Hox gene complement in Phoronida
Similar to the results of the investigation of P. australis genome, we identified eight Hox genes in Ph. harmeri, which represent single copies of the conserved orthologues of the spiralian Hox genes (Figs. 1C, 2). Luo et al.(89) reported that P. australis lacks scr and post1 orthologues and we also did not identify orthologues of those two genes in the transcriptome of Ph. harmeri, strengthening the idea they were already absent in the common ancestor of all phoronids.
In their paper Luo et al.(89) suggested that scr, which is expressed in the shell forming tissues of brachiopods(16, 40), might be lost in Phoronida due to the evolutionary reduction of the shell in this clade. Such interpretation is in accordance with paleontological data, as a fossil cambrian tommotiid, Eccentrotheca sp., which has been proposed as a stem group phoronid(98, 99), possessed a mineralized external tube-shaped skeleton. Recent studies favor a sister group relationship between phoronids and ectoprocts(52-55, 100), the latter of which possess a mineralized external skeleton, similar to brachiopods. However, the Hox gene survey using degenerate polymerase chain reaction primers in the ectoproct Crisularia (Bugula) turrita did not retrieve a scr sequence(101), which questions the possible correlation between loss of this gene and the reduction of shell secreting tissues in phoronid lineage. Yet, since it is difficult to recover the full hox complement with degenerate polymerase chain reaction, further studies on bryozoan hox genes, utilizing genomic or transcriptomic data, are needed to ascertain whether scr is truly missing.
The gene that was identified as lox2 by Luo et al.(89) in the genome of P. australis (and its orthologue in Ph. harmeri) was recovered in our gene orthology analysis as orthologue of antp (Fig. 2). Inspection of the phylogenetic tree available in Luo et al. shows that the assessment of the orthology of this gene was tentative, since the gene was actually placed outside of the well-defined clade of lox2 in their analysis(89). Identification of this gene as antp instead of lox2 is further supported by its position in the genome of P. australis, which corresponds to the antp position in the spiralian species with conserved, organised Hox clusters (Fig. 1C). Additionally, alignment of those phoronid genes with antp and lox2 shows that they lack typical signatures of lox2(92) and instead are more similar to the antp sequence (Additional File 1: Fig. S1). Consequently, both phoronid species lack an orthologue of lox2, an absence, which is apparently shared by Phoronida with other Lophophorata(16, 89, 90, 101) as well as with some other Spiralia – i.e. Rotifera(34, 102) and Platyhelminthes(42, 103). Lox2 was originally described from leeches(104, 105) and later proposed as an evolutionary innovation of Lophotrochozoa ((92), sensu = Spiralia(106)). However, its orthologues are so far identified only in annelids (e.g. (27, 46, 92, 104, 105, 107, 108)), nemerteans(89), molluscs (e.g. (30, 36, 41, 92, 107, 109-112)) and possibly kamptozoans(113) (however, in the latter the lox2-like sequence lacks most of the residues considered as lox2 signature; Additional File 1: Fig. S1). This indicates that lox2 evolved only after split of the common ancestor of those clades from remaining Spiralia and does not belong to the ancestral hox complement of all Spiralia(16). Whether the absence of lox2 in lophophorates is plesiomorphic or represents an evolutionary reversal depends on the position of Lophophorata within Spiralia, which is still debatable and not fully resolved(52-55, 100).
Hox genes in Phoronida do not show traces of collinear expression
When assuming the presence of a similar gene order in the Hox cluster of Ph. harmeri as in P. australis then the former does not show any traces of temporally or spatially collinear expression of Hox genes (Fig. 4). This is in stark contrast to other Spiralia, in which at least some of the Hox genes show staggered expression along A-P axis (e.g. (16, 23, 27, 31, 35-37, 39, 41, 45)). The lack of collinear Hox expression in phoronids is especially intriguing taking into account that P. australis has highly organized Hox cluster and collinear expression (especially in its temporal aspect) has been proposed as a main evolutionary factor responsible for conservation of Hox cluster organization(9, 11-16, 49). Therefore, either another mechanism is responsible for Hox cluster conservation in Phoronida or the two discussed phoronid species vary greatly in the cluster organization and/or Hox gene expression patterns.
