Systematic genome-wide annotation of spliceosomal proteins reveals differential gene family expansion

Abstract

Although more than 200 human spliceosomal and splicing-associated proteins are known, the evolution of the splicing machinery has not been studied extensively. The recent near-complete sequencing and annotation of distant vertebrate and chordate genomes provides the opportunity for an exhaustive comparative analysis of splicing factors across eukaryotes. We describe here our semiautomated computational pipeline to identify and annotate splicing factors in representative species of eukaryotes. We focused on protein families whose role in splicing is confirmed by experimental evidence. We visually inspected 1894 proteins and manually curated 224 of them. Our analysis shows a general conservation of the core spliceosomal proteins across the eukaryotic lineage, contrasting with selective expansions of protein families known to play a role in the regulation of splicing, most notably of SR proteins in metazoans and of heterogeneous nuclear ribonucleoproteins (hnRNP) in vertebrates. We also observed vertebrate-specific expansion of the CLK and SRPK kinases (which phosphorylate SR proteins), and the CUG-BP/CELF family of splicing regulators. Furthermore, we report several intronless genes amongst splicing proteins in mammals, suggesting that retrotransposition contributed to the complexity of the mammalian splicing apparatus.

Figure 1.

Schematics of the computational pipeline flow. Sources, software, and parameters are represented in blue and species in red.

Figure 2.

Evolutionary relationship among the protein members of hnRNP F/H family in several eukaryotes, i.e., human (Hsap), mouse (Mmus), rat (Rnor), chicken (Ggal), Fugu (Frub), zebrafish (Drer), Tetraodon (Tnig), Ciona intestinalis (Cint), fruit fly (Dmel), mosquito (Agam), C. elegans (Cele), Arabidopsis (Atha), and Trypanosoma (Tcru). Vertebrate factors are highlighted in blue, red, and shades of green. (A) Rooted Neighbor-Joining phylogenetic tree generated using ClustalW (1000 bootstraps), based on amino-acid alignment generated by T-Coffee. Bootstrap values are shown. Branch lengths are scaled in arbitrary units. (B) Rooted Gamma-corrected Maximum-Likelihood phylogenetic tree generated using GAMMA and the Phylip program Proml, based on amino-acid alignment generated by T-Coffee. Branch lengths are scaled in arbitrary units.

Figure 3.

Evolutionary relationship among the RNA-recognition motifs (RRM) of members of the family SRp30c-ASF for several eukaryotes, i.e., human (Hsap), mouse (Mmus), chicken (Ggal), Fugu (Frub), Ciona (Cint), fruit fly (Dmel), C. elegans (Cele), Arabidopsis (Atha), and Plasmodium (Pfal) (for simplicity only one rodent, one teleost, and one insect are shown). Amino-acid positions of each domain within the protein are also indicated in the domain identification. The unrooted Neighbor-Joining phylogenetic tree was generated using ClustalW (1000 bootstraps) based on amino-acid alignment generated by T-Coffee. Bootstrap values are shown. Branch lengths are scaled in arbitrary units. RRM1 in Ggal_ASF and Cint_ASFb corresponds to RRM2 in the other proteins as their sequences are truncated in the N-terminal. Pfal_SF is found to have only one RRM. Atha_RS31A can be technically considered an ortholog of the Hsap_SR30C (reciprocal BLAST hit) but exhibits a considerably lower degree of identity (36%) with the human factor than its Arabidopsis paralogs (e.g., 53% for Atha_SRp30).

Figure 4.

Evolutionary relationship among the RNA-binding K-Homology (KH) domains of members of the family hnRNP-E/PCB for several metazoans, i.e., human (Hsap), mouse (Mmus), chicken (Ggal), Fugu (Frub), Ciona (Cint), fruit fly (Dmel), and C. elegans (Cele) (for simplicity only one rodent, one teleost, and one insect are shown). Amino-acid positions of each domain within the protein are also indicated in the domain identification. The unrooted Neighbor-Joining phylogenetic tree was generated using ClustalW (1000 bootstraps) based on amino-acid alignment generated by T-Coffee. Bootstrap values are shown. Branch lengths are scaled in arbitrary units.