Children develop robust and sustained cross-reactive spike-specific immune responses to SARS-CoV-2 infection

Abstract

SARS-CoV-2 infection is generally mild or asymptomatic in children but a biological basis for this outcome is unclear. Here we compare antibody and cellular immunity in children (aged 3-11 years) and adults. Antibody responses against spike protein were high in children and seroconversion boosted responses against seasonal Beta-coronaviruses through cross-recognition of the S2 domain. Neutralization of viral variants was comparable between children and adults. Spike-specific T cell responses were more than twice as high in children and were also detected in many seronegative children, indicating pre-existing cross-reactive responses to seasonal coronaviruses. Importantly, children retained antibody and cellular responses 6 months after infection, whereas relative waning occurred in adults. Spike-specific responses were also broadly stable beyond 12 months. Therefore, children generate robust, cross-reactive and sustained immune responses to SARS-CoV-2 with focused specificity for the spike protein. These findings provide insight into the relative clinical protection that occurs in most children and might help to guide the design of pediatric vaccination regimens.

Fig. 1. Children and adults develop coordinated antibody responses to SARS-CoV-2.

a, SARS-CoV-2 antibody levels measured by MSD assay in children (n = 91) and adults (n = 154). Serostatus was assigned based on spike serology and used to divide the cohorts into seropositive (red/blue) and seronegative (light red/light blue) (seropositive/negative children n = 43/48, adults n = 91/63, respectively). The dotted lines represent cutoff values for serostatus. Fold change indicates the difference between the GMTs in seropositive children and adults. The bars indicate the geometric mean with 95% confidence interval (CI). b, The level of the spike- and nucleocapsid-specific antibody response was correlated within individual donors and revealed a coordinated response to both proteins. a.u., arbitrary unit. Source data

Fig. 2. Antibody responses to hCoVs are back-boosted by SARS-CoV-2 in children.

a,b, Antibody titers to the seasonal hCoV coronaviruses (a) and other respiratory viruses (b) in children (red) and adults (blue) based on SARS-CoV-2 serostatus (dark, seropositive, light, seronegative; seropositive/negative children n = 43/48, adults n = 91/63, respectively). Fold change indicates the difference between the GMTs in seropositive children and adults. The bars indicate the geometric mean with the 95% CI. Only significant differences are shown. Brown–Forsythe and Welch’s ANOVA with Dunnett’s T3 multiple comparison tests were used. Source data

Fig. 3. SARS-CoV-2 S2 domain antibodies cross-react with hCoV.

Plasma from SARS-CoV-2 seropositive children (n = 21) was assessed for binding to the spike protein of the 4 hCoVs or the spike or nucleocapsid regions of SARS-CoV-2. Plasma was either applied neat (control) or after preabsorption with either recombinant spike S1 domain (spike 1 block) or spike S2 domain (spike 2 block). S1 preabsorption markedly reduced binding to SARS-CoV-2 spike with no effect on hCoV, while S2 preabsorption reduced binding to OC43 and HKU-1. One-way repeated measures ANOVA with Holm–Sidak’s multiple comparison test or Friedman test with Dunn’s multiple comparisons test were used as appropriate. Source data

Fig. 4. Spike-specific T cell responses in SARS-CoV-2 seropositive and seronegative children.

a, SARS-CoV-2-specific T cell responses in children (n = 57, red) and adults (n = 83, blue) based on SARS-CoV-2 serostatus (dark; seropositive, light; seronegative). SARS-CoV-2 serostatus was 37/20 seropositive or negative in children and 64/29 seropositive or negative in adults, respectively. The assay used IFN-γ ELISpot using pepmixes containing overlapping peptides to spike, N/M or influenza and is shown in relation to serostatus. b, The magnitude of the spike-specific cellular response was compared to that against N/M and displayed as a ratio in seropositive and seronegative adults and children, as indicated. The bars indicate the mean. Brown–Forsythe and Welch’s ANOVA with Dunnett’s T3 multiple comparisons tests were used. c, Proportions of individuals within each cohort who demonstrated a cellular response to S or N/M peptides from SARS-CoV-2. d, Cytokine concentration within supernatants from the ELISpot cultures (n = 12 children, red; n = 8 adults, blue). The bars indicate the mean ± s.d. e, hCoV-specific cellular responses showed equivalent expansion after stimulation of PBMCs from SARS-CoV-2 seronegative children with the SARS-CoV-2 S2 domain pepmix (n = 11). Cultures were stimulated for 9 d and then assessed by IFN-γ ELISpot to the pepmix of the S2 domain from SARS-CoV-2 or the Alpha (OC43 and HKU-1) or Beta (NL63 and 229E) hCoV. Expansion is shown relative to unstimulated control cultures. The lines indicate the median. f, SARS-CoV-2-specific T cell response in PBMC samples taken from children before the COVID-19 pandemic (n = 4). Dotted lines in a and f indicate pre-defined positive thresholds. Source data