Six out of eight identified Hox genes are expressed in the metasomal sac (pb and lox4 being the only two, which expression was not detected in the structure) and already at the stage of 8-tentacle actinotrocha some of those genes (lab, dfd, antp, post2) show differentiated expression in a particular region of the sac (Fig. 5), although without any clear pattern along the future A-P axis. However, it is possible that in the competent larvae (at the 24-tentacle stage, when the metasomal sac is a fully formed, elongated structure; (81, 82)), the expression of particular Hox genes is restricted to the different regions of the trunk rudiment and shows some traces of staggered expression along the future A-P axis of the worm body. Hence, the future investigation of Hox expression in competent larvae and freshly metamorphosed juveniles can reveal spatial collinearity obliterated in the early stages of metasomal sac development or eventually confirm a lack of collinear Hox expression throughout entire development of phoronids.
Germ layer-specific expression of Hox genes in Spiralia
Although Hox genes in Bilateria are predominantly expressed in the ectoderm (including nervous system) and their ectodermal expression is often considered as an ancestral feature(14, 28, 34), in various spiralian species certain Hox genes are also expressed in mesoderm, endoderm and clade-specific structures like chaetal sacs or shell fields (e.g. (16, 23, 24, 27, 29, 31, 35, 36, 39-41, 46); Tab. 1). Inclusion of the data on Hox expression in Phoronida gives some new insight into the understanding of the evolution of germ-layer specific Hox expression in Spiralia. Ph. harmeri, similar to two investigated brachiopod species(16, 40), seems to lack expression of any of the Hox genes in the nervous system, a peculiarity that might actually represent an apomorphy of Lophophorata (Tab. 1). Three of the Hox genes – pb, hox3 and dfd – were shown to be differentially expressed along the A-P axis in the mesoderm of brachiopod larvae(16). Out of those three genes, only pb (which mesodermal expression is actually lacking in craniiformean Novocrania anomala(16)) is expressed mesodermally in Ph. harmeri, indicating that cooption of hox3 and dfd into mesoderm patterning occurred after the split of brachiopods and phoronids. Comparison of Hox gene expression across Spiralia (Tab. 1) allows the observation that pb is mesodermally expressed in many species and it is likely that mesodermal expression of pb represents an ancestral condition in Lophotrochozoa (sensu stricto(106)). On the other hand, the expression of lox4 in the digestive system of Ph. harmeri is a peculiar and derived feature as this gene is expressed in other Spiralia in ectoderm, nervous system or mesoderm. In general, among investigated Spiralia, the Hox genes are rarely expressed in the digestive system (Tab 1.).
Table 1. Expression of Hox genes in spiralian species.
