Background

Diversification of morphological traits during evolution also requires diversification at the molecular level. Apart from the emergence of entirely new genes [1,2,3], diversification of gene function can be achieved either through changes at the regulatory level or by alterations in the coding region of the gene. Most genes have more than one function (pleiotropy), thus changes in their coding region and to a lesser extent in their regulatory regions will not only change a single feature of its function, but usually cause multiple (and mostly negative) effects in different organs or tissues (reviewed in [4, 5]). One possibility for overcoming this problem and facilitating evolutionary changes in pleiotropic genes is gene duplication: one copy preserves the original functions of the pleiotropic gene and thus allows the second copy to evolve in an (at least theoretically) almost unrestricted fashion.

Indeed, there is an increasing number of studies that provide evidence for a prominent role of gene duplication in the evolution of physiological or morphological novelty. Especially in plants where duplications of the entire genome are frequent, gene duplications have been implicated in the evolution of morphological novelties [6,7,8]. But also in animals, gene duplications have been linked to evolutionary novelties, for example in the evolution of spermatogenesis, the immune system, protective embryonic membranes, the physiology of digestion, and the morphology of the exoskeleton in Drosophila species [9, 10], sexual dimorphism in stalk-eyed flies [11], the visual system in crustaceans and insects [12], sex determination in bees [13], electric organs and the bulbus arteriosus in fish [14, 15], or the silk producing apparatus and leg morphology in spiders [16, 17].

In the classical model of gene duplication ([18], discussed in [19, 20]), the new copy evolves fast, because it is redundant and therefore is not subject to selection. It thus diverges quickly from the other copy which remains evolutionarily conserved, because of stabilizing selection on the old function. However, if the new copy becomes essential either by subfunctionalisation or neofunctionalisation, it is no longer redundant and from that time point it is also evolutionarily conserved because of its new functionality. The typical outcome of this process is that duplicated genes within a species (paralogs) differ in their expression profile, but the expression pattern of each paralog is evolutionarily conserved between different species.

The genomes of arachnids (including spiders) contain a large number of duplicated genes, and it has already been suggested that these may have fueled the evolution of many arachnid-specific novelties [16, 21, 22]. Previous work has shown that the homothorax (hth) gene is duplicated in the spider species Cupiennius salei (C. salei) and Acanthoscurria geniculata (A. geniculata) [23, 24]. The hth genes are widely conserved in the Metazoa and homologs have been described from all major arthropod clades [25], onychophorans [26], and from vertebrates, where the gene is called Meis [27]. The hth gene of Drosophila melanogaster (D. melanogaster) codes for a transcription factor with two highly conserved protein domains. The MEIS domain is a protein-protein binding domain that mediates the binding between Hth and a co-factor named Extradenticle (Exd) [28,29,30]. In addition to the MEIS domain the Hth protein also contains a modified homeodomain (TALE-HD) that facilitates the binding of Hth to DNA (e.g. [28]). In many cellular contexts Hth first binds to Exd in the cytoplasm, and the protein pair is then translocated to the nucleus where they bind DNA together with Hox proteins [31,32,33,34,35]. Apart from Exd and Hox proteins, Hth is also known to bind other DNA binding proteins (e.g. [36]). Therefore, hth is a highly pleiotropic gene and is involved in the formation and function of a large number of organs and tissues, for example leg and antenna specification, eye development, renal tubule growth, muscle fiber identity, and regulation of neuron differentiation (e.g. [31, 37,38,39,40,41]).

The hth duplicates in the spiders C. salei and A. geniculata show a strikingly different expression pattern, especially in the legs. One paralog, termed hth1, is expressed in a uniform pattern in the proximal and medial parts of the legs in both species [23, 24, 42]. This expression pattern is similar to hth expression in other arthropods and therefore likely is the conserved original expression of the gene before gene duplication [23]. The second paralog, termed hth2, is expressed in repeated rings along the proximal-distal axis of the legs in both species. This patterning is not seen in hth genes from other animals, and therefore likely represents a novel patterning function after divergence and neofunctionalisation of this second duplicate [23]. Therefore, at first glance, the duplicated hth genes in spiders show the typical behaviour of duplicated genes after neofunctionalisation: divergence of expression patterns between the paralogous genes, but strong conservation of the old and the new expression patterns between species. However, the ring pattern of hth2 is not entirely identical in C. salei and A. geniculata as noted by Pechmann et al. [43]. We have studied hth genes from two additional spider species and show that the expression patterns of hth2 indeed differ quite substantially between the species. We hypothesise that the hth2 gene has gained a new patterning function after gene duplication, but has then undergone a second phase of diversification of its new role. The hth2 gene is apparently a spider specific duplication and not related to other hth duplication events of non-spider arachnids. Diversification of hth2 expression patterns would thus be an interesting example for a duplicated gene that has not only provided genetic diversity through neofunctionalisation, but beyond that has been able to escape evolutionary conservation after neofunctionalisation thus becoming the basis for further genetic diversification.

