The sequencing of the genome of the pea aphid Acyrthosiphon pisum indicated an unusual expansion of the small non coding RNA machinery specific to miRNAs [15, 17]. The discovery of these gene duplications opens new perspectives of research about the regulation of gene expression by miRNAs in aphids, which might play a crucial role in the striking polyphenism displayed by these insects. Several scenarios may occur after gene duplications: non-functionalization, subfunctionalization and, more rarely, neofunctionalization . Given that the miRNA machinery appears to be a fundamental mechanism in metazoa, with genes very conserved and as unique copies in most animal genomes known to date, it was particularly interesting to evaluate the different fates of these duplications. More precisely, we tried in this study to determine the age, evolutionary rates, and expression-specificities of the gene duplicates of dicer-1 and argonaute-1 described for A. pisum. The preliminary characterization of the expansion of the miRNA machinery in the pea aphid pointed out a strong divergence of ago-1b in A. pisum, a feature that suggested a change in function for AGO-1b protein in aphids .
The sequencing of ago-1a and ago-1b in different aphid species in the present study has revealed the presence of both copies of this gene in all the species analyzed, including the two tribes of the aphid subfamily Aphidinae: Aphidini and Macrosiphini. This result indicates that ago-1 was duplicated in an ancestor of this subfamily (Figure 1). By contrast, our analysis provided a different estimation of the timing of duplication of dcr-1. We found a dcr-1b copy only in the three species of the genus Acyrthosiphon, which suggests a more recent duplication event. However due to the moderate bootstrap support values, we cannot presently propose a solid conclusion on the timing of this duplication. However, scenarios of a more recent duplication than found for ago-1 (scenarios in which the duplication of dcr-1 would only concern Macrosiphini) seem more parsimonious, as they imply less events of gene loss (or failure to amplify one copy in several species). Future genome sequencing projects for different aphids species will give the opportunity of estimating whether the duplication of dcr-1 may have occurred earlier during aphid evolution.
Aphid genomes seem to have been affected by recurrent gene duplications through their evolution, with the pea aphid genome showing the highest number of gene families among the insect genomes sequenced to date [15, 21]. In a previous study, we found similar pairwise synonymous distances between duplicates of genes of the miRNA machinery in the pea aphid, which suggested that the duplications of ago-1 and dcr-1 had occurred roughly simultaneously . In contrast, in this study, with more phylogenetic evidence, we found that the duplication of ago-1 probably preceded that of dcr-1. Indeed, we have also found that dcr-1 has been further duplicated two independent times after the dcr-1a/b duplication in the genus Acyrthosiphon, although these new copies have most likely become pseudogenes.
The analysis of the patterns of non-synonymous and synonymous substitutions in the duplicates of ago-1 and dcr-1 in aphids has also provided, for each of the genes, a similar picture of a slow evolving copy subject to purifying selection (ago-1a and dcr-1/1a) and a fast evolving copy characterized by relaxed selection (ago-1b and dcr-1b). The difference among copies was much more pronounced for ago-1 than for dcr-1. The evolution of ago-1a has clearly been driven by strong purifying selection, as detected by branch models in PAML, with extremely low values of ratios of non-synonymous to synonymous substitution rates (dN/dS or ω) and near-absence of replacements for the aphids of the subfamily Aphidinae. Such strong purifying selection implies sequence stability for this protein over a period of approximately 50–70 millions of years, the estimated age of the split between Aphidini and Macrosiphini . Contrastingly, the ago-1b copies have been characterized by a strong acceleration of evolutionary rates, with different intensity in the three regions analyzed (Region 1: ω=0.8621; Region 2: ω=0.6236; Region 3: ω=0.2476). Branch models for the analysis of selective pressures estimate a dN/dS value that is averaged over the entire sequence analyzed. However, the gene sequence may contain both positions with very low or no variability due to purifying selection and positions in which change is promoted by positive selection. Site-models allow the detection of positive selection acting on specific positions of the sequence that might be unseen by branch models. Indeed, site and branch-site models led us to detect the signature of positive selection acting on Regions 1 and 2 of ago-1b. Branch-site models in particular showed that positive selection has played a role in the evolution of the ago-1b copies, with many codon positions with dN/dS ratios significantly higher than 1 (54 and 74 in Regions 1 and 2 respectively, which represent 22% and 24% of the codons of each region).
