The study cohort was followed from the early phase of the pandemic (May-July 2020), before the emergence of SARS-CoV-2 VOCs alpha, beta, gamma, delta and omicron. All seropositive children had participated in a population-based SARS-CoV-2 seroprevalence study (27). These children and their families were followed up for 15 months. Serology and personal interviews with parents and children allowed classification of asymptomatic or mild infection. In the present study, 50 children were included to investigate whether asymptomatic or mild SARS-CoV-2 infection confers specific antibody and T cell responses lasting for 15 months. Of the 50 children, 26 were seropositive by SARS-CoV-2 RBD antibody testing and 24 were consistently seronegative throughout the follow up.

All work was conducted in accordance with the Declaration of Helsinki in terms of informed consent and approval by an appropriate institutional board. Blood samples were obtained after donors and their parents consented to participate in this study. The ethics committee of the Medical University of Vienna, Austria, approved the study protocol (approval no. 2104/2020).

PBMCs were isolated from blood samples using Ficoll-Paque Plus™ (GE Healthcare) and cryopreserved in liquid nitrogen. Serum samples were stored at −20°C.

SARS‐CoV‐2‐specific antibodies were quantified using the Wantai SARS-CoV-2 IgG ELISA (Quantitative) kit (Beijing Wantai Biological Pharmacy Enterprise) as described previously (27). The assay is calibrated according to the WHO International standard for SARS-CoV-2 immunoglobulin, enabling the quantification of antibody concentrations in BAU/ml. Results were interpreted using a cut-off of 5.4 BAU/ml, as recommended by the manufacturer. The SARS-CoV-2 antibody profile against spike RBD, S1, S2 and nucleocapsid was assessed using the SARS-CoV-2 ViraChip assay (Viramed, Viramed Biotech Ag, Germany) as described previously (28). The SARS-CoV-2 ViraChip assay is calibrated according to the WHO International standard for SARS-CoV-2 immunoglobulin, enabling the quantification of antibody concentrations in BAU/ml. Results were interpreted using the following cut-offs, as recommended by the manufacturer: RBD≥15 BAU/ml; S1≥25 BAU/ml; S2≥72 BAU/ml; N≥16 BAU/ml.

The preparation of the D614G virus stock (GISAID accession number: EPI_ISL_438123) has been described previously (29). Following the same protocol, omicron BA.1 (GISAID accession number: EPI_ISL_9110894) and BA.2 (GISAID accession number: EPI_ISL_11110193) were isolated from nasopharyngeal swabs from COVID-19 patients, passaged on Vero E6-TMPRSS2 cells (kindly provided by Anna Repic), and subjected to next-generation sequencing. The isolates were controlled to be free of other respiratory viruses by PCR as described in Koblischke et al, 2020 (29) and were tested negative for mycoplasma with the MycoAlertTM Mycoplasma Detection Kit (Lonza Group Ltd, Basel, Switzerland).

The neutralization assay was carried out essentially as described previously (29). Briefly, serial dilutions of serum samples were incubated with 50-100 TCID50 SARS-CoV-2 for 1 hour at 37°C. The mixtures were then transferred to Vero E6 cells (ECACC 85020206) and incubation was continued for three (D614G) or five (omicron) days. NT titers were expressed as the reciprocal of the serum dilution required for protection against virus-induced cytopathic effects. NT titers ≥10 were considered positive.

For T cell stimulation, PepMixTM SARS-CoV-2 peptide pools (product codes: PM-WCPV-VEMP, PM-WCPV-VME, PM-WCPV-S, PM-WCPV-NCAP, and PM-SARS2-SMUT08-1) were purchased from JPT (Berlin, Germany). The pools comprise 15mer peptides overlapping by 11 amino acids and cover the entire sequences of the SARS-CoV-2 wildtype structural proteins: envelope (E), membrane (M), spike (S) composed of two sub-pools S1 (aa 1-643) and S2 (aa 633-1273), and nucleoprotein (N), and S from the omicron variant B.1.1.529. Peptides were dissolved in dimethyl sulfoxide and diluted in AIM-V medium for use in PBMC expansion and intracellular cytokine staining (ICS) assays.

