As well as these spatial domains of expression and their association with particular tissues and organs there is a notable correlation between the timing of gene activation and the genomic organization of the ParaHox genes. This is now clearly exemplified by the comparison between the echinoderms S. purpuratus and P. miniata. In the intact ParaHox cluster of P. miniata the first gene to be activated is PmCdx, followed by PmLox, with the final gene PmGsx not being expressed in the early larval stages (the bipinnaria) examined by Annunziata et al. [2] (note, the low level PmGsx expression observed by Annunziata et al. is from a maternal contribution and so is not provided from the embryonic ParaHox cluster and is regulated via a different mechanism than whatever produces activation of the embryonic ParaHox genes). This order of expression (Cdx first, Xlox second and Gsx last) matches exactly that of chordates like amphioxus and Xenopus [5], and so presumably reflects the order of expression in the last common ancestor of the deuterostomes. This contrasts with the situation in the purple sea urchin, in which the ParaHox cluster has broken apart and SpLox is activated first, followed by SpGsx and finally SpCdx (see Figure five of [6]).
Such a pattern of an intact ParaHox cluster coinciding with temporal colinearity, or similarly a broken ParaHox cluster corresponding to absence of temporal colinearity, is now looking ever more robust. Particularly so since the hemichordate P. flava now provides us with an example of an intact, ordered cluster that does not have complete spatial colinearity, but does have temporal colinearity [3]. This hemichordate thus highlights the tighter relationship of temporal rather than spatial colinearity with intact, ordered clusters. Intriguingly, this pattern of intact clusters correlating with the presence of temporal colinearity also seems to extend to the Hox gene cluster. This may well reflect the paralogous relationship between the Hox and ParaHox clusters and potentially results from the mechanism that is responsible for temporal colinearity being homologous between the Hox and ParaHox clusters. Obviously more data are required to test this hypothesis and exclude the alternatives: either Hox and ParaHox temporal colinearity arose from distinct mechanisms, or, if there is a common mechanism, then it was co-opted into Hox regulation independently of its co-option into ParaHox regulation. Regardless of which of these alternative evolutionary scenarios is accurate, it seems extremely likely that understanding Hox regulatory mechanisms will inform our understanding of ParaHox mechanisms, and vice versa.
There are already some intriguing similarities, particularly centered on the role of retinoic acid (RA) signaling. Some of the earliest data on regulation of Hox genes revealed a role for RA in sequential temporal activation (for example, in human cell culture [7]), and the direct regulation of Hox genes by RA is well established. Intriguingly, RA regulates all of the ParaHox genes in amphioxus [5]. A link between RA signaling and intact Hox clusters has been proposed [8], which could just as well extend to the ParaHox genes.
Elaborating the precise mechanisms of RA signaling and its role in the regulation of Hox and ParaHox regulation thus has the potential to reveal the basis for temporal colinearity and the evolutionary forces that constrain the integrity of both Hox and ParaHox clusters, possibly entwined with chromatin regulation and progressive movement of cluster regions between inactive and active conditions [9]. We must tread with caution, however, as a distinction must be made between global, pan-cluster regulatory mechanisms as distinct from gene-specific, local mechanisms. And RA may well be involved with both. For example, it is clear that RA regulates Hox1 in Ciona intestinalis, whose Hox cluster is largely dispersed, but Cañestro and Postlethwaite [8] propose that this represents a secondarily derived mode of Hox regulation in Ciona, which is clearly acting at a gene-specific level. We also need to tease apart the mechanisms producing spatial and temporal control of Hox/ParaHox genes, which can be distinct at least in some contexts in mice [10], and understand these mechanisms in a variety of species. Potentially RA is involved in distinct mechanisms and had an ancient role, since the genes involved in RA signaling are now known to be widespread across the animal kingdom [11]. With this new sea star and hemichordate data the prospect is raised that ambulacrarians could be key systems contributing to this endeavor, with their relatively wide accessibility, abundant embryo and larval material, and a variety of intact versus disorganized and dispersed Hox and ParaHox clusters.