Fig. 5. Immune responses are maintained in children at least six months after infection.

a, Antibody responses in children (n = 35, red) and adults (n = 81, blue) who were seropositive at first testing and therefore at least 6 months post-primary infection. The bars indicate the GMT ± 95% CI. The dotted lines indicate seropositive cutoffs. Fold change indicates increment in children’s GMT compared to adults. The black dots indicate individuals who were seronegative for spike but retained a nucleocapsid-specific antibody response. b, Proportion of individuals retaining antibody responses to spike, nucleocapsid or RBD at ≥6 months. c, Antibody binding to spike, RBD or nucleocapsid in children ≥6 months (n = 35, red) or ≥12 months (n = 16, dark red) after infection. The bars indicate the GMT ± 95% CI. The dotted lines indicate seropositive cutoffs. A one-way Kruskal–Wallis with Dunn’s multiple comparisons test was used. d, Paired antibody levels for children (n = 6) at ≥6 and ≥12 months post-primary infection. The dotted lines indicate positive cutoffs. e, Spike-specific IFN-γ ELISpot in children (n = 27, red) and adults (n = 52, blue). The black dots indicate individuals who lacked a spike-specific response but retained a nucleocapsid-specific response. Brown–Forsythe and Welch’s ANOVA with Dunnett’s T3 multiple comparisons tests were used. SFC, spot-forming cell. f, Proportion of each cohort scored as responding to SARS-CoV-2 by ELISpot. g, Spike-specific IFN-γ ELISpot in children ≥6 months (n = 27, red) or ≥12 months (n = 14, dark red) after infection. The dotted lines indicate seropositive cutoffs. h, Paired ELISpot results for children at ≥6 and ≥12 months post-primary infection (n = 6). The dotted lines indicate seropositive cutoffs. Source data

Fig. 6. Superior antibody binding of SARS-CoV-2 variants in children and comparable neutralization.

a,b, Antibody binding to spike (a) and RBD proteins (b) from SARS-CoV-2 variants using plasma from children (n = 19) or adults (n = 18). The bars indicate the geometric mean ± 95% CI. Kruskal–Wallis with Dunn’s multiple comparisons tests were used. c,d, Live virus neutralization assays on SARS-CoV-2 variants displayed as maximal neutralization of infection (c) and titer at 50% neutralization (d) using plasma from children (n = 28) or adults (n = 43). The bars indicate the median ± 95% CI. Source data

Fig. 7. Model of adaptive immunity to coronaviruses in children and adults.

Children develop hCoV spike S2-specific antibodies and cellular responses that can cross-react with SARS-CoV-2. Robust S1-specific adaptive responses develop after SARS-CoV-2 infection. Cross-reactive S2-specific responses probably contribute to immune control in children.

Extended Data Fig. 1. MSD assay specificity and sensitivity.

(a) MSD results using convalescent plasma samples from 35 children with PCR-confirmed SARS-CoV-2 infection. Bars indicate Geometric Mean±95%CI. (b) Assay cut-offs were tested using plasma samples from nine children (red) and 50 adults (blue) taken prior to COVID-19. Dotted lines indicate the cut-off used. Source data

Extended Data Fig. 2. CD8 + T cells with a IL-2⁻TNF⁺IFN-γ⁺ phenotype dominate the spike-specific T cell response in children.

PBMC from SARS-CoV-2 seropositive children were stimulated for 6hrs in the presence of spike peptide pool and then cytokine production analysed by flow cytometry by intracellular cytokine staining (ICS). (a) Gating strategy for analysis. (b) Representative example of ICS staining from one child at six months post SARS-CoV-2 infection showing TNF + IFN-γ + CD8 + T cell response representing 0.48% of the global CD8 + T cell repertoire. Source data

Extended Data Fig. 3. Antibody binding, RBD-ACE2 inhibition and pseudo-neutralisation of SARS-CoV-2 VOC in plasma from children or adults.

(a) Antibody titres normalised to the Wuhan sequence as determined by MSD to total spike and (b) RBD from VOC as indicated in children (n = 19) and adults (n = 18) ≥6 months post primary infection. Bars indicate Median±95%CI. (c) Antibody inhibition in an RBD-ACE2 competitive binging assay using plasma ≥6 months post infection in children (n = 12) and adults (n = 15). Results from pre-pandemic adult samples are also shown (n = 10). Bars indicate geometric mean ±95% CI. (d) Results from pseudo-virus neutralisation assays displayed as titre at 50% neutralisation from children (n = 28) and adults (n = 43) ≥6 months post primary infection. Bars indicate Median±95%CI. Source data