Abstract
Temporal collinearity is often regarded as the force preserving Hox clusters in vertebrate genomes. Studies that combine genomic and gene expression data in invertebrates would allow generalizing this observation across all animals, but are scarce, particularly within Lophotrochozoa (e.g., snails and segmented worms). Here, we use two brachiopod species –Terebratalia transversa, Novocrania anomala– to characterize the complement, cluster and expression of their Hox genes. T. transversa has an ordered, split cluster with ten genes (lab, pb, Hox3, dfd, scr, lox5, antp, lox4, post2, post1), while N. anomala has nine (missing post1). Our in situ hybridization, qPCR and stage specific transcriptomic analyses show that brachiopod Hox genes are neither strictly temporally nor spatially collinear; only pb (in T. transversa), Hox3 and dfd (in both brachiopods) show staggered mesodermal expression. The spatial expression of the Hox genes in both brachiopod species correlates with their morphology and demonstrates cooption of Hox genes in the chaetae and shell fields, two major lophotrochozoan morphological novelties. The shared and specific expression of a subset of Hox genes, Arx and Zic orthologs in chaetae and shell-fields between brachiopods, mollusks, and annelids supports the deep conservation of the molecular basis forming these lophotrochozoan hallmarks. Our findings challenge that collinearity alone preserves lophotrochozoan Hox clusters, indicating that additional genomic traits need to be considered in understanding Hox evolution.
Introduction
Hox genes are transcription factors that bind to regulatory regions via a helix-turn-helix domain to enhance or suppress gene transcription [1, 2]. Hox genes were initially described in the fruit fly Drosophila melanogaster [3, 4] and later on in vertebrates [5–7] and the nematode Caenorhabditis elegans [8]. In all these organisms, Hox genes were shown to provide a spatial coordinate system for cells along the anterior-posterior axis [9]. Remarkably, the Hox genes of these organisms are clustered in their genomes and exhibit a staggered spatial [3] and temporal [10, 11] expression during embryogenesis that corresponds to their genomic arrangement [3, 12, 13]. These features were used to classify Hox genes in four major orthologous groups –anterior, Hox3, central and posterior Hox genes– and were proposed to be ancestral attributes to all bilaterally symmetrical animals [1, 13, 14].
However, the study of the genomic arrangements and expression patterns of Hox genes in a broader phylogenetic context has revealed multiple deviations from that evolutionary scenario. Hox genes are prone to gains [15–17] and losses [18–21], and their arrangement in a cluster can be interrupted, or even completely disintegrated [22–25]. Furthermore, the collinear character of the Hox gene expression can fade temporally [24, 26, 27] and/or spatially [28]. Hox genes have also diversified their roles during development, extending beyond providing spatial information [29]. In many bilaterian embryos, Hox genes are expressed during early development, well before the primary body axis is patterned [26, 30–32]. They are also involved in patterning different tissues [33] and have been often recruited for the evolution and development of novel morphological traits, such as vertebrate limbs [34, 35], cephalopod funnels and arms [28], and beetle horns [36].
It is thus not surprising that Hox genes show diverse arrangements regarding their genomic organization and expression profiles in the Spiralia [37], a major animal clade that includes a high disparity of developmental strategies and body organizations [38–42]. A striking example is the bdelloid rotifer Adineta vaga, which belongs to the Gnathifera, the possible sister group to all remaining Spiralia [41, 42]. As a result of their reduced tetraploidy, its Hox complement includes 24 genes, albeit it lacks posterior Hox genes and a Hox cluster [43]. The freshwater flatworms Macrostomum lignano and Schmidtea mediterranea also lack a Hox cluster [44, 45] and parasitic flatworms have undergone extensive Hox gene losses, likely associated with their particular life style [21]. Interestingly, the limpet mollusk Lottia gigantea [16] shows a well-organized Hox cluster. Other mollusks (e.g. the pacific oyster Crassostrea gigas) and the segmented annelid Capitella teleta exhibit organized split Hox clusters [46, 47]. On the other hand, the cephalopod mollusk Octopus bimaculiodes has lost several Hox genes and lacks a Hox cluster [22]; and the clitellate annelids Helobdella robusta and Eisenia fetida do not show a Hox cluster and have greatly expanded some of the Hox classes [16, 17].
