Date Published: November 8, 2016
Publisher: Public Library of Science
Author(s): Jennifer Risso-Ballester, José M. Cuevas, Rafael Sanjuán, Siobain Duffy.
Rates of spontaneous mutation determine the ability of viruses to evolve, infect new hosts, evade immunity and undergo drug resistance. Contrarily to RNA viruses, few mutation rate estimates have been obtained for DNA viruses, because their high replication fidelity implies that new mutations typically fall below the detection limits of Sanger and standard next-generation sequencing. Here, we have used a recently developed high-fidelity deep sequencing technique (Duplex Sequencing) to score spontaneous mutations in human adenovirus 5 under conditions of minimal selection. Based on >200 single-base spontaneous mutations detected throughout the entire viral genome, we infer an average mutation rate of 1.3 × 10−7 per base per cell infection cycle. This value is similar to those of other, large double-stranded DNA viruses, but an order of magnitude lower than those of single-stranded DNA viruses, consistent with the possible action of post-replicative repair. Although the mutation rate did not vary strongly along the adenovirus genome, we found several sources of mutation rate heterogeneity. First, two regions mapping to transcription units L3 and E1B-IVa2 were significantly depleted for mutations. Second, several point insertions/deletions located within low-complexity sequence contexts appeared recurrently, suggesting mutational hotspots. Third, mutation probability increased at GpC dinucleotides. Our findings suggest that host factors may influence the distribution of spontaneous mutations in human adenoviruses and potentially other nuclear DNA viruses.
DNA viruses have been traditionally viewed as slowly-evolving entities, but this notion has been challenged in the last decade after the discovery of several highly diverse and fast-evolving DNA viruses [1–6]. The pace of evolution should be dependent on the rate at which new spontaneous mutations are produced, yet it is currently accepted that DNA virus mutation rates are typically much lower than those of RNA viruses . However, as opposed to RNA viruses, few mutation rate estimates have been obtained for DNA viruses, which include four bacteriophages (φX174, m13, λ, and T4), herpes simplex virus, and human cytomegalovirus [7–11]. Moreover, these estimates were derived from indirect, phenotype-based methods of mutation detection, used very small portions of the viral genome, or suffered from bias due to selection acting on population mutation frequencies. Therefore, we currently lack an unbiased, genome-wide view of how spontaneous mutations are produced in DNA viruses. Although next-generation sequencing (NGS) has made it possible to analyze genetic variation in full-length DNA virus genomes with unprecedented detail, its relatively low per-read accuracy has prevented detection of rare variants, including new spontaneous mutations. This problem has been solved in recently-developed methods that increase the accuracy of NGS by orders of magnitude [12,13], now permitting an in-depth characterization of DNA virus spontaneous mutation rates.
Per-base mutation rates correlate negatively with genome sizes over a broad range of DNA microorganisms including viruses, bacteria, and unicellular eukaryotes [7,24,25]. As a result, the genomic mutation rate varies weakly, and a quasi-constant rate of approximately 0.003 mutations per genome per round of copying was suggested . For HAdv5, our calculated mutation rate per cell infection cycle is 1.3 × 10−7 or, equivalently, 0.0046 per 35.9 kbp genome, in good agreement with the suggested rule. In recent work with human cytomegalovirus, de novo mutations were identified in longitudinal patient samples and, using the estimated duration of the cell infection cycle for this virus in vivo, the calculated mutation rate was 2.0 × 10−7 . Previous work with murine cytomegalovirus gave a very similar estimate of 1.4 × 10−7, although this mutation rate was measured per day instead of per cell infection cycle . In herpes simplex virus, the mutation rate was estimated by scoring null mutations in the tk gene using ganciclovir . This yielded an estimated rate of 5.9 × 10−8 per cell infection cycle . Therefore, mutation rates for different human DNA viruses measured by widely different methods vary within approximately twofold around 10−7 mutations per base per cell infection cycle. Genome sizes range from 35.9 kbp for HAdV to 150 kpb for herpes simplex virus and 230–236 kbp for cytomegaloviruses. As a result, genomic mutation rates vary by approximately an order of magnitude, and are substantially lower than the 0.003 expected value for the largest DNA viruses (Fig 4).