Methods

Animal cultivation and gene cloning

Parasteatoda tepidariorum (P. tepidariorum) and Pholcus phalangioides (Ph. phalangioides) embryos were obtained from our laboratory stocks in Göttingen, and were treated as described previously [17, 44]. For gene cloning, we used SMARTer RACE cDNA (Clontech, Mountain View, CA, USA) synthesized from a combination of all embryonic stages of the different spider species (for details see [17]). Initial fragments of Pp-hth1 and Pt-hth2 were amplified using nested PCR with degenerated primers as described in [23]. Larger gene fragments for Pp-hth1 were subsequently obtained by RACE PCR using the following primers Pp-hth-3RACE: GGA CAT CCA TTG TTT CCA TTG TTG G, Pp3RaceHthnew: CAG GGA CTG CGA CGG GGG CCT C. Pp-hth2 was cloned with the following gene specific primers that were designed based on sequence information of a preliminary de novo assembled Ph. phalangioides transcriptome (unpublished data):

Pp_hth2_for: CGGTTATCGGGTGGACTTCGG.

Pp_hth2_rev: GGTCCATGATGTTCGGAGGCGAA.

Phylogenetic analysis

For the phylogenetic analysis of hth we searched for larger/complete open reading frames in the transcriptomes of Ph. phalangioides (unpublished data) and P. tepidariorum [45] using D. melanogaster and C. salei hth homologs as search queries. With the same query sequences we also searched in the published genome resource of the scorpion Mesobuthus martensii (M. martensii) [21], which resulted in three fragments with similarity to hth, named MMa38611, MMa00258, and MMa39038 (note that MMa00258 is very short and thus might be an incompletely assembled fragment). For the protein and nucleotide alignments of the hth orthologs we used previously published hth-related sequences from various metazoan species plus all hth sequences identified in the transcriptomes of P. tepidariorum and Ph. phalangioides and the genome of M. martensii (see list of species and accession numbers in Additional file 1).

Sequences were aligned with Clustal Omega [46] using default settings (the alignments are available in Nexus format in Additional file 2 and Additional file 3). Phylogenetic trees were calculated using the parallel version of MRBAYES (version 3.2.5) [47]. For the protein alignment, after testing mixed amino acid substitution models the Jones model was chosen to generate topological convergence using Metropolis coupling for 179,000 generations after reaching an average standard deviation of split frequencies below 0.01. A total number of 718 trees were written in 2 files (each file contained 359 trees of which 270 were sampled). In the nucleotide sequence analysis (using Metropolis coupling and the 4by4 nucleotide substitution model) an average standard deviation of split frequencies of 0.023742 was reached after 9 million generations. A total of 36,002 trees were written in 2 files (18,001 trees per file of which 13,501 were sampled). The 50% majority rule consensus trees shown in Fig. 1 and Additional file 4 were visualized with Geneious (version 9.1.3) [48].

Fig. 1
figure 1

Phylogenetic analysis of hth and related protein sequences from diverse Metazoa. Unrooted 50% majority rule consensus tree after Bayesian Markov chain Monte Carlo analysis. Branch lengths in the phylogram give the expected substitutions per site. Numbers at the tree edges are clade credibility values, which are a measure of the probability of each clade in the tree. The monophyletic clades formed by all Meis sequences from vertebrates, the hth sequences from mandibulate arthropods, and those from chelicerates are indicated by coloured circles (red, blue and green, respectively), Within the Chelicerata clade hth sequences from non-spider chelicerates (horse-shoe crab, scorpions, harvestman and mite) are indicated in green, whereas spider sequences are highlighted in orange). Homologs of spider hth1 and hth2 are indicated by black circles. For species abbreviations and sequence accession numbers please see Additional file 1