Although Hox gene expression is known for a handful of spiralian species [26, 44, 46, 48–58], the relationship between genomic organization and expression domains is known for only three of them, namely the annelids C. teleta and H. robusta, and the planarian S. mediterranea. Consistent with the lack of a Hox cluster, H. robusta and S. mediterranea show neither temporal nor spatial collinearity [44, 54–56]. Conversely, C. teleta, which has an organized, broken cluster, does exhibit these features [46]. In general, these observations suggest that the presence of collinearity – in particular, temporal collinearity– could be associated with the retention of a more or less intact spiralian Hox cluster, as it seems the case for the vertebrate cluster [14, 23, 59, 60]. However, more studies combining genomic and expression information, and including the vast spiralian morphological diversity, are essential to draw robust conclusions about Hox gene evolution and regulation in Spiralia and Metazoa [61]. These studies would also allow to test if hypotheses about the correlation between collinearity and cluster organization as observed in deuterostomes [23] stand true for protostomes.
Here, we present a comprehensive study of the genomic arrangement and expression of Hox genes in Brachiopoda, a lineage of the Spiralia whose origins date back to the Lower Cambrian [62]. Brachiopods are marine, sessile, filter-feeding animals. They are protected by two dorsoventral mineralized shells and reproduce by external fertilization, often developing through an intermediate, free-living larval stage [63]. In this study, we use two brachiopod species –the ‘articulate’ Terebratalia transversa and the ‘inarticulate’ Novocrania anomala– that respectively belong to the two major brachiopod lineages, thus allowing the reconstruction of putative ancestral characters for Brachiopoda as a whole (Figure 1A). By transcriptomic and genomic sequencing we demonstrate that the Hox complement consists of ten Hox genes in T. transversa and nine in N. anomala. In addition, the ten Hox genes of T. transversa are ordered in a split Hox cluster that differs from the genomic arrangement reported for the brachiopod Lingula anatina [personal communication, Luo and 64]. We show that Hox gene expression is restricted to the ‘trunk’ region of the larva, and is overall neither temporally nor spatially collinear. However, the genes pb (only in T. transversa), Hox3 and dfd show spatially collinear expression in the mesoderm of both brachiopod species. Additionally, the Hox genes lab, scr, antp and post1 appear to be associated with the development of two brachiopod features: the chaetae and the shell-forming epithelium. Altogether, our findings demonstrate that the presence of a split Hox cluster in the Brachiopoda is likely not associated with a temporally collinear expression of Hox genes, which differs from the hypothesized correlation between temporal collinearity and the retention of the vertebrate Hox cluster [14, 23, 59, 60] and suggests that alternative/additional genomic forces might shape Hox clusters during spiralian evolution, such as low genomic rearrangement frequency.
Results
The Hox gene complement of T. transversa and N. anomala
Transcriptomic and genomic searches resulted in the identification of ten Hox genes in T. transversa. In the brachiopod N. anomala, we identified seven Hox genes in the transcriptome and two additional fragments corresponding to a Hox homeodomain in the draft genome assembly. Attempts to amplify and extend these two genomic sequences in the embryonic and larval transcriptome of N. anomala failed, suggesting that these two Hox genes might be expressed only during metamorphosis and/or in the adult brachiopod. Maximum likelihood orthology analyses resolved the identity of the retrieved Hox genes (Figure supplementary 1). The ten Hox genes of T. transversa were orthologous to labial (lab), proboscipedia (pb), Hox3, deformed (dfd), sex combs reduced (scr), lox5, antennapedia (antp), lox4, post2 and post1. The nine Hox genes identified in N. anomala corresponded to lab, pb, Hox3, dfd, scr, lox5, antp, lox4, and post2.
Genomic organization of Hox genes in T. transversa and N. anomala
We used the draft assemblies of T. transversa and N. anomala genomes to investigate the genomic arrangement of their Hox genes. In T. transversa, we identified three scaffolds containing Hox genes (Figure 1B). Scaffold A spanned 81.7 kb and contained lab and pb in a genomic region of 15.4 kb, flanked by other genes with no known linkage to the Hox cluster in other animals. Scaffold B was the longest (284.8 kb) and included Hox3, dfd, scr, lox5, antp, lox4 and post2, in this order (Figure 1B) including the micro RNA mir-10 between dfd and scr. As in scaffold A, other genes flanked the Hox genes, which occupied a genomic region of 76.2 kb. Finally, post1 aligned to various short scaffolds. We could not recover any genomic linkage between the identified Hox genes in N. anomala due to the low contiguity (N50 of 3.5 kb) of the draft genome assembly. Altogether, these data demonstrate that T. transversa has a split Hox cluster broken into three sub-clusters, each of them with an organized arrangement. Importantly, the potential genomic disposition of these three sub-clusters is similar to that observed in other spiralians, such as C. teleta and L. gigantea (Figure 1C), which suggests that the lineage leading to the brachiopod L. anatina experienced genomic rearrangements that modified the ordered and linkage of the Hox genes.
