Genomewide analysis of reassortment and evolution of human influenza A(H3N2) viruses circulating between 1968 and 2011

Abstract

Influenza A(H3N2) viruses became widespread in humans during the 1968 H3N2 virus pandemic and have been a major cause of influenza epidemics ever since. These viruses evolve continuously by reassortment and genomic evolution. Antigenic drift is the cause for the need to update influenza vaccines frequently. Using two data sets that span the entire period of circulation of human influenza A(H3N2) viruses, it was shown that influenza A(H3N2) virus evolution can be mapped to 13 antigenic clusters. Here we analyzed the full genomes of 286 influenza A(H3N2) viruses from these two data sets to investigate the genomic evolution and reassortment patterns. Numerous reassortment events were found, scattered over the entire period of virus circulation, but most prominently in viruses circulating between 1991 and 1998. Some of these reassortment events persisted over time, and one of these coincided with an antigenic cluster transition. Furthermore, selection pressures and nucleotide and amino acid substitution rates of all proteins were studied, including those of the recently discovered PB1-N40, PA-X, PA-N155, and PA-N182 proteins. Rates of nucleotide and amino acid substitutions were most pronounced for the hemagglutinin, neuraminidase, and PB1-F2 proteins. Selection pressures were highest in hemagglutinin, neuraminidase, matrix 1, and nonstructural protein 1. This study of genotype in relation to antigenic phenotype throughout the period of circulation of human influenza A(H3N2) viruses leads to a better understanding of the evolution of these viruses.

Importance: Each winter, influenza virus infects approximately 5 to 15% of the world's population, resulting in significant morbidity and mortality. Influenza A(H3N2) viruses evolve continuously by reassortment and genomic evolution. This leads to changes in antigenic recognition (antigenic drift) which make it necessary to update vaccines against influenza A(H3N2) viruses frequently. In this study, the relationship of genetic evolution to antigenic change spanning the entire period of A(H3N2) virus circulation was studied for the first time. The results presented in this study contribute to a better understanding of genetic evolution in correlation with antigenic evolution of influenza A(H3N2) viruses.

FIG 1

ML trees of all segments of A(H3N2) viruses circulating between 1968 and 2011. The ML trees of PB2, PB1, PA, HA, NP, NA, M, and NS were generated with 286 nucleotide sequences per segment. Scale bars roughly represent 5% of nucleotide substitutions between close relatives. The color coding of viruses is based on the antigenic clusters of HA (13, 14) and is consistent between all trees. Trees were rooted on A/Hong Kong/1/1968.

FIG 2

Reassortment events between segments of A(H3N2) viruses circulating between 1968 and 2011. Tanglegrams are displayed with the ML HA tree on the left side and the mirrored trees of PB2 (A), PB1 (B), PA (C), NP (D), NA (E), M (F), and NS (G) on the right side. Twines between both trees are color coded according to the antigenic clusters of HA (see the legend to Fig. 1). GiRaF was used to detect reassortment events between all segments. Reassortment events with a support of 100% between HA and any other segment are depicted as bold twines, and numbers correspond to reassortment events in Table 1. The arrows indicate the introduction of new segments by reassortment into the population that persisted until the last sampled isolate.

FIG 3

Time of circulation of the MRCA of A(H3N2) viruses of each antigenic cluster. The time of circulation of the MRCA of each genomic segment is shown. The values shown represent the means (spheres) and 95% highest posterior density intervals (error bars) for the times of circulation of the MRCAs estimated across trees sampled using Bayesian MCMC analyses. Dashed lines show the date of the first sampled isolate of each antigenic cluster. Arrows show the time span from the first sampled isolate until the last sampled isolate of the data set for each antigenic cluster. Asterisks represent two data points on top of each other (i.e., BE89/WU95 and SY97/FU02). Color coding was done according to the antigenic clusters of HA (see the legend to Fig. 1).

FIG 4

Rates of amino acid evolution of all segments and ORFs of A(H3N2) viruses. Phylogenetic trees were generated with 286 amino acid sequences for PB2, PB1, PB1-N40, PB1-F2, PA, PA-N155, PA-N182, PA-X, HA0, HA1, HA2, NP, NA, M1, M2, NS1, and NEP. For all A(H3N2) viruses, the amino acid distance of each ORF to A/Hong Kong/1/68 was calculated from the phylogenetic tree and was plotted as a function of time. The color coding of viruses is based on the antigenic clusters of HA and is consistent between all plots (see the legend to Fig. 1). Note that the vertical axes differ between proteins with lower rates of amino acid substitution (four upper rows) and proteins with higher rates of amino acid substitution (two bottom rows).