What was told by polychaetes about UrBilateria?

Our attempts to reconstruct traits of bilaterian’s common ancestor with the aid of modern life model objects are like the wish to reconstruct acorn’s appearance with the aid of oak’s saw cut. However, the search of similarities and differences in development programs of polychaetes, arthropods and vertebrates isn’t useless. Within the gene regulatory network we can separate circuits, which are related with bilateral body architecture too closely and so cannot change. One of such hyperconservative functional control lever for bilaterian development is the Hox-cluster. Its most prominent feature is the colinear transcription. The spatial colinearity or its trace is detected almost always, even if the cluster is scattered (atomized) within a genome (Seo et al., 2004). The temporal colinearity is strictly associated with cluster’s continuity (Monteiro and Ferrier 2006) and is obviously shown only in vertebrates with their compact and ordered clusters (Denis Duboule called them “organized” (Duboule, 2007)). Not only colinearity order, but even local exceptions can tell about ancestral condition of the cluster and about control ways at the dawn of Bilateria.

In basal polychaete Chaetopterus there was shown spatial and temporal transcription colinearity of first five Hox-genes in larval segments. In Capitella spatial colinearity of Hox-expression is held at all ontogenesis stages, but strict temporal colinearity is doubtful because its development staging has time boundaries that are too wide to analyze a dynamical Hox-pattern in details. Besides Capitella has gap in the cluster (Frobius et al., 2008). In nereid polychaetes there was revealed a division of Hox-expression at two stages. Part of the genes (Hox1, Hox2, Hox3, Hox4, Hox5, Lox5 and Post2) are patterning larva’s four-segmented body, colinearly in space but not in time. These genes together with remaining Lox7, Lox4 and Lox2 begin to work in juvenile nereid by absolutely different rules (Kulakova et al., 2007; Bakalenko et al., 2013). If we assume Chaetopterus basality, then it is a secondary situation. From the other side, we don’t know, how acts all Hox-genes complex of Chaetopterus at the stage of segmented larva, and we know nothing about CH-cluster work in postlarval segments (tagma C).

An ontogenetic difference in nereid’s Hox-genes expression patterns is too obvious and provokes a question. Are there functional traces, which connect a structural Hox-cluster evolution with an evolution of development programs?

According to molecular phylogeny data, a last common Bilateria ancestor had at least seven Hox-genes: five anterior (PG1-5), one central (Hox6-8/Antp/Hox7) and one posterior (Hox9-14/Abd-B/Post2-1) (deRosaetal., 1999; Hueber et al., 2013). If this scenario is true, then segmented larva forming in Avi and Pdu is accompanied by expression of Hox-genes, which belong to most ancient paralog groups (Kulakova et al., 2007).

In polychaete’s larvae were found single exceptions of common colinearity rule and additional domains of some Hox-orthologs transcription. Such secondary gene’s involvement into the unusual morphological program is conventionally called “co-option” (True and Carroll, 2002). For example, Post1 is expressed in chaete-bearing sacs in nereids and in Capitella (Kulakova et al., 2002; Frobius et al., 2008). The evolutionary age of this co-option is comparable with the age of common ancestor for brachiopods and annelids, because in brachiopod Terebratalia transversa the Hox gene Post1 also was transiently detected in late gastrula stages in the four ectodermal chaetae sacs of larva (Schiemann et al., 2017).

The second important exception for common spatial colinearity rule is a floating boundary of PG2 expression. It is known, that CH-Hox2 has an anterior expression boundary rostral to that of CH-Hox1. The same relation exists for PG2-orthologs in Capitella, Alitta and Platynereis (Frobius et al., 2008; Kulakova et al., 2007; Bakalenko et al., 2013). In amphioxus Branchiostoma floridae (Cephalochordata) Hox-genes are activating consecutively, but Amphi-Hox2 violates the spatial colinearity rule and its anterior expression border is shifted forward relatively to Amphi-Hox1 (Wada et al., 1999). This feature was noted for Hox2-orthologs of all examined vertebrates (Krumlauf, 1993; Prince et al., 1998). In Drosophilapb domain also gets out of common register, but its anterior boundary lies more posteriorly than lab and Dfd (Rogers and Kaufman, 1997). It is interesting, that Hox-genes studies in Cnidaria (Hydrozoa, Clytia) suggest, that PG2 and PG3 may be appeared earlier than PG1 (Quiquand et al., 2009). If so, then early and more anterior expression of Hox2 and Hox3 may be a trace of their true ancient function. Some anterior Hox-genes of polychaetes take part in foregut development program. Surprisingly, but in vertebrates HoxB1 - HoxB5 genes are expressed in overlapping domains in the developing foregut (Bogue et al., 1996) and lancelet’s Amphi-Hox2 is expressed preoral pit (Wada et al., 1999).

Thus, some Hox-genes have several conservative pan-bilaterian functions, which aren’t related directly with their axial regionalizing activity within unite vector tool – the Hox-cluster. It is possible, that while calling these functions as a “co-options”, we bend the truth, because they could emerge at the earliest stage of Bilateria evolution. If it is true, then co-option of these genes is a coordinated expression within Hox-cluster, but not a specification of separate cell lines. To all seeming, UrBilateria already had time- and space-ordered Hox-genes transcription, and it depended upon cluster’s continuity. If this wasn’t the case, we wouldn’t find undivided clusters in different animal lines. On the other side, this coordinated common regulation was accompanied by individual regulation of separate genes, which are probably most ancient. All exceptions of colinear rule (including complete Hox-function change or additional pan-bilaterian transcription domain) most often are revealed for PG1-PG5.

There is another important question: in what germinal layer began its activity the ’Hox-cluster of UrBilateria? It is known that in echinoderms and in vertebrates Hox-genes start their work in mesoderm (Arenas-Mena et al., 2000; Barak et al., 2012), whereas in arthropods - in ectoderm (Hughes and Kaufman, 2002). In some cases Hox-genes begin expression in ectoderm and in mesoderm almost simultaneously (Brena et al., 2006). Spiralia could shed light upon this question. During the Capitella development, Hox-genes are sequentially activated in the larval ectoderm (Frobius et al., 2008). In the case of Chaetopterus, Hox patterning is more complex. Hox-genes are activated in the posterior part of the L1 larva, according to the temporal colinearity rule, and later Hox expression translates into spatial colinear pattern in the nervous system as the larva elongates (Irvine and Martindale, 2000). Recently we discovered early cryptic expression of Hox2, Hox4 and Lox5 in mesodermal bands of Avi (Kulakova et al., 2017). This short-time expression is initiated in embryo and in early trochophore (Fig.2c). It is colinear in space and in time and comes before same orthologs expression in ectoderm. The difference between time of initiation for each of three genes is approximately two hours, and a synchronous culture of slowly developing larvae is necessary “to catch” the transcription begin. It isn’t impossible that in other polychaetes a similar early transcription stage can be “caught”. Besides the early staggered expression of the Hox-genes pb, Hox3, and Dfd was found along the anteroposterior axis of the developing larval mesoderm in two brachiopod species - Terebratalia transversa and Novocrania anomala (Schiemann et al., 2017). In brachiopods Hox-expression colinearity signs are tracked only at mesoderm level.