Hox gene expression in T. transversa
To investigate the presence of temporal and/or spatial collinearity in the expression of the clustered Hox genes in T. transversa, we first performed whole-mount in situ hybridizations in embryos from blastula to late, competent larval stages (Figure 2).
Anterior Hox genes
The anterior Hox gene lab was first detected in the mid gastrula stage in two faint bilaterally symmetrical dorsal ectodermal domains (Figure 2Ad, Ae). In late gastrulae, lab expression consisted of four dorsal ectodermal clusters that corresponded to the position where the chaetae sacs form (Figure 2Af, Ag). In early larva, the expression was strong and broad in the mantle lobe (Figure 2Ah, Ai), and in late larvae it became restricted to a few mantle cells adjacent to the chaetae sacs (Figure 2Ij, Ik). These cells do not co-localize with tropomyosin, which labels the muscular mesoderm of the larva (Figure 3A). This suggests that lab expressing cells are likely ectodermal, although we cannot exclude localization in non-muscular mesodermal derivates.
The Hox gene pb was first detected asymmetrically on one lateral of the ectoderm of the early gastrula (Figure 2Bb, Bc). In the mid gastrula, the ectodermal domain located dorsally and extended as a transversal stripe (Figure 2Bd, Be). Remarkably, this domain disappeared in late gastrula embryos, where pb was detected in the anterior mantle mesoderm (Figure 2Bf, Bg). This expression was kept in the early and late larva (Figure 2Bh–Bk; Figure 3B)
Hox3
The gene Hox3 was detected already in blastula embryos in a circle of asymmetric intensity around the gastral plate (Figure 2Ca). In early gastrulae, Hox3 is restricted to one half of the vegetal one, which is the prospective posterior side (Figure 2Cb, Cc). With axial elongation, Hox3 becomes expressed in the anterior mantle mesoderm and in the ventral ectoderm limiting the apical and mantle lobe (Figure 2Cd, Ce). This expression is maintained in late gastrula stages and in the early larva (Figure 2Cf–Ci). In the late larva, Hox3 is detected in part of ventral, internal mantle ectoderm and in the most anterior part of the pedicle mesoderm (Figure 2Cj, Ck; Figure 3C)
Central Hox genes
The Hox gene dfd was asymmetrically expressed on one side of the vegetal pole of the early gastrula of T. transversa (Figure 2Db, Dc). This expression was maintained in the mid gastrula, and corresponded to the most posterior region of the embryo (Figure 2Dd, De). In the late gastrula, dfd becomes strongly expressed in the posterior mesoderm (Figure 2Df, Dg). In the early larva, the expression remained in the pedicle mesoderm, but new domains in the posterior ectoderm and in the anterior, ventral pedicle ectoderm appear (Figure 2Dh, Di). These expression domains are also observed in the late larva (Figure 2Dj, Dk; Figure 3D).
The central Hox gene scr was first expressed in the medial dorsal ectoderm of the mid gastrula (Figure 2Ed, Ee). In late gastrula stages, the expression expanded towards the ventral side, forming a ring (Figure 2Ef, Eg). In the early larva, scr was detected in a ring encircling the most anterior ectoderm of the pedicle lobe and extending anteriorly on its dorsal side (Figure 2Eh, Ei). With the outgrowth of the mantle lobe in the late larva, the expression became restricted to the periostracum, the internal ectoderm of the mantle lobe that forms the shell (Figure 2Ej, Ek; Figure 3E).
The Hox gene Lox5 is expressed on one side of the early gastrula (Figure 2Fb, Fc). During axial elongation, the expression became restricted to the most posterior ectoderm of the embryo (Figure 2Fd–Fg). This domain remained constant in larval stages, where it was expressed in the whole posterior ectoderm of the pedicle lobe (Figure 2Fh–Fk).
The antp gene is weakly detected at the mid gastrula stage, in one posterior ectodermal domain and one dorsal ectodermal patch (Figure 2Gd, Ge). In the late gastrula, the posterior expression is maintained and the dorsal domain extends ventrally, encircling the embryo (Figure 2Gf, Gg). These two domains remained in the larvae: the ectodermal anterior-most, ring-like domain localized to the periostracum, and the posterior domain limited to the most posterior tip of the larva (Figure 2Gh–Gk).
