Novel and Conserved Features of the Hox Cluster of Entoprocta (Kamptozoa)

Hox genes are highly conserved developmental genes involved in the patterning of the anterior-posterior axis of nearly all metazoan animals. While Hox genes have been characterized for many bilaterians, several cryptic taxa, often comprising microscopic specimens, have hitherto been neglected. We here present the first combined transcriptomic and genomic Hox gene study for Entoprocta (=Kamptozoa), a phylum of microscopic, sessile, tentacle-bearing animals with unresolved phylogenetic affinities. We identified 10 of the 11 Hox genes commonly found in other lophotrochozoans. The analyses of transcriptomic data of different developmental stages of three species (regenerating stages of the colonial species Pedicellina cernua, budding stages of the solitary species Loxosomella vivipara and embryos of the solitary species Loxosomella murmanica/atkinsae) yielded the Hox genes Labial, Hox3, Lox5, and Post2 in all species. Pb and Dfd were only found being expressed in the colonial species P. cernua. Lox4 was uniquely expressed in the solitary species L. vivipara and L. murmanica/atkinsae. Other homeobox genes belonging to the ANTP-class genes, e.g., ParaHox and NK-like genes, were also found. Thus, in addition to newly identified Hox genes (PceLox2-like & LviPost2-like), Entoporocta show the typical lophotrochozoan Hox pattern besides the loss of the posterior class Hox gene Post1.

Its members are microscopic, sessile, colonial or solitary, mostly marine animals. Their bodies can be subdivided into calyx, stalk and foot [33][34][35]. The calyx comprises the characteristic tentacle crown, which surrounds both, mouth and anus, the U-shaped gut, one pair of protonephridia, the reproductive organs and the cerebral ganglion.
They reproduce asexually by budding or sexually, whereby two different larval types can be found: the lecithotrophic and supposedly basal creeping larval type and the more common planktotrophic trochophore-like swimming larval type [36][37][38][39]. So far, approximately 150 species are known from four families: the solitary Loxosomatidae and the colonial Loxokalypotidae, Barentsiidae and Pedicellinidae [36,40]. Due to environmental conditions and injuries the calyx of Pedicellinidae and Barentsiidae can die off and a new "head" forms from the remaining stalk; alternatively, parts of the stalk are rebuilt prior to calyx regeneration [41][42][43]. For the Loxosomatidae, so far only one species, Loxosomella antarctica, is known to have regeneration capabilities comparable to colonial entoprocts [42].
The phylogenetic position of Entoprocta is still a matter of debate.
Classical morphological and some molecular studies favor a grouping of entoprocts with ectoprocts as sistergroup [37][38]. Other molecular studies comprise entoprocts and cycliophorans as a sistergroup to ectoprocts to form the monophyletic Polyzoa [44][45]. In contrast, the so-called Tetraneuralia-concept (also Sinusoida or Lacunifera) places mollusks and entoprocts as sistergroups, since the creeping-type larva resembles a mosaic of larval and adult molluscan characters, such as the tetraneury of the longitudinal nerve cords or the number of flaskshaped cells in the apical organ [46][47][48][49][50][51].
So far, Hox genes have not been characterized for any entoproct species. However, Hox genes play an important role in determining the body plan, may be used to study and analyze both, the early development in embryos and regeneration processes in adults (see above), e.g. by in situ hybridization experiments, and are also useful characters for phylogenetic studies. We therefore sequenced three transcriptomes of regeneration stages of the colonial species Pedicellina cernua, budding stages of the solitary Loxosomella vivipara, and embryonic stages of the solitary Loxosomella murmanica, in order to reveal the expression of Hox genes during the different developmental processes in these species. In addition, we mind the genome of P. cernua to identify the entire entoproct Hox gene

Journal of Phylogenetics & Evolutionary Biology
cluster in order not to overlook any non-or less expressed Hox genes in species that were analyzed by transcriptomic data only.

