A selective sweep in the Spike gene has driven SARS-CoV-2 human adaptation

Abstract

The coronavirus disease 2019 (COVID-19) pandemic underscores the need to better understand animal-to-human transmission of coronaviruses and adaptive evolution within new hosts. We scanned more than 182,000 severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) genomes for selective sweep signatures and found a distinct footprint of positive selection located around a non-synonymous change (A1114G; T372A) within the spike protein receptor-binding domain (RBD), predicted to remove glycosylation and increase binding to human ACE2 (hACE2), the cellular receptor. This change is present in all human SARS-CoV-2 sequences but not in closely related viruses from bats and pangolins. As predicted, T372A RBD bound hACE2 with higher affinity in experimental binding assays. We engineered the reversion mutant (A372T) and found that A372 (wild-type [WT]-SARS-CoV-2) enhanced replication in human lung cells relative to its putative ancestral variant (T372), an effect that was 20 times greater than the well-known D614G mutation. Our findings suggest that this mutation likely contributed to SARS-CoV-2 emergence from animal reservoirs or enabled sustained human-to-human transmission.

Keywords: COVID-19; SARS-CoV-2; emergence; molecular virology; selective sweep; spillover; viral adaptation.

Graphical abstract

Figure 1

Selective sweeps analysis (A) Selective sweep regions (shown as red blocks) identified in 182,792 SARS-CoV-2 genomes using OmegaPlus (blue lines) and RAiSD (yellow lines). The common outliers (0.05 cutoff, purple dots) from the two methods were used to define selective sweep regions. (B) Non-synonymous difference (Thr372Ala) between SARS-CoV-2 and four other Sarbecovirus members found in the putative selective sweep region (22,529–22,862).

Figure 2

Structure-based analysis of SARS-CoV-2 S protein variants (A) Visualization of the T372 and D614G mutants. The structure of S protein (PDB: 7A94) is displayed as a cartoon and colored by RBD (green), N-terminal domain (NTD; orange), central helix (CH;blue), FP (yellow), and connector domain (CD;pink). Glycans are displayed as spheres colored hot pink. The top panel shows the WT (A372) and T372 mutant, the center panel displays a glycosylated N370 T372 S protein with various rotamers of the GlcNAc-glycosylated N370, and the bottom panel shows the WT and G614 mutant. (B and C) Surface map of the WT S protein (B) and the N370-glycosylated T372 S protein (C), colored by the residue side-chain properties: green for hydrophobic, blue for positively charged, red for negatively charged, teal for polar uncharged, and gray for neutral. (D) Predicted N-glycosylated residues identified by Schrödinger-Maestro’s BioLuminate (v.2020-2) Reactive Residue package with percent solvent-accessible surface area (SASA) exposure of each residue. (E) Predicted N-glycosylated residues identified by the NetNGlyc 1.0 server with the probability of being glycosylated.

Figure 3

Decreased binding of the A372T mutant to human ACE2 (a) Functional ELISA was used to determine the binding affinity of different S protein receptor-binding domains (RBDs). Plates were coated with recombinant human ACE2 receptor (2 μg/mL at 100 μL/well) and then probed with varying concentrations (0.256–4000 ng/mL) of purified RBDs from WT SARS-CoV-2 (S A372), A372T, and N501Y (positive control). To determine EC₅₀ values, the absorbance values (450 nM) were fit to a sigmoidal, 4PL nonlinear model using Prism 9 (GraphPad). The experiment was repeated in two independent replicates with four total technical replicates per sample. Error bars represent standard deviation of the mean. (B) The EC₅₀ values were compared by one-way ANOVA with Dunnett’s multiple comparisons test. ^∗∗∗∗p < 0.0001 compared with WT SARS-CoV-2 (A372). Error bars represent standard deviation of the mean.

Figure 4

A372T substitution decreases SARS-CoV-2 replication on human lung epithelial cells (A) The S T372 SARS-CoV-2 mutant was generated by making a single G-to-A substitution. The mutant nucleotide is presented in red, and the altered codon is highlighted in a yellow box. (B) Plaque morphology of WT and mutant viruses. Plaques were visualized 2 days post-infection (dpi) on Vero E6 cells. (C and D) Viral replication on Vero E6 (C) and Calu-3 (D) cells following infection at an MOI of 0.05. The sample at 0 dpi was collected immediately after infection to ensure cells were exposed to similar levels of virus, and then samples were collected at 24-h intervals. (E and F) Kinetics of thermal stability. A solution of 10⁵ PFU of each virus was incubated at the indicated temperature for different lengths of time. Infectious virus was measured by plaque assay on Vero E6 cells. Statistical comparisons were made using two-way ANOVA with Dunnett’s multiple comparisons test. ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. Error bars represent standard deviation of the mean.