The Hox gene Lox4 is first detected in the dorsal, posterior end of the late gastrula and early larva (Figure 2Hf–Hi). In the late larva, Lox4 is expressed dorsally and posteriorly, although it is absent from the most posterior end (Figure 2Hj, Hk).
Posterior Hox genes
The posterior Hox gene post2 was first detected in mid gastrula stages at the posterior tip of the embryo (Figure 2Id, Ie). This expression was maintained in late gastrulae (Figure 2If, Ig). In early larva, post2 expression extended anteriorly and occupied the dorso-posterior midline of the pedicle lobe (Figure 2Ih, Ii). In late, competent larvae, post2 was detected in a T-domain in the dorsal side of the pedicle ectoderm (Figure 2Ij, Ik).
The Hox gene post1 was transiently detected in late gastrula stages in the four mesodermal chaetae sacs (Figure 2Jf, Jg).
We verified the absence of temporal collinearity in the expression of the Hox genes in T. transversa by quantitative real-time PCR and comparative stage-specific RNA-seq data (Figure supplementary 2).
Hox gene expression in N. anomala
In order to infer potential ancestral Hox expression domains for the Brachiopoda, we investigated the expression of the nine Hox genes of N. anomala during embryogenesis and larval stages (Figure 4).
Anterior Hox genes
The Hox gene lab was first detected at the mid gastrula stage in three bilaterally symmetrical ectodermal cell clusters that appear to correlate with the presumptive site of chaetae sac formation (Figure 4Ad, Ae). The expression in the most posterior pair was stronger than in the two most anterior ones. This expression was maintained in the late gastrula (Figure 4Af, Ag). In larval stages, lab was detected in the two most anterior chaetae sacs of the mantle lobe (Figure 4Ah, Ai), expression that fainted in late larvae (Figure 4Aj, Ak).
The Hox gene pb was asymmetrically expressed already at blastula stages, in the region that putatively will rise to the most posterior body regions (Figure 4Ba). With the onset of gastrulation, the expression of pb extended around the vegetal pole, almost encircling the whole blastoporal rim (Figure 4Bb, Bc). During axial elongation, pb was first broadly expressed in the region that forms the mantle lobe (Figure 4Bd, Be) and later on the ventral mantle ectoderm of the late gastrula (Figure 4Bf, Bg). In early larvae, pb was detected in the anterior ventral mantle ectoderm (Figure 4Bh, Bi). This domain was not detected in late, competent larvae (Figure 4Bj, Bk).
Hox3
The Hox gene Hox3 was asymmetrically detected around half of the vegetal pole of the early gastrulae (Figure 4Cb, Cc). In mid gastrulae, the expression almost encircled the whole posterior area and the blastoporal rim (Figure 4Cd). In addition, a domain in the mid-posterior mesoderm became evident (Figure 4Ce). By the end of the axial elongation, Hox3 was strongly expressed in the posterior mesoderm and weakly in the ventral posterior mantle ectoderm (Figure 4Cf, Cg). Noticeably, the posterior most ectoderm did not show expression of Hox3. This expression pattern was maintained in early and late larval stages (Figure 4Ch–Ck).
Central Hox genes
The central Hox gene dfd was first detected in the posterior ectodermal tip of mid gastrulae (Figure 4Dd, De). In late gastrula stages, dfd was expressed in the posterior ectodermal end (Figure 4Df) and in the posterior mesoderm (Figure 4Dg). Early larvae showed expression of dfd in the posterior mesoderm and posterior mantle ectoderm (Figure 4Dh, Di). This expression remained in late larvae, although the most posterior ectodermal end was devoid of expression (Figure 4Dj, Dk).
The Hox gene scr was only detected in late larval stages, in a strong dorsal ectodermal domain (Figure 4Ej, Ek).
The gene Lox5 was detected asymmetrically around half of the blastoporal rim in early gastrula stages (Figure 4Fb, Fc). During axial elongation, the expression progressively expanded around the blastoporal rim (Figure 4Fd, Fe) and limited to the ventral midline (Figure 4Ff, Fg). In the larvae, Lox5 was expressed in the ventral, posterior-most midline (Figure 4Fh–Fk).
