The genomic rate of molecular adaptation of the human influenza A virus

Abstract

Quantifying adaptive evolution at the genomic scale is an essential yet challenging aspect of evolutionary biology. Here, we develop a method that extends and generalizes previous approaches to estimate the rate of genomic adaptation in rapidly evolving populations and apply it to a large data set of complete human influenza A virus genome sequences. In accord with previous studies, we observe particularly high rates of adaptive evolution in domain 1 of the viral hemagglutinin (HA1). However, our novel approach also reveals previously unseen adaptation in other viral genes. Notably, we find that the rate of adaptation (per codon per year) is higher in surface residues of the viral neuraminidase than in HA1, indicating strong antibody-mediated selection on the former. We also observed high rates of adaptive evolution in several nonstructural proteins, which may relate to viral evasion of T-cell and innate immune responses. Furthermore, our analysis provides strong quantitative support for the hypothesis that human H1N1 influenza experiences weaker antigenic selection than H3N2. As well as shedding new light on the dynamics and determinants of positive Darwinian selection in influenza viruses, the approach introduced here is applicable to other pathogens for which densely sampled genome sequences are available, and hence is ideally suited to the interpretation of next-generation genome sequencing data.

FIG. 1.

The accumulation of adaptive substitutions in each gene of human influenza A subtype H3N2 during the period 1977–2009. Three plots are shown for each of the envelope proteins HA and NA (grey boxes): One for the whole gene and one for each of the two structural partitions (see Materials and Methods). In each panel, black circles show the estimated number of adaptive substitutions at each time point to which a linear regression model is fitted (black line). Gray lines show the bootstrap distribution of regression lines (1,000 replicates). Error bars represent the 95% bootstrap percentiles at each time point.

FIG. 2.

The accumulation of adaptive substitutions in each gene of human influenza A subtype H1N1 during the period 1977–2009. See figure 1 legend for details.

FIG. 3.

The rate of adaptive substitutions “per codon” per year in human influenza A genes. Rates for subtype H1N1 are in orange and those for H3N2 in blue. Circles indicate the estimated adaptation rates, and error bars show the 95% bootstrap percentiles.

FIG. 4.

The rate of adaptive substitution per codon for H1N1 genes as a function of the threshold value used to define the low and intermediate site-frequency classes (see Methodological Framework). The results in figures 1–3 correspond to a threshold value of 0.15. Error bars show the 95% bootstrap percentiles.