Pediatric T cell and B cell responses to SARS-CoV-2 infection

Abstract

BACKGROUNDUnderstanding age-associated differences in acute and memory adaptive immunity to SARS-CoV-2 and how they contribute to more favorable outcomes in children is critically important.METHODSWe evaluated SARS-CoV-2-specific T cell, B cell, and antibody responses in 329 peripheral blood samples collected from nonhospitalized children, adolescents, and adults at 3 time points, including acute and memory time points.RESULTSMost children produced robust CD4+ T cell responses during infection and developed memory CD4+ T cells; however, young children less than 4 years old often had undetectable CD4+ T cell responses compared with older children and adults. Young children also generated reduced frequencies of memory B cells; despite this, they mounted substantial and durable neutralizing antibody responses. CD4+ T cell responses in children were biased toward non-spike epitopes, especially in asymptomatic cases. Memory B cells in children were preferentially classical memory or, paradoxically, CXCR3+.CONCLUSIONThese findings support the concept that the kinetics and composition of T and B cell responses shift across age groups and may be associated with milder COVID-19 outcomes in children.FUNDINGNIH National Institute of Allergy and Infectious Diseases (NIAID) award AI142742, the Duke University School of Medicine, and grants from the Children's Miracle Network Hospitals, the Translating Duke Health Children's Health and Discovery Initiative, the NIH NIAID (R01-AI161008-02), and the Defense Advanced Research Projects Agency N66001-09-C-2082. NIH Career Development Awards (K23-AI135090 and K01-AI173398). NIH contract 75N93019C00065.

Keywords: B cells; COVID-19; Clinical Research; Immunology; Infectious disease; T cells.

Figure 1. Virus-specific CD4⁺ T cell responses to SARS-CoV-2 infection in children and adults.

(A) Representative flow cytometry plots showing SARS-CoV-2–specific CD4⁺ T cells assessed by AIM assay. (B) Frequencies of SARS-CoV-2 spike-specific, (C) non-spike-specific, and (D) combined spike and non-spike CD4⁺ T cells. (E) Paired longitudinal spike-specific and (F) non-spike-specific CD4⁺ T cell responses. (G) Bar graph indicating peak responses at acute or 2-month time points, tabulating spike- and non-spike-specific CD4⁺ T cell responses from 14 children and 6 adults (n = 25, n = 12 responses, respectively). (H) Correlations of spike (left) and non-spike (right) CD4⁺ T cell frequencies during acute infection with age in children (top) and adults (bottom). Triangles represent asymptomatic donors. (I) Frequencies of SARS-CoV-2 spike- (left) and non-spike-specific (right) CD4⁺ T cells. Connecting lines represent the geometric mean (GeoMean); numbers below the graphs denote the number of donors for each time point. Center lines and error bars in B–D represent the GeoMean ± geometric SD. Dotted line in B–F, H, and I indicates the LOQ; solid gray line indicates the lower limit of detection (LOD). Uninfected participants indicated as Un; 2-month and 6-month time points as 2m and 6m. AIM⁺ denotes the frequencies of OX40⁺CD40L⁺ cells among total CD4⁺ T cells. N, %, and Gm in B–D represent number of donors, percent responders, and GeoMean for each group, respectively. Ped in G represents the pediatric group. P values for B–D were calculated by Kruskal-Wallis test with Dunn’s correction for comparisons within the pediatric and adult groups and by Mann-Whitney test with Holm-Šídák correction for comparisons across age groups, for E and F by Wilcoxon’s test, and for G by Fisher’s exact test and are indicated as *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. r values in H indicate Spearman correlation coefficient.

Figure 2. SARS-CoV-2 infection in children induces a bias toward acute non-spike CD4⁺ T cell responses.

(A) Correlation of spike-specific CD4⁺ T cell responses with non-spike-specific CD4⁺ T cell responses. Triangles denote asymptomatic infections. (B) Paired spike- and non-spike-specific CD4⁺ T cell responses at indicated time points; children (n): Un = 19, Acute = 75, 2m = 45, 6m = 46; adults (n): Un = 8, Acute = 18, 2m = 29, 6m = 23. (C) Comparisons of ratios of non-spike/spike-specific CD4⁺ T cell responses between asymptomatic (Asym) and symptomatic (Sym) pediatric donors at indicated time points; Acute: Asym = 6, Sym = 36; 2m: Asym = 10, Sym = 24; 6m: Asym = 8, Sym = 23. Center lines in C represent the GeoMean. Dotted line in A and B indicates the LOQ of the assay; solid gray line indicates the lower LOD. Uninfected participants are indicated as Un; 2-month and 6-month time points as 2m and 6m, respectively. AIM⁺ denotes the frequencies of OX40⁺CD40L⁺ among total CD4⁺ T cells. r values in A indicate Spearman correlation coefficient. P values for B were calculated by Wilcoxon’s test with Holm-Šídák correction and for C by Mann-Whitney test and are indicated as **P < 0.01.