The Hox gene antp was first expressed asymmetrically in one lateral side of the early gastrula (Figure 4Gj, Gk). In the mid gastrula, antp was detected in the dorsal ectodermal mantle in a cross configuration: dorsal midline and the mantle cells closer to the apical-mantle lobe boundary (Figure 4Gd, Ge). In late gastrulae, antp was only expressed in a mid-dorsal ectodermal region (Figure 4Gf, Gg). This expression pattern was also observed in early larval stages, although the size of the domain reduced (Figure 4Gh, Gi). In late larvae, antp was detected in a small mid-dorsal patch and a weak ventro-posterior ectodermal domain (Figure 4Gj, Gk).
We could neither identify nor amplify Lox4 in a transcriptome and cDNA obtained from mixed embryonic and larval stages, suggesting that either it is very transiently and weakly expressed during embryogenesis or it is only expressed in later stages (metamorphosis and adulthood).
Posterior Hox genes
The only posterior Hox gene present in N. anomala, post2, could not be amplified in cDNA obtained from mixed embryonic and larval stages, suggesting that it is not expressed –or at least expressed at really low levels– during these stages of the life cycle. The absence of larval expression of Lox4 and post2 could be related to the lack of the pedicle lobe of craniiform brachiopod larva, which is a characteristics of the lineage [65, 66].
Discussion
The brachiopod Hox complement and the evolution of Hox genes in Spiralia
Our findings on T. transversa and N. anomala reveal an ancestral brachiopod Hox gene complement consistent with what has been hypothesized to be ancestral for Spiralia and Lophotrochozoa on the basis of degenerate PCR surveys [15, 67–69]. This ancient complement comprises eight Hox genes – lab, pb, Hox3, Dfd, Scr, Lox5, Lox4 and Post2 – and has been confirmed by genomic sequencing of representative annelids and mollusks [16, 22, 47], rotifers and platyhelminthes [21, 43–45] and the linguliform brachiopod L. anatina [64]. While T. transversa and L. anatina (N. Satoh and Y.-J. Luo, personal communication) have retained this ancestral Hox complement, N. anomala has apparently lost Post1 (Figure 1).
Our genomic information shows that the Hox cluster of T. transversa is split in three parts, with lab and pb separate from the major cluster and Post1 also on a separate scaffold (Figure 1B). Overall, the cluster extends over 100 kb, which is significantly shorter than those of other lophotrochozoans, such as C. teleta (~345kb) [46] and L. gigantea (~455 kb) [16]. Its compact size is related to short intergenic regions and introns, comparable to the situation observed in vertebrate Hox clusters [23]. The order and orientation of the Hox genes in T. transversa is preserved and more organized than in the Hox cluster reported for the brachiopod L. anatina, which exhibits a genomic rearrangement that placed a portion of the cluster upstream lab and in reverse orientation [64]. Indeed, the split Hox clusters reported so far in lophotrochozoan taxa exhibit all different conformations, indicating that lineage-specific genomic events have shaped Hox gene clusters in Spiralia.
Non-collinearity of Hox expression in T. transversa despite the presence of a split cluster
The analysis of Hox clustering in different animal species together with the temporal and spatial expression patterns of their Hox genes grounded the hypotheses that the regulatory elements required for their collinearity –mostly temporal– maintain the clustered organization of the vertebrate Hox genes and possibly other animals [13, 23, 59–61, 70, 71] (Figure supplementary 3). Although there are cases in which spatial collinearity is displayed in the absence of a cluster, as in the appendicularian chordate O. dioica [24], all investigated clustered Hox genes show at least one type of collinearity that could account for their genomic organization [23, 61] (Figure 6). Since there are exceptions to the spatial collinearity in vertebrates, for instance Hoxa2 and Hoxb2 are expressed more anteriorly than Hox1 genes in the vertebrate hindbrain [72], temporal collinearity is seen as a manifestation of Hox clustering. But whether temporal collinearity is the agent keeping the cluster together, e.g. through enhancer sharing [73], is still subject of debate.
Within Spiralia, this evolutionary scenario appears to be supported by the staggered temporal and spatial expression of the Hox genes in the split cluster of the annelid C. teleta [46]. In the other investigated spiralians, there is only either genomic information (e.g. the mollusks L. gigantea and C. gigas) or expression analysis (e.g. the mollusks G. varia, Haliotis asinina) [16, 47, 52, 58]. Most of these gene expression studies have demonstrated coordinated spatial or temporal expression of Hox genes along the anteroposterior axis of the animal [48, 49, 74] or in organ systems, such the nervous system [52, 58]. However, the absence of studies that can reveal a correlation between the expression of Hox genes and their genomic organization in these animals hampers the reconstruction of the putative mechanisms that preserve Hox clusters in Lophotrochozoa, and thus prevent generalizations about possible scenarios of Hox cluster evolution across all animals.
