Origin and diversification of the basic helix-loop-helix gene family in metazoans: insights from comparative genomics

Abstract

Background: Molecular and genetic analyses conducted in model organisms such as Drosophila and vertebrates, have provided a wealth of information about how networks of transcription factors control the proper development of these species. Much less is known, however, about the evolutionary origin of these elaborated networks and their large-scale evolution. Here we report the first evolutionary analysis of a whole superfamily of transcription factors, the basic helix-loop-helix (bHLH) proteins, at the scale of the whole metazoan kingdom.

Results: We identified in silico the putative full complement of bHLH genes in the sequenced genomes of 12 different species representative of the main metazoan lineages, including three non-bilaterian metazoans, the cnidarians Nematostella vectensis and Hydra magnipapillata and the demosponge Amphimedon queenslandica. We have performed extensive phylogenetic analyses of the 695 identified bHLHs, which has allowed us to allocate most of these bHLHs to defined evolutionary conserved groups of orthology.

Conclusion: Three main features in the history of the bHLH gene superfamily can be inferred from these analyses: (i) an initial diversification of the bHLHs has occurred in the pre-Cambrian, prior to metazoan cladogenesis; (ii) a second expansion of the bHLH superfamily occurred early in metazoan evolution before bilaterians and cnidarians diverged; and (iii) the bHLH complement during the evolution of the bilaterians has been remarkably stable. We suggest that these features may be extended to other developmental gene families and reflect a general trend in the evolution of the developmental gene repertoires of metazoans.

Figure 1

Phylogenetic relationships between the species used in this study. The tree is based on the current view of the phylogeny of the metazoans [73,74]. The total number of bHLHs and the number of represented bHLH metazoan families of orthologs in each genome is indicated. For the number of represented families, a range is indicated for most species, due to the uncertainty about the presence of some families in these species (see additional file 1 for details). The names of representative phylogenetic groups are indicated on the left of the nodes that define these different groups and along some of the terminal branches. The Ciona intestinalis data come from [12] and have not been reanalysed in this study.

Figure 2

Phylogenetic analysis of the cnidarian bHLHs related to the Twist and Atonal superfamilies. The represented tree is a NJ tree, which has been rooted using a human bHLH sequence from the MyoD family as an outgroup. This tree is based on a multiple alignment that only includes the bHLH sequences of Drosophila melanogaster (in yellow) and Homo sapiens (in orange) which constitute the families belonging to the Atonal and Twist superfamilies [11] and their relatives in Nematostella vectensis (in light green) and Hydra magnipapillata (in dark green). The Atonal superfamily includes the Atonal, Neurogenin, NeuroD, Net, Oligo, Beta3, Delilah, and Mist families; the Twist superfamily includes the Twist, Paraxis, Hand, PTFa, PTFb, MyoRa, MyoRb, SCL, and NSCL families. Similar relationships (with similar statistical supports) were found when we used the entire set of bHLH genes of these 4 aforementioned species. The different metazoan families of orthologs are indicated in blue. Numbers above the internal branches are their statistical support values obtained with different methods of phylogenetic reconstruction: first number = bootstrap support in neighbour-joining analysis (10,000 bootstrap replicates); second number = bootstrap support in maximum-likelihood analysis (150 bootstrap replicates); third number = posterior probabilities in Bayesian inference-based analysis. Only statistical support values >50% are shown except for a few cases. Other internal branches (with statistical support <50%) should be considered unreliable. Statistical supports for the existence of the different families are shown in blue. We can see that 7 families (indicated by red asterisks) out of the 18 that are shown on this tree, have cnidarian members and that several cnidarian bHLHs cannot be assigned to any of these families. Some of these bHLHs form monophyletic groups comprising sequences from either one or both of the two cnidarian species (indicated by dark green numbers, from 1 to 6). Group 1 is comprised of only Nematostella vectensis sequences and groups 2 to 9 are comprised of at least one representative from both Nematostella vectensis and Hydra magnipapillata. Group 2 may correspond to Delilah genes, group 3 to Oligo/Beta3/Mist genes, and group 4 to NeuroD/Neurogenin genes (see additional files 13 and 14 for more details).

Figure 3

Phylogenetic analysis of the bHLHs from Amphimedon queenslandica. The represented tree is a NJ tree whose rooting should be considered as arbitrary. This tree is based on a multiple alignment that includes all the Amphimedon queenslandica bHLH (named Amq1 to Amq16) sequences (in blue) and one representative sequence (from Homo sapiens) for each of the bilaterian families of orthologs (in orange). Similar relationships (with similar statistical supports) are found when we used the whole set of bHLHs genes from Homo sapiens. Numbers above the internal branches are as in Figure 2. We only show the statistical support for the internal branches that correspond to monophyletic groups concerning the Amphimedon queenslandica bHLHs. Higher-order groups (A to F) are shown. The Amphimedon queenslandica bHLHs that can be assigned to a family are in grey filled boxes (light grey denotes cases for which there is an uncertainty). Amphimedon queenslandica bHLHs that are associated with more than one family (and the concerned families) are in open black boxes. The single 'orphan' Amphimedon queenslandica bHLH is underlined.

Figure 4

Physical linkages detected between Nematostella bHLH genes. The bHLH genes are in red, putative non-bHLH intervening genes are in blue. A = gw.168.63.1, gene similar to the uncharacterized Drosophila CG13990 gene; B = estExt_fgenesh1_pm.C_570004, gene similar to Q9QXA6 (Glycoprotein-associated amino acid transporter b0+AT1) from Mus musculus; C = fgenesh1_pg.scaffold_57000051, gene similar to Drosophila CG18497. See text and additional file 16 for details.

Figure 5

A model for the evolution of the bHLH complement in metazoans. A simplified phylogenetic tree of the opisthokonts is represented. The three main groups of opisthokonts – the fungi, the choanoflagellates and the metazoans – are indicated. For the metazoans, we have represented their main subdivisions: the two probable monophyletic groups of 'sponges', the demosponges + hexactinellids, and the calcareous sponges; the cnidarians; and the two main groups of bilaterians, the protostomes and the deuterostomes. Last common ancestors of opisthokonts (Uropisthokonta), metazoans (Urmetazoa), eumetazoans (Ureumetazoa), and bilaterians (Urbilateria) are represented by black polygons. The minimal number of bHLHs inferred to have been present in these ancestors is shown (light grey boxes). We also indicate the minimal inferred number of families for the different higher-order groups present in each last common ancestor.

Figure 6

Amphimedon queenslandica bHLHs that are associated with more than one bHLH family. We took, as an example, the case of the ARNT and Bmal families. In Amphimedon queenslandica, there is a single bHLH that clusters with both families as an outgroup to them (Figure 3). There are two main scenarios that can explain this situation. (A) The duplication that gives rise to the two families occurred after the divergence between demosponges and the other metazoans. No duplication occurred in Amphimedon queenslandica, which displays a single gene as in the ancestral situation. (B) The duplication that gives rise to the two families occurred before the divergence between demosponges and the other metazoans. Amphimedon queenslandica displays a single gene because one of the duplicates was lost. The remaining gene became quite divergent in such a way that it cannot be confidently related to either of the two families.