Full size image

Whole mount in situ hybridization and imaging

Fixation and in situ hybridization of embryos were performed according to standard protocols with minor modifications [49]. Digoxygenin-labeled RNA probes were synthesized from RACE PCR fragments for Pp-hth1, the entire Pp-hth2 and Pt-hth2 fragments, and from the previously published fragment of Pt-hth1 (GenBank accession number HE608682 [50]). The location of the probes is indicated in Additional file 5. This shows that the probes overlap highly conserved regions and thus in principle could produce unwanted cross-hybridization. However, given the distinct expression patterns detected with the used probes, we think that cross-hybridization was minimal. After in situ hybridization and cell nuclei staining with SYTOX-Green (Invitrogen), whole embryos were imaged with a Leica M205 FA binocular equipped with a QImaging Micropublisher 5.0 RTV camera using combined UV and white light. In addition appendages were dissected and mounted in 80% Glycerol in PBST and images were captured with a Zeiss Axioplan-2 microscope equipped with an Intas digital camera. For colour and brightness correction of all images Adobe Photoshop CS5 Extended or CS6 for Apple Macintosh was used. Staging of embryos was performed on the basis of the published staging schemes for P. tepidariorum [51], C. salei [52], and Ph. phalangioides [44].

Results

Duplicated hth genes in P. tepidariorum and Ph. phalangioides

Previous studies in the entelegyne spider C. salei [23] and the bird spider A. geniculata [25] have shown that hth is duplicated in these species, and that the genes apparently form two separate phylogenetic groups, termed hth1 and hth2, that differ in spatio-temporal expression. To investigate whether the duplication of hth is a general feature present in all spider groups, we have studied two additional spider species that belong to two remotely related spider groups, the cellar spider Ph. phalangioides and the common house spider P. tepidariorum, representing a haplogyne and another entelegyne spider, respectively. At the beginning of the present study transcriptome resources were not available for these species, therefore hth-related sequences were amplified from cDNA preparations with degenerated primers and more sequence information was obtained via RACE-PCR (see Methods). As expected from previous studies in spiders we were able to isolate partial sequences of two hth paralogs in P. tepidariorum. For Ph. phalangioides we only identified the hth1 homolog. When transcriptome resources became available, we also searched for hth-related sequences in the now available developmental transcriptomes of P. tepidariorum [45] and Ph. phalangioides (unpublished data). These sequencing resources confirmed the presence of two hth homologs in P. tepidariorum, and also revealed the presence of a second hth homolog in Ph. phalangioides that had escaped previous conventional gene cloning attempts. The fact that duplicated hth genes are now known from representatives of most major spider lineages including entelegyne spiders, haplogyne spiders and mygalomorph spiders, suggests that this gene is duplicated in all spiders.

Phylogenetic analysis reveals duplication of hth early in spider evolution

We next asked if these duplicates derive from a single duplication event before the radiation of the spider lineages, or if hth has been duplicated several times independently in the different spider groups. A previous phylogeny with all then available spider sequences had suggested that the duplicated hth genes cluster together according to duplicate and not according to species, suggesting a duplication before the split between mygalomorph and entelegyne spiders [24]. We therefore reconstructed the phylogeny of hth and related genes using published hth sequences from chelicerates, mandibulates, onychophorans, and the vertebrate Meis genes (see Additional file 1 for a full list of species and sequence accessions). Protein and nucleotide sequences were analysed separately and resulted in phylogenetic trees with very similar topology (Fig. 1, Additional file 4). The vertebrate Meis sequences form a monophyletic group in the trees and are clearly separated from the hth sequences from arthropods and the onychophoran. The insect and crustacean sequences also form a monophyletic group that is well separated from the group that contains the chelicerate sequences. Interestingly, the chelicerate sequences are divided into two separate monophyletic groups. The first group (denoted as hth1 in Fig. 1 and Additional file 4) contains all sequences from non-spider chelicerates (including three hth fragments (MMa38611, MMa00258, and MMa39038) found in the genome of the scorpion M. martensii, see Materials and Methods), the previously described hth1 copy from C. salei and A. geniculata, and one hth copy each from P. tepidariorum and Ph. phalangioides, that we therefore also denote as hth1. The second group (denoted as hth2 in Fig. 1 and Additional file 4) exclusively contains the hth2 copy from C. salei and A. geniculata, and the second copy from P. tepidariorum and Ph. phalangioides (therefore also denoted as hth2).

