HOX genes in the sepiolid squid Euprymna scolopes: Implications for the evolution of complex body plans

Abstract

Molluscs display a rich diversity of body plans ranging from the wormlike appearance of aplacophorans to the complex body plan of the cephalopods with highly developed sensory organs, a complex central nervous system, and cognitive abilities unrivaled among the invertebrates. The aim of the current study is to define molecular parameters relevant to the developmental evolution of cephalopods by using the sepiolid squid Euprymna scolopes as a model system. Using PCR-based approaches, we identified one anterior, one paralog group 3, five central, and two posterior group Hox genes. The deduced homeodomain sequences of the E. scolopes Hox cluster genes are most similar to known annelid, brachiopod, and nemertean Hox gene homeodomain sequences. Our results are consistent with the presence of a single Hox gene cluster in cephalopods. Our data also corroborate the proposed existence of a differentiated Hox gene cluster in the last common ancestor of Bilaterians. Furthermore, our phylogenetic analysis and in particular the identification of Post-1 and Post-2 homologs support the Lophotrochozoan clade.

The molluscs constitute one of the most successful animal phyla. The earliest fossils appear in the Tommotian and Atdabanian stages of the early Cambrian about 530 million years ago (1, 2). Between 50,000 and 100,000 extant mollusc species are divided into seven classes: Aplacophora, Polyplacophora, Monoplacophora, Scaphopoda, Bivalvia, Gastropoda, and Cephalopoda (3). Species of the various classes display a rich diversity of body plans ranging from a wormlike appearance in the aplacophorans to the complex and specialized anatomy found in the cephalopods. Despite the very diverse mollusc adult morphologies, the following synapomorphies have been proposed for the average mollusc: a ventral locomotory foot, a dorsal shell secreted by the mantle that also defines a cavity that houses the gills, a chitinous radula for feeding, a hemocoel, a dorsal heart, and a coelom mostly restricted to the pericardial and gonadal domains (4, 5). However, the very different forms observed in bivalves, gastropods, and cephalopods suggest that the basic molluscan body plan can be modified to the extent that all but the most fundamental characteristics are obscured. Contrary to the situation with the arthropods, nematodes, echinoderms, and chordates, where the mechanisms that underlie embryonic development have been well studied (6–10), surprisingly little is known about developmental mechanisms in molluscs.

Key determinants for anteroposterior body axis formation are a subset of homeobox-containing genes, the Hox genes. Homeobox genes encode transcription factors containing a highly conserved DNA-binding motif, the homeodomain (11). Hox genes initially were discovered in Drosophila but have since been identified in all animal phyla where they were studied (reviewed in refs. 11–14). Clustered organization of Hox genes has been found in a number of chordates including human, mouse (15), zebrafish (16), and amphioxus (17). Nonchordate Hox clusters have been described for Drosophila (18), ribbonworm (19), sea urchin (20), and Caenorhabditis elegans (21–23). Phylogenetic analysis of the available Hox gene information indicates that the last common ancestor of protostomes and deuterostomes had a cluster of at least six or seven Hox genes (14, 24). It appears contradictory that, despite their remarkable conservation, Hox genes could be responsible for the dramatic differences in body plans within and between phyla. Several hypotheses have been put forward to resolve this question. It has been proposed that the divergence between classes or phyla is accompanied by molecular changes in the Hox gene cluster or in the deployment of individual members. These changes may encompass duplications of the Hox gene cluster itself, selective gain or loss of Hox genes, changes in Hox gene expression levels, and changes in regulatory interactions between Hox proteins and their targets. Comparative genome analyses support the hypothesis that seemingly drastic developmental differences between phyla are brought about by differential deployment of functionally and structurally conserved Hox gene products in development rather than by the de novo invention of unique proteins (6, 10, 24). The molecular mechanisms contributing to this evolutionary “tinkering” (25, 26) are thought to have been established during the pre-Cambrian period, ultimately contributing to the “Cambrian explosion” of morphological novelties (1, 24).

