Adaptive Proteome Diversification by Nonsynonymous A-to-I RNA Editing in Coleoid Cephalopods

Abstract

RNA editing by the ADAR enzymes converts selected adenosines into inosines, biological mimics for guanosines. By doing so, it alters protein-coding sequences, resulting in novel protein products that diversify the proteome beyond its genomic blueprint. Recoding is exceptionally abundant in the neural tissues of coleoid cephalopods (octopuses, squids, and cuttlefishes), with an over-representation of nonsynonymous edits suggesting positive selection. However, the extent to which proteome diversification by recoding provides an adaptive advantage is not known. It was recently suggested that the role of evolutionarily conserved edits is to compensate for harmful genomic substitutions, and that there is no added value in having an editable codon as compared with a restoration of the preferred genomic allele. Here, we show that this hypothesis fails to explain the evolutionary dynamics of recoding sites in coleoids. Instead, our results indicate that a large fraction of the shared, strongly recoded, sites in coleoids have been selected for proteome diversification, meaning that the fitness of an editable A is higher than an uneditable A or a genomically encoded G.

Keywords: RNA editing; adaptation; evolution.

Fig. 1.

The harm-permitting and adaptive editing models. Recoding sites may be fixated into the genome due to random genomic drift, even though their editing does not provide any selective advantage and may be even slightly deleterious. These randomly fixated sites (A-preferring sites) are not expected to be enriched in nonsynonymous editing (left). In a second class of recoding sites, editing does provide a selective advantage as it replaces an inferior A allele by the preferred G allele (middle). For these G-preferring sites, editing does increase the fitness of the organism over an uneditable A, but having a genomically encoded G is equally beneficial or even better (middle). In both cases, fitness does not depend on the protein diversity and flexibility provided by recoding. The HPM asserts that all (or almost all) recoding sites belong to these two categories. In contrast, according to the adaptive editing model some of the recoding sites belong to a third class, where having a recodable codon is functionally important, and the editable A provides a selective advantage over both an uneditable A and a genomically encoded G (right).

Fig. 2.

Restorative and diversifying editing. (a) Phylogeny of the eight species studied here: six coleoids and two outgroup mollusks (see Materials and Methods). (b) Incidence of weak editing (<10%) in synonymous, restorative, and diversifying sites, for each of the six coleoid species. P values for the difference between the incidence rate of restorative and diversifying sites are indicated (Fisher’s exact test). (c) Same as (b), for strong (>10%) editing sites. (d) Box plots showing the distribution of editing levels for synonymous, restorative, and diversifying sites, per species. The boxes represent the first-to-third quartiles range, the horizontal line within the box indicates the median, and the whiskers extend to the most extreme values within a window sized 1.5 times the box size, centered at the median. P values for the difference between the editing levels of restorative and diversifying sites are indicated (Mann–Whitney test). (e) Fraction of sites edited in C1 (LCA of the six coleoids studied) that were mutated into a genomic G in at least one of the six descendant species. Significance of difference for each pair of groups is indicated (Fisher’s exact test). P values < 0.05 are marked with an asterisk. Error bars represent SEM.

Fig. 3.

The number of expected A > G mutations of ancestrally strongly edited sites based on the HPM. According to HPM, all ancestral strong editing sites are either A-preferring or G-preferring. A-preferring sites are neutral or slightly deleterious, and thus an upper bound to their incidence rate at the ancestral node (lower black circle) is obtained from the incidence rate of synonymous sites at the same node. The remaining sites are G-preferring (a lower bound). Mutations are expected to accumulate along the evolutionary path from the above ancestral node to each of the descendants (thick black line). A-preferring sites are expected to mutate along this path at a rate higher than the background nonsynonymous mutation rate, and G-preferring sites are expected to mutate at a rate higher than the neutral (synonymous) rate. Together, one obtains a conservative estimate to the number of mutations expected based on HPM, to be compared with the observed numbers.

Fig. 4.

A > G mutations are depleted at strong editing sites. Panels Compare the actual numbers of A > G mutations in ancestral strongly edited sites (vertical dashed line, numbers indicated in legend) with the distributions based on HPM assumptions and background transcriptomic data. Note that the eight paths (and thus the eight statistical tests) are not independent. Distributions plotted based on 10⁶ samples of the statistical model. Vertical axis represents the number of instances generated in the above process within each bin range. P values calculated by direct comparison with the distribution. P values < 0.05 are marked with an asterisk.