Expression of duplicated hth genes during embryogenesis

Our phylogenetic analysis suggests that the duplicated hth genes in P. tepidariorum and Ph. phalangioides are orthologs of the hth1 and hth2 genes previously described from C. salei and A. geniculata, and should therefore also show the diverged expression patterns in the legs. To test this assumption, we have studied the expression of the two hth genes in P. tepidariorum and Ph. phalangioides with a focus on the expression during embryonic leg development.

The expression of hth1 is virtually identical in both species throughout embryonic development, especially in the prosoma. It is strongly expressed throughout the ventral neuroectoderm and in the anterior part of the head. During brain differentiation the expression in the head becomes restricted to the non-neurogenic ectoderm, the labrum, the stomodeum and two stripes next to the stomodeum (Figs. 2d and 3d). When the prosomal limb buds start developing, hth1 is expressed in the entire prosomal appendages excluding the most distal part (Figs. 2a and 3a). This strong expression in the prosomal appendages remains until late dorsal closure stages (Figs. 2b, c and 3b, c), but then decreases slightly, especially in the distal portion, mainly in the metatarsus (Figs. 4f and 5f). We found minimal differences in the opisthosomal expression patterns between the species. In P. tepidariorum, hth1 is expressed in the entire opisthosoma at uniform levels except for the posterior end and one spot next to the limb bud of the second opisthosomal segment (O2; Fig. 2e). In Ph. phalangioides the expression in the opisthosoma is more restricted to presumptive ventral tissue and no expression can be detected in the developing heart tissue (Fig. 3e).

Fig. 2
figure 2

Expression of hth paralogs in Parasteatoda tepidariorum. Expression of Pt-hth1 in the prosoma of embryos (a) during limb bud elongation, (b) inversion, and (c) dorsal closure, all in lateral-ventral view. d Frontal view of Pt-hth1 expression in the head of an embryo during inversion. e Lateral-ventral view of opisthosomal expression of Pt-hth1 during inversion. Prosomal expression of Pt-hth2 (f) during limb bud elongation, (g) inversion, and (h) dorsal closure, all in lateral-ventral view. i Frontal view of the head during inversion: no expression of Pt-hth2 is present in the head. j Opisthosomal expression of Pt-hth2 during inversion, lateral-ventral view. All embryos shown with anterior to the left, except (d) and (i) which are in frontal aspect. Abbreviations: hl, head lobes; lr, labrum; ch, chelicera; pp., pedipalps; L1-L4, walking legs pairs 1–4; O1-O5, opisthosomal segments 1–5; pe, posterior end of the germband

Full size image
Fig. 3
figure 3

Expression of hth paralogs in Pholcus phalangioides. Expression of Pp-hth1 in the prosoma of embryos (a) during limb bud elongation, (b) inversion, and (c) dorsal closure, all in lateral-ventral view. d Frontal view of Pp-hth1 expression in the head of an embryo during inversion. e Ventral view of opisthosomal expression of Pp-hth1 during inversion. Prosomal expression of Pp-hth2 (f) during limb bud elongation, (g) inversion, and (h) dorsal closure, all in lateral-ventral view. i Frontal view of the head during inversion: no expression of Pp-hth2 is present in the head. j Opisthosomal expression of Pp-hth2 during inversion, lateral-ventral view. All embryos shown with anterior to the left, except (d) and (i) which are in frontal aspect. Abbreviations: see Fig. 2

Full size image
Fig. 4
figure 4

Expression of hth paralogs in the prosomal appendages of Parasteatoda tepidariorum. a-f Expression of Pt-hth1 is shown in the panels on the left side. g-o Expression of Pt-hth2 is shown in the panels on the right side in the figure. The top row of panels shows dissected chelicerae (che) (a, g) at early inversion, (h) late inversion, and (b, i) dorsal closure, the center row shows dissected pedipalps (ped) (c, j) at early inversion, (k) late inversion and (d, l) dorsal closure, and the bottom row shows dissected walking legs (e, m) at early inversion, (n) late inversion, and (f, o) dorsal closure. The asterisks in (j-o) denote distal expression rings (one in the pedipalps, two in the leg). The arrows in (k, l, n, o) denote an additional medial ring