Animals and fixation
Adults of the colonial species P. cernua live epizooically on the ectoproct Bugula sp. or the ascidian Styela sp., which inhabit the wharfs of the island Neeltje Jans, The Netherlands. Individuals of P. cernua were removed from their hosts and maintained in glass dishes on a shaker in seawater at a temperature of approximately 16°C. Cultured animals were fed once a week and water was changed ~24 h after feeding. For the collection of different regeneration stages, approximately 60 animals were decapitated and collected after a period of four, six, eight, ten, twelve and fourteen days, fixed in RNAlater and stored at -18°C. For genomic analyses, animals were transferred into 100% ethanol.
Specimens of L. vivipara live on the alga Amphiroa fragilissima in 1.5 m depth in the southern reef of Heron Island, Queensland, Australia. Adults with buds were removed and relaxed in a 1:1 dilution of seawater and 7,14% MgCl 2 for 10 min, since they immediately glue themselves with their foot onto the glass wall of the dish. After relaxation ~100 animals were transferred into RNAlater and stored at -18°C.
Loxosomella murmanica (and L. atkinsae) can be found on Phascolion strombus. This sipunculid species resides in empty shells of the scaphopod Antalis sp. or the gastropod Turritella sp. Thus, the entoprocts were collected by dredging shells from 30 m depth at Gåsö Ränna, Gullmarsfjord closely located to the Kristineberg Marine Research Station (Sweden). Approximately 150 brooding animals were removed from their host and transferred into RNAlater and stored at -18°C.
The adult gross morphology of L. murmanica and L. atkinsae is quite similar. During sampling, a determination of the two species was only possible through their different larval types: L. murmanica develops via the creeping-type larva and L. atkinsae via the swimmingtype larva. According to the amount of animals clearly identified through the larval type and the amount of species used with ambiguous determination, we assume that at least 85% were L. murmanica.

RNA extraction, sequencing and analyses
After storage, extraction of total-RNA of all probes (~50 to 100 individuals per probe) was performed following the instruction manual with the miRCURY RNA Isolation Kit-Tissue (Exiqon A/S, Denmark). DNase I treatment was skipped for minimizing the loss of RNA during additional washes. For the genomic analyses, DNA extraction of approximately 60 individuals of P. cernua was done with the NucleoSpin Tissue XS-Kit (Macherey-Nagel, Germany) following the instruction manual. Quantity and quality of the probes were determined with the Agilent 2100 Bioanalyzer (Agilent Technologies, USA). In preparation for sequencing, cDNA libraries were synthesized for all RNA probes and samples were sequenced paired end with an Illumina Hiseq 2000 (GENterprise Genomics Mainz, Germany). Transcriptome and genome data were analyzed with Geneious version 5.6.6 [52]. Prior to sequence analyses, a database was generated for each sample. Then, sequence search was performed against the amino acid sequence of the Drosophila melanogaster Hox gene Antp (Acc.-Nr. AAA70216.1; 1000 Hits, WordSize 3, Max E-value 1e-1), and the nucleotide sequences of all hits were downloaded and assembled. Hox fragments were identified through GeneBank search (National Center for Biotechnology Information). Longer gene fragments were built with the `map to reference` program of the Geneious software. Still incomplete gene fragments of P. cernua were elongated with the Genome Walker Universal Kit (Clontech) following the instruction manual. Gene fragments of L. vivipara were tried to be extended with the GeneRacer Kit L1502-01 (Invitrogen). Therefore, 4,3 µg of total RNA was used and RACE-ready cDNA was synthesized following the instruction manual. For the 5´-and 3´-RACE a nested PCR was performed with the Dream Taq PCR Master Mix (2X) (Thermo Scientific, Germany), two gene specific primers, and the GeneRacerTM 5´ (Nested) Primer and 3´ (Nested) Primer. The amplification product was gel purified and extracted with the GeneJET Gel Extraction Kit (Thermo Scientific, Germany), and cloned with the StrataClone PCR Cloning Kit (Agilent Technologies, Germany) following the manufacturer´s instructions. Plasmids of relevant clones were purified with the GeneJET Plasmid Miniprep Kit (Thermo Scientific, Germany) and sequenced (StarSEQ, Germany Mci). For this reason, all sequences were brought into the same translation frame. Only entoproct sequences were allowed to have incomplete homeobox sequences. The alignment was converted into Phylip format using the data converter of phylogeny.fr [53]. The ML analysis was done with raxmlGUI version 1.3 [54,55] using GTR + GAMMA model parameters with 5.000 bootstrap replications.

