Nonsense-mediated decay enables intron gain in Drosophila

Abstract

Intron number varies considerably among genomes, but despite their fundamental importance, the mutational mechanisms and evolutionary processes underlying the expansion of intron number remain unknown. Here we show that Drosophila, in contrast to most eukaryotic lineages, is still undergoing a dramatic rate of intron gain. These novel introns carry significantly weaker splice sites that may impede their identification by the spliceosome. Novel introns are more likely to encode a premature termination codon (PTC), indicating that nonsense-mediated decay (NMD) functions as a backup for weak splicing of new introns. Our data suggest that new introns originate when genomic insertions with weak splice sites are hidden from selection by NMD. This mechanism reduces the sequence requirement imposed on novel introns and implies that the capacity of the spliceosome to recognize weak splice sites was a prerequisite for intron gain during eukaryotic evolution.

Figure 1. The uneven distribution of novel introns across Drosophila species.

307 novel introns were identified across a set of 3,593 genes with a full-length ortholog in each species. Dotted lines indicate branches with a greater number of novel than lost introns (Figure S2). Branch lengths are drawn proportionally to the rate of intron gain. The numbers of novel introns is indicated above each branch. 350 events occurred at the root of the tree and could not be classified as either intron gain or loss.

Figure 2. Intron gain in response to low complexity sequence in the gene CG42594.

While the exact sequence of this highly variable region in the common ancestor of D. melanogaster and D. ananassae is not known, it is plausible that a single nucleotide deletion within the QSGQSG amino acid repeat (blue shading) generated the canonical 5′ splice site CAG | GTGAGT used by this phase 0 intron. Similarly, the CAG repeat (encoding poly-Q sequence) is a potent 3′ splice acceptor site . Sequence conservation across all species is indicated with light shading. The novel intron (denoted by < >) is highly length variable across all species of the melanogaster group.

Figure 3. PTC as a backup for weak splice sites in novel introns.

(A) The percentage of introns that use the most common 5′ and 3′ splice site motifs. Significantly fewer novel introns use the canonical GT(A/G)AGT motif at position +1 to +6 of the 5′ splice site. Likewise, fewer novel introns use CAG at −3 to −1 of the 3′ site. Error bars represent the 95% confidence intervals generated by resampling 307 introns with replacement 10,000 times (Figure S5). (B) A logistic regression identified a significant deficiency of 3n PTC-free introns within conserved introns (conserved 3n versus conserved 3n+1and2 - bottom contrast) confirming the finding of Jaillon et al. (2008) that selection acts against introns that would remain invisible to the NMD pathway upon intron retention. This effect is significantly stronger amongst novel introns (novel 3n versus novel 3n+1and2 - top contrast) and significantly stronger in a direct comparison between novel and conserved introns (second contrast) (95% CI that do not include one indicate a significant deficiency of 3n PTC-free introns).

Figure 4. NMD conceals weakly spliced novel introns from selection.

A new insertion in exonic sequence (or UTRs) that has the potential to undergo weak splicing but also disrupts the coding sequence (due to frame-shift or an in-frame PTC) will lead to a population of spliced and unspliced transcripts. NMD is expected to remove any unspliced transcript, leading to the translation of only the correct protein product. If sufficient protein is produced, the new insertion might be hidden from selection, thus allowing subsequent mutations to improve splicing and reducing the requirement for NMD. A new insertion that does not evoke NMD (3n PTC-free) will not enjoy this advantage and must encode strong splice sites from the beginning.