species
|
clade
|
reference
|
Hox gene
|
lab
|
pb
|
hox3
|
dfd
|
scr
|
lox5
|
antp
|
lox4
|
lox2
|
post2
|
post1
|
Phoronopsis harmeri
|
Phoronida
|
this study
|
ectoderm, mesoderm
|
mesoderm,
nephridia
|
ectoderm
|
ectoderm
|
gene absent
|
ectoderm, mesoderm
|
ectoderm
|
intestine
|
gene absent
|
ectoderm
|
gene absent
|
Terebratalia transversa
|
Brachiopoda
|
(16, 40)
|
chaetal sac
|
ectoderm, mesoderm
|
ectoderm, mesoderm
|
ectoderm, mesoderm
|
shell field
|
ectoderm
|
ectoderm
|
ectoderm
|
gene absent
|
ectoderm
|
chaetal sac
|
Novocrania anomala
|
Brachiopoda
|
(16)
|
chaetal sac
|
ectoderm
|
ectoderm, mesoderm
|
ectoderm, mesoderm
|
shell field
|
ectoderm
|
ectoderm
|
unknown
|
gene absent
|
unknown
|
gene absent
|
Capitella teleta
|
Annelida
|
(46)
|
ectoderm, nervous system
|
ectoderm, nervous system
|
ectoderm, nervous system
|
ectoderm, nervous system
|
ectoderm, nervous system
|
ectoderm, nervous system
|
ectoderm, nervous system
|
ectoderm, nervous system
|
ectoderm, nervous system
|
ectoderm, nervous system
|
chaetal sac
|
Alitta virens
|
Annelida
|
(27, 31)
|
ectoderm, chaetal sac, nervous system
|
ectoderm, mesoderm
|
ectoderm
|
ectoderm, nervous system
|
ectoderm, nervous system
|
ectoderm, nervous system
|
ectoderm, nervous system
|
ectoderm, nervous system, mesoderm
|
ectoderm, nervous system
|
ectoderm, nervous system, mesoderm, endoderm
|
chaetal sac
|
Chaetopterus variopedatus
|
Annelida
|
(23)
|
ectoderm, nervous system
|
ectoderm, nervous system, mesoderm
|
ectoderm, nervous system
|
ectoderm, nervous system
|
ectoderm, nervous system
|
unknown
|
unknown
|
unknown
|
unknown
|
unknown
|
unknown
|
Acanthochitona crinita
|
Mollusca
|
(35, 36)
|
ectoderm, mesoderm
|
mesoderm
|
mesoderm
|
ectoderm, mesoderm
|
mesoderm
|
ectoderm, mesoderm
|
mesoderm
|
ectoderm, mesoderm
|
mesoderm
|
ectoderm, mesoderm
|
gene absent
|
Antalis entalis
|
Mollusca
|
(39)
|
shell field, nervous system
|
ectoderm, nervous system
|
ectoderm, shell field, nervous system, mesoderm
|
nervous system, mesoderm
|
nervous system, mesoderm
|
ectoderm, nervous system
|
gene absent
|
ectoderm, nervous system, mesoderm
|
unknown
|
ectoderm, nervous system
|
ectoderm, shell field, nervous system
|
Lottia goschimai
|
Mollusca
|
(41)
|
shell field, nervous system
|
shell field, nervous system
|
shell field, nervous system
|
shell field, nervous system
|
shell field, nervous system
|
shell field, nervous system
|
nervous system
|
shell field, nervous system
|
ectoderm, shell field, nervous system
|
shell field, nervous system
|
nervous system
|
Brachionus manjavacas
|
Rotifera
|
(34)
|
gene absent (?)
|
nervous system
|
nervous system
|
nervous system
|
unknown
|
nervous system
|
gene absent (?)
|
gene absent (?)
|
gene absent (?)
|
gene absent (?)
|
gene absent (?)
|
Hox gene expression and the nature of actinotrocha larvae
We showed that in Ph. harmeri Hox genes are not expressed during embryogenesis, when the larval body is formed, but instead they are expressed mainly in prospective adult structures, namely in the metasomal sac (which will contribute to the adult trunk epidermis), posterior mesoderm (which contributes to the mesodermal structures in the adult trunk), the small posterior portion of the endoderm (which during metamorphosis descent into the trunk rudiment forming the loop of the U-shaped intestine) and the larval telotroch. In most of the investigated Bilateria, Hox genes are already expressed during early developmental stages and, if a biphasic life cycle is present, they are involved in the formation of both larval and adult body plans (e.g. (16, 27, 29-31, 40, 41, 45, 46, 48)). However, there are some animals that, similar to phoronids, deviate from this general pattern. Specifically, in pilidiophoran nemerteans(37) and indirectly developing hemichordates(38), the larvae develop without expressing any of the Hox genes, which instead patterns only the adult body rudiment.
Two evolutionary processes have been proposed to explain these observations. According to the first hypothesis, based on the results from pilidiophoran nemerteans, the new larval form, a pilidium, was intercalated into to the ancestral life cycle of gradually developing nemertean(37, 45). This intercalation of a larval form caused Hox gene patterning to only be retained during development of the adult worm. In contrast the new larval form, in which the body axis is not aligned with the adult one, uses another molecular mechanism to provide primary positional information to the cells of the developing body(37, 45).