Expression pattern analysis
For each of the three transcriptome data bases, sequence search was performed against the amino acid sequence of the Drosophila melanogaster Hox gene Antp (AAA70216.1), and the nucleotide sequence of all hits were downloaded and assembled. In addition, only blast-hits were considered for this analysis, fitting exactly within the homeodomain. With this restriction we assumed to retrieve approximately one hit per gene expression (that would not be the case if overlaps were allowed; note: incomplete homeodomain sequences of the respective species such as PceHox3 or LviLox5 are excluded by this restriction). We assembled the resulting hits and determined the

Results
The transcriptomic analyses of the three investigated entoprocts resulted in sequences of the Hox genes Labial (LmuLab KP691959, LviLab KP691967, PceLab KP691974), Hox3 (LmuHox3 KP691958, LviHox3 KP691966, PceHox3 KP691972), Lox5 (LmuLox5 KP691961, LviLox5 KP691968, PceLox5 KP691976) and Post2 (LmuPost2 KP691962, PcePost2 KP691978, LviPost2 KP691982) (Figure 1). The respective Labial, Hox3 and Post2 sequences could be clearly identified through an initial search against the NCBI database for non-redundant protein sequences (nr) using blastx and phylogenetic analyses ( Figure  2). Lox5 could be characterized by the "KLTGP"-motif, a C-terminal parapeptide flanking the homeodomain only found in Lophotrochozoa [30,56].  2 and 3). 75 base pairs of the homeobox of an unidentified Hox gene sequence were sequenced by RACE. The corresponding amino acid sequence matches to 100% with the unidentified Hox gene sequence of P. cernua. The Post2 gene of L. vivipara, LviPost2, could be unambiguously identified by GenBank analyses and also by our phylogenetic analyses (Figure 2). An additional posterior class Hox gene, LviPost2-like (KP691984), was uniquely found in L. vivipara. Since the homeodomains of LviPost2 and LviPost2-like have only 43 identical sites (~72%), we assume that LviPost2-like most probably belongs to the Post1 genes. However, LviPost2-like groups together with the Post2 genes and not with Post1 (cf. Figures 2 and 3). An additional Lox4 cognate, LmuLox4B, was found in the transcriptome of L. murmanica/atkinsae. We could not obtain the complete homeodomain sequence of LmuLox4B, but of 46 detected sites, 44 amino acids were identical with Lox4 (~96%).
The genome data of P. cernua supplemented the transcriptome data set with the Hox8 orthologue PceLox4 (KP691975). PceLox4 is separated by an intron of approximately 800bp length. A cognate of PceHox3, PcePost2, PceHox3B (PKP691973) and PcePost2B (KP691979), respectively, could additionally be identified. The homeoboxes of PceHox3 and PceHox3B have 137 identical sites  Lmu

Discussion
The Hox gene cluster of Entoprocta Hitherto, nothing was known about Hox genes in Entoprocta. Here we present the first Hox gene sequences for this phylum. For our analyses, we generated and investigated both, transcriptome data and genomic sequences to avoid any possibility not to obtain the complete set of entoproct Hox genes due to any transcriptional or sequencing bias. In addition, we discuss possible differences in the expression pattern of regeneration, budding and embryonic stages. To this end, we collected up to 150 individuals of three entoproct species and analyzed the corresponding transcriptomes in regenerating, budding and embryonic stages.
Accordingly, we could identify and assign 10 orthologues of the 11 Hox genes known for Lophotrochozoa to Entoprocta. In addition, we detected a so far unidentified Hox gene, Lox2-like, present in two entoproct species, as well as an unknown posterior class Hox gene. The latter unknown posterior Hox gene was solely expressed in budding stages. Thus, this novel Hox gene might be involved in clonal reproduction by budding.

Different patterns of Hox gene expression during different developmental processes in Entoprocta
While several Hox genes (Lab, Hox3, Lox5, Post2) were expressed in all three species, the Hox genes Pb and Dfd could only be found in the transcriptome data of the regenerating stages of P. cernua (cf. Figures 1  and 2). The expression of the PceHox3 cognate PceHox3B during regeneration is questionable, since an assembly of the PceHox3B sequence with the transcriptome data yielded no result. L. vivipara shows an additional posterior class Hox gene, LviPost2-like, which could not be characterized further, as well as one additional central class Hox gene, most probably representing an orthologue of Hox7.
The reason for this individual gene expression pattern might have its origin in the variable expression during the different developmental processes: Pb, Dfd, and Post2 seem to play a central role during regeneration events, Hox3 and Post2-like are highly expressed in budding stages and more than 50% of the expressed Hox genes in embryos belong to Lox4. In any case, only in situ hybridization experiments of numerous developmental stages will show the sites of expression of Hox genes involved in regeneration, embryogenesis or budding, or the persistent expression of individual Hox genes in adult tissues.

