Vaccination has minimal impact on the intrahost diversity of H3N2 influenza viruses

Abstract

While influenza virus diversity and antigenic drift have been well characterized on a global scale, the factors that influence the virus' rapid evolution within and between human hosts are less clear. Given the modest effectiveness of seasonal vaccination, vaccine-induced antibody responses could serve as a potent selective pressure for novel influenza variants at the individual or community level. We used next generation sequencing of patient-derived viruses from a randomized, placebo-controlled trial of vaccine efficacy to characterize the diversity of influenza A virus and to define the impact of vaccine-induced immunity on within-host populations. Importantly, this study design allowed us to isolate the impact of vaccination while still studying natural infection. We used pre-season hemagglutination inhibition and neuraminidase inhibition titers to quantify vaccine-induced immunity directly and to assess its impact on intrahost populations. We identified 166 cases of H3N2 influenza over 3 seasons and 5119 person-years. We obtained whole genome sequence data for 119 samples and used a stringent and empirically validated analysis pipeline to identify intrahost single nucleotide variants at ≥1% frequency. Phylogenetic analysis of consensus hemagglutinin and neuraminidase sequences showed no stratification by pre-season HAI and NAI titer, respectively. In our study population, we found that the vast majority of intrahost single nucleotide variants were rare and that very few were found in more than one individual. Most samples had fewer than 15 single nucleotide variants across the entire genome, and the level of diversity did not significantly vary with day of sampling, vaccination status, or pre-season antibody titer. Contrary to what has been suggested in experimental systems, our data indicate that seasonal influenza vaccination has little impact on intrahost diversity in natural infection and that vaccine-induced immunity may be only a minor contributor to antigenic drift at local scales.

Fig 1. Viral shedding by vaccination status.

Genome copy number per μl of transport media was determined by RT-qPCR all samples from the 2007–2008 season. (A) Copy number by vaccination status. IIV, inactivated influenza vaccine; LAIV, live attenuated influenza vaccine. (B) Copy number by day of infection (onset of symptoms is day 0) and vaccination status. Circle, IIV; Square, LAIV; Triangle, Placebo. (C) Copy number by HAI titer (D) Copy number by NAI titer. There were no differences among any of the groups by one-way ANOVA with Bonferroni correction.

Fig 2. Phylogenetic trees of HA and NA consensus sequences from the 2007–2008 season.

Maximum likelihood trees of HA (left) and NA (right) with tips coded by vaccine status and pre-season HAI (left; blue >40, magenta <40) or NAI (right; blue ≥40, magenta <40) titer. HAI (left) and NAI (right) titers are shown on tips as well. Outgroups are HA (EU103823.1) and NA (CY114383.1) for the vaccine strain A/Wisconsin/67/2005. Bootstrap values (n = 1000 bootstraps) are shown and nodes with bootstrap values <50 are collapsed for easier visualization.

Fig 3. Intrahost diversity in samples from the 2007–2008 season.

(A) Histogram of the number of iSNV at a given frequency. Bin width = 0.01. (B) Histogram of the number of samples in which each iSNV is found. Arrows indicate bars with one SNV, which are hard to discern in the histogram. These polymorphic SNV, at PB2 position 900 and PA position 515 respectively, were found at 4–6% frequency within hosts and in similar numbers of individuals across vaccination groups. (C) Number of HA iSNV per sample stratified by pre-season HAI titer. >40 = serologically immune, <40 = not serologically immune. (D) Number of NA iSNV per sample stratified by pre-season NAI titer. ≥40 = serologically immune, <40 = not serologically immune.

Fig 4. Temporal patterns of intrahost diversity.

Number of genome-wide iSNV per sample (y-axis) by day of symptoms (x-axis) stratified by (A) recipients of IIV, magenta; LAIV, blue; placebo, white (B) HAI >40, magenta; HAI <40, white (C) NAI ≥40, magenta; NAI <40, white. Mean number of iSNV in each group is indicated (bar).

Fig 5. Structural mapping of HA variants.

A homology model of the A/Brisbane/10/2007 (H3N2) HA trimer is shown, with each monomer represented by a different color (purple, grey, and teal) and HA1 and HA2 designated by lighter and darker shades of the same color, respectively. (A) A side view of HA. All identified non-synonymous mutations and known antigenic amino acid positions are shown as balls on the grey monomer. Those variants colored red are within known antigenic sites, while light orange mutations are not. (B) A side and top view of HA. All amino acid positions within known antigenic sites are displayed as balls, with the antigenic mutations identified here shown in red on all three monomers.