Chaetognath phylogenomics: a protostome with deuterostome-like development

Traditional textbook phylogeny splits bilaterians into protostomes and deuterostomes according to whether their mouth derives from the blastopore or not. This scheme has been largely confirmed by small subunit ribosomal RNA (SSU) molecular phylogeny. However, some phyla, such as the lophphorate phyla Phoronida and Brachiopoda as well as the Chaetognatha exhibit classical deuterostome embryological features such as formation of the mesoderm from the gut (enterocoely) and secondary opening of the mouth. However, their SSU sequences do not classify them as deuterostomes [1]. For example, using SSU based phylogeny the lophophorates are now grouped with protostome trochozoans (Annelida and Mollusca) to form lophotrochozoans [1,2]. But in the case of chaetognaths, a phylum of small marine predators, phylogenetic placement has been more problematic. Indeed, although they exhibit some protostome features, such as ventral nerve cords and circum-oesophageal fibers, SSU phylogeny first led to them being positioned wrongly as stem bilaterians because their fast evolving SSU resulted in a a long branch attraction artefact [3]. More recently, analysis of several mitochondrial markers showed affinities of chaetognaths to protostomes [4]. In order to address the discrepancy of morphological characters and molecular phylogeny, we attempted to resolve the phylogenetic position of chaetognaths using genomic data. Here, we clearly position chaetognaths among protostomes, likely as a sister-group of all other protostome phyla using a ribosomal protein dataset including hemichordate genomic data. We sequenced 11,526 expressed sequence tags (ESTs) from a juvenile cDNA library of the benthic chaetognath Spadella cephaloptera (Busch, 1851) (Supplemental data). BLAST comparisons of transcriptome similarities between chaetognaths and various species (specific databases: Drosophila, Lumbricus, Homo) or clades (composite databases: Deuterostomes, Lophotrochozoans, Ecdysozoans) showed conservation of a large set of genes within bilaterians (Table S1) as previously observed by Kortschak et al. [5]. Among these genes, ribosomal proteins were retrieved as valuable markers for phylogenomic analysis because of their conservation among eukaryotes and relative abundance in the EST collections of different phyla we used. We determined that BLAST searches provide significantly higher similarity between input and matched sequences when performed against a composite database that gathers all the available sequences for the various species belonging to a clade, rather than against a one-species database. As current problems in phylogenetic reconstruction are often related to long branch attraction [6], this composite database strategy was employed to shorten branches. For each validated monophyletic group (i.e. phylum), we retrieved the sequences of phylogenetic markers from a composite …