Our findings robustly demonstrate that the Hox genes of the split Hox cluster of T. transversa overall show neither strictly spatial nor temporal collinearity (Figures 2, 3), and lack quantitative collinearity [61], as it has been shown for example in mouse [75]. These observations are also supported by the absence of a coordinated spatial and temporal expression of the Hox genes in N. anomala (Figure 4). Although a general trend of spatial collinearity is present (e.g. the posterior Hox genes are expressed in posterior tissues), the early expression of Hox3 breaks temporal collinearity in T. transversa, while it is pb that becomes first expressed in N. anomala. In both species, the gene Lox5 is also expressed before Scr, as it is also the case in the annelid N. virens [74]. Ectodermal spatial collinearity is absent in the two brachiopods even when considering the future location of the larval tissues after metamorphosis [76, 77]. The most anterior class gene lab is exclusively expressed in the chaetae of T. transversa and N. anomala, and thus is not affiliated with anterior neural or foregut tissues as in other lophotrochozoans, such as annelids [46, 78]. Similarly, the most posterior Hox gene, Post1, is very transiently expressed in the chaetae sacs, which occupy a mid-position in the larval body. We only detected a strict spatial collinearity in the staggered expression of the Hox genes pb, Hox3 and Dfd along the anterior-posterior axis of the developing larval mesoderm in both T. transversa and N. anomala (Figure 5).
Altogether, the absence of a strict temporal and spatial collinearity in the brachiopod T. transversa, albeit the presence of a split Hox cluster, indicates that temporal collinearity is likely not the underlying factor keeping spiralian Hox genes clustered, as it seems to be the case in vertebrates [14, 23, 59–61]. Therefore, alternative mechanisms might need to be considered. In this regard, why do Hox clusters split in different positions between related species, as seen for instance in brachiopods (this study) and drosophilids [79], but still display similar expression profiles? This might indicate that the control of expression in large split Hox clusters relies more on gene-specific short-range transcriptional control than on a global, coordinated regulation, as seen in the small Hox vertebrate clusters [23, 75, 80]. The conservation of Hox clusters in many animals could then be a consequence of the general conservation of syntenic relationships in their genomes (Figure 6). Our findings thus highlight the necessity of further detailed structure-function analyses of spiralian Hox clusters to better understand the intricate evolution of the genomic organization and regulation of Hox genes in metazoans.
Recruitment of Hox genes for patterning lophotrochozoan chaetae and shell fields
The bristle-like chaetae (or setae) of annelids and brachiopods, and shell valves in mollusks and brachiopods are the most prominent hard tissues found in lophotrochozoan spiralians [63] and provide fossilized hallmarks of the Cambrian explosion [81]. It has been already recognized that the ultrastructural morphology of the brachiopod and annelid chaetae is nearly identical [82–84] and with the placement of brachiopods as close relatives of annelids and mollusks [85], the homology of these structures appeared more likely [86]. In this context, the anterior Hox gene lab is expressed in the chaetae of Chaetopterus sp. [26] and Post1 is expressed in the chaetae of C. teleta, P. dumerilii and N. virens [46, 74]. Our results show that similarly, lab and Post1 are expressed specifically in the chaetal sacs of the brachiopods T. transversa and N. anomala (Figures 2, 4) and follow the different arrangement of the chaetae in both species. Further evidence of a common, and probably homologous, molecular profile comes from the expression of the homeodomain gene Aristaless-like (Arx) and the zinc finger Zic. These genes are expressed at each chaetae sac territory in the Platynereis larva [87], in Capitella teleta [88], and also in the region of the forming chaetae sac territories in T. transversa (Figure supplementary 4). Therefore, the expression of the Hox genes lab and Post1 and the homeodomain gene Arx indicate that similar molecular signature underlays the development of chaetae in annelids and brachiopods. This, together with the evident and striking morphological similarities shared by brachiopod and annelid chaetae, support considering these two structures homologous, and thus, common lophotrochozoan innovations. This would be consistent with placing the iconic Cambrian fossil Wiwaxia, which contains chaetae, as a stem group lophotrochozoan [89].
