius fragilis; Em, Ephydatia muelleri; Hb, Halichondria sp.; Sd, Suberites domuncula; Sl, Spongilla lacustris; Sr, Sycon raphanus; Ta, Tethya aurantia; and Th, Trochospongilla horrida; Arthropoda—Dm, Drosophila melanogaster; Dv, Drosophila virilis; and Vertebrata—Dr, Danio rerio; Gg, Gallus gallus; Hs, Homo sapiens; Mm, Mus musculus; Rn, Rattus norvegicus; and Xl, Xenopus laevis. Accession numbers for the sequences are as follows: Gbx family—Gbx2-Xl, U04867; unp-Dm, U35427; and anth-hbxE-Nv, AAG37789; Evx family—eveC-Af, 228960; Evx1-Mm, NM007966; and eve-Dm, M14767; En family—en-Dv, CAA28436; and en2-Xl, CAA44724; Not family—flh-Dr, NP571130; Gnot1-Gg, NP990685; and Cnot-Hv, CAB88387; Emx family—ems-Dm, X66270; EMX1-Mm, CCA48752; and Cn-Ems-Hs, CAA72534; Hlx family—H2-Dm, Y00843; and Hlx1-Mm, AAH50146; Sycox family—Sycox1-Sr, AF197139; Sycox2-Sr, AF197140; and Sycox3-Sr, AF197141; Nk2 family—SrNkxB-Sr, AF197144; SrNkxA-Sr, AF197143; SrNkxC-Sr, AF197145; SrNkxD-Sr, AF197146; SrNkx2-Sr, AF197142; NK2-Dm, S78691; TTF1-Rn, X53858; CnNK2-Hv, AF012538; Nkx3.1-Mm, U88542; bagpipe-Dm, L17133; and prox1-Ef, L10984; Barx family—BarH1-Dm, M73259; Xbh1-Xl, AAG14450; BarX1-Mm, Y07960; Cnox3-Cv, X64627; and BarBsh-Hb, AY328027; Bsx family—Bsx-Mm, NM178245; and bsh-Dm, L06475; Dlx family—dll-Dm, S47947; Dlx2-Mm, NM008553; and Dlx-Hv, AJ252183; Lbx family—Lbe-Dm, Y08821; and Lbx1-Mm, X90829; Hex family—Hex-Dm, AAF55844; HEX-Hs, Z21533; and cnHex-Hv, CAB88389; Tlx family— Tlx-Pp, CAEA5768; Tlx3-Gg, AF071875; and Hox11-Dm, CAA80535; Demox family—EmH-3-Em, U97664; EmH-3-Efr, AY300029; EmH-3-Sl, AY300030; EmH-3-Th, AY300031; Prox2-Ef, S43226; HOXa1-Sd, AJ493056; and Spox-Ta, X79265; and Msx family—Msx3-Mm, X96518; Msh-Dm, X85331; Msh-Hv, CAB88390; and Prox3-Ef, L23476. CnOtx-Hv (ADD20830), Otx1-Mm (CAA48754), otd-Dm (CAA41732), and the POU gene from Xenopus laevis (Pou1-Xl, S23248) were incorporated as outgroups.

Phylogenetic analyses of these amino acid alignments were performed using the neighbor-joining (NJ) distance method with the PAUP 4.0b10 software (Swofford 2000), with uncorrected distances, and the maximum-likelihood (ML) method with the PhyML software (Guindon and Gascuel 2003). The chosen model of amino acid substitutions was JTT (Jones et al. 1992). Among site rate variation was estimated using a discrete approximation to the gamma distribution with eight rate categories. Parameters (proportion of invariant sites and gamma shape parameters) were estimated during the search. There were no gaps in the sequences, but some sequences were incomplete; lacking positions were scored as missing data (see the alignment, available upon request from the corresponding author). For both methods, the statistical robustness of the nodes was evaluated by bootstrapping (Felsenstein 1985) (500 bootstraps for NJ, 200 bootstraps for ML).