We have analyzed the Hox genes of the Hawaiian sepiolid squid Euprymna scolopes (Cephalopoda, Dibranchiata, Sepioloidea), of which adults and embryos can be reared readily under laboratory conditions, providing the basis for systematic developmental studies (27) (Fig. 1). We have identified one anterior group gene, one ortholog of paralog group 3 (PG3), five central group genes, and two posterior group genes. Our data support the idea of a single putative Hox cluster in E. scolopes encoding a minimum of nine Hox paralogs, and support earlier hypotheses about the ancient origin of the Hox gene cluster (14, 28). The presence of the diverged posterior group genes provides additional support for the Lophotrochozoan assemblage (29).

Figure 1.

Dorsal and ventral views of E. scolopes ), subadult specimen.

Materials and Methods

DNA and RNA Extraction.

Genomic DNA was prepared from brains and arms of two medium-sized adult E. scolopes. Only this tissue was used to prevent contamination of the preparation with DNA from ingested prey. Genomic DNA was isolated by equilibrium density centrifugation in cesium chloride gradients (30).

Total RNA was prepared from 25 embryos each from Arnold stages 19–20 and 21–23 (6 and 8 days old, respectively) (27) by using TriReagent following the manufacturer's protocols (Molecular Research Center, Cincinnati). Poly(A)⁺ mRNA was isolated by using an Oligotex Direct mRNA Isolation Kit (Qiagen, Valencia, CA) following the manufacturer's protocols.

Cloning Homeobox Fragments and Hox Genes.

PCR primers.

We used the previously described primer pairs A/B (31) and E/F (32). For the amplification of the Post-1 and Post-2 homeobox fragments, specific primer pairs were designed that match the consensus sequence of the amino terminus of the homeodomain of Lophotrochozoan Post genes (14) and have the following sequence: Post-1 (5′) AA(A/G)TA(T/C)CA(A/G)AT(T/C/A)GCNGA(A/G)(C/T)TNGA(A/G)(C/A)GNGA(A/G)TA(T/C)G corresponding to amino acid sequence KYQIAELEREY; Post-1 (3′) (T/C)TT(T/C)TT(T/C)TC(T/C)TTCATNC(G/T)NC(G/T)(A/G)TT(T/C)TG(A/G)AACCA corresponding to amino acid sequence WFQNRRMKEKK; Post-2 (5′) (C/A)GNTA(T/C)CA(A/G)ACNATGGTN(C/T)TNGA(A/G)AA(T/C)GA(A/G)TT(T/C) corresponding to amino acid sequence RYQTMVLENEF; Post-2 (3′) (T/C)TT(T/C)TTNC(G/T)(T/C)TTCATNC(G/T)NC(G/T)(A/G)TT(T/C)TG(A/G)AACCA corresponding to amino acid sequence WFQNRRMKRKK.

PCR conditions.

Amplification with the A/B primer set in a reaction volume of 100 μl used 0.45 μg of genomic DNA, 0.5 μM primers, 1.5 mM Mg²⁺, 10 mM dNTPs, and 2.5 units Taq polymerase. Cycling conditions were one cycle (94°C for 3 min), 35 cycles (94°C for 1 min; 37°C for 1 min; 72°C for 30 sec), and one cycle (72°C for 5 min). Amplification with the primer E/F primer pair in a reaction volume of 100 μl used 0.45 μg of genomic DNA, 0.5 μM primers, 1.5 or 5 mM Mg²⁺, 10 mM dNTPs, and 2.5 units Taq polymerase. Cycling conditions were one cycle (94°C for 3 min), 35 cycles (94°C for 1 min; 37°C for 1 min, ramping time 0.5°C per sec, or 40°C for 1 min with maximum ramping; 72°C for 30 sec), and one cycle (72°C for 5 min). Amplification of a Post-1 homeobox in a reaction volume of 50 μl used 0.38 or 0.57 μg genomic DNA, 5 mM Mg²⁺, 0.2 mM dNTPs, 0.2 μM primers, and 2.5 units Taq polymerase. Cycling conditions were one cycle (94°C for 3 min), 30 cycles (94°C for 1 min; 57°C or 60°C for 1 min; 72°C for 30 sec), and one cycle (72°C for 5 min). Amplification of a Post-2 homeobox in a reaction volume of 50 μl used 0.1 or 0.38 μg genomic DNA, 5 mM Mg²⁺, 0.2 mM dNTPs, 0.2 μM primers, and 2.5 units Taq polymerase. Cycling conditions were one cycle (94°C for 3 min), 30 cycles (94°C for 1 min; 55°C or 56°C for 1 min; 72°C for 30 sec), and one cycle (72°C for 5 min).

