Role of Host-Driven Mutagenesis in Determining Genome Evolution of Sigma Virus (DMelSV; Rhabdoviridae) in Drosophila melanogaster

Abstract

Sigma virus (DMelSV) is ubiquitous in natural populations of Drosophila melanogaster. Host-mediated, selective RNA editing of adenosines to inosines (ADAR) may contribute to control of viral infection by preventing transcripts from being transported into the cytoplasm or being translated accurately; or by increasing the viral genomic mutation rate. Previous PCR-based studies showed that ADAR mutations occur in DMelSV at low frequency. Here we use SOLiD^TM deep sequencing of flies from a single host population from Athens, GA, USA to comprehensively evaluate patterns of sequence variation in DMelSV with respect to ADAR. GA dinucleotides, which are weak targets of ADAR, are strongly overrepresented in the positive strand of the virus, consistent with selection to generate ADAR resistance on this complement of the transient, double-stranded RNA intermediate in replication and transcription. Potential ADAR sites in a worldwide sample of viruses are more likely to be "resistant" if the sites do not vary among samples. Either variable sites are less constrained and hence are subject to weaker selection than conserved sites, or the variation is driven by ADAR. We also find evidence of mutations segregating within hosts, hereafter referred to as hypervariable sites. Some of these sites were variable only in one or two flies (i.e., rare); others were shared by four or even all five of the flies (i.e., common). Rare and common hypervariable sites were indistinguishable with respect to susceptibility to ADAR; however, polymorphism in rare sites were more likely to be consistent with the action of ADAR than in common ones, again suggesting that ADAR is deleterious to the virus. Thus, in DMelSV, host mutagenesis is constraining viral evolution both within and between hosts.

Keywords: ADAR; RNA virus; immunity; mutation; phylogeny; quasispecies.

Fig. 1.—

Maximum likelihood phylogenetic tree based on the General Time Reversible model, taking into account the gamma distribution and the proportion of invariable sites (GTR + G+I). Numbers on the branches represent the bootstrap support (out of 100 bootstrap replications). Only the bootstrap values above 50% are shown. Each sequence is identified by its GenBank accession number and color-coded according to the place of origin. BC strains are shown in red, while European, African, and North American strains are shown in blue, green, and black, respectively.

Fig. 2.—

Observed GA and UC frequencies for GQ451694 as plotted against the corresponding null distributions for its 1,000 random sequences.

Fig. 3.—

Observed GA and UC frequencies for X91062 as plotted against the corresponding null distributions for its 1,000 random sequences.

Fig. 4.—

Schematic depiction of a DMelSV lifecycle. Larger arrows indicate higher activity or amounts, for example more negative, genomic copies are made from the positive antigenomes and there is little reverse synthesis of positive antigenomes from these new genomes; similarly, more of the new negative genomes are used for production of transcripts than virion production. Note the central role played by the initial full-length positive antigenomes. Any editing of these antigenomes would affect not only the genomes packaged in progeny virions, but all secondary transcription as well.