Salivary IgA and vimentin differentiate in vitro SARS-CoV-2 infection: A study of 290 convalescent COVID-19 patients

Abstract

SARS-CoV-2 initially infects cells in the nasopharynx and oral cavity. The immune system at these mucosal sites plays a crucial role in minimizing viral transmission and infection. To develop new strategies for preventing SARS-CoV-2 infection, this study aimed to identify proteins that protect against viral infection in saliva. We collected 551 saliva samples from 290 healthcare workers who had tested positive for COVID-19, before vaccination, between June and December 2020. The samples were categorized based on their ability to block or enhance infection using in vitro assays. Mass spectrometry and enzyme-linked immunosorbent assay experiments were used to identify and measure the abundance of proteins that specifically bind to SARS-CoV-2 antigens. Immunoglobulin (Ig)A specific to SARS-CoV-2 antigens was detectable in over 83% of the convalescent saliva samples. We found that concentrations of anti-receptor-binding domain IgA >500 pg/µg total protein in saliva correlate with reduced viral infectivity in vitro. However, there is a dissociation between the salivary IgA response to SARS-CoV-2, and systemic IgG titers in convalescent COVID-19 patients. Then, using an innovative technique known as spike-baited mass spectrometry, we identified novel spike-binding proteins in saliva, most notably vimentin, which correlated with increased viral infectivity in vitro and could serve as a therapeutic target against COVID-19.

Fig. 1

Schematic of methods used to study mucosal correlates of protection against in vitro SARS-CoV-2 infection in saliva. Created with Biorender.com. ACE2 = angiotensin-converting enzyme 2.

Fig. 2

Analysis of antibodies against Spike, RBD and Nucleocapsid antigens in health care workers after recovery from COVID-19 in absence of prior vaccination. (A) Distribution of antibody responses to SARS-CoV-2 spike (S), RBD and nucleocapsid antigens for salivary IgA (n = 488) and serum IgG (n = 172). IgA is displayed as pg/μg total protein as measured by ELISA, IgG is displayed as MSD chemiluminescent assay titer. (B) Left panel shows the correlation matrix showing the association of anti-SARS-CoV-2 saliva IgA and systemic IgG by antigen across all timepoints as determined using the cor function of ggcorrplot package using R version 2023.03.1+446. Right panel shows a chord diagram of the results of the correlation matrix. Data excludes Ab responses <1, n = 172 donors. * represents significant correlation p < 0.05. C, Correlation of serum IgG and salivary IgA across different clinic visits. D, Salivary IgA and serum IgG responses in longitudinal samples at sequential clinic visits. E, Distribution of salivary IgA and serum IgG against SARS-CoV-2 antigens for self-reported symptomatic versus asymptomatic participants. Ab = antibody; ELISA = enzyme-linked immunosorbent assay; Ig = immunoglobulin; MSD = Meso Scale Discovery; RBD = receptor binding domain.

Fig. 3

Viral neutralization and RBD-ACE2 inhibition by saliva from health care workers after recovery from confirmed COVID-19 in absence of prior vaccination. (A) Distribution of relative infection of VeroE6 cells (fold-change versus virus-only controls) after pre-incubation of virus with saliva (n = 395). Shapiro-Wilk test of normality, where p value < 0.05 indicates significant deviation from a normal distribution. (B) The relationship between IgA level and relative SARS-CoV-2 infection of VeroE6 cells for S (red), RBD (blue) and N (green). (C) Relative infection analyzed against anti-RBD IgA titer, with solid black bars denoting the median per group. (D) Longitudinal reproducibility of neutralization for donors who provided multiple samples, ordered by average log2 fold change in infectivity per donor, and separated into symptomatic and non-symptomatic groups. (E) Distribution of saliva ability to inhibit the interaction of SARS-CoV-2 RBD with human ACE2 receptor, relative to positive and negative controls (n = 490). (F) A linear model fit was used to show the relationship between saliva IgA and inhibition of in vitro RBD-ACE2 binding for antigens S (red), RBD (blue) and N (green). (G) Inhibition of RBD-ACE2 binding analyzed against anti-RBD IgA titer, with solid black bars denoting the median per group. ACE2 = angiotensin-converting enzyme 2; RBD = receptor binding domain.

Fig. 4

Functional saliva subsets by effect on SARS-CoV-2 infectivity. (A) A graphical representation of sample collection broken down by function. Protective effect on infectivity represents FC >+0.5, detrimental effect represents FC <−0.5. RBD-ACE2 inhibition represents >50% relative inhibition. (B) The relative VeroE6 infection data for each of the samples in subgroups A, B, and C selected for proteomic analysis (n = 10 per group), compared to all other UG samples (n = 365). (C) The relative RBD-ACE2 inhibition data for each of the samples in subgroups A, B, and C selected for proteomic analysis (n = 10 per group), compared to all other UG samples (n = 365). (D) The distribution of salivary IgA concentrations (pg/μg total protein) for the saliva samples in each subgroup, compared to all other UG samples. (E) The distribution of matched serum IgG titers (MSD units) for the saliva samples in each subgroup, compared to UG samples. ACE2 = angiotensin-converting enzyme 2; MSD = Meso Scale Discovery; RBD = receptor binding domain; UG = ungrouped.

Fig. 5

Proteins associated with SARS-CoV-2 infectivity of saliva functional subgroups. (A) Graphical representation of SARS-CoV-2 spike-baited mass spectrometry methodology. Created with Biorender.com. (B) Orthogonal Projections to Latent Structures Discriminant Analysis (OPLS-DA) models between functional groups showing the score plot of the salivary proteins that bind to SARS-CoV-2 spike. (C) Cluster dendrogram of all spike binding proteins detected in saliva. (D) Dot plot showing comparative mean abundance of significantly elevated proteins in groups A–C. (E) Violin plot showing the abundance of most differentially expressed spike-binding proteins detected in Group C (detrimental). Estimated concentration (μg/ml) calculated from mass spectrometry abundance normalized against known mass of spike bait protein (n = 10).

Fig. 6

Vimentin concentration in detrimental saliva is associated with enhanced in vitro SARS-CoV-2 infectivity. (A) The effect of increasing concentrations of recombinant proteins on relative SARS-CoV-2 infection of VeroE6 cells (n = 3; ** = p < 0.005, analysis of variance). (B) Predicted effect of target proteins on relative infection of VeroE6 cells for vimentin abundances in Group C (detrimental) saliva samples as calculated from mass spectrometry (n = 10). (C) Immunofluorescence staining of infected air-liquid interface airway epithelial cells, with cell nuclei (red), spike antigen (blue) and vimentin (white), and a plot of signal co-localization between SARS-CoV-2 spike and cellular vimentin. (D) Graphical representation of proposed mechanisms for enhanced viral entry caused by vimentin. Left panel shows that surface-expressed vimentin may act as a co-receptor for SARS-CoV-2 and ACE2 binding to promote viral entry. Right panel shows extracellular vimentin in mucosal secretions such as saliva may bind to SARS-CoV-2 via the spike protein, stabilizing binding interactions with ACE2 during cell infection. Created with Biorender.com. ACE2 = angiotensin-converting enzyme 2.