Transmission of SARS-CoV-2 from humans to animals and potential host adaptation

Abstract

SARS-CoV-2, the causative agent of the COVID-19 pandemic, can infect a wide range of mammals. Since its spread in humans, secondary host jumps of SARS-CoV-2 from humans to multiple domestic and wild populations of mammals have been documented. Understanding the extent of adaptation to these animal hosts is critical for assessing the threat that the spillback of animal-adapted SARS-CoV-2 into humans poses. We compare the genomic landscapes of SARS-CoV-2 isolated from animal species to that in humans, profiling the mutational biases indicative of potentially different selective pressures in animals. We focus on viral genomes isolated from mink (Neovison vison) and white-tailed deer (Odocoileus virginianus) for which multiple independent outbreaks driven by onward animal-to-animal transmission have been reported. We identify five candidate mutations for animal-specific adaptation in mink (NSP9_G37E, Spike_F486L, Spike_N501T, Spike_Y453F, ORF3a_L219V), and one in deer (NSP3a_L1035F), though they appear to confer a minimal advantage for human-to-human transmission. No considerable changes to the mutation rate or evolutionary trajectory of SARS-CoV-2 has resulted from circulation in mink and deer thus far. Our findings suggest that minimal adaptation was required for onward transmission in mink and deer following human-to-animal spillover, highlighting the 'generalist' nature of SARS-CoV-2 as a mammalian pathogen.

Fig. 1. Multiple emergences and onward transmission of SARS-CoV-2 in animals.

a Subsampled Audacity tree (n = 16,911) comprising 10 human isolates per PANGO lineage, and all animal isolates shown in Table 1, illustrating the global context of SARS-CoV-2 infections in animals. b Maximum-likelihood tree of all 928 mink isolates, with manually curated cluster names (see “Methods”) and country of isolation annotated.

Fig. 2. Homoplasy and allele frequency analysis.

Scatter plot of putatively adaptive non-synonymous mutations in a mink and b deer. Point size represents the minimum number of independent emergences for each mutation in a phylogeny reconstructed from 928 mink or 95 deer isolates. Human isolates with matching PANGO lineages, from the same countries, and that were sampled within the range of sampling dates of mink (n = 835) or deer isolates (n = 94), were used to compute the human background allele frequencies (human background 1). The dotted red lines and solid black lines, indicate where the allele frequencies in each animal host are two-fold that in humans, and where the human and animal allele frequencies are equal, respectively. Heatmap visualising the proportions of mutation-carrying SARS-CoV-2 isolates within manually curated phylogenetic clusters in c mink and d deer. e Allele frequencies of 20 mink and 34 deer candidate mutations in human background 2. The strongest candidate non-synonymous and synonymous mutations satisfying criteria (A)–(E) are indicated by red and blue boxes, respectively. The genomic region associated with each mutation is given by the colour in panel b.

Fig. 3. Timeline of the COVID-19 pandemic.

a The key events of the pandemic from the estimated emergence of SARS-CoV-2 in humans to the sampling dates of the first isolates for each VoC are annotated in the lowest panel. The coloured rectangles in the upper first and second panels indicate the range of sampling dates of animal-associated SARS-CoV-2 sequences in the different countries. The sampling dates of the earliest human isolates carrying each candidate mutation are annotated along the timeline are indicated by black points. Panels b and c show the temporal distributions of candidate mutations in human SARS-CoV-2 isolates collected prior to 17th March 2022. Red and blue dashed lines indicate the sampling date of the first mink-associated isolate in the Netherlands and deer-associated isolate in the USA, respectively. For panel b, country names were omitted, and the number of countries where the candidate mutations were found in human isolates are annotated. For panel c, countries where human-to-mink transmission has been documented are highlighted in yellow.

Fig. 4. Host-specific genomic landscapes.

a Nucleotide-nucleotide transition frequencies (x-axis) against average mutations observed per isolate in human and animal hosts (as indicated by symbols), and b principal components analysis of all dinucleotide frequencies, stratified by host.

Fig. 5. Host-specific substitution rate variation.

Raincloud plots of terminal branch lengths stratified by a host, and b by both host and country. These plots comprise Gaussian kernel probability density, scatter and box-and-whisker plots (centre line, median; box limits, upper and lower quartiles; whiskers, 1.5x interquartile range). Multiple mink-human maximum-likelihood phylogenies of mink and human background 1 isolates were reconstructed and used for tip-calibration. Isolates that did not have complete dates or that were duplicate sequences were removed prior to analysis. The final number of isolates in each stratum that were used for tip-calibration, Mann–Whitney U-statistics and their associated p-values (based on a two-sided test), are annotated.

Fig. 6. Predicted effects of candidate mutations.

a HADDOCK scores for the Spike:ACE2 complexes. More negative values relative to the WT-Spike:ACE2 complexes (highlighted in grey) indicate stronger binding energy of the complex. b mCSM-PPI2 predicted changes in binding energy (ΔΔG). Negative ΔΔG values are associated with destabilisation of the complex following mutation of the residue and positive values with stabilisation of the complex. Values in blue and red indicate predicted increases or decreases in complex stability respectively.