Molecular evolution of the ependymin-related gene epdl2 in African weakly electric fish

Abstract

Gene duplication and subsequent molecular evolution can give rise to taxon-specific gene specializations. In previous work, we found evidence that African weakly electric fish (Mormyridae) may have as many as three copies of the epdl2 gene, and the expression of two epdl2 genes is correlated with electric signal divergence. Epdl2 belongs to the ependymin-related family (EPDR), a functionally diverse family of secretory glycoproteins. In this study, we first describe vertebrate EPDR evolution and then present a detailed evolutionary history of epdl2 in Mormyridae with emphasis on the speciose genus Paramormyrops. Using Sanger sequencing, we confirm three apparently functional epdl2 genes in Paramormyrops kingsleyae. Next, we developed a nanopore-based amplicon sequencing strategy and bioinformatics pipeline to obtain and classify full-length epdl2 gene sequences (N = 34) across Mormyridae. Our phylogenetic analysis proposes three or four epdl2 paralogs dating from early Paramormyrops evolution. Finally, we conducted selection tests which detected positive selection around the duplication events and identified ten sites likely targeted by selection in the resulting paralogs. These sites' locations in our modeled 3D protein structure involve four sites in ligand binding and six sites in homodimer formation. Together, these findings strongly imply an evolutionary mechanism whereby epdl2 genes underwent selection-driven functional specialization after tandem duplications in the rapidly speciating Paramormyrops. Considering previous evidence, we propose that epdl2 may contribute to electric signal diversification in mormyrids, an important aspect of species recognition during mating.

Keywords: electric fish; ependymin; gene duplication; molecular evolution.

Fig. 1.

Graphical summary of the bioinformatic pipeline we leveraged to identify epdl2 genes in the filtered amplicons from each species, see Methods for details. Same-colored connector arrows represent an analysis with a specific value for c (cd-hit's sequence identity threshold parameter, range analyzed 0.84–0.91). Magenta connector arrows represent the analysis with the c value chosen with the selection criteria.

Fig. 2.

Evolutionary relationships between EPDR vertebrate genes. Gene color legend applies to all panels. a) Chordate EPDR gene tree based on Genomicus v03.01 (Fam016630). Stars represent duplications that led to genes supported by our analysis (orange: ancestral EPDR in early vertebrates, red: epdl in early teleosts, black: epd in early clupeocephalans). b) Vertebrate epdl gene tree based on Bayesian inference, only posterior probabilities <1 are shown. Supported epdl genes are color coded in branches. c) Simplified PhyloView alignment (Genomicus v03.01) of the epdl teleost homologs in select taxa (rows) aligned to one bowfin epdl gene. Black pentagons in each row denote epdl genes, including three tandem copies (black rectangle) in bowfin (top row). Pentagons represent the position and orientation of genes syntenic to the epdl gene in each taxon. Colored pentagons highlight genes present in a taxon and in the bowfin reference. Taxa labels (right column) are color coded by epdl gene.

Fig. 3.

Epdl2 genomic region in P. kingsleyae, showing three epdl2 paralogs in tandem. DNA sequence is represented by the black line, numbers above indicate base positions, annotations are shown below the line. Pentagons represent a gene's position and orientation, flanking genes are depicted in blue, epdl2 paralogs in red (start to stop codon) and purple (CDSs). Orange blocks mark the Sanger-sequenced regions.

Fig. 4.

Mormyrid epdl2 gene tree based on Bayesian inference, only posterior probabilities <1 are shown. epdl2 paralogs are classified based on our best hypothesis of epdl2 duplications.

Fig. 5.

Distribution of detected epdl2 genes across the phylogenetic topology of the M. ntemensis + Paramormyrops clade. Potential paralogs losses could have occurred at nodes A, B, or C; see main text for details.

Fig. 6.

Topology of the osteoglossiform epdl2 gene tree used in the selection tests. Mormyrid lineages with (orange) and without (blue) epdl2 duplications are indicated. This branch partition was used in the RELAX and Contrast-FEL tests. Thick orange branches were tested for positive selection with aBSREL, and significant branches from this test are labeled A and B.

Fig. 7.

Paramormyrops sp. SZA Epdl2 as a representative Epdl2 protein. a) Predicted amino acid sequence, annotated with (1) salient structural properties: signal peptide (pink), conserved cysteine residues (orange) forming disulfide bonds (orange lines), N-glycosylation sites (magenta), a proline residue highly conserved across EPDRs (cyan); (2) secondary structure adopted in the 3D model: β strands (gray arrows) and α helixes (gray cylinders); and (3) the 10 positively selected sites with increased ω rates in epdl2 paralogs (red and blue, numbers indicate sites’ positions derived from the homologous codon alignment). b) Two views of our Epdl2 (minus the signal peptide) 3D model (gray) superimposed on its best structural analog, X. tropicalis Epdr1 (PDB entry 6JL9, purple backbone trace). c) Our 3D model from c), showcasing structural properties and select residues as color coded in a), and predicted functional regions: ligand-binding pocket (red cloud) and dimerization surface (blue cloud). The side chains (colored sticks) of the residues depicted in red and cyan point inwards the pocket, and side chains of blue residues are oriented toward the dimerization surface.