Respiratory Syncytial Virus whole-genome sequencing identifies convergent evolution of sequence duplication in the C-terminus of the G gene

Abstract

Respiratory Syncytial Virus (RSV) is responsible for considerable morbidity and mortality worldwide and is the most important respiratory viral pathogen in infants. Extensive sequence variability within and between RSV group A and B viruses and the ability of multiple clades and sub-clades of RSV to co-circulate are likely mechanisms contributing to the evasion of herd immunity. Surveillance and large-scale whole-genome sequencing of RSV is currently limited but would help identify its evolutionary dynamics and sites of selective immune evasion. In this study, we performed complete-genome next-generation sequencing of 92 RSV isolates from infants in central Tennessee during the 2012-2014 RSV seasons. We identified multiple co-circulating clades of RSV from both the A and B groups. Each clade is defined by signature N- and O-linked glycosylation patterns. Analyses of specific RSV genes revealed high rates of positive selection in the attachment (G) gene. We identified RSV-A viruses in circulation with and without a recently reported 72-nucleotide G gene sequence duplication. Furthermore, we show evidence of convergent evolution of G gene sequence duplication and fixation over time, which suggests a potential fitness advantage of RSV with the G sequence duplication.

Figure 1

Bayesian maximum clade credibility trees for RSV-A (A) and RSV-B (B) G gene sequences. Strain names are colored by the presence (red) or absence (blue) of the large G gene duplication, with study samples in darker shades of red and blue. Multiple co-circulating lineages of RSV were observed during the 2012–2013 RSV season. These phylogenies and related analyses suggest that the G gene duplication occurred convergently in two separate genotypes of RSV-A. Bayesian posterior probability >0.9 are provided for key nodes.

Figure 2. Times to most recent common ancestors (tMCRAs) and mean evolutionary rate estimates inferred by Bayesian analyses.

This dataset includes a subset of the available GenBank whole-genome sequences along with the study samples. Estimates are provided for RSV-A (purple) and RSV-B (green) for the whole genome (WG) and each individual gene. (A) Evolutionary rates (substitutions/site/year) for RSV-A and RSV-B datasets and (B) mean tMRCAs for RSV-A and RSV-B datasets are provided with 95% HPD intervals. The whiskers in each plot extend to the full 95% HPD interval, and the boxes indicate the 25–75% interquartile range of the posterior distribution, thus describing its central tendency. Mean whole-genome tMRCA estimates are indicated with arrows: 1951 for RSV-A and 1967 for RSV-B.

Figure 3. Consensus N- and O-linked glycosylation patterns for the seven study genotypes.

The seven genotype-specific consensus glycosylation patterns for O- and N-linked (bars and dots, respectively) glycans are displayed in rows. RSV-A and RSV-B genotypes are indicated with purple and green bars to the right. Each genotype displays a unique glycosylation pattern and duplication status.

Figure 4. Divergence time estimates from a Bayesian divergence dating analysis of the RSV-A G gene sequences.

The GA2.1 clade consists of ON1 genotypes containing only sequences with the G gene duplication, TN1 genotypes containing sequences with mostly non-duplicated G genes and four interleaved G gene duplication sequences, and TN2 genotypes containing only sequences lacking the G gene duplication. Divergence estimates suggest clade GA2.1 originated from a non-duplicated ancestor, with the duplication being convergently gained first in genotypes ON1 and then in TN1. This hypothesis of convergent G gene duplications is supported by divergence estimates that largely do not overlap between genotypes ON1 and TN1. The whiskers in each plot extend to the full 95% HPD interval, and the boxes indicate the 25–75% interquartile range of the posterior distribution, thus describing its central tendency.