The protective shell is a mineralized tissue present in brachiopods and mollusks. In the gastropod mollusk G. varia, the Hox genes lab, Post1 and Post2 are first expressed in the shell field, and later is Dfd [57]. In H. asinina also lab and Post2 are related to shell formation [52]. In brachiopods, Dfd is associated to the adult shell in L. anatina [64]. During embryogenesis of T. transversa and N. anomala, however, only Scr and Antp are expressed in the shell fields, but not lab or Post1, which are expressed in the chaetae sacs. This could support the homology of the chitin-network that is formed at the onset of brachiopod and mollusk shell fields. However, the different deployment of Hox genes in the shell fields of brachiopods and mollusks might indicate that these genes do not have an ancient role in the specification of the shell-forming epithelium. However, their consistent deployment during shell development might reflect a more general, conserved role in shaping the shell fields according to their position along the anterior posterior axis.
Conclusions
In this study, we characterize the Hox gene complement of the brachiopods T. transversa and N. anomala, and demonstrate the last common ancestor to all brachiopods likely had ten Hox genes (lab, pb, Hox3, dfd, scr, Lox5, antp, Lox4, post2, post1). Noticeably, brachiopod Hox genes do not show global temporal and spatial collinearity, albeit T. transversa exhibits an ordered, split Hox cluster. Only the genes pb (in T. transversa), Hox3 and dfd (in both brachiopods) show spatial collinearity in the ‘trunk’ mesoderm. In addition, the Hox genes lab and post1, as well as the homeobox Arx, are expressed in the developing chaetae, as also described for other annelid species [46, 53, 74]. These molecular similarities, together with evident morphological resemblances [83], support considering brachiopod and annelid chaetae homologous structures and reinforce considering the fossil Wiwaxia as a stem group lophotrochozoan [89]. Altogether, our findings challenge a scenario in which temporal collinearity is the major force preserving Hox clusters [12, 14, 23, 60, 61], and indicate that alternative/additional genomic mechanisms might account for the great diversity of Hox gene arrangements observed in extant animals.
Material and Methods
Animal cultures
Gravid adults of Terebratalia transversa (Sowerby, 1846) were collected around San Juan Island, Washington, USA and Novocrania anomala (Müller, 1776) around Bergen, Norway. Animal husbandry, fertilization and larval culture were conducted following previously published protocols [90–92].
Hox cluster reconstruction in T. transversa and N. anomala
Male gonads of T. transvesa and N. anomala were preserved in RNAlater (Life Technologies) for further genomic DNA (gDNA) isolation. Paired end and mate pair libraries of 2 kb and 5 kb insert sizes of T. transversa gDNA were sequenced using an Illumina HiSeq2000 platform. First we trimmed Illumina adapters with Cutadapt 1.4.2 [93]. Then, we assembled the paired end reads into contigs, scaffolded the assembly with the mate pair reads, and closed the gaps using Platanus 1.21 [94]. The genomic scaffolds of T. transversa including Hox genes are published on GenBank with the accession numbers KX372775 and KX372776. Paired end libraries of N. anomala gDNA were sequenced using an Illumina HiSeq2000 platform. We removed Illumina adapters as above and assembled the paired end reads with MaSuRCA 2.2.1 [95].
Gene isolation
Pooled samples of T. transversa and N. anomala embryos at different developmental stages (cleavage, blastula, gastrula, mid gastrula, late gastrula, early larva, and late/competent larva) were used for RNA isolation and Illumina sequencing (NCBI SRA; T. transversa accession SRX1307070, N. anomala accession SRX1343816). We trimmed adapters and low quality reads from the raw data with Trimmomatic 0.32 [96] and assembled the reads with Trinity 2.0.6 [97]. Hox genes were identified by BLAST searches on these transcriptomes and their respective draft genomes (see above). First-strand cDNA template (SuperScript™, Life Technologies) of mixed embryonic stages was used for gene-specific PCR. RACE cDNA of mixed embryonic stages was constructed with SMARTer RACE cDNA Amplification Kit (Clontech) and used to amplify gene ends when necessary. All fragments were cloned into the pGEM-T-Easy vector (Promega) and sequenced at the University of Bergen sequencing facility. T. transversa and N. anomala Hox gene sequences were uploaded to GenBank (accession numbers KX372756–KX372774).
Orthology analyses
Hox gene sequences of a representative selection of bilaterian lineages (Supplementary Table S1) were aligned with MAFFT v.7 [98]. The multiple sequence alignment, which is available upon request, was trimmed to include the 60 amino acids of the homeodomain. ProtTest v.3 [99] was used to determine the best fitting evolutionary model (LG+G+I). Orthology analyses were conducted with RAxML v.8.2.6 [100] using the autoMRE option. The resulting trees were edited with FigTree and Illustrator CS6 (Adobe).
