Phylodynamics of H1N1/2009 influenza reveals the transition from host adaptation to immune-driven selection

Abstract

Influenza A H1N1/2009 virus that emerged from swine rapidly replaced the previous seasonal H1N1 virus. Although the early emergence and diversification of H1N1/2009 is well characterized, the ongoing evolutionary and global transmission dynamics of the virus remain poorly investigated. To address this we analyse >3,000 H1N1/2009 genomes, including 214 full genomes generated from our surveillance in Singapore, in conjunction with antigenic data. Here we show that natural selection acting on H1N1/2009 directly after introduction into humans was driven by adaptation to the new host. Since then, selection has been driven by immunological escape, with these changes corresponding to restricted antigenic diversity in the virus population. We also show that H1N1/2009 viruses have been subject to regular seasonal bottlenecks and a global reduction in antigenic and genetic diversity in 2014.

Figure 1. Evolution of human H1N1/2009 viruses and structural mapping of positively selected sites on the HA molecule.

(a) Maximum likelihood phylogeny of 2,280 human H1-HA sequences from 2009 to 2014 with branches coloured by year of virus isolation. Representative amino-acid substitutions are mapped at the major tree nodes. Sites marked with asterisks (*) indicate positively selected sites identified by MEME method, with the significance level at P<0.05. Scale bar represents number of substitutions per site. (b) Mapping of positively selected amino-acid sites identified by MEME method onto the three-dimensional structure of the HA glycoprotein of H1N1/2009 virus (Protein Data Bank code: 3LZG) for the pandemic (2009–2010) and post-pandemic (2011–2014) periods. The monomer shows the HA1 subunit in light blue and the HA2 subunit in green. The numbers in coloured circles denote codon alignment number and their locations in the three-dimensional structure are indicated with red arrows.

Figure 2. Selection pressures on human H1N1/2009 viruses over time.

Yearly estimates of global d_N/d_S ratios for each gene using SLAC method, with the significance level at P<0.05. Each gene data set per epidemic year consisted of up to 500 randomly selected global sequences. The means of d_N/d_S ratios are indicated by diamond-shaped symbols with error bars representing 95% confidence intervals. Coloured bars correspond to different genes in the following order: polymerase basic 2 (PB2), polymerase basic 1 (PB1), polymerase acidic (PA), haemagglutinin (HA), nucleoprotein (NP), neuraminidase (NA), two matrix proteins (M1 and M2), and two non-structural proteins (NS1 and NS2).

Figure 3. Antigenic evolution of human H1N1/2009 viruses.

(a) Maximum likelihood phylogeny of the HA gene of Australian viruses used in the HI assay. Coloured branches represent the year of virus isolation on the right. Scale bar represents number of substitutions per site. (b) Antigenic cartography of H1N1/2009 viruses reconstructed based on HI assays of pandemic and post-pandemic strains against a panel of polyclonal antisera. Coloured dots correspond to the year of virus isolation. The position of viruses isolated in 2009 is indicated by the black arrow.

Figure 4. Temporal phylogenies and epidemic patterns of human influenza viruses.

Evolution of the HA genes of (a) H1N1/2009 viruses (2009–2014), (b) seasonal H1N1 viruses (2000–February 2009), (c) H3N2 viruses (2009–2014) and (d) influenza B viruses (2009–2014). Phylogenies were inferred using the uncorrelated lognormal relaxed clock model with branches coloured by year of virus isolation and relative genetic diversity estimated using the Gaussian Markov Random Field (GMRF) model. Solid black lines in the GMRF plot represent mean relative genetic diversity while the corresponding grey shades indicate the 95% HPD intervals.

Figure 5. Comparative phylogenetic analyses and population dynamics of H1-HA viruses circulating in different geographical regions, 2009–2014.

Phylogenies were inferred using the uncorrelated lognormal relaxed clock model and relative genetic diversity estimated using the Gaussian Markov Random Field (GMRF) model. Solid lines in the GMRF plot represent the mean relative genetic diversity through time, while the corresponding shaded areas indicate the 95% HPD intervals.

Figure 6. Proportional ancestral location states on the phylogenetic tree trunk estimated for each location over time.

The waiting time between geographic location transitions was inferred using the continuous-time Markov chain model. The trunk reward proportion for each geographic location from 2009 through 2014 was determined from an analysis of 2,225 HA sequences of H1N1/2009 viruses. Shaded areas represent the trunk proportions over time for the nine geographical locations included in the analysis.

Figure 7. Spatial diffusion of H1N1/2009 viruses.

Spatial diffusion pathways and histograms of total number of state transitions for (a,b) 2009–2010, (c,d) 2010–2011, (e,f) 2011–2012, (g,h) 2012–2013 and (i,j) 2013–2014. Global epidemic year was defined as October to the following September based on observed epidemics in Fig. 3a. Significant epidemiological unidirectional pathways from one location to another are indicated on the maps. Thickness of coloured lines represents statistically supported migration rates with a mean indicator of >0.5: pink arrows, decisive rates with BF≥1,000; blue arrows, very strongly supported rates with 100≤BF<1,000; dark green arrows, strongly supported rates with 10≤BF<100; and light green arrows, supported rates with 3≤BF<10. AF, Africa; AUS, Australia; EA, East Asia; EU, Europe; NAm, North America; SEA, Southeast Asia; SAm, South America; SA, South Asia.

Figure 8. Genomic reassortment and global genetic diversity of H1N1/2009 viruses.

(a) Multidimensional scaling plot of uncertainty of TMRCA between samples of 500 trees for each segment of pandemic H1N1/2009 viruses sampled between 2009 and 2014. The space occupied by human H3N2 viruses is indicated by the dashed circle. (b) The TMRCA of each genomic segment of H1N1/2009 viruses circulating in each year. The TMRCAs of all segments in 2014 were dated more recently than 2009–2013 lineages, indicative of reduced genetic diversity across the whole genome. Values shown indicate the mean TMRCAs and the 95% HPD intervals from the Bayesian MCMC analysis. The diagonal line is traced along January of each year.