Cloning and sequencing of PCR products.

Genomic PCR products were gel-purified by using either the Mermaid or Geneclean procedure (Bio 101) and cloned into the pCR2.1 vector (Invitrogen). PCR fragments obtained by rapid amplification of cDNA ends were either gel-purified or directly ligated into the pCRII-TOPO vector (Invitrogen). Cloned PCR products were sequenced by dideoxy chain termination (33) with SEQUENASE 2.0 (United States Biochemical) or by using an automated sequencer (Applied Biosystems). Forty-six clones derived from the Hox A/B primer set were analyzed, and another 75 were selected from the Hox E/F experiment. Four Post-1 clones, and 15 clones obtained with the Post-2 gene-specific primer set, were sequenced.

Rapid amplification of cDNA ends–PCR.

PCR-mediated rapid amplification of cDNA ends (Marathon cDNA Amplification Kit, CLONTECH) was used to clone the 5′ and 3′ sequences of the E. scolopes Hox cDNAs starting from E. scolopes embryonic poly(A)⁺ mRNA following the manufacturer's protocols. Gene-specific nested primers were synthesized for each Hox paralog (primer sequences are available on request) and used in conjunction with the nested adaptor-specific primers provided in the Marathon cDNA amplification kit. Touch-down PCR cycling parameters were: one cycle (94°C for 1 min), five cycles (94°C for 30 sec; 72°C for 4 min), five cycles (94°C for 30 sec; 70°C for 2.5 min), and 20 cycles (94°C for 20 sec; 68°C for 2.5 min). A second 50-μl reaction with the nested primers and 1 μl of first-round PCR products as the template was performed by using the same cycling parameters as described above.

DNA and Protein Sequence Analysis.

Nucleotide sequences and their conceptual translation products were entered into the blastx and blastp programs of the National Center for Biotechnology Information to identify putative homologs and paralog groups and to determine the similarity between cognate genes (34). The selection of posterior group gene sequences for amino acid substitution calculations and phylogenetic tree construction reflected available representatives from the major phyla and similarity scores obtained from blastx and blastp analyses. We further concentrated on phyla for which comprehensive data sets are available to increase the confidence in paralogy assignments, and the sequences of the entire homeodomains were applied to the analysis. The sequences were aligned by using CLUSTAL X 1.8.1 (35–37) with manual modifications performed to optimize matches. Alignments were used to calculate amino acid substitution numbers between sequence pairs with the correction of multiple substitutions by using Kimura's method (38, 39). For distance-based phylogenetic analyses and tree construction, programs from the oden package were applied. Reconstruction of phylogenetic trees by the maximum-likelihood method was performed by using PUZZLE 4.0.2 (40, 41), applying the BLOSUM 62 substitution model (42) and 1,000 puzzling steps. Final editing of the resulting trees was performed with the program treeview (43).

Nomenclature.

The E. scolopes Hox genes were named in accordance with previous suggestions and in keeping with the current literature (12, 14). The gene name is prefixed with the letters Esc denoting the species initials for E. scolopes, followed by the paralogy assignment.

Results

Using PCR amplification with degenerate homeobox-specific primers we were able to recover 11 unique homeobox gene fragments from genomic DNA of E. scolopes in a total of 116 informative sequences. Two of these sequence classes represented the non-Hox cluster genes caudal and gbx (results not shown; GenBank accession numbers AF127341–2). A presumed labial (PG1) ortholog was recovered with the HOX E/F primer set only, whereas PG3 and central homeoboxes were identified at various frequencies with both primer sets (data not shown). The E. scolopes Post-1 and Post-2 homeoboxes were recovered with the Post-specific primers. Subsequently, complete homeobox sequences and flanking sequences were obtained for all Hox genes except for the Deformed gene.

Anterior Group Orthologs.

The putative homeodomain sequence of the E. scolopes labial ortholog (Esc-lab) is very similar to that of the nemertean Lineus (19) and polychaetes [Ctenodrilus CTs-lab (44); Chaetopterus CHv-Hb3 (45); Nereis virens Nvi-lab (14)] and brachiopods (Lingula anatina, ref. 14). We were unable to identify a sequence with significant homology to paralog group 2 (proboscipedia orthologs).