Figure 3. T helper subset distribution in SARS-CoV-2–infected children and adults.

(A) Representative flow cytometry plots showing T helper (T_H) subset distribution among total CD4⁺ T cells (black) and among SARS-CoV-2 antigen-specific CD4⁺ T cells (blue) in children (left) and adults (right). (B) Proportion plots showing distribution of T_H subsets among SARS-CoV-2 spike- (left) and non-spike-specific (right) CD4⁺ T cells at the acute phase of infection. (C and D) Correlations of CCR6⁺ (top) and CXCR3⁺ frequencies (bottom) with age; y-axis represents percentages of (C) total non-naive CD4⁺ T cells and (D) SARS-CoV-2–specific CD4⁺ T cells. Columns and error bars in B represent the mean and SEM, respectively. Ped in B represents the pediatric group. P values for B were calculated by Mann-Whitney test with Holm-Šídák correction and are indicated as **P < 0.01, ****P < 0.0001. r in C and D indicates Spearman correlation coefficient.

Figure 4. SARS-CoV-2–specific CD4⁺ T cell cytokine responses in children.

(A) Frequencies of SARS-CoV-2 spike-specific, (B) non-spike-specific, and (C) spike and non-spike combined IFN-γ–producing CD4⁺ T cells. (D) Bar graph indicating peak responses at acute or 2-month time points, tabulating spike- and non-spike-specific IFN-γ⁺CD4⁺ T cell responses from 14 children and 7 adults (n = 27, n = 12 responses, respectively). (E) Pie charts showing the proportion of spike-specific cytokine-producing CD4⁺ T cells bearing 2, 3, or 4 functions. (F) Correlations of spike (left) and non-spike (right) CD40L⁺IFN-γ⁺CD4⁺ T cell frequencies at the acute phase of infection with age. Triangles represent asymptomatic donors. (G) Frequencies of SARS-CoV-2 spike-specific (left) and non-spike specific (right) CD40L⁺IFN-γ⁺CD4⁺ T cells. Each connecting line represents the GeoMean; numbers below the graphs denote the number of donors for each time point. Center lines and error bars in A–C represent the GeoMean ± geometric SD. Dotted line in A–C, F, and G indicates the LOQ; solid gray line indicates the lower LOD. Uninfected participants are indicated as Un; 2-month and 6-month time points as 2m and 6m. N, %, and Gm in A–C represent number of donors, percent responders, and GeoMean for each group, respectively. Ped in D represents the pediatric group. P values for A–C were calculated by Kruskal-Wallis test with Dunn’s correction for comparisons within the pediatric and adult groups and by Mann-Whitney test with Holm-Šídák correction for comparisons across pediatric and adult age groups and for D by Fisher’s exact test and are indicated as *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. r in F indicates Spearman correlation coefficient.

Figure 5. SARS-CoV-2–specific CD4⁺ T cell cytokine secretion in children.

(A) Spike- and non-spike-specific CD4⁺ T cell secreted cytokine responses. In the uninfected group, blue circles represent pediatric samples, and white circles represent adult samples. (B) Paired spike- and non-spike-specific IFN-γ secretion; children n: Un = 6, Acute = 24, 2m = 24; adults n: Un = 6, Acute = 6, 2m = 6. (C) Correlation between secreted cytokine measured by cytokine bead array and the frequency of T_H subset determined by hybrid AIM assay. Indicated populations shown as percentage of OX40⁺CD40L⁺CD4⁺ for children and adults in aggregate. Solid black line in A and B indicates the lower LOD; dashed black line indicates the LOQ; dashed red line indicates the median of all DMSO-negative controls. Uninfected participants are indicated as Un; 2-month and 6-month time points as 2m and 6m. AIM⁺ in C denotes the frequencies of OX40⁺CD40L⁺ cells as percentage of CD4⁺ T cells. P values for A were calculated by Kruskal-Wallis test with Dunn’s correction, and for B by Wilcoxon’s test with Holm-Šídák correction and are indicated as *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. r in C indicates Spearman correlation coefficient.

Figure 6. SARS-CoV-2–specific cT_FH and antibody response in children.

