Origins and Evolution of Human Tandem Duplicated Exon Substitution Events

Abstract

The mutually exclusive splicing of tandem duplicated exons produces protein isoforms that are identical save for a homologous region that allows for the fine tuning of protein function. Tandem duplicated exon substitution events are rare, yet highly important alternative splicing events. Most events are ancient, their isoforms are highly expressed, and they have significantly more pathogenic mutations than other splice events. Here, we analyzed the physicochemical properties and functional roles of the homologous polypeptide regions produced by the 236 tandem duplicated exon substitutions annotated in the human gene set. We find that the most important structural and functional residues in these homologous regions are maintained, and that most changes are conservative rather than drastic. Three quarters of the isoforms produced from tandem duplicated exon substitution events are tissue-specific, particularly in nervous and cardiac tissues, and tandem duplicated exon substitution events are enriched in functional terms related to structures in the brain and skeletal muscle. We find considerable evidence for the convergent evolution of tandem duplicated exon substitution events in vertebrates, arthropods, and nematodes. Twelve human gene families have orthologues with tandem duplicated exon substitution events in both Drosophila melanogaster and Caenorhabditis elegans. Six of these gene families are ion transporters, suggesting that tandem exon duplication in genes that control the flow of ions into the cell has an adaptive benefit. The ancient origins, the strong indications of tissue-specific functions, and the evidence of convergent evolution suggest that these events may have played important roles in the evolution of animal tissues and organs.

Keywords: alternative splicing; exon duplication; function; ion channels; proteomics; tissue specificity.

Fig. 1.

Physicochemical properties of resides in UHP regions. In (A), the distribution of relative solvent accessible area (RSA) for amino acid residues in loops, sheets, and helices in three distinct types of UHP regions (N-terminal, Internal, and C-terminal), compared with the background distribution form the whole set of AlphaFold models. Panel (B) shows the percentage of ligand binding, ordered (globular), and disordered residues that are conserved between two UHP regions, that change, or that align to a gap. In (C), the average gain in McLachlan Matrix score for each type of amino acid in the alignments of the UHP regions. The amino acids are taken from the UHP region that is modeled by AlphaFold in each case. The gain is calculated by subtracting the average McLachlan Matrix score from the average score of the substitutions in the UHP regions.

Fig. 2

Tandem exon duplications and functional residues. Resolved structures of isoforms of (A) CAMK2D (PDB (Burley et al. 2017) structure: 6ayw) in spacefill mode showing the protein surface, (B) KHK (2hqq) in spacefill mode, (C) ACOX1 (7q86) in cartoon mode with ligand binding residues shown as sticks, and (D) ACTN1 (2n8y) also in cartoon mode with ligand binding residues shown as sticks. For all panels, the UHP regions are highlighted in red if the amino acids are conserved between UHP regions, and in yellow when not conserved. Ligand binding regions in panels (A) and (B) are shown in purple. Ligand binding regions in panels (C) and (D) are shown as sticks. The teal residue in panel (B) is arginine 108, part of both the ligand binding region and the UHP region. The teal structure in panel (C) is the bound flavin-adenine dinucleotide ligand. The ball in panel (D) is the bound calcium. All images were created with PyMol.

Fig. 3

Tissue specificity of UHP regions. In panel (A), the percentage of tandem duplicated substitution events for which we detected both UHP regions in large-scale proteomics experiments broken down by the age of the event, and the percentage of tandem duplicated substitution events that were both detected and tissue specific at the protein level. Peptide evidence for both UHP regions was found for almost 80% of Bilaterian events, and more than 50% were also found to be tissue-specific. In panel (B), the AlphaFold [47] model of one of the two tissue-specific FYN isoforms (FynB in this case). The UHP region that spans the two domains is coloured. The conserved residues in the UHP region that interact with the adenosine triphosphate (ATP) ligand are in red.

Fig. 4

CACNA family homologous exons in Drosophila melanogaster and human. Tandem exon duplication substitution events mapped onto the structure of rabbit CACNA1S (PDB: 5gjv) and colored by gene family. Panel A: side view of CACNA1S showing the helices that cross the membrane. Yellow regions are from D. melanogaster gene Cacophony, orange regions from D. melanogaster gene Ca-alpha1D, magenta from D. melanogaster gene Ca-alpha1T, red from human genes CACNA1C and CACNA1D, and green from human genes CACNA1A, CACNA1B, and CACNA1E. Panel B: top view of the transporter showing the transmembrane pore. Colors identical to A. Panel C: a single tandem exon duplication substitution region mapped onto the structure of rabbit CACNA1S. The structure colored in blue is the protein from CACNB1, which interacts with one of the regions coded by a tandem exon duplication substitution event in D. melanogaster gene Cacophony. Panel D: cross-species alignments of regions coded by the tandem exon duplication from arthropod species for the tandem exon duplication substitution event in section C.

Fig. 5

Tandem duplicated exon events in Caenorhabditis elegans, Drosophila melanogaster, and human. Events are coloured by species as per the legend. (A) Tandem exon duplication substitution events for ABCC9, and D. melanogaster and C. elegans orthologues (mrp and mrp-1), mapped onto the structure of human cystic fibrosis transmembrane conductance regulator (6o1v) colored by species. mrp and mrp-1 have unrelated events that coincide in the same region of the structure. The event unique to mrp has eight interchangeable homologous exons. (B) Tandem exon duplication substitution events mapped onto the structure of human KCNMA1 (3naf). Both D. melanogaster slo and C. elegans slo-1 have unrelated events that coincide in the same region of the structure. (C) Tandem exon duplication substitution events mapped onto the structure of human TPM1 (1c1g). D. melanogaster Tm1 and C. elegans lev-11 have apparently non-orthologous tandem exon duplication substitution events, as do TPM1 and Tm1, and TPM1 and lev-11. TPM1, lev-11, and Tm1 all have three tandem duplicated 3' exons that can produce homologous C-terminals, but the exons appear not to have evolved from a common ancestor. A further unrelated tandem exon duplication substitution generates distinct C-terminals in D. melanogaster Tm2. Only the N-terminal tandem exon duplication substitution event seems to be conserved between the three species. In this event, the first two exons are swapped for a homologous 5′ exon which produces a shorter isoform. This event evolved in a common ancestor almost 700 million years ago. Only one exon (the penultimate) is not involved in a tandem exon duplication substitution event in at least one of the three species. Mapping to 3D structures was carried out using HHPred (Zimmermann et al. 2018).