PG3 Group Orthologs.

Esc-Hox3 is a member of PG3 (Table 1 and Fig. 2), which is identified by the Drosophila gene zerknüllt (zen). The closest relatives of the E. scolopes sequence are the Hox3 paralogs of representative species of the Lophotrochozoa clade. These include the mollusc Patella vulgata (Pvu-Hox3; ref. 14), the nemertean Lineus (LsHox-3; ref. 19), the polychaetes Chaetopterus (ChvHb-5; ref. 45), Ctenodrilus (CTs-Hox3; ref. 44), Nereis (Nvi-Hox3; ref. 14), and the brachiopod Lingula (Lan-Hox3; ref. 14).

Table 1.

Species names and taxonomic affinities used in sequence analyses

Genus/species	Phylum/class	Abbreviation	References
Euprymna scolopes	Mollusca/Cephalopoda	Esc	This article
Haliotis rufescens	Mollusca/Gastropoda	Hru	46
Chaetopterus variegatus	Annelida/Polychaeta	Chv	45
Nereis virens	Annelida/Polychaeta	Nvi	14
Helobdella robusta	Annelida/Hirudinea	Hro	14
Helobdella triserialis	Annelida/Hirudinea	Htr	14
Priapulus caudatus	Priapulida	Pca	14
Lineus sanguineus	Nemertea	Lsa	19
Lingula anatina	Brachiopoda	Lan	14
Dugesia japonica	Platyhelminthes/ Turbellaria	Dja	48
Drosophila melanogaster	Arthropoda/  Hexapoda	Dme	18
Tripneustes gratilla	Echinodermata	Tgr	57
Ciona intestinalis	Chordata	Cin	17
Branchiostoma floridae	Chordata	Bfl	17

Figure 2.

Comparisons of known Hox orthologs and homeobox-containing genes with predicted E. scolopes homeodomains. Only representative homeodomains and short flanking sequences from individual classes or phyla are shown to simplify the diagrams and to highlight phyletic distances. Emphasis was placed on organisms for which an array of Hox sequences are known to facilitate the orthology comparisons. The Drosophila Antennapedia homeodomain sequence is shown at the top of the alignments. Nonidentical amino acids are shown, and dashes in the subject sequences indicate identity. Paralog-specific amino acids (light gray) and paralog-specific signature motifs (dark gray) are shaded. See Table 1 for abbreviations.

Central Group Orthologs.

We recovered five distinct sequences that are representative for Antennapedia-like homeoboxes of the central group. The central group orthologs were assigned to paralogy groups based on conserved residues within the homeodomain and on signature motifs for particular paralogs found in regions flanking the homeodomain (14).

Esc-Dfd displayed the highest sequence similarity with paralog group 4 genes from both Ecdysozoan and Lophotrochozoan phyla. The Drosophila Deformed gene is the namesake for this paralog group. Although we were unable to isolate 3′ flanking sequences of this homeobox, which may contain Dfd-specific signature motifs (see Fig. 2) the recovered sequence still allowed a confident paralogy assignment.

Esc-Scr represents a paralog group 5 member, identified by the Drosophila Sex combs reduced gene. It is most similar to Hru-Hox5 from the archaegastropod Haliotis (46) and additional Lophotrochozoan Hox genes, including the brachiopod Lingula and the polychaete Nereis (14). It also shows significant sequence similarity to Amphi-Hox5 (17), and the leech Hox genes Lox20 and Lox6 (47).

Esc-Antp displayed significant similarity with Antennapedia orthologs, including Priapulus Pca-HB2, the nemertean Ls-Hox7 homeodomain, and the brachiopod Lan-HB1 sequence. The inferred sequence of Esc-Hox7 is identical to Drosophila Antennapedia.

