A non-tree-based comprehensive study of metazoan Hox and ParaHox genes prompts new insights into their origin and evolution

Abstract

Background: Hox and the closely-related ParaHox genes, which emerged prior to the divergence between cnidarians and bilaterians, are the most well-known members of the ancient genetic toolkit that controls embryonic development across all metazoans. Fundamental questions relative to their origin and evolutionary relationships remain however unresolved. We investigate here the evolution of metazoan Hox and ParaHox genes using the HoxPred program that allows the identification of Hox genes without the need of phylogenetic tree reconstructions.

Results: We show that HoxPred provides an efficient and accurate classification of Hox and ParaHox genes in their respective homology groups, including Hox paralogous groups (PGs). We analyzed more than 10,000 sequences from 310 metazoan species, from 6 genome projects and the complete UniProtKB database. The HoxPred program and all results arranged in the Datab'Hox database are freely available at http://cege.vub.ac.be/hoxpred/. Results for the genome-scale studies are coherent with previous studies, and also brings knowledge on the Hox repertoire and clusters for newly-sequenced species. The unprecedented scale of this study and the use of a non-tree-based approach allows unresolved key questions about Hox and ParaHox genes evolution to be addressed.

Conclusions: Our analysis suggests that the presence of a single type of Posterior Hox genes (PG9-like) is ancestral to bilaterians, and that new Posterior PGs would have arisen in deuterostomes through independent gene duplications. Four types of Central genes would also be ancestral to bilaterians, with two of them, PG6- and PG7-like that gave rise, in protostomes, to the UbdA- and ftz/Antp/Lox5-type genes, respectively. A fifth type of Central genes (PG8) would have emerged in the vertebrate lineage. Our results also suggest the presence of Anterior (PG1 and PG3), Central and Posterior Hox genes in the cnidarians, supporting an ancestral four-gene Hox cluster. In addition, our data support the relationship of the bilaterian ParaHox genes Gsx and Xlox with PG3, and Cdx with the Central genes. Our study therefore indicates three possible models for the origin of Hox and ParaHox in early metazoans, a two-gene (Anterior/PG3--Central/Posterior), a three-gene (Anterior/PG1, Anterior/PG3 and Central/Posterior), or a four-gene (Anterior/PG1--Anterior/PG3--Central--Posterior) ProtoHox cluster.

Figure 1

HoxPred classification approach. A. Generalised profile construction. A multiple alignment is built from a set of non-redundant homeodomain sequences that belong to a given homology group (PG9 for this illustration). This alignment then serves as input to a program from the pftools suite [62], which generates the corresponding generalised profile. This profile is a scoring matrix that allows to assign a score to a sequence, based on its similarity with the profile. Contrary to more simple pattern search technique, a profile can provide scores for residues that were not originally found at a given position of the motif. These scores are residue-specific, and extrapolated by using a substitution matrix when building the profile. B. HoxPred classification principle. The sequence to classify is scored by an optimal combination of profiles. The resulting vector of scores then serves as input to a discriminant function that has been previously trained to classify such a vector of scores into a specific class (eg PG4). C. Linear discriminant classifier training. The training phase aims at generating the discriminant function. The training dataset comprise sequences for which the class is known. They can be HOX, RANDOM or HOMEO sequences (see Materials and methods). All sequences are scored by the profiles, so that each sequence is represented by a vector of scores. The classifier is then trained to classify such vector of scores into their associated class (specified on the right). CTL is the control class (see Materials and methods).

Figure 2

Genomic organization of the Hox genes identified with HoxPred in the genome-scale analyses. Hox and ParaHox genes are depicted with arrows indicating transcription orientation, over black lines representing the scaffolds. This representation takes into account the relative distance between the genes. The transcription orientation is the same as provided by the JGI genome browser. The color of the arrows relates to HoxPred classification (see color code on the left); white squares are non-Hox genes. The Hox cluster of Strongylocentrotus is from [46] and the ParaHox genes are from SpBase [63]. The Hox cluster of Branchiostoma is from [43], with the additional Branchiostoma Hox15 gene found in the genome assembly. Hellobdella genes are not indicated as they span many scaffolds, probably due to poor genome assembly. When available, gene names are specified: in black (from published studies [16,42,43,46]) or in blue (from the JGI genome browser or SpBase). An additional putative Hox gene, showing sequence similarities with Sp-Hox11/13c, lies outside the Hox cluster in Strongylocentrotus. See Additional file 2, Table S4 for the genomic coordinates.

Figure 3

Models for the evolution of Posterior and Central Hox genes in bilaterians. A. Posterior Hox genes. The predicted PGs for each phylogenetic group are indicated with colors in the tables. Inside these table, the names of the genes are indicated when HoxPred predictions differ from their current annotation. The possible emergence of individual PGs are indicated on the schematic tree with vertical bars (only the PG content is considered, not the actual number of genes belonging to each PG, i.e. lineage-specific duplication and losses of individual genes are not indicated). Given that both protostomes and deuterostomes have PG9 predictions, it seems that a Hox9 gene was already present in Urbilateria. PG10 would have emerged in deuterostomes, in the lineage leading to chordates. After the divergence of cephalochordates, the lineage leading to urochordates and vertebrates would have acquired PG12. PG14 appeared in vertebrates. With respect to PG11, this group could have emerged either before or after the split between urochordates and vertebrates. Considering that both Ciona intestinalis and Oikopleura dioica have disintegrated clusters and likely miss PGs, we cannot exclude a possible loss of PG11 in urochordates. The emergence of PG13 is uncertain due to the prediction of the amphioxus Hox15 gene as PG13. It could either be early in the chordate lineage, or in the last common ancestor of urochordates and vertebrates. B. Central Hox genes. The possible emergence and loss of individual PGs are indicated on the schematic tree with vertical bars and crosses, respectively. Four Central PGs were present in Urbilateria (PG4, PG5, PG6 and PG7). PG6 and PG7 would have been independently lost within deuterostomes. PG8 emerged in vertebrates.

Figure 4

Summary of HoxPred predictions in cnidarians. The predicted homology group are indicated with colors in the table. ANT, CENT and POST predictions were obtained with the "Bilateria_relaxed" version, while the PG predictions were obtained with the "Vertebrate_relaxed" version.

Figure 5

Models for the early evolution of Hox and ParaHox genes. The predicted homology groups for each phylogenetic group are indicated with colors in the table. The uncertainty of the phylogenetic position of placozoans is indicated by a dashed line [64,65]. The cnidarian/bilaterian ancestor inferred Hox-ParaHox repertoire is depicted in a double box. Posterior genes are depicted with two colors to reflect the uncertainty of predictions into Central and Posterior groups. This repertoire would result from three equally parsimonious scenarios: a two-gene ProtoHox cluster composed of ancestral Anterior/PG3 and Central (or Central/Posterior) genes, undergoing two or three duplications; a three-gene ProtoHox cluster composed of ancestral Anterior/PG1, Anterior/PG3 and Central (or Central/Posterior) genes undergoing one gene loss, and one or two duplication; or a four-gene ProtoHox cluster composed of ancestral Anterior/PG1, Anterior/PG3, Central and Posterior genes, undergoing two gene losses and a possible duplication.