Evaluating the adaptive hypothesis of A-to-I RNA editing in filamentous ascomycete fungi

A-to-I RNA editing enzymatically converts adenosine (A) to inosine (I) in RNA molecules. During sexual reproduction in several filamentous ascomycete fungi, hundreds to tens of thousands of protein-coding sites are edited from A to I, mostly read as guanine (G) by ribosomes. A previous study reported a higher frequency of nonsynonymous than synonymous editing and inferred that A-to-I editing is adaptive in these fungi. However, this inference was based on ∼1% of all editing sites due to methodological limitations, and an alternative nonadaptive explanation-the harm-permitting model-was not considered. Here, we develop a method to test the adaptive hypothesis of RNA editing while accounting for sequence motifs associated with editing, thereby enabling the inclusion of all detected editing events. We apply this method to genomic and transcriptomic data from Fusarium graminearum, Neurospora crassa, and N. tetrasperma. Our analyses suggest that nonsynonymous A-to-I RNA editing in these species is frequently adaptive and that, for at least some nonsynonymous editing events, the benefit primarily arises from the production of multiple distinct proteins from a single gene. Nonetheless, not all nonsynonymous editing is adaptive. Sequence motifs prone to nonsynonymous editing have been selectively depleted at specific genomic locations in genes expressed in sexual reproduction, and a subset of editing events exhibits patterns consistent with the harm-permitting model. In summary, both adaptive and nonadaptive nonsynonymous editing exist in filamentous ascomycetes.