Poriferan ANTP genes: primitively simple or secondarily reduced?

Abstract

Using in silico techniques, Larroux et al. (2007) have recently shown that the demosponge Amphimedon queenslandica has only seven different NK genes, but no Hox, Parahox, or EGH genes, consistent with all previous PCR surveys that have been performed on demosponges (reviewed in GarciaFernandez 2005). They further argued that this reduced set of ANTP genes is primary, and is not the result of secondary loss. Here, we argue instead that their gene tree, as does ours, supports the opposite conclusion, namely that demosponges in particular, and possibly most basal animals in general, have secondarily lost myriad transcription factors including many NK genes, and at least one Hox/Parahox/EGH gene. This observation has implications for our understanding of the genetic complexity of the last common ancestor of all living metazoans, the monoversus paraphyly of sponges, and the early evolutionary history of animals. In order to decide between the two competing hypotheses, primitively simple versus secondarily reduced, it is necessary to distinguish between the gene duplication events that gave rise to the individual paralogy groups versus the speciation events that led to extant demosponges and eumetazoans (cnidarians, protostomes, and deuterostomes). As outlined by Simionato et al. (2007; see also Ryan et al. 2006), and as shown here (Fig. 1, A–D), each hypothesis makes different predictions about the order of gene duplication versus speciation events. In the former hypothesis, that of primitive simplicity, gene duplication events follow the speciation event between sponges and eumetazoans giving two genes per eumetazoan taxon (primitively) for each single demosponge gene (Fig. 1A). When drawn as a gene phylogeny, with the nodes representing either gene duplication events (magenta and lime green) or speciation events (black), the two gene duplication events clearly follow the speciation event (Fig. 1B). If, however, the gene duplication events preceded the speciation event between sponges and eumetazoans, and was followed by secondary loss of some of these genes in the sponge, as illustrated in Fig. 1C, then the resulting gene tree should show sponge orthologues to some, but not all, of the eumetazoan genes (Fig. 1D). This is made especially clear when one of the deuterostomes genes is secondarily lost (indicated on Fig. 1, A and C with an ‘‘X’’)Fnote that the cnidarian green gene is now the sister gene of its protostome orthologue, and given the species tree there is no way that deuterostomes are primitive with respect to the absence of this gene. Simionato et al. (2007) showed a clear example of gene duplication following the poriferan/eumetazoan speciation with the basic Helix–Loop–Helix (bHLH) genes: ARNT and Bmal. In their example (see their Fig. 6), the single gene from A. queenslandica is equally related to the two eumetazoan genes, ARNT and Bmal. This is seen with other bHLH genes including the TF4/MLX group, and the HIF/Sim/Trh group. A similar pattern is found with the NK2 cluster. Eumetazoans primitively have three NK2 genes:NK2.1, NK2.2, and NK2.5, whereas the demosponge has a single gene equally related to all three (Fig. 1E, magenta; see also Fig. S3 in Larroux et al. 2007). Hence, rather than sponges losing NK2 genes, it appears that the gene duplication events followed the speciation event between demosponges and eumetazoans, as illustrated in Fig. 1, A and B, and the presence of a single NK2 gene in calcisponges (Manuel and Le Parco 2000) is consistent with this hypothesis. However, Amphimedon also shows clear examples of secondary loss of both bHLH genes and ANTP genes. With respect to the NK genes, demosponge orthologues of Msx, NK6, and Hex are supported with varying degrees of precision, whereas three others, Tlx, BarX1, and BarH, are found but not supported450% by bootstrap analysis or495% by posterior probabilities (Fig. 1E; Fig. S3 in Larroux et al. 2007). Nonetheless, in both our analysis and in the analysis of Larroux et al. (2007), each individual demosponge NK gene clusters with a single eumetazoan NK gene, and in no case does the sponge gene group with more than one eumetazoan NK gene (Fig. 1E, red and brown). The point is well exemplified with NK2 and NK3. Because the presence of a single NK2 gene in the demosponge is moderately well supported, as is the sister grouping between the NK2 and NK3 families (Fig. 1E, and Larroux et al. 2007), the sponge must have secondarily lost the NK3 gene because the gene duplication event giving rise to NK2 and NK3 preceded the speciation event between demosponges and eumetazoans (see Fig. 1, C and D). Similarly, given that an Msx gene is found in the EVOLUTION & DEVELOPMENT 9:5, 405–408 (2007)