In vitro expansion of PBMCs was performed as described previously (30). In brief, cryopreserved PBMCs were thawed in pre-warmed StableCell RPMI-1640 medium (Sigma) containing 10% FBS (FBS 12-A, Capricorn), 10 mM Hepes (Sigma), 50 IU/ml Pen-Strep (Sigma) and 50 IU IL-2 (Peprotech) at a concentration of 1×10⁶ cells/ml. PBMCs were pulsed with peptides (1 μg/ml) and cultured for 10 days adding 100 IU IL-2 on day 5. In vitro expanded cells were analyzed by intracellular cytokine and cell surface marker staining. PBMCs were incubated with 2 μg/ml of peptide and 1 μg/ml anti-CD28/49d antibodies (L293 and L25, Becton Dickinson) or with anti-CD28/49d antibodies alone (negative control) for 6h. After 2h, 0.01 μg/ml brefeldin A (Sigma) was added. Staining was performed using APC/H7 anti-human CD3 (SK7, Becton Dickinson), Pacific Blue anti-human CD4 (RPA-T4, Becton Dickinson), PE anti-human CD8 (HIT8a, Becton Dickinson), FITC anti-human IFN-γ (25723.11, Becton Dickinson), and Fix/Perm kit (Invitrogen). Viable cells were determined using Live/Dead Fixable Aqua Dead (Invitrogen). The gates for detection of cytokines in peptide-stimulated cell samples were set in the samples with no antigen stimulation. Antigen-specific responses were determined as the frequency of IFN-γ positive T cells in stimulated samples with background subtraction from paired unstimulated controls (30). The cut-off was determined by background staining (no peptide) of all negative controls (31). For all experiments combined, the mean background was 0.44%, 0.16% and 0.26% for samples expanded with spike (composed of subpools S1 and S2), membrane or nucleocapsid peptide pools, respectively. To exclude non-specific or background responses, responses >3-fold above the background of the respective peptide pools were considered positive.

Subject characteristics were described by medians and interquartile ranges (IQR) for continuous variables and frequencies and percentages for categorical variables. Statistical analysis was performed with GraphPad Prism Version 9.2.0. Significance between paired groups was assessed using the two-sided Wilcoxon signed rank test, or paired t test. Groups were compared with Kruskal-Wallis test and Dunn´s multiple comparison test, or Mann-Whitney U test. For neutralizing antibody titers < 10, a value of 5 was used for the analyses as well as the fold changes shown in Figures 3A, B . Ratios of neutralizing antibodies to SARS-CoV-2 IgG were calculated for samples which were positive in both assays. P values <0.05 were considered significant; significance values are indicated as *P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001.

The cohort consisted of 26 children diagnosed with SARS-CoV-2 infection who had participated in a population-based SARS-CoV-2 seroprevalence study (27), conducted in the early phase of the pandemic (May-July 2020) before the emergence of VOCs, and 24 SARS-CoV-2 age-matched seronegative children. Age (median [IQR], 13.5 [1.6] vs 12.2 [5.4] years) and sex (13m/13f vs 13m/11f) were similar in both groups ( Table 1 ). In the seropositive group, 23% (6/26) had no symptoms, while 77% (20/26) experienced COVID-19 symptoms ( Supplementary Table 1 ). The first serum sample was collected at a median interval of 3.6 [IQR 0.7] months after symptom onset. Further samples were collected 6, 9, and 12 months after serodiagnosis (visits 1-3). Furthermore, we were able to study the effects of vaccination in infection-naïve children (n=10) compared to those with pre-existing immunity (n=10).

Quantification of SARS-CoV-2 spike-specific IgG antibodies was performed with an enzyme-linked immunosorbent assay (ELISA) based on the SARS-CoV-2 RBD. In seropositive children, antibody concentrations declined over 6 months post serodiagnosis, but after this decrease, remained relatively stable between 6 and 12 months ( Figure 1A ). The stratification of children for symptomatic or asymptomatic infection revealed no significant differences in antibody concentrations (labeled in Figure 1 ). To evaluate the kinetics of SARS-CoV-2 spike- and nucleocapsid-specific antibodies, we used a commercial microarray assay based on spike RBD, S1, S2 and nucleocapsid. In line with the ELISA results, antibodies against the RBD were present in all convalescent children and were retained at 12 months after serodiagnosis ( Supplementary Figure 1 ). In contrast, we observed a decline of nucleocapsid-specific antibodies to undetectable levels in 6 of 26 children ( Supplementary Figure 1 ).