Another concept was proposed to explain the phenomenon observed during larval development of a hemichordate Schizocardium californicum(38, 91). Although metamorphosis in this species is not so drastic(114) and the body axes of both stages are congruent, the larva develops without expression of any Hox genes. Instead, they are expressed only late during larval development and only in the most posterior region of the competent larvae, from which the trunk of the juvenile worm will develop during metamorphosis(38, 114). Because the larva expresses genes that are usually expressed in the bilaterian head throughout its body, the so-called “head larva”-hypothesis was proposed which states that the larval body represents the homologue of only the head region of the future animal, while the trunk is added later during post-larval development(38). It has been proposed that ancestrally in Bilateria Hox genes were involved only in the patterning of the trunk, while head developed from the anterior, Hox-free region, the condition, which is still retained in numerous bilaterian lineages(38, 45, 89, 93, 94). That would explain why tornaria, as a larva composed solely of the head, develops without expression of the Hox genes, which become activated only after the onset of trunk development and pattern only the adult body(38).
Both of those hypotheses (intercalation and “head-larva”) might be applied to explain the Hox expression patterns we observed in Ph. harmeri. According to the first hypothesis, the specific actinotrocha larva would represent an evolutionary novelty in the life cycle of phoronids, which was intercalated in the phoronid lineage and that is why it is not patterned by an ancestral Hox gene system. Such an idea is supported by the fact, that the actinotrocha body plan does not bear obvious homology to those of any other spiralian larvae(80, 115-117). Additionally, similar to the case of pilidium, most of the larval tissues are lost during the drastic metamorphosis event and the larval A-P axis is not aligned with the juvenile one(60, 72, 77, 81, 82). Moreover, the actinotrocha is lacking in P. ovalis(60), which is the sister species to all remaining phoronids(62-64), suggesting that the actinotrocha was not even present in the most recent ancestor of all Phoronida, but instead appeared after the split between P. ovalis and the remaining phoronids.
On the other hand, from the morphological point of view, the tentacles of actinotrocha larvae correspond, in case of Ph. harmeri, to the tentacles of the lophophore in the adult worm ((73, 82, 116); Fig. 1B), and the adult lophophore exhibits the molecular signature of a bilaterian head(89). As tentacles are positioned posteriorly in the early actinotrocha, one can conclude that on a morphological basis the early actinotrocha is mostly composed of the head region. Following such interpretation, all of the Hox genes are expressed in the structures that will contribute to the adult trunk tissues but are not expressed in the developing future head (and hence in the largest portion of the larval body). Accordingly, based on a body region specific transcriptome, it has been demonstrated that in adults of P. australis Hox genes are not expressed in the lophophore, while their expression is detectable in the trunk and posterior ampulla(89). Similarly, in rhynchonelliformean and craniiformean brachiopods none of the Hox genes are expressed in the larval anterior lobe(16, 40), which contributes to the lophophore after metamorphosis(40, 116). A lack of Hox expression in the adult lophophore tissue (as opposed to the remaining body regions) was also shown for the linguliformean Lingula anatina, based on the tissue specific transcriptomics(89). Additionally, our study shows that two of the Hox genes (lox5 and post2) are expressed in the telotroch, which represent a truly larval structure, that is lost during metamorphosis(73, 82), therefore Hox genes are indeed, albeit to only a limited degree, involved in larval development. Hox gene expression in the larval telotroch is a result of the telotroch representing a truly “posterior” structure, which belongs to the post-head body region even in the earliest, “head dominated” actinotrocha. The “head larva” interpretation is additionally strengthened by our results of the expression of several head-specific genes in Ph. harmeri. Those genes are broadly expressed in the early larvae and 8-tentacle stage, but only in the structures located anteriorly to the Hox-expressing territory (Fig. 6Z), resembling conditions in developing tornaria(38).