Cell type-specific adaptation of the SARS-CoV-2 spike

Abstract

Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) can infect various human tissues and cell types, principally via interaction with its cognate receptor angiotensin-converting enzyme-2 (ACE2). However, how the virus evolves in different cellular environments is poorly understood. Here, we used experimental evolution to study the adaptation of the SARS-CoV-2 spike to four human cell lines expressing different levels of key entry factors. After twenty passages of a spike-expressing recombinant vesicular stomatitis virus (VSV), cell-type-specific phenotypic changes were observed and sequencing allowed the identification of sixteen adaptive spike mutations. We used VSV pseudotyping to measure the entry efficiency, ACE2 affinity, spike processing, TMPRSS2 usage, and entry pathway usage of all the mutants, alone or in combination. The fusogenicity of the mutant spikes was assessed with a cell-cell fusion assay. Finally, mutant recombinant VSVs were used to measure the fitness advantage associated with selected mutations. We found that the effects of these mutations varied across cell types, both in terms of viral entry and replicative fitness. Interestingly, two spike mutations (L48S and A372T) that emerged in cells expressing low ACE2 levels increased receptor affinity, syncytia induction, and entry efficiency under low-ACE2 conditions. Our results demonstrate specific adaptation of the SARS-CoV-2 spike to different cell types and have implications for understanding SARS-CoV-2 tissue tropism and evolution.

Keywords: SARS-CoV-2 Spike; SARS-CoV-2 variants; cell–cell fusion; experimental evolution; viral tropism; virus–host interactions.

Figure 1.

Experimental evolution of SARS-CoV-2 spike reveals cell type-specific adaptations. (A) Schematic representation of the experimental evolution. (B) Expression of ACE2, TMPRSS2, FURIN, and CTSL mRNAs in the four cell lines used in the experimental evolution was measured by RT-qPCR. Each dot represents an independent experiment (n = 1–2). (C) Representative images of infection with the founder and evolved lineages (passage 20) in their own cell line. A549-A: 48 hpi, A549-AT: selected times indicated in each image, Huh-7: 16 hpi, IGROV-1: 16 hpi. Scale bar: 400 µm. (D) Mutations fixed in each lineage at passage 20 were identified by Sanger sequencing. Amino acid positions are indicated at the top in gray numbers and colored boxes depict functional domains.

Figure 2.

Functional characterization of mutations observed after 20 passages. (A) Heat map showing cell type-dependent effects of each mutation on viral entry. VSV pseudotypes were titrated in the four cell lines. Within each cell line, titers were first normalized to the average titer of all mutants (log(titer)—mean(log(titer))). For each mutant, normalized titers were then normalized to their average across the four cell lines. Results represent the average of 2–3 independent experiments. Mutations are colored according to the cell line in which they were detected. (B) Anti-S2 Western blot of pseudotyped VSV particles (top panel). Spike processing was calculated as the proportion of S2 over total spike (bottom panel). Each dot represents an independent blot (n = 2). (C) TMPRSS2 usage of pseudotyped VSV particles was measured as the ratio of the titer in Vero E6-T to the titer in Vero E6 cells. Asterisks indicate the P-values obtained from a randomized-block one-way ANOVA with Dunnett’s correction for multiple comparisons, comparing all mutants to the Wuhan-Hu-1 reference. Data are shown as mean ± SEM, each dot corresponding to an independent experiment (n = 3). (D) Infection inhibition reached with a 100 µM of camostat mesylate or E-64d dose. Data are shown as mean ± SEM (n = 3 independent experiments). Asterisks indicate the P-values obtained from a one-way ANOVA with Dunnett’s correction for multiple testing, comparing all mutants to the Wuhan-Hu-1 reference. (E) Cell–cell fusogenicity of the S mutants. A schematic of the GFP-Split assay is presented on top. The assay was performed in the presence (top graph) or absence (bottom graph) of human ACE2 overexpression. Data are shown as the log10 of the GFP confluence:cell confluence ratio normalized to that of the Wuhan-Hu-1 reference (mean ± SEM, n = 3–4 independent experiment). Asterisks show the P-values obtained from an ordinary one-way ANOVA with Dunnett’s multiple test correction, comparing all mutants to the Wuhan-Hu-1 reference. Representative images from statistically significant results are shown at the bottom. Scale bar: 800 µm. (F) Infection inhibition assay by soluble ACE2 against WT, L48S or A372T VSV pseudotypes. Each dot is the average of three technical replicates, lines correspond to a sigmoidal 2-parameter fit, and shaded areas correspond to 95 per cent confidence intervals.

Figure 3.

Cell type-dependent effect of spike mutations on the fitness of recombinant viruses. (A) Representative images of infection of the four cell lines with VSVs expressing Wuhan-Hu-1 or mutant spikes. Images were acquired at different time points for each cell line (92 hpi in A549-A, 9 hpi in A549-AT, 48 hpi in IGROV-1, and 76 hpi in Huh-7). Scale bar: 800 µm. (B) Viral spread in the cell culture was quantified by measuring AUC over the course of infection. Data are mean ± SEM are shown, each dot corresponding to an independent experiment (n = 3). Asterisks correspond to P-values obtained from a randomized-block one-way ANOVA with Dunnett’s correction for multiple tests, comparing all mutants to the Wuhan-Hu-1 reference. (C) Titration of supernatants at 12 hpi (A549-AT), 48 hpi (IGROV-1), or 72 hpi (A549-A and Huh-7). Data are shown as mean ± SEM, each dot corresponding to an independent experiment (n = 3). Asterisks correspond to P-values obtained from a randomized-block one-way ANOVA with Dunnett’s correction for multiple tests, comparing all mutants to the Wuhan-Hu-1 reference.