Full size image
Fig. 5
figure 5

Expression of hth paralogs in the prosomal appendages of Pholcus phalangioides. a-f Expression of Pp-hth1 is shown in the panels on the left side. g-l Expression of Pp-hth2 is shown in the panels on the right side in the figure. The top row of panels shows dissected chelicerae (che) (a, g) at inversion, and (b, h) shortly before ventral closure, the center row shows dissected pedipalps (ped) (c, i) at inversion, and (d, j) shortly before ventral closure, and the bottom row shows dissected walking legs (e, k) at inversion, and (f, l) shortly before ventral closure

Full size image

Expression of hth2 differs from hth1 in terms of timing and pattern. In both species, the gene is expressed later and in a more restricted fashion than hth1. At germ band stages hth2 is expressed more ventrally than hth1 in the neuroectoderm and an expression domain in the head tissue or the developing appendages is absent (Figs. 2f, i and 3f, i). First expression of hth2 in the prosomal appendages is present at early inversion stages, which then differentiates until ventral closure stages (see next chapter), but is always different from hth1 expression (Figs. 2g, h, and 3g, h). The expression of hth2 in the opisthosoma differs significantly between Ph. phalangioides and P. tepidariorum. In P. tepidariorum hth2 is expressed adjacent to the opisthosomal limb buds, and dorsally in the tissue of the developing heart (Fig. 2j), but in Ph. phalangioides hth2 is not expressed in heart tissue and expression is restricted to the ventral side of the neuroectoderm and a spot ventral to each opisthosomal limb bud (Fig. 3j).

Hth expression in the prosomal appendages of different spider species

The largest expression differences between the hth duplicates exist in the prosomal appendages (chelicerae, pedipalps and walking legs). To analyse these differences in more detail we studied dissected prosomal appendages of P. tepidariorum and Ph. phalangioides. Expression of hth1 is very similar in P. tepidariorum and Ph. phalangioides. Strong expression of hth1 can be observed within almost the entire chelicera throughout embryonic development in both species (Figs. 4 a, b and 5 a, b). Throughout most of development hth1 is also strongly expressed in the proximal and medial domain of the pedipalps and the walking legs, and is excluded from the tip of both appendage types. In embryos undergoing dorsal closure the medial expression becomes weaker and patchy (Figs. 4 c-f and 5 c-f).

Expression of hth2 is quite different from the pattern of hth1 and also differs more significantly between the species. In the chelicera of P. tepidariorum hth2 expression starts as a diffuse domain (Fig. 4g), that later forms two stripes or patches (Fig. 4h, i). In the pedipalps of P. tepidariorum hth2 expression starts as a weak ring at the distal end of the medial portion of the pedipalp (asterisk Fig. 4j). A second ring appears later in development (arrow in Fig. 4k). In the legs Pt-hth2 is activated as two weakly expressed rings (asterisks Fig. 4m). An additional ring appears proximal to the initial two rings later in development (Fig. 4n, o (arrows)).

In Ph. phalangioides the hth2 expression pattern also comprises ring-shaped domains, but they differ in number and expression dynamics from hth2 in P. tepidariorum. In the chelicera Pp-hth2 is first expressed as a ring in the medial part, which partially dissolves during development (Fig. 5g,h). In the pedipalp Pp-hth2 expression starts as one ring in the distal part, and a number of weak segmental rings appear more proximally later in development (Fig. 5i, j). In the walking legs Pp-hth2 expression also starts as a ring in the distal part (Fig. 5k) and later during development additional weakly expressed rings appear near the future leg joints (Fig. 5l).

We have also re-studied here the expression profile of hth2 in C. salei, because in the original publication, only the expression pattern in the fully formed embryonic leg was shown [23], but the expression profile of this gene is actually more dynamic. Cs-hth2 has a diffuse expression in the chelicera with a faint medial ring (Fig. 6a). In young pedipalps it is weakly expressed in two rings (arrowhead and arrow, Fig. 6b). Later the proximal of these two rings broadens (arrow, Fig. 6c, d), the medial ring divides (arrowhead(s), Fig. 6c, d), and an additional distal ring appears de novo (asterisk, Fig. 6c, d). In young walking legs Cs-hth2 is expressed in a proximal, medial and distal ring (arrow, arrowhead and asterisk, respectively, Fig. 6e). All three rings broaden during development (indicated by bars in Fig. 6f, g), and then split into two rings (Fig. 6h). This results in the presence of one ring near each of the developing leg joints in late stages of embryonic development.

