SARS-CoV-2 Infection Severity Is Linked to Superior Humoral Immunity against the Spike

Abstract

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is currently causing a global pandemic. The antigen specificity of the antibody response mounted against this novel virus is not understood in detail. Here, we report that subjects with a more severe SARS-CoV-2 infection exhibit a larger antibody response against the spike and nucleocapsid protein and epitope spreading to subdominant viral antigens, such as open reading frame 8 and nonstructural proteins. Subjects with a greater antibody response mounted a larger memory B cell response against the spike, but not the nucleocapsid protein. Additionally, we revealed that antibodies against the spike are still capable of binding the D614G spike mutant and cross-react with the SARS-CoV-1 receptor binding domain. Together, this study reveals that subjects with a more severe SARS-CoV-2 infection exhibit a greater overall antibody response to the spike and nucleocapsid protein and a larger memory B cell response against the spike.IMPORTANCE With the ongoing pandemic, it is critical to understand how natural immunity against SARS-CoV-2 and COVID-19 develops. We have identified that subjects with more severe COVID-19 disease mount a more robust and neutralizing antibody response against SARS-CoV-2 spike protein. Subjects who mounted a larger response against the spike also mounted antibody responses against other viral antigens, including the nucleocapsid protein and ORF8. Additionally, this study reveals that subjects with more severe disease mount a larger memory B cell response against the spike. These data suggest that subjects with more severe COVID-19 disease are likely better protected from reinfection with SARS-CoV-2.

Keywords: SARS-CoV-2; humoral immunity; infection severity; memory B cells; neutralizing antibodies.

FIG 1

Antibody specificity and kinetics in SARS-CoV-2-infected subjects. (a) Proportion of subjects in the acutely infected and convalescent cohorts who had detectable antibodies (total Ig) to one or more SARS-CoV-2 antigens. Number in center represents the number of subjects tested in each cohort. (b) Proportion of subjects in the acutely infected (n = 35) and convalescent (n = 105) cohorts with total Ig binding spike, N protein, ORF7a, ORF8, or at least one NSP antigen. (c and d) Total Ig endpoint titers of antibodies targeting spike, N protein, ORF7a, and ORF8 in the acutely infected cohort (c) (n = 35) and convalescent cohort (d) (n = 105). Lines connect titers across one subject. Data in panel b were analyzed using Fisher’s exact tests for statistical analyses. Data in panels c and d were analyzed using paired nonparametric Friedman tests. Dashed lines in panels c and d are the limit of detection.

FIG 2

Acutely infected subjects with longer hospitalizations have a higher antibody response against N protein. (a) Heatmap of hierarchical clustering of acutely infected subjects (n = 35) based on antibody binding specificity and antibody isotype/subclass. Subjects clustered into three distinct clusters: high (n = 15), mid (n = 7), and low (n = 13) responders. (b and c) Days since symptom onset (high, n = 14; mid, n = 7; low, n = 10) (b) and length of hospitalization (c) among subjects in the high, mid, and low responder clusters. (d) PCA biplot of subjects clustering based on distinct antibody binding features. (e and f) Total Ig antibody titers against the spike (e) and N protein (f) among the high, mid, and low responder clusters. Data in panels b, c, e, and f were analyzed using unpaired nonparametric Kruskal-Wallis tests. Dashed lines in panels e and f are the limit of detection. Bars in panels b, c, e, and f represent the median. Unless noted otherwise (b), number of subjects per responder group is the following: high (n = 15), mid (n = 7), and low (n = 13) responders. Data in panels b, c, e, and f are presented as the median with interquartile range.

FIG 3

Convalescent subjects with higher antibody responses against multiple SARS-CoV-2 antigens tended to have a more severe infection. (a) Heatmap of hierarchical clustering of convalescent subjects (n = 105) based on antibody binding specificity and antibody isotype/subclass. Subjects clustered into three distinct clusters: high (n = 23), mid (n = 42), and low (n = 40) responders. (b to d) Infection severity (b), age (c), and sex (d) of subjects in the high, mid, and low responder clusters. (e) PCA biplot of subjects clustering based on distinct antibody binding features. (f) Total Ig antibody titers against the spike among the high, mid, and low responder clusters. (g) Neutralization titer, as determined by viral cytopathic effect, of 20 randomly selected samples from each of the high, mid, and low responder clusters. Data in panels f and g were analyzed using unpaired nonparametric Kruskal-Wallis tests. For panels b to d, data were analyzed using Fisher’s exact tests. Dashed lines in panels f and g are the limit of detection. Bars in panels f and g represent the median. Data in panels c, f, and g are presented as the median with interquartile range.

FIG 4

MBC response is largely driven against the spike. (a and b) PBMCs from convalescent donors were polyclonally stimulated, and ELISpots were performed to assess the number of antigen-specific MBCs. (a) Number of IgG/IgA⁺ MBCs (antigen-specific MBCs per 10⁶ cells) targeting the spike, N protein, or ORF8 (n = 36). Lines connect antigen-specific MBCs across subjects. (b) Number of spike-targeting IgG/IgA⁺ MBCs among the high (n = 14), mid (n = 15), and low (n = 11) responder clusters. (c) Spearman correlation of the number of anti-spike IgG/IgA⁺ MBCs and anti-spike endpoint titers by individual (n = 40). Data in panel a were analyzed using paired nonparametric Friedman tests. Data in panel b were analyzed using unpaired nonparametric Kruskal-Wallis tests. Data in panel c were analyzed by a nonparametric two-tailed Spearman correlation. Data in panel b are presented as the median with interquartile range.

FIG 5

Antibody cross-reactivity to G614 spike mutant and SARS-CoV-1 and MERS-CoV RBD. (a and b) Endpoint titers of total Ig antibodies binding to the WT (D614) and mutant (D614G) SARS-CoV-2 spike protein from the acute (a) (n = 35) and convalescent (b) (n = 105) cohorts. (c) Correlation of total Ig endpoint titers against the WT (D614) and mutant (D614G) spike from the convalescent cohort (n = 105). (d and e) Correlation between SARS-CoV-2 RBD total Ig endpoint titers and SARS-CoV-1 RBD (d) or MERS-CoV RBD (e) total Ig endpoint titers from convalescent subjects (n = 105). (f and g) SARS-CoV-1 RBD total Ig endpoint titers among the high, mid, and low responder clusters from the acutely infected cohort (f) (high, n = 23; mid, n = 42; and low, n = 40) and the convalescent cohort (g) (high, n = 23; mid, n = 42; and low, n = 40). Data in panel a were analyzed using a two-tailed Wilcoxon matched-pairs signed-rank test. Data in panel b were analyzed using a two-tailed paired t test. For panels c to e, data were analyzed using a two-tailed Pearson correlation. Data in panels f and g were analyzed using unpaired nonparametric Kruskal-Wallis tests. Dashed lines in panels a, b, f, and g are the limit of detection. Bars in panels f and g represent the median. Data in panels f and g are presented as the median with interquartile range.