Impact of ageing on homologous and human-coronavirus-reactive antibodies after SARS-CoV-2 vaccination or infection

Abstract

The endemic human coronaviruses (HCoVs) circulate worldwide yet remain understudied and unmitigated. The observation of elevated levels of HCoV reactive antibodies in COVID-19 patients highlights the urgent necessity of better understanding of HCoV specific immunity. Here, we characterized in-depth the de novo SARS-CoV-2 specific antibody responses and the boosting of HCoV-reactive antibodies after SARS-CoV-2 vaccination or infection in individuals up to 98 years old. All the vaccinees were home-dwelling with no documented SARS-CoV-2 infection before receiving the COVID-19 mRNA vaccine (BNT162b2). The first two vaccine doses elicited potent SARS-CoV-2 spike binding antibodies in individuals up to 80 years. The third dose largely boosted the previously low S2 domain binding and neutralizing antibodies in elderly 80-90 years old, but less so in those above 90 years. The endemic betacoronavirus (HKU1 and OC43) reactive antibodies were boosted in all vaccinees, although to a lesser extent in those above 80 years old. COVID-19 patients had potent elevation of alpha- and betacoronavirus (229E, NL63, HKU1 and OC43) reactive antibodies. In both patients and vaccinees, S2 domain specific antibody increases correlated with SARS-CoV-2 neutralizing and HCoV-reactive antibody responses in all ages, indicating S2 domain as a candidate for future universal coronavirus vaccine design.

Fig. 1. SARS-CoV-2 specific antibody responses vary across ages.

a Illustration of the study design. b, c SARS-CoV-2 spike binding IgG concentrations in the vaccinees (b) and patients (c) were measured against the full-length spike (top), S1 (middle) and S2 domain (bottom) from the ancestral Wuhan-Hu-1 strain of SARS-CoV-2. d, e Neutralizing antibody titers in the vaccinees (d) and patients (e) were measured against the live SARS-CoV-2 Bergen-1 virus. f–i Probability of subjects having indicated fold-induction (FI) in spike binding IgG (f, g) and neutralizing antibodies (h, i). From all vaccinees (f and h) fold-inductions were calculated after the first (left), second (center), and third dose vaccines (right). From the infected patients (g and i) after the convalescent (conv.) phase. The antibody concentrations (binding antibody unit (BAU)/ml) and titers were calculated as geomean of the nearest 5 ages in (b, e). The probabilities were calculated for the group of subjects at the nearest 5 ages in (f–i). Nonlinear fitting curves are plotted. Duplicates were performed in all experiments.

Fig. 2. Different kinetics of homologous antibody increases in adult, elderly vaccinees and patients.

a Fold-inductions and b fold-waning of SARS-CoV-2 full-length spike (top), S1 (middle) and S2 domain (bottom) binding IgG. c Fold-inductions and d fold-waning of neutralizing antibodies in the vaccinees and patients. The fold-changes are shown in boxplot, in which center line represents the median, bounds of the box represent 25–75 percentile, and whiskers represent 10–90 percentile. *P < 0.05, **P < 0.01, ***P < 0.001, ns not significant, conv. convalescent (Antibody fold-inductions and fold-wanings were Ln transformed in statistical analyses. Unpaired t test was performed between vaccinees below (n = 25) and above 80 years old (n = 55) at each time point. Ordinary one-way ANOVA and Dunnett’s multiple comparisons were performed between COVID-19 patients (n = 29) and the vaccinees at different time points, where * is marked in red when fold-inductions and fold-wanings were higher in the patients than vaccinees, and in black otherwise). The subjects with no detectable neutralizing antibody titer after 2nd dose were excluded in fold-waning analyses. Duplicates were performed in all experiments.

Fig. 3. Antibody responses cross-reactive to endemic HCoV.

a Phylogenetic tree shows the genetic divergence among spike proteins from the SARS-CoV-2 and four endemic human coronaviruses (HCoV) tested in enzyme-linked immunosorbent assay (ELISA) and neutralization assays. Phylogenetic analyses were performed at ngPhylogeny.fr. b, d Endemic HCoVs spike binding IgG concentrations in the vaccinees and patients were measured against the full-length spike proteins from alphacoronavirus (b, 229E and NL63) and betacoronavirus (d, HKU1 and OC43). c, e HCoV specific neutralizing antibody titer in the vaccinees and patients were measured against the genuine NL63 and OC43 viruses. f, g Landscape plotting of spike binding IgG (f) and neutralizing antibody (g) fold-inductions in the vaccinees below (top, n = 25), above 80 years old (middle, n = 55) and patients (bottom, n = 29). The geometric mean titers are shown as bars, and each symbol represents one subject in (b–e). The geometric means of fold-induction are connected as curves in (f, g). *P < 0.05, **P < 0.01, ***P < 0.001, conv. convalescent (Antibody concentrations (binding antibody unit (BAU)/ml) and titers and fold-inductions were Ln transformed in statistical analyses. RM two-way ANOVA with the Geisser-Greenhouse correction and Turkey’s multiple comparisons were performed among time points in (b–e). * is marked in red when the antibody level was lower after vaccination, and in black otherwise). The horizontal dotted lines indicate fold-induction of 1 in (f, g). Duplicates were performed in all experiments.

Fig. 4. Associations between SARS-CoV-2 homologous and endemic HCoV cross-reactive antibody increases in patients.

a, b Correlation between the four endemic HCoV spike binding IgG fold-inductions and the SARS-CoV-2 full-length (FL, a) spike and S2 domain (b) binding IgG increase in the patients. Each symbol represents one subject. (Spike binding IgG and neutralizing antibody fold-inductions were Ln transformed before performing Pearson correlation analyses. Linear fitting curves are plotted when Pearson P < 0.05. Pearson r and P values are noted in each correlation.).

Fig. 5. Correlations between vaccine specific and cross-reactive antibody responses in vaccinees.

a Correlations between SARS-CoV-2 S2 domain binding IgG fold-induction and full-length (FL) spike binding IgG or neutralizing antibody fold-inductions in vaccinees below (left) and above 80 years old (right). b Matrix of Pearson correlation coefficients r (top) and P values (bottom) among SARS-CoV-2 full-length (FL) spike, S2 domain binding IgG and neutralizing antibody fold-inductions in vaccinees below (left) and above 80 years old (right). c Correlations between SARS-CoV-2 S2 domain binding IgG and endemic HCoV HKU1 and OC43 spike binding IgG fold-inductions in vaccinees below (left) and above 80 years old (right). d Matrix of Pearson correlation coefficients r (top) and P values (bottom) among SARS-CoV-2 full-length (FL) spike, S2 domain binding IgG and endemic HCoV HKU1 and OC43 spike binding IgG fold-inductions in vaccinees below (left) and above 80 years old (right). (IgG and neutralizing antibody fold-inductions were Ln transformed in statistical analyses. Pearson correlation analyses were performed, and linear fitting curves are plotted as solid lines when Pearson P < 0.05, and as dashed lines when Pearson 0.05 ≤ P < 0.2.