The relationships within the Demospongiae were inferred using nucleotide sequences. Multiple sequence alignments were generated with Clustal W (Thompson et al. 1994) and checked by eye. There were no ambiguities and no gaps, but some of the sequences were incomplete and lacking positions were scored as missing data (see the alignment, available upon request from the corresponding author). Trees were built with the NJ distance method, with uncorrected distances, and the maximum parsimony (MP) method, with the PAUP 4.0b10 software (Swofford 2000). For MP analyses, characters were treated as unordered and equally weighted. MP trees were computed using heuristic searches with 20 replicates of random taxon addition sequence and TBR branch swapping. For the ML analyses of nucleotide data, a GTR (general time-reversible) + G + I model was used. This model was chosen because it is one of the most general models available for phylogenetic reconstruction from nucleotide data using ML. Among site variation was estimated using a discrete approximation to the gamma distribution with eight rate categories. Parameters were estimated with PAUP 4.0b10 from the result of a MP heuristic search, using the command Lscores. The estimated parameters were shape = 0.485787, pinvar = 0.0855665, and rmatrix = (1.68043 4.63832 1.65267 1.04942 6.49495 1.00000). Branch support was tested with bootstrapping (500 replicates for NJ and MP). Bootstraps were not performed for the ML analysis due to excessive computing time.

Results and Discussion

Characterization of the Demox Sequences

Amplified PCR products of the expected length encoding a HD were recovered with the different sets of degenerate and nondegenerate primers. A total of 24 new homeobox-containing sequences, called Demox, were isolated, all of them from the class Demospongiae. They were identified in 15 families corresponding to the major orders of the Demospongiae sensu stricto (Borchiellini et al. 2004) with the exception of the order Astrophorida (Table 1). They were not found in Homoscleromorpha and Calcarea. The deduced amino acid sequences shared a high degree of identity/similarity with the EmH-3 HD from E. muelleri (Richelle-Maurer et al. 1998), ranging from 95% to 100% for freshwater sponges and from 92% to 97% for marine sponges. Similarly to EmH-3, the Demox sequences harboured the characteristic residues of the Tlx/Hox11 family-specific signature (Fig. 1), R1, P4, R10, K24, S28, K39, T47, Q59, and S60 (Gauchat et al. 2000). However, identity with clear members of this family was limited (62%–66%), much lower than between Tlx/Hox11 representatives (85%–96%) or with EmH-3 (92%–100%). Almost all marine sequences possessed an E22, while freshwater sequences possessed a D22, which can thus be considered as a molecular diagnostic character of the Spongillina. Marine Haplosclerida seemed to be characterized by the sequence T21-E22-R23 and Dictyoceratida by an S39, while a G36R37 allied the Agelas species (Agelasida) with Axinella corrugata (Halichondrida). As many as 83% of the HD residues were invariant, corresponding mainly to the three helices. There were only a few variable residues situated at positions 1, 4, 15, 21, 22, 23, 29, 36, 37, and 57 of the HD (Fig. 1). Moreover, most of the substitutions were conservative. Identity remained high even outside the HD for the freshwater sponges but decreased for the marine sponges. However, a 12-amino acid, highly conserved sequence (100% similarity) was observed at the C-terminal of the HD in the new Demox sequences (Demox-Bint and Demox-Psp) and in the previously published genes EmH-3, prox2, and HOXa1 (Fig. 1; R1 underlined). This region shares no similarities with any other protein but might have regulatory functions. Actually, conserved sequences downstream of the HD have been observed in Engrailed (eh5) and Tlx (TH3) homeoproteins and it has been shown that the eh5 region contributed to repression (Smith and Jaynes 1996; Logan et al. 1998). It is worth noting that in all the published EmH-3-related genes for which the N-terminus is available, i.e., EmH-3Efr, EmH-3Sl, EmH-3Th, prox2, and HOXa1, a third conserved region (R2; Fig. 1) is found at the beginning of the gene. This region, which comprises a 7-amino-acid core, displays high similarity with the eh1 repressor domain of the En, Gsx, and Tlx/Hox11 families (Smith and Jaynes 1996; Logan et al. 1998). The presence of the eh1 motif is of particular interest as this motif is crucial for repression activity in several classes of HD proteins (Yasuo and Lemaire 2001; Bae et al. 2003; Kurata and Ueno 2003) and thus raises the possibility that EmH-3 and related genes may act as transcriptional repressors.

Figure 1
figure 1

Alignment of the new Demox deduced amino acid sequences to that of EmH-3 (Ephydatia muelleri) allowing deletions/insertions. Demox sequences are compared to the other published related homeodomains (HDs) from the freshwater sponges Eunapius fragilis (EmH-3Efr), Spongilla lacustris (EmH-3Sl), Trochospongilla horrida (EmH-3Th [Richelle-Maurer et al. 2004]), and E. fluviatilis (prox2 [Seimiya et al. 1994]) and from the marine sponge Suberites domuncula (HOXa1 [Wiens et al. 2003]). The HD and the eh1 motif (R2) are boxed; the conserved sequence (R1) downstream from the HD is underlined. The 7-amino-acid core in the eh1 motive is in boldface. Amino acid identities are indicated by dashes; gaps, by equal signs (=); and variable amino acid positions in the HD, by asterisks. The residues of the Tlx/Hox11 family-specific signature are highlighted (reversed colors). The length of the published sequences is indicated numerically.