Fig. 6
figure 6

Expression of hth2 in prosomal appendages of Cupiennius salei. a Weak expression of Cs-hth2 in the chelicera during late inversion. Expression of Cs-hth2 in the pedipalp (b) before inversion, (c) at early inversion and (d) at late inversion. The arrow points to the proximal Cs-hth2 ring which broadens during inversion (indicated by the bar). The arrowhead points to the medial Cs-hth2 ring, which broadens during inversion (bar in (c)) and splits at late inversion (two arrowheads in (d)). The asterisks mark the distal Cs-hth2 ring. Expression of Cs-hth2 in the walking legs (e) before inversion, (f) at early inversion, (g) at late inversion, and (h) at dorsal closure. The arrows mark the proximal Cs-hth2 ring and its descendent. The arrowheads mark the splitting medial Cs-hth2 ring and the asterisks mark the splitting distal Cs-hth2 ring. Bars always indicate broadening of the ring expression before splitting

Full size image

Discussion

The hth2 gene has originated by a spider specific hth duplication

Duplicated hth-related genes exist in vertebrates, scorpions, and spiders. Our phylogenetic analysis indicates that these gene copies do not trace from a single duplication event at the base of the metazoan phylogenetic tree, but have arisen in three independent gene duplication events (Fig. 1; Additional file 4). The ancestral state is the presence of a single hth-related gene and this is preserved in the genomes of onychophorans and pancrustaceans. Also a number of chelicerate species have only a single copy of hth, which clusters with the hth1 copy of spiders. This strongly suggests that the spider hth1 copy preserves the ancestral properties of hth, which is also supported by the gene expression pattern (see also below).

Larger numbers of duplicated genes are found in several chelicerate groups including spiders, scorpions and horseshoe crabs [16, 21, 22, 53, 54], but it is still unclear if these trace from a single large-scale gene or genome duplication event in the stem-line of the Chelicerata or if multiple gene duplications have occurred in several chelicerate lineages independently. We have previously shown that the duplication of the dachshund gene (dac) is an old duplication and has apparently occurred before the diversification of the arachnid orders [17]. Our phylogenetic analysis of hth shows that in contrast to dac, the hth duplicates in scorpions and spiders are placed in separate clades in the phylogenetic tree. Thus, the hth duplicates identified in spiders and scorpions are unlikely to trace from the same duplication event and therefore must have been duplicated independently after the split between scorpions and spiders. Genes of the hth2 type are only found in spiders and therefore represent a relatively young hth duplication event that is specific to spiders. The alternative hypothesis that hth has been duplicated at the base of the chelicerate or arachnid tree is less parsimonious, because a single hth gene is present in other chelicerate groups (including mites and harvestmen) and for these groups additional independent losses must then be assumed.

Evidence for rapid diversification of hth2 regulation after gene duplication

Our analysis of the gene expression patterns shows that the regulation of hth1 is highly conserved, resulting in virtually identical expression patterns in all four spider species. As noted previously [42], this highly conserved expression pattern is comparable with hth expression in other arthropods (e.g. [55,56,57,58]), onychophorans [26] and vertebrates (e.g. [59]). In addition, not only the regulation, but also the protein coding region of the hth1 genes is highly conserved: sequence alignments of the spider protein sequences show that the Hth1 amino acid sequences align with high overall identity (Additional file 6). This observation is further corroborated by the phylogenetic sequence analysis that shows that the spider Hth1 sequences group closely together, separated by few expected substitutions per site and thus are closely related phylogenetically (Fig. 1). Taken together, these data strongly suggest that the hth1 copy retained most of the original gene functions after gene duplication in spiders.