Esc-Lox5 appears to be a member of the Antennapedia group of Hox genes, most similar to paralog group 6 homeobox genes from other Lophotrochozoa. Members of this paralog group share conserved amino acids Q1, T4, and G39 of the homeodomain (14), features that separate them from the Ecdysozoan and deuterostome paralog group 6 members. These features were first defined in leech homeobox genes and are therefore frequently grouped as Lox5 genes. In addition, a motif C terminal to the homeodomain (KLTGP) characteristic for Lox5 orthologs strongly supports the notion that this is a member of paralog group 6/Lox5. Because established paralog groups are based on vertebrate gene sequences, and because the central group Hox genes have no true orthologs among the invertebrate phyla (see ref. 14), we elected to name this gene Esc-Lox5.

Esc-Lox4 is most similar to Lox4 from the brachiopod Lingulina anatina. Esc-Lox4 shares two of the three conserved residues, Q-19, H-22, and K-27, that identify the Lox4 homeodomain, in addition to a Lox4-specific motif C terminal of the homeodomain. Hence, we identify Esc-Lox4 as the Lox4 homolog from E. scolopes.

Posterior Group Orthologs.

Complete homeodomain sequences for Post-1 and Post-2 class genes were obtained through the combination of genomic PCR and rapid amplification of cDNA ends. The Esc-Post-1 homeodomain contains most of the previously assigned conserved residues (14). Based on the additional sequence information now available, we have designated additional conserved residues (>50% majority)(see Fig. 2). Similarly, the Esc-Post-2 homeodomain displays very high sequence similarity to the previously identified Post-2 sequences, in particular to the nemertean Lineus sanguineus Hox9 gene (14, 19, 48). Both neighbor-joining analysis (Fig. 3) and maximum-likelihood analysis (not shown) provided support for the distinction between the posterior genes of the Lophotrochozoan Post-1 and Post-2 groups, and an Ecdysozoan Abdominal-B group, as was previously observed (14). The neighbor joining raised the interesting possibility that the Post-2 and Post-1 groups may be sister groups of the Ecdysozoan Abd-B and deuterostome posterior genes, respectively. More sequence information will be required to provide this intriguing possibility with the necessary bootstrap support.

Figure 3.

Phylogram analysis of E. scolopes Post-1 and Post-2 homeodomain fragments and other known posterior homeodomain sequences, as revealed by neighbor-joining analysis. Existing representative sequences from Lophotrochozoa, Ecdysozoa, and deuterostomes were selected for the comparison. The Drosophila Hox genes were used as an outgroup. One thousand replicate data sets were generated. Internal edge labels refer to the number of bootstraps supporting the partition. Bootstrap values below 300 were not included in the phylogram. High support is found for the Lophotrochozoan Post-1 and Post-2, and Ecdysozoan Abd-B genes. The scale shows the number of amino acid substitutions per site.

Discussion

The identification of nine unique Hox genes (one anterior group, one PG3 group, five central group, and two posterior) from the sepiolid squid E. scolopes, a cephalopod, provides a systematic analysis of Hox-cluster genes in a mollusc. These data will enable us in future experiments to further refine the phylogenetic relationship within the Lophotrochozoa, analyze in detail the genomic organization and the sequence of a Lophotrochozoan Hox cluster, and study the spatiotemporal expression pattern and the role of the Hox genes in the formation of the very complex morphology in cephalopods.

Cephalopod Hox Genes and the Relationship with Other Lophotrochozoa.

Recent molecular data derived from the comparative analysis of 18S rRNA genes indicate that the bilaterian phylogenetic tree is divided into three clades, the Lophotrochozoa, Ecdysozoa, and Deuterostomia (29, 49). The Lophotrochozoa comprise Annelida, Mollusca, Sipunculida, Echiurida, Plathelminthes, Nemertea, Brachiopoda, and Ectoprocta, whereas the Ecdysozoa include Arthropoda, Nematoda, and other molting phyla (49). Our data on Hox cluster genes in the cephalopod E. scolopes, and in particular the identification of Lox4, Lox5, Post-1 and Post-2 homologs, are consistent with this subdivision of the protostomes.

A Single Hox Cluster in Cephalopods?