Full size image

Variability was higher at the nucleotide level, partly due to the degeneration of the code, so that identical amino acid sequences corresponded to different nucleotide sequences in different species except for Lubomirskia baikalensis and Baikalospongia intermedia, Agelas clathrodes, and A. dispar. Nucleotide identities varied from 82% to 100% for freshwater sponges, from 72% to 98% for marine sponges, and from 72% to 84% between freshwater and marine sponges. The Baikalian freshwater sponges L. baikalensis and B. intermedia exhibited high nucleotide identity (91%–95.6%) with E. muelleri, E. fluviatilis, and S. lacustris, contrary to Potamolepis sp. (82%–84%). High nucleotide identity (92%) was also observed between the Agelas species and Axinella corrugata.

Phylogenetic Analyses

The phylogenetic tree constructed by the ML method and based on Demox amino acid sequences and HDs from all non-Hox families of the Antp class clearly indicated that the Demox sequences together with EmH-3 genes, prox2, and HOXa1 formed a well-supported clade (ML bootstrap, 62%; NJ bootstrap, 82%) loosely related to the other non-Hox metazoan families (bootstrap, <50%), each family being strongly supported (Fig. 2). Interestingly, in the ML tree (Fig. 2), the Demox family was the sister group not of the Tlx family, but of a clade comprising the Tlx and Hex families (with no bootstrap support). In contrast, a clade Demox + Tlx was found in the NJ tree, but with no support (result not shown). The Sycox family (from Sycon raphanus [Manuel and Le Parco 2000]) is the sister group of the Hlx family (with no bootstrap support in ML and 60% of bootstraps in NJ).

Figure 2
figure 2

Phylogenetic relationships among 87 Antp-class non-Hox genes, including all Demox sequences reported in this study, were inferred from the amino acid sequences of the full homeodomain (60 amino acids) using the maximum-likelihood (ML) method (–logL = 3689.18522) (see Materials and Methods). The designation codes are listed under Materials and Methods. The tree is rooted with the diverging homeodomain Pou1 from Xenopus laevis. Bootstrap values (above the branches, ML; below the branches, in boldface, neighbor joining) are indicated when >50%. The scale bar indicates an evolutionary distance of 0.5 substitution per site.

Full size image

Together, these data do not permit the Demox sequences to be considered as true members of Tlx/Hox11 family as previously thought (Coutinho et al. 1998, 2003; Galliot 2000; Gauchat et al. 2000; Richelle-Maurer et al. 2004). On the contrary, our data supported the view of a new family of homeobox-containing genes, designated the Demox family, with no identifiable orthologues in other metazoans and possibly specific to the Demospongiae sensu stricto (even if it remains to look for them in Hexactinellida and in other Homoscleromorpha and Calcispongia). Among the Antp class, the Demox family seemed to be evolutionarily close to the Tlx and Hex families. It can be hypothesized that some of the duplications from which these families originated have occurred in the eumetazoan branch, after the divergence of the sponge lineages, since there are no known sponge members of the Tlx and Hex families. However, this could also simply be due to lack of data, and this question will be resolved only when complete sponge genomes become available. The presence in all the members of the Demox family of the Tlx/Hox11 family-specific signature indicates that the Demox and Tlx/Hox11 are closely related families. After the early branching of Porifera, the Demox and the Tlx/Hox11 families would have diverged, each for their part, maintaining the Tlx/Hox11 signature and some other regions necessary to their function. It should be noted that the Tlx protein from Ctenophora (Tlx-Pp), assigned to the Tlx/Hox11 family by Martinelli and Spring (2005), also belongs to this family in our study but with no bootstrap support. However, contrary to the Demox family, the Tlx-Pp protein does not exhibit the Tlx/Hox11 family-specific signature but, interestingly, contains an eh1 repressor motif very similar to the eh1/Tlx motif (Martinelli and Spring 2005).