The difference of rates between the two copies of dcr-1 was less pronounced than for the two copies of ago-1. The branch model of best fit estimated that purifying selection has characterized the dcr-1a copy of the Acyrthosiphon spp and the unique dcr-1 copy in species where the duplication was not found, (ω=0.1247), although the sequence was less conserved than the slow-evolving ago-1a. By contrast, the selective pressures on dcr-1b appeared to be more relaxed (ω=0.3577) and similar to the value found for Region 3 in ago-1b. As for Region 3 of ago-1b, no sign of positive selection was detected in dcr-1b.
We finally evaluated differences of expression of the duplicated copies among different reproductive morphs characteristic of the aphid life-cycle, which alternates sexual and asexual reproduction. The display of such polyphenism must involve a finely tuned mechanism of regulation of gene expression from their gene families-rich genomes, in which miRNAs could play a key role, along with other phenomenon like alternative transcription. In this report, we have obtained a first insight on the functional characterization of the gene duplicates of ago-1 and dcr-1 by means of semiquantitative PCRs carried out on four morphs of the life cycle of the pea aphid. Our results have shown a differential regulation of the expression of ago-1a, ago-1b and dcr-1b among the morphs. The most striking pattern is that of dcr-1b, which seems to be only expressed in the sexupara, the parthenogenetic female that gives birth to the sexual morphs. Although no signature of positive selection could be detected for dcr-1b, the structure of its sequence gives an interesting clue supporting a possible change in function. The three species of Acyrthosiphon where dcr-1b was sequenced share a deletion of 47 amino acids in the protein sequence with respect to dcr-1a, which is located inside the first RNase III domain. This deletion does not affect any of the active sites, but has probably changed the distance between them inside the RNAse IIIa domain, as well as the distance between the RNAse IIIa and RNase IIIb domains. Interestingly, the two RNase III domains in Drosha and Dicer seem to interact with each other to make an intramolecular dimer with the two catalytic sites located close to each other, and the distance between them seems to determine the length of the small RNA produced after cleavage [14, 23]. It is tempting to hypothesize that the deletion in DCR-1b, which has been evolutionarily conserved in the genus Acyrthosiphon, may have determined a change in the distance between catalytic centers and consequently in its function, leading to the production of small non coding RNAs of different length. All active sites in the two RNase III domains of DCR-1 have remained conserved in the two copies of the protein found in aphids and also in the three outgroup species except for two changes. By contrast, the region of the RNase IIIa domain that contains the deletion in DCR-1b is highly variable and shows other deletions in the outgroup species, which could affect the regulation of gene expression by small non coding RNAs. Consistent with this hypothesis, the alternative transcription described for dcr-1a in A. pisum would yield two translated DCR-1a proteins differing by the absence or presence of 19 aa located inside the second set of functional sites of the first RNase III domain of the protein, which might also alter the length between the catalytic centers and allow the production of small non coding RNAs of different length. The same alternative transcription observed in A. gossypii and R. padi entails the loss of the RNase IIIb domain in one of the alternative transcripts, which probably leads to an non-functional protein, given that the interaction of both RNase III domains seem to be needed for cleavage.
Future experiments will allow to study whether the gene duplications of the miRNA machinery play a primary role in the sexual/asexual polyphenism in aphids. It is important to note that cyclical parthenogenesis is an ancient and ubiquitous trait in these insects, and that most of its subfamilies display this same reproductive polyphenism with only one copy of DCR-1. This means that the duplication of the miRNA machinery genes was not essential for settling the sexual/asexual alternation of reproductive modes that define the aphid life-cycle. But the different copies might have evolved differently and specific roles at different steps of the life-cycle. The research on the implications of the miRNA machinery in the aphid polyphenism will benefit not only from comparative genomics analysis that will allow knowing precisely the phylogenetic distribution of its gene duplications but also comparative functional studies between aphids having and not having the gene duplicates. The strong purifying selection acting on ago-1a and dcr-1a suggests that these proteins probably have retained their function in the miRNA machinery. On the other side, both relaxed and positive selection acting on DCR-1b and AGO-1b might have led these copies to evolve a new function. Further studies will be needed to explore their potential new role in the miRNA system and at the different steps of the reproductive cycle of these organisms.