Exon-phase symmetry and intrinsic structural disorder promote modular evolution in the human genome

Abstract

A key signature of module exchange in the genome is phase symmetry of exons, suggestive of exon shuffling events that occurred without disrupting translation reading frame. At the protein level, intrinsic structural disorder may be another key element because disordered regions often serve as functional elements that can be effectively integrated into a protein structure. Therefore, we asked whether exon-phase symmetry in the human genome and structural disorder in the human proteome are connected, signalling such evolutionary mechanisms in the assembly of multi-exon genes. We found an elevated level of structural disorder of regions encoded by symmetric exons and a preferred symmetry of exons encoding for mostly disordered regions (>70% predicted disorder). Alternatively spliced symmetric exons tend to correspond to the most disordered regions. The genes of mostly disordered proteins (>70% predicted disorder) tend to be assembled from symmetric exons, which often arise by internal tandem duplications. Preponderance of certain types of short motifs (e.g. SH3-binding motif) and domains (e.g. high-mobility group domains) suggests that certain disordered modules have been particularly effective in exon-shuffling events. Our observations suggest that structural disorder has facilitated modular assembly of complex genes in evolution of the human genome.

Three reasons protein disorder analysis makes more sense in the light of collagen

“…If disorder levels are considered for the entire protein, rather than for individual exons, differences in phase symmetry are larger. Of exons found in mostly disordered proteins, 53.03% are symmetric, compared with 41.97% of exons in mostly ordered proteins ( P = ∼0), which is comparable to previous work . Excluding collagen‐encoding proteins from this analysis reduces this difference (44.41 and 41.88% for mostly disordered and mostly ordered respectively), though it does remain statistically significant, suggesting there is a modest preference for symmetric exons within proteins that are mostly disordered.…”

“…Exons in collagens typically encode either fully ordered (14.1%) or fully disordered (64.9%) protein, as shown in Figure (A). However in line with previous literature, we classify exons as mostly ordered and mostly disordered when <30%, or >70%, of residues are predicted disordered . These thresholds correspond to 19.9 and 73.7% of exons in collagen‐containing proteins, respectively.…”

“…It was previously reported that higher levels of phase symmetry are found in exons that encode mostly disordered protein regions . We have found that this effect is due to collagen‐encoding proteins, as summarized in Table .…”

“…Furthermore, the locations of splice junctions themselves may be constrained by the tolerance for disorder‐promoting amino acids in the translated protein product . Finally, it was recently shown that symmetric exons are also enriched for protein disorder, with the suggestion that this symmetry has aided the modular evolution of the human proteome . This previous work includes a graph of the distribution of the lengths of exons that encode disordered protein regions (Fig.…”

The RIDL hypothesis: transposable elements as functional domains of long noncoding RNAs

“…This process is strictly limited by the requirement that the newly inserted exon be in the same open reading frame as the host gene, limiting the frequency with which such events give rise to a viable protein. In the case of an inserted internal exon, for example, just one in three insertions will result in a viable protein (Marsh and Teichmann 2010;Schad et al 2013). Similarly, although TEs have been shown to occasionally contribute novel exonic sequence to protein-coding genes, the insertion of a TE within a coding exon, or else the spliced inclusion of an entire TE-derived exon, only has a one-in-three probability of creating a viable protein, and even then it would likely be a stretch of nonsense protein (Sela et al 2010).…”