As for the internal phylogeny of the class Demospongiae, most of the interorder relationships could not be resolved due to the high conservation and short length of the Demox sequences leading to a large polytomy. Nevertheless, two clades were supported by bootstrap values greater than 50% (Fig. 2): the marine Haplosclerida (S.O. Haplosclerina and Petrosina) except Demox-Hsp (ML bootstrap, 86%; NJ bootstrap, 85%), on one hand, and the Agelas species (Agelasida) and Axinella corrugata (Demox-Acor, Hadromerida; bootstrap, 60% in ML and 62% in NJ), on the other hand.

Phylogenetic analyses applied to the Demox nucleotide sequences, with prox1 and prox3 (the two other known Ephydatia Antp-class homeobox genes) as outgroups, led to no better resolution of the relationships among the Demox clade (Fig. 3A). However, the monophyly of freshwater sponges (Spongillina) was confirmed, in both the MP analysis (Fig. 3A) and the ML analysis (Fig. 3B) (MP bootstrap, 67%; NJ bootstrap, 80%). Gene pairs coming from the same species (Demox-Sves1 and Demox-Sves2, Demox-Xmut1 and Demox-Xmut2) or from congeneric species (Demox-Msp1 and Demox-Msp2, Demox-Acon, Demox-Acla and Demox-Adis, Demox-Ndig and Demox-Namo) formed well-supported clades. The sequence from Axinella corrugata grouped with the three Agelas sequences in the MP tree (bootstrap, 69%), the NJ tree (bootstrap, 86%), and the ML tree.

Figure 3
figure 3

Trees computed from the nucleotide sequences corresponding to amino acid positions 1 to 60 of the Demox homeodomain. The tree is rooted with prox1 (NK-2 family) and prox3 (Msx family) from Ephydatia fluviatilis. A Strict consensus of 21 minimal trees (390 steps) from the maximum parsimony (MP) analysis. Bootstrap values (above the branches, MP; below the branches, neighbor joining) are indicated when >50%. B Result from the maximum-likelihood analysis (–logL = 1782.51337). The scale bar indicates an evolutionary distance of 0.2 substitution per site.

Full size image

These data provide independent corroboration of the current phylogenies of Porifera based on 18S, 28SrDNA, and other molecular and biochemical data, in particular, of the close relationship between some axinellid taxa and the order Agelasida (Itskovich et al. 1999; Schröder et al. 2003; Borchiellini et al. 2004; Erpenbeck et al. 2004, 2005a, b; Nichols 2005).

Expression in Sponges

In terms of expression, RT-PCR experiments carried out with five different species from the orders Hadromerida, Poecilosclerida, and Haplosclerida revealed the presence of transcripts of the expected size in the freshly dissociated cells from adult animals (Fig. 4). Their sequence was identical to that obtained by PCR with the same primers. The levels of expression were high, similar to those obtained with the positive control gene EmH-3 from adult sponges, especially when specific primers were used (Figs. 4c and d). These data are consistent with the high levels of expression of EmH-3 observed at the moment of hatching to the formation of the aquiferous system and throughout the sponge’s life (Richelle-Maurer and Van de Vyver 1999a). They are also in agreement with HOXa1 expression in adult tissues as well as in reaggregated cells (primmorphs) and during the formation of canals in the marine sponge S. domuncula (Wiens et al. 2003). Our data should be useful for future functional studies on the Demox genes of various demosponge species.

Figure 4
figure 4

Expression of the Demox genes in adult sponges (RT-PCR analysis). Gel electrophoresis shows the amplified products obtained with different sets of primers. Lane a: EmH-3 (Ephydatia muelleri) used as positive control, Em9/Em6. Lane b: Demox-Namo (Niphates amorpha), Em9/Em6. Lane c: Demox-Namo, Em9/Sp10R. Lane d: Demox-Namo, Sp10F/Sp10R. Lane e: Demox-Hsp (Haliclona sp.), Ev1/Ev3. Lane f: Demox-Msp1 (Mycale sp.), Ev1/Ev3. Lane g: Demox-Sves1 (Spheciospongia vesparium), Ev1/Ev3. Lane h: negative control without template. Molecular sizes are indicated by arrows.

Full size image

In conclusion, our data extend to 31 the number of published genes related to EmH-3. They show that the Demox sequences, EmH-3, prox2, and HOXa1 genes are not members of the Tlx/Hox11 family but belong to a novel family of the Antp class, designated the Demox family. Nevertheless, the presence in all the members of the Demox family of the Tlx/Hox11 family-specific signature indicates that the Demox and Tlx/Hox11 families are closely related.