Positive selection, genetic recombination, and intra-host evolution in novel equine coronavirus genomes and other members of the Embecovirus subgenus

Abstract

There are several examples of coronaviruses in the Betacoronavirus subgenus Embecovirus that have jumped from an animal to the human host. Studying how evolutionary factors shape coronaviruses in non-human hosts may provide insight into the coronavirus host-switching potential. Equids, such as horses and donkeys, are susceptible to equine coronaviruses (ECoVs). With increased testing prevalence, several ECoV genome sequences have become available for molecular evolutionary analyses, especially those from the United States of America (USA). To date, no analyses have been performed to characterize evolution within coding regions of the ECoV genome. Here, we obtain and describe four new ECoV genome sequences from infected equines from across the USA presenting clinical symptoms of ECoV, and infer ECoV-specific and Embecovirus-wide patterns of molecular evolution. Within two of the four data sets analyzed, we find evidence of intra-host evolution within the nucleocapsid (N) gene, suggestive of quasispecies development. We also identify 12 putative genetic recombination events within the ECoV genome, 11 of which fall in ORF1ab. Finally, we infer and compare sites subject to positive selection on the ancestral branch of each major Embecovirus member clade. Specifically, for the two currently identified human coronavirus (HCoV) embecoviruses that have spilled from animals to humans (HCoV-OC43 and HCoV-HKU1), we find that there are 42 and 2 such sites, respectively, perhaps reflective of the more complex ancestral evolutionary history of HCoV-OC43, which involves several different animal hosts.IMPORTANCEThe Betacoronavirus subgenus Embecovirus contains coronaviruses that not only pose a health threat to animals and humans, but also have jumped from animal to human host. Equids, such as horses and donkeys are susceptible to equine coronavirus (ECoV) infections. No studies have systematically examined evolutionary patterns within ECoV genomes. Our study addresses this gap and provides insight into intra-host ECoV evolution from infected horses. Further, we identify and report natural selection pattern differences between two embecoviruses that have jumped from animals to humans [human coronavirus OC43 and HKU1 (HCoV-OC43 and HCoV-HKU1, respectively)], and hypothesize that the differences observed may be due to the different animal host(s) that each virus circulated in prior to its jump into humans. Finally, we contribute four novel, high-quality ECoV genomes to the scientific community.

Keywords: coronavirus; equine coronavirus; natural selection; tropism shift.

Fig 1

Phylogenetic tree inputs to MEME analyses. Figure (A) is the ECoV-specific spike tree used to infer ECoV-specific selection; all branches were tested. This process was repeated for all ECoV-specific protein-coding sequences analyzed herein. Figure (B) demonstrates how MEME was used to test for the presence of site-specific positive selection on branches of the phylogeny inferred from alignment set two, containing ECoV and other representative Embecovirus member sequences. In the example above, the PHEV ancestral branch (dashed purple line) was selected, and MEME tested for the presence of positive selection on that branch. This process was repeated for all 11 Embecovirus member clades analyzed herein. Several clades have been collapsed and labeled accordingly. All other trees and associated alignments used for the MEME analyses can be found here: https://data.hyphy.org/web/ECoV/ECoV-data/. Trees are displayed with iTOL (52).

Fig 2

An example phylogenetic input for the Contrast-FEL analyses. ECoV and PHEV sequences in trees inferred from each of the four RFPs from alignment set two were analyzed. These two branch sets were used to test for codon-specific selection pressure differences between ECoV and its sister taxon PHEV. Several clades have been collapsed and labeled accordingly. The three other trees and associated alignments used for the Contrast-FEL analyses can be found here: https://data.hyphy.org/web/ECoV/ECoV-data/. This tree is displayed with iTOL (52).

Fig 3

An unrooted, ECoV-specific genome-wide phylogeny reconstructed from full-length genome sequences using RAxML-ng (ML) with 500 bootstrap replicates (bootstrap values are denoted at each node). All recombinant fragments within the genome were identified using RDP5 and then removed, in order to infer a recombinant free phylogeny. This phylogeny demonstrates the intra-taxon relatedness from all publicly available ECoV genomes. Leaf nodes are annotated with accession number and collection location.

Fig 4

All 12 supported ECoV-specific genome-wide recombination events (numbered gray rectangles) as identified by RDP5 mapped. Events 1 through 11 fell within the ORF1ab boundary, and Event 12 began in the S2 subunit of Spike and continued until the end of the genome. ECoV genes represented include (nucleotide boundaries in parentheses as identified in accession no. EF441165.1): ORF1a (210..13478), ORF1b (13479..21595), NS2 accessory protein [NS2] (21610..22446), Hemagglutinin-esterase protein [HE] (22458..23729), Spike protein [S] with the putative cleavage motif highlighted in a red dashed line with red scissors (23744..27835), 4.7 kDa protein (27825..27947), 12.7 kDa protein (28076..28405), Small envelope protein [E] (28392..28646), Membrane protein [M] (28661..29353), Nucleocapsid [N] (29363..30703), and I protein [I] (29424..30044).

Fig 5

Sites under positive selection in the ECoV spike protein. Sites highlighted in black correspond to positively selected sites in an ECoV-specific analysis identified by the MEME method; orange sites indicate those under positive selection on the ancestral branch of ECoVs identified by MEME, and the magenta site indicates the site evolving differently between ECoVs and its sister-taxa (PHEV), identified by the Contrast-FEL method. Site 84 was identified by both the MEME and Contrast-FEL methods. Site 519 was also identified on the HKU23 (camel coronaviruses) ancestral branch, 544 on the HCoV-OC43 ancestral branch, and 764 (within the S1/S2 proteolytic cleavage site) on the HKU14 (rabbit coronaviruses) ancestral branch. Spike S1 and S2 subunits are separated by the putative proteolytic cleavage domain highlighted by both a red dashed line and scissors.

Fig 6

Clade-defining (blue) and intra-clade (orange) sites under positive selection identified in 825 HCoV-OC43 sequences, mapped to HCoV-OC43 (accession no. AAT84354.1). All sites mapped correspond to ungapped amino acid positions. The Spike protein is separated into S1 and S2, where S1 is subdivided into the NTD (Domain A) and CTD (Domain B, C and D). The signal peptide (SP), S1/S2 cleavage site (red dashed line and scissors), and transmembrane domain (TM) are all highlighted.