An archetypic Lophotrochozoan Hox gene cluster has been proposed to consist of two anterior genes, one PG3 ortholog, five central genes, and two posterior class genes (14). This number gives an estimate of the number of Hox cluster genes we can expect to find in E. scolopes. Of these, we identified all except a paralog group 2 homolog (proboscipedia). Previous studies on several Lophotrochozoan species have variably reported the presence of paralog 2 and posterior group genes. A paper on the primitive gastropod Haliotis suggested the existence of three members of the anterior group of Hox genes, including a Hox2 paralog, and three members of the central group (50). One anterior group gene and five central group genes have been identified in the gastropod Patella (14). Hox2/proboscipedia orthologs have also been identified in two polychaetes (14, 44). One explanation for these differing results using PCR methods could be the presence of an intron (19). In addition to Hox2/proboscipedia orthologs, earlier studies yielded different results with respect to posterior group genes. Two groups failed to identify Abd-B homologs during comprehensive surveys of two polychaetes, Chaetopterus variopedatus (45) and Ctenodrilus serratus (44). A posterior class gene most similar to AbdB was found in oligochaeta (44), and two posterior Hox genes were found in the polychaete Nereis (14). A member of the posterior group of Hox genes in the ribbonworm could not be identified with PCR methods, but eventually was isolated by using low-stringency hybridization methods (19). The identification of posterior group genes in Lophotrochozoan and Ecdysozoan protostomes and in deuterostomes supports the existence of the posterior group of Hox genes in the last common ancestor of deuterostomes and protostomes. A number of recent analyses indicate that the posterior class even predates the last common ancestor of Cnidarians and Bilateria. Our data and recent data from other Lophotrochozoa (14) show the existence of divergent posterior group genes in a number of protostome phyla, which helps explain the apparent limitations of the PCR-based approach in identifying this paralog group. The use of specific primers as shown in this article or genomic walking should enable the characterization of all posterior group genes where they are present. Eventually, the identification of additional posterior Hox genes from animals of different phyla should help resolve whether the posterior Hox genes were inherited from a common ancestor, or whether they are the result of independent duplication events.

In conclusion, based on our results and by extrapolation of the data sets mentioned above, we propose that an archetypical mollusc Hox cluster contained at least 10 genes (two anterior group genes, one PG3 ortholog, five central group genes, two posterior group genes), and that a cephalopod Hox cluster contains at least nine of those genes. It remains possible that additional, divergent posterior group genes will be discovered. Furthermore, we have no evidence for multiple genes of any given paralogy group. Therefore, our data are consistent with the existence of a single copy of a presumed Hox cluster in cephalopods.

Hox Genes and Mollusc Body Plan Evolution.

Hox genes are one group among many transcriptional regulators that qualify as “master regulator genes” of development (51–53), and they determine regional specializations along the anteroposterior body axis. The Hox cluster duplications observed in vertebrates have given rise to a complex interacting network of Hox genes controlling many aspects of skeletal, muscle, and neuronal differentiation. On the other hand, there are animals, which possess a single Hox gene cluster, yet display a sophisticated morphological organization and complexity, as in Drosophila, and the cephalopod E. scolopes, as shown in this article. The molluscs comprise classes with species ranging from microscopic wormlike representatives to the largest and behaviorally most sophisticated invertebrates known. Yet, our data and previous reports (14, 50) strongly suggest that molluscs have only one Hox cluster. We therefore propose that the rich diversity in body plans observed among molluscs is more likely the consequence of the recruitment or intercalation of different target genes into existing regulatory networks (intercalary evolution, ref. 54), and that subtle changes in the control regions of both Hox genes and their downstream effectors have led to significant alterations of developmental programs (10). In a recent study comparing Hox sequences between distantly related metazoans, the authors came to the conclusion that many of the amino acid replacements used as diagnostic criteria for particular paralog groups may represent functionally significant substitutions because they are likely to be localized on the surface of the respective proteins (55). Protein–protein interactions between homeodomains and between homeodomains and paired domains have recently been demonstrated (56). Thus, specific amino acid substitutions may signify the gain or loss of a particular interaction between proteins, and thereby represent indicators of morphological change. The use of Hox gene expression pattern analysis to identify homologous structures and the isolation of target genes will allow the characterization of changes in gene regulation that were instrumental in mollusc evolution. In addition, the identification of the Hox gene complement in other mollusc species and the identification of differentially regulated target genes will contribute to our understanding of the morphological diversification within the phylum Mollusca, and by extension also across animal phyla.

Abbreviation

PG3

paralog group 3