(A) Representative flow cytometry plots showing SARS-CoV-2–specific CD4⁺ T cells (blue) overlaid on total CD4⁺ T cells (gray); CXCR5⁺ gate represents cT_FH cells. Numbers in blue indicate frequencies of SARS-CoV-2–specific cT_FH among total CD4⁺ T cells. PD-1, programmed cell death 1. (B) Frequencies of spike-specific and (C) non-spike-specific cT_FH cells. (D) Paired longitudinal spike- (left) and non-spike-specific (right) cT_FH responses. (E) Bar graph indicating peak responses at acute or 2-month time points, tabulating spike- and non-spike-specific cT_FH responses from 12 children and 6 adults (n = 21, n = 12 responses, respectively). (F) Correlations of spike- (left) and non-spike-specific (right) cT_FH frequencies at acute phase of infection with age in children (top) and adults (bottom). Triangles represent asymptomatic donors. (G) Frequencies of spike-specific (left) and non-spike-specific (right) cT_FH cells. Each connecting line represents the GeoMean; numbers below the graphs denote number of donors for each time point. Center lines and error bars in B and C represent the GeoMean ± geometric SD. Dotted line in B–D, F, and G indicates the LOQ; solid gray line indicates the lower LOD. Uninfected participants indicated as Un; 2-month and 6-month time points as 2m and 6m. AIM⁺ denotes frequencies of OX40⁺CD40L⁺ cells among total CD4⁺ T cells. N, %, and Gm in B and C represent number of donors, percent responders, and GeoMean for each group, respectively. Ped in E represents the pediatric group. P values for B and C were calculated by Kruskal-Wallis test with Dunn’s correction for comparisons within pediatric and adult groups and by Mann-Whitney test with Holm-Šídák correction for comparisons across age groups, for D by Wilcoxon’s test, and for E by Fisher’s exact test, indicated as *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. r in F indicates Spearman correlation coefficient.

Figure 7. Antigen-specific B cell memory to SARS-CoV-2 infection in children and adults.

(A) Representative flow cytometry plots showing spike (left), RBD (middle), and nucleocapsid (right) probe binding B_Mem in children (n = 50) and adults (n = 12) at 6 months postinfection. (B) Frequencies of spike (left), RBD (middle), and nucleocapsid-binding (right) B_Mem shown as percentages of total B cells (CD19⁺CD20⁺). (C) Correlations of spike (left) and RBD-binding (right) B_Mem frequencies with age, in children (blue) and adults (black). (D) Relative frequencies of IgG, IgM, IgA, and other isotypes of B_Mem indicated as percentage of spike (left), nucleocapsid (middle), and total (right) B_Mem. Center lines in B represent the GeoMean. Center lines and error bars in D represent the mean ± SD. P values for B were calculated by Mann-Whitney test and for D by Mann-Whitney test with Holm-Šídák correction, indicated as *P < 0.05, **P < 0.01. r in C indicates Spearman correlation coefficient; r values in black indicate Spearman correlation coefficient for the combined pediatric and adult data; r values in blue indicate Spearman correlation coefficient for the pediatric data alone.

Figure 8. SARS-CoV-2–specific B_Mem in children are predominantly classical B_Mem and express CXCR3.

(A) Representative flow cytometry plots showing B_Mem subsets based on differential CD27 and CD21 staining, gated on total B_Mem (left) and probe-binding B_Mem (right). (B) Frequencies of B_Mem subsets indicated as percentage of SARS-CoV-2–specific or total B_Mem. (C) Representative flow cytometry plot showing CXCR3 staining on SARS-CoV-2–specific B_Mem. FSC, forward scatter. (D) Frequencies of CXCR3⁺ B_Mem indicated as percentage of spike- or nucleocapsid-binding B_Mem; spike: n = 49 pediatric group, n = 12 adult group; nucleocapsid: n = 35 pediatric group, n = 5 adult group. Center lines and error bars in B represent the mean ± SD. Center lines in D represent the mean. Ped in B and D represents the pediatric group. P values for B were calculated by Mann-Whitney test with Holm-Šídák correction and for D by Mann-Whitney test and are indicated as *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 9. Correlation of pediatric spike B_Mem with humoral responses.

(A) Correlation of spike IgG titers as determined by binding antibody multiplex assay (BAMA) at the acute (left) and 4- to 6-month (right) time points with spike B_Mem indicated as percentage of total B cells in children (blue) and adults (black). (B) Correlation of Neut ID₅₀ titers at the acute (left) and 4- to 6-month (right) time points with spike B_Mem indicated as percentage of total B cells in children. (C) Correlation of spike IgG titers at the acute (left) and 4- to 6-month (right) time points with classical B_Mem indicated as percentage of total B_Mem in children (blue) and adults (black). (D) Correlation of Neut ID₅₀ titers at the acute (left) and 4- to 6-month (right) time points with classical B_Mem indicated as percentage of total B_Mem in children. (E) Comparison of B_Mem responses in children less than 4 years old, older children, and adults. Center lines in E represent the GeoMean. In A and C 4- to 6-month time points, r values in black indicate Spearman correlation coefficient for the combined pediatric and adult data; r values in blue indicate Spearman correlation coefficient for the pediatric data alone. P values in E were calculated by Kruskal-Wallis test with Dunn’s correction, indicated as *P < 0.05.