Species-specific sequence variation in the homeodomain and cognates
The homeodomain is a 60 amino acid long peptide motif of Hox genes, highly conserved among nearly all metazoans [5]. In all three of the investigated entoproct species, the homeodomain sequence of respective Hox genes shows modifications, similarly but also uniquely found within the Lophotrochozoa. At position 37, labial shows a methionine (M) instead of an alanine (A) in all entoprocts Besides some exceptions coming from some annelids, this alanine is present in all other lophotrochozoan species (blue marks, Figure 3).
The sequence of Hox3/3B also unravels two amino acids uniquely found in Entoprocta. At position 11, a serine (S) is present instead of an alanine (A), and at position 37, a highly conserved leucine (L) is replaced by a methionine (M) or a threonine (T), respectively (see also labial; blue marks, Figure 3).
The Lox4 sequences of Lophotrochozoa and of D. melanogaster usually possess an aromatic tyrosine (Y) or phenylalanine (F) at position 22. In Entoprocta, this aromatic residue is replaced by a nonpolar leucine (L). Within the same sequence, at the positions 9 and 29, respectively, a serine (S) is exchanged by a threonine (T), and a Lysine (K) is replaced by an arginine (R). At the positions 11 and 59, respectively, within the Post2 sequences of the investigated entoprocts, a tyrosine (Y) is 'replaced' by an phenylalanine (F), and a leucine (L) is 'replaced' by an isoleucine (I) (blue and red marks, Figure 3). Remarkably, exchanges in Lox4 at positions 9 and 29 and exchanges in Post2 (position 11) are not common for Lophotrochozoa, but instead are typical for D. melanogaster (Ecdysozoa). But, due to the similar chemico-physiological characteristics of the latter mentioned exchanges (Y→F, S→T, K→R), these exchanges most probably may not affect any functionality instead of just representing isofunctional More strikingly, however, the exchanges observed within labial (A/S/T→M) or Pb (A→S, L→M/T) might affect the functional characters of these Hox genes. While more studies are needed to further assess functional issues, these unique features represent an apomorphy of Entoprocta, which might also be useful for further phylogenetic inferences [61][62][63][64][65][66].

Conclusions
We analyzed the transcriptomes of three entoproct species, one colonial and two solitary forms. In total, we detected 11 different Hox gene sequences and we also identified other homeobox-genes, which belong to the ANTP class homebox genes (Extended Hox, ParaHox, and NK-like). A definite assignment of a Lox2 orthologue was not possible. Instead, we found a closely related Lox2-like gene. A Post1 orthologue could not be found, even by screening the genomic data of P. cernua. Thus, we assume that Post1 was lost before, during or after the evolutionary emergence of Entoprocta. Nevertheless, the presence of typical lophotrochozoan Hox genes such as Lox5 and Post2 further corroborates Entoprocta as being a member of the Lophotrochozoa.
While Labial, Hox3, Lox5, and Post2 were present in the transcriptomes of all investigated species, we only found Pb and Dfd in the transcriptome of the colonial species P. cernua and Lox4 in the transcriptomes of both solitary species, L. vivipara and L. murmanica/ atkinsae. Our findings clearly reflect the specificity and accuracy of the controlled expression pattern and recruitment of different Hox genes for different processes also in Entoprocta.
In P. cernua and L. vivipara, we additionally found a yet unidentified Hox gene. We termed this gene PceLox2-like, because our phylogenetic analyses unraveled its closest relationship to Lox2. Accordingly, PceLox2-like most probably represents a novel central class Hox gene that to date is unique to Entoprocta. In addition to PceLox2-like, we also detected a so far unknown posterior class Hox gene, which is highly expressed in budding stages of L. vivipara. We therefore assume that this gene, besides others (e.g. Hox3), most probably plays a major role during the budding processes and thus should be investigated more intensely in the near future. The detailed comparisons of the individual entoproct Hox genes revealed some intriguing substitutions within the homeodomain of the three investigated entoproct species that are unique among the Lophotrochozoa. Whether this might have been a driving force for Entoprocta splitting off from its lophotrochozoan sister group or whether this constitutes a later event that occurred after the establishment of the phylum remains a matter of further studies.