By contrast, the hth2 copy has a markedly different expression profile in most tissues during embryonic development, especially in the prosomal appendages. The pattern in the pedipalps and legs consists of several segmental rings, suggestive of a role of hth2 in the development of the joints between the limb segments. Surprisingly this role is apparently not required in all segments of all species and each species investigated so far shows a unique pattern of segmental rings (summary in Fig. 7). This pattern does not correlate with the phylogenetic position of the species (e.g. the patterns in A. geniculata and C. salei are most similar, although both belong to distantly related clades in spider phylogeny). The pattern also does not correlate with the length of the adjacent segments and thus there is no evidence for a role in allometric limb growth. One possibility is that the divergent patters arose by neutral evolutionary processes, e.g. drift, that deleted/reduced different rings in different species. Alternatively, we suggest that expression/lack of expression of hth2 at the joints might correlate with specific features at certain joints, but the comparative morphology of the limb joints in spiders has not been investigated in such detail to provide direct evidence for this hypothesis. Interestingly, most of the segmental rings of hth2 are located within the hth1 expression domain, i.e. hth1 and the rings of hth2 (where present) are co-expressed at the limb joints. Thus, apart from a possible new function of hth2 at the limb joints that would be independent of hth1, there is also the possibility that hth1 and hth2 act redundantly or cooperatively at selected limb joints to produce a certain morphology. As already mentioned in the introduction, the Hth protein usually binds to the Exd protein to be transferred to the nucleus; however, previous studies have shown that exd expression in C. salei is restricted to the proximal limb portion and a single distal ring [23]. Thus, most of the hth2 rings in C. salei (and by inference also in the other species) are not co-expressed with exd and thus would require other means for their function. We have currently no further data on possible alternative binding factors for hth2 in spiders, but we note that there is recent evidence that Hth can indeed function independently from Exd [60].

Fig. 7
figure 7

Diversification of hth2 expression during spider evolution. The drawings show a schematic summary of hth2 expression in the walking legs in four different spider species (Acanthoscurria geniculata after [24]). Black stripes indicate strong expression in the walking legs, light gray bars represent weak expression. The cladogram at the left shows the phylogenetic relationships between the species (phylogeny simplified after [64]). The black box indicates the spider specific duplication event of hth that resulted in the origin of hth2 at the base of the spider phylogenetic tree

Full size image

The gain of a novel role of a hth-related gene after gene duplication would not be unprecedented. Recent work in bats has shown that a duplicate of Meis, Meis2, is implicated in the growth of the wing membrane, a morphological novelty specific to bats [61]. However, the subsequent diversification of the expression pattern after the gain of the new function as seen in the expression patterns of hth2 in spiders is unusual. Normally genes are expected to conserve their new function after neofunctionalisation and therefore should not show rapid diversification after neofunctionalisation [18]. Our data suggest that genes can escape this pressure of conservation after neofunctionalisation, and thus further diversify. In addition, genuine hth2 duplicates have so far only been found in opisthothelid spiders that originated in the Triassic period approximately 240 million years ago [62]. This makes the duplication event relatively young in terms of evolutionary age (relative to the age of the entire chelicerata clade which is at least 500 million years old [63]) and suggests that diversification of hth2 regulation must have occurred rapidly during spider radiation. Interestingly, also the Hth2 proteins (i.e. the coding sequence of the hth2 genes) from all four spider species studied so far are less similar among themselves than the Hth1 proteins. In sequence alignments the Hth2 amino acid sequences align with less overall similarity (Additional file 6 and Additional file 7), and in the phylogenetic analysis the spider Hth2 sequences are separated by more expected substitutions per site and thus are less closely related phylogentically than are the Hth1 sequences. This suggests that after gene duplication subsequent diversification of hth2 not only affected its regulatory region, as evidenced by the diverse expression patterns, but to a certain extent also its protein coding region.

Conclusions

Our study of the duplicated homothorax genes in spiders shows that this gene duplication is specific to spiders and that the two copies differ significantly in their expression patterns. In addition, the expression pattern of the hth2 copy is weakly conserved and shows a different expression pattern in each species studied so far. We conclude from these data that after duplication of hth at the base of the spider lineage the hth1 copy preserved the original functions and therefore shows a pattern similar to the unduplicated hth gene in other arthropods. The hth2 gene on the other hand shows a strongly divergent expression pattern, likely indicating new functionality not shared with the hth1 copy. We hypothesise that hth2 after neofunctionalization underwent a second phase of diversification of gene expression that led to the different expression patterns in each species. Thus, hth2 may serve as an interesting example for a duplicated gene that not only acquired new functionality, but also rapidly diversified this new functionality during evolution.