Recombination events are concentrated in the spike protein region of Betacoronaviruses

Abstract

The Betacoronaviruses comprise multiple subgenera whose members have been implicated in human disease. As with SARS, MERS and now SARS-CoV-2, the origin and emergence of new variants are often attributed to events of recombination that alter host tropism or disease severity. In most cases, recombination has been detected by searches for excessively similar genomic regions in divergent strains; however, such analyses are complicated by the high mutation rates of RNA viruses, which can produce sequence similarities in distant strains by convergent mutations. By applying a genome-wide approach that examines the source of individual polymorphisms and that can be tested against null models in which recombination is absent and homoplasies can arise only by convergent mutations, we examine the extent and limits of recombination in Betacoronaviruses. We find that recombination accounts for nearly 40% of the polymorphisms circulating in populations and that gene exchange occurs almost exclusively among strains belonging to the same subgenus. Although experimental studies have shown that recombinational exchanges occur at random along the coronaviral genome, in nature, they are vastly overrepresented in regions controlling viral interaction with host cells.

Fig 1. Detecting recombination within Betacoronavirus subgenera based on ratios of homoplastic (h) to non-homoplastic (m) polymorphisms.

Within each bivariate plot, black dots and the grey-shaded area denote the median and standard deviation of h/m values of the indicated number of subsampled combinations of genomes; and red dots and pink-shaded area denote the median h/m values and standard deviation for simulated data in which all homoplasies are introduced by convergent mutations. Differences between the distributions of observed and simulated h/m values indicate the extent to which polymorphisms are attributable to recombination. Analyses were performed on the three subgenera for which genomes were sampled to sufficient depths to provide robust results.

Fig 2. Extent of recombination evident for each gene in Sarbecovirus genomes.

For each annotated gene (numbered), we calculated h/m ratios based on available sequences (grey) and on simulated datasets (pink). Box-and-whiskers plots of each gene show median (black line), interquartile range IQR (box), and 1.5*IQR (whiskers), and asterisks (and gene numbers highlighted in red) mark instances in which observed and simulated datasets differed significantly (p < 10⁻⁵, one-sided Wilcoxon test with Bonferroni correction) due to an excess of polymorphisms introduced by recombination.

Fig 3. Signal of recombination in strains of SARS-Cov-2.

Pattern of gene flow based on h/m ratios was analyzed in SARS-CoV-2 using 218 sequenced strains. As in Fig 1, black dots and the grey-shaded area denote the median and standard deviation of h/m values of the indicated number of subsampled combinations of genomes; and red dots and pink-shaded area denote the median h/m values and standard deviation for simulated data in which all homoplasies are introduced by convergent mutations.