Genetic drift and purifying selection shape within-host influenza A virus populations during natural swine infections

Abstract

Patterns of within-host influenza A virus (IAV) diversity and evolution have been described in natural human infections, but these patterns remain poorly characterized in non-human hosts. Elucidating these dynamics is important to better understand IAV biology and the evolutionary processes that govern spillover into humans. Here, we sampled an IAV outbreak in pigs during a week-long county fair to characterize viral diversity and evolution in this important reservoir host. Nasal wipes were collected on a daily basis from all pigs present at the fair, yielding up to 421 samples per day. Subtyping of PCR-positive samples revealed the co-circulation of H1N1 and H3N2 subtype IAVs. PCR-positive samples with robust Ct values were deep-sequenced, yielding 506 sequenced samples from a total of 253 pigs. Based on higher-depth re-sequenced data from a subset of these initially sequenced samples (260 samples from 168 pigs), we characterized patterns of within-host IAV genetic diversity and evolution. We find that IAV genetic diversity in single-subtype infected pigs is low, with the majority of intra-host single nucleotide variants (iSNVs) present at frequencies of <10%. The ratio of the number of nonsynonymous to the number of synonymous iSNVs is significantly lower than under the neutral expectation, indicating that purifying selection shapes patterns of within-host viral diversity in swine. The dynamic turnover of iSNVs and their pronounced frequency changes further indicate that genetic drift also plays an important role in shaping IAV populations within pigs. Taken together, our results highlight similarities in patterns of IAV genetic diversity and evolution between humans and swine, including the role of stochastic processes in shaping within-host IAV dynamics.

Figure 1.. Overview of sampling and sequencing of pigs at the county fair.

(A) Sampling, testing, and sequencing effort over the course of the fair. The decrease in the number of samples collected on days 5 and 6 stems from departure of a subset of pigs prior to the fair’s conclusion – all pigs still present were sampled. (B) Distribution of the number of sequenced samples per pig. Pigs with no sequenced samples either never tested positive for IAV or tested positive at some point but positive samples all had Ct values exceeding 31.0.

Figure 2.. Haplotype networks of IAV segments support co-circulation of two distinct lineages at the county fair.

Consensus sequences from all successfully sequenced samples were inferred and used to construct haplotype networks. Each circle represents a unique haplotype (excluding differences caused by ambiguous bases), with circle size of a haplotype being proportional to the number of consensus sequences that belong to it. Numbers in parentheses refer to the number of differentiating single nucleotide polymorphisms (dSNPs) that distinguish the dominant haplotypes from each lineage in each internal gene segment. Because of the high genetic divergence between the HA of lineage I and the HA of lineage II, and between the NA of lineage I and the NA of lineage II, dotted lines rather than solid lines are shown between the lineages.

Figure 3.. Classification of gene segments and samples into lineages.

(A) Mapping of sample reads on to reference H1, N1, H3, and N2 gene segments for four selected samples. Based on this mapping, a sample’s HA gene segment was classified as belonging to either lineage I (with reads mapped to H1; vial#14SW6024), lineage II (with reads mapped to H3; vial#14SW6331 and vial#14SW5957), or coinfected (with reads mapped to both H1 and H3; vial#14SW6164). Similarly, a sample’s NA gene segment was classified as belonging to either lineage I (with reads mapped to N1; vial#14SW6024), lineage II (with reads mapped to N2; vial#14SW6331 and vial#14SW5957), or coinfected (with reads mapped to both N1 and N2; vial#14SW6164). (B) Detection of dSNPs characteristic of lineages I and II for each of the six internal gene segments of the four selected samples. (C) The proportion of samples classified as lineage I, II, or coinfected, by gene segment. Segments were labeled as unknown when there was insufficient coverage across the focal gene segment to determine nucleotide identities at the dSNP sites or to robustly map to the HA and NA gene segments. Samples (‘All’) were classified as belonging to lineage I, lineage II, coinfected, or reassortant based on their constituent segments. A sample was defined as a lineage I or II sample if at least five of its gene segments were successfully classified and found to be either singly infected lineage I or singly infected lineage II, with no evidence of coinfection. A sample was considered a reassortant if at least five segments were successfully classified and a combination of lineage I singly infected and lineage II singly infected segments were present in the same sample. A sample was considered coinfected if at least one of its gene segments was classified as coinfected. In panels (A) and (B), each row corresponds to a different sample classification. Vial#14SW6024 corresponds to a sample with a lineage I virus, vial#14SW6331 corresponds to a sample with a lineage II virus, vial#14SW6164 corresponds to a coinfected sample, and vial#14SW5957 corresponds to a reassortant virus.

Figure 4.. Singly infected samples harbor relatively low levels of genetic diversity and show evidence of purifying selection.

(A) Distribution of the number of iSNVs identified per sample, across samples. (B) The proportion of detected iSNVs that fall below a given frequency, as specified on the x-axis. Approximately 70% iSNVs are detected at frequencies below 10%. Results are shown by sample lineage (I or II) and separately for nonsynonymous and synonymous iSNVs. (C) The ratio of the number of nonsynonymous to synonymous iSNVs, by gene segment and lineage. These ratios are shown alongside the neutral expectation, given by the ratio of nonsynonymous to synonymous sites. Black whiskers show the 95% confidence interval of the ratio.

Figure 5.. Longitudinal dynamics of iSNVs in singly infected swine.

(A) The number of iSNVs detected in each sample according to the number of days since that animal had its first PCR positive sample, stratified by IAV lineage. (B) Changes in the frequency of synonymous (left), and non-synonymous (right) iSNVs according to the number of days between samples. iSNVs that were detected at both time points are shown in blue. (C) Cumulative distribution showing the proportion of iSNVs that persist between one day and the next with a frequency change that is less than or equal to the frequency change shown on the x-axis. Red line shows results from the swine data; the blue line shows results from humans IAV infections, as calculated from the data provided in [19]. For both datasets, we called iSNVs as present if found at frequencies of ≥3% and ≤97%.