Gene expression analyses
T. transversa and N. anomala embryos at different embryonic and larval stages were fixed in 4% paraformaldehyde in sea water for 1 h at room temperature. All larval stages were relaxed in 7.4% magnesium chloride for 10 min before fixation. Fixed samples were washed several times in phosphate buffer saline (PBS) with 0.1% tween-20 before dehydration through a graded methanol series and storage in 100% methanol at −20 °C. Single colorimetric whole mount in situ hybridization were carried out following an established protocol (detailed protocol available in Protocol Exchange: doi:10.1038/nprot.2008.201) [101, 102]. Double fluorescent in situ hybridizations were conducted as described elsewhere [103]. Representative stained specimens were imaged with bright field Nomarski optics using an Axiocam HRc connected to an Axioscope Ax10 (Zeiss). Fluorescently labeled embryos were mounted in Murray’s clearing reagent (benzyl alcohol: benzyl benzoate, 1:2) and imaged under a SP5 confocal laser-scanning microscope (Leica). Images and confocal z-stacks were processed with Fiji and Photoshop CS6 (Adobe) and figure panels assembled with Illustrator CS6 (Adobe). Contrast and brightness were always adjusted to the whole image, and not to parts of it.
Quantitative Hox gene expression in T. transversa
Thousands of synchronous T. transversa embryos collected at 14 specific stages (oocytes, 8h mid blastula, 19h late blastula, 24h moving late blastula, 26h early gastrula, 37h asymmetric gastrula, 51h bilateral gastrula, 59h bilobed, 68h trilobed, 82h early larva (first chaetae visible), 98h late larva (long chaetae, eye spots), 131h competent larva, 1d juvenile, 2d juvenile) were pooled together and preserved in RNAlater (Life Technologies). Total RNA was isolated with Trizol Reagent (Life Technologies). For quantitative real time PCR, total RNA was DNAse treated and preserved at −80 °C. Gene specific primers bordering an intron splice-site and defining an amplicon of 80-150 bp sizes were designed for each gene (Supplementary Table S2). Expression levels of two technical replicates performed in two biological replicates were calculated based on absolute quantification units. For comparative stage-specific transcriptomic analyses, total RNA was used for constructing Illumina single end libraries and sequenced in four lanes of a HiSeq 2000 platform. Samples were randomized between the lanes. To estimate the abundance of transcripts per stage, we mapped the single end reads to the transcriptome of T. transversa with Bowtie, calculated expression levels with RSEM, and generated a matrix with TMM normalization across samples by running Trinity’s utility scripts. Expression levels obtained after quantitative real-time PCR and comparative stage-specific transcriptomics were plotted with R.
Author contributions
A.H. designed the study. A.H., S.M.S. and J.M.M.D. conducted the gene isolation and in situ hybridization studies. J.M.M.D. performed the gene orthology analyses. A.H., J.M.M.D., Y.P. and B.V. collected the stage-specific samples of T. transversa embryos. A.B. and J.M.M.D. isolated the genomic DNA of T. transversa and N. anomala. J.M.M.D. and B.V. did the draft genome assemblies and S.M.S. analyzed the Hox genomic organization. J.M.M.D. performed the stage-specific RNA isolations; A.B. did the quantitative real time PCR experiments, and B.V. conducted the analysis of the stage-specific transcriptomes. A.H. and J.M.M.D. wrote the manuscript. All authors discussed the data and edited the text.
Acknowledgements
We thank the crew of the “Centennial” boat and office stuff at Friday Harbor Laboratories (USA) and the crew of the “Hans Brattström” and “Aurelia” boats at the Espeland Marine Station (Norway) for their invaluable help during animal collections. We also thank Daniel Thiel and Anlaug Boddington for their help with animal collections and spawnings, Daniel Chourrout for his valuable comments on early versions of this manuscript, and Kevin Kocot for the access to entoproct transcriptomes. We are indebted to Yi-Jyun Luo and Nori Satoh for sharing unpublished data from the Lingula anatina genome assembly Version 2. The trip to Friday Harbor Laboratories was funded by a Meltzer Fond grant. The research conducted in this study was funded by the Sars Centre core budget.
References
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.
- 20.
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.
- 40.
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.
- 51.
- 52.↵
- 53.↵
- 54.↵
- 55.
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.↵
- 97.↵
- 98.↵
- 99.↵
- 100.↵
- 101.↵
- 102.↵
- 103.↵