To analyze the neutralizing activity of antibodies, we performed live-virus neutralization assays using either an isolate from the early pandemic with the D614G mutation (wildtype) or the omicron variant (B.1.1.529) (see Materials and Methods). In seropositive children, neutralizing antibody titers for wildtype were relatively stable over time. As illustrated in Figure 1B , neutralizing antibodies decreased to undetectable levels at V1 in two children, one of whom remained undetectable neutralizing antibodies at V2 and V3, and one was lost for follow-up. However, 95% (19 of 20) still having neutralizing antibodies 12 months after serodiagnosis. In agreement with the ELISA results, neutralizing antibodies were not detected in any of the 24 seronegative children (data not shown).

To determine the specific neutralizing activity, we calculated the ratios of neutralizing antibodies to SARS-CoV-2 IgG for the different time points. Notably, the ratios continued to increase over 9 months (V2) and revealed no further change until 12 months (V3) after serodiagnosis, indicating that antibodies improved their neutralizing activity to the original D614G over 9 months ( Figure 1C ). In contrast, there was largely no neutralization of the omicron variant ( Figure 1D ).

In addition to antibodies, T cells may contribute to protection by limiting viral dissemination in the host (17). Moreover, CD4 and CD8 T cell responses were recently reported to be less affected by omicron than antibody responses (32–34). We therefore assessed CD4 and CD8 T cell responses in 26 recovered children at 6 and 12 months after serodiagnosis. For sensitive detection of SARS-CoV-2 T cells, antigen-specific T cells were expanded by stimulating PBMCs with peptide pools derived from SARS-CoV-2 wildtype spike, membrane and nucleocapsid for ten days, followed by restimulation with the same antigens and analysis of IFN-γ using flow cytometry intracellular cytokine staining (30) ( Supplementary Figure 2 ). Among seropositive children, T cell responses to one or more SARS-CoV-2 peptide pools were present in 81% (21 of 26) for CD4 T cells and 85% (22 of 26) for CD8 T cells. CD4 T cells were most frequently directed against spike, followed by membrane and nucleocapsid, while CD8 T cells were mostly directed against nucleocapsid, followed by spike and membrane proteins ( Supplementary Figure 3 ). In line with these results, CD8 T cell responses directed against nucleocapsid and/or membrane proteins, but not against the spike were observed in 8 children, while only two children mounted nucleocapsid- and/or membrane-specific CD4 T cell responses, but lacked a spike-specific reactivity. There were no significant differences in T cell levels between children with symptomatic or asymptomatic infection (labeled in Figure 2 ). Comparison of responses between 6 and 12 months revealed no significant difference in CD4 T cell levels, while CD8 T cell responses at 12 months were substantially reduced (spike = 10-fold, membrane = 8-fold, nucleocapsid = 7-fold) ( Figure 2 ). The specificity of responses at 12 months displayed a similar distribution as at 6 months, with CD4 T cells recognizing spike in 60%, and membrane or nucleocapsid in 30%, respectively, whereas CD8 T cells were directed against nucleocapsid in 30%, followed by spike (20%) and membrane (5%) ( Figure 2 and Supplementary Figure 3 ).

We analyzed CD4 and CD8 T cell responses in PBMC samples from 12 of the 24 seronegative children ( Supplementary Figure 4 ). In line with previous studies from SARS-CoV-2 unexposed populations (8, 35–40), we observed SARS-CoV-2 reactivity in 17% (2/12) of CD4 and 42% (5/12) of CD8 T cell responses. Consistent with recent data (35), responses in seronegative individuals were focused on spike but lacked reactivity to membrane and nucleocapsid antigens, and are most likely pre-existing cross-reactive T cells induced by endemic common cold coronaviruses or other human viruses (35, 37–43). Together, the data indicate strong CD4 and CD8 T cell responses in recovered individuals with durable CD4 T cell immunity for at least 12 months.

Moreover, we analyzed T cell responses to omicron versus wildtype peptides in PBMCs from 10 of the 26 children at V3. A CD4 T cell response to wildtype versus omicron spike was present in 60% (6/10) versus 50% (5/10), while a CD8 T cell response was detected in 40% (4/10) versus 40% (4/10), respectively ( Supplementary Figures 5A, B ), indicating that there were no significant differences in CD4 or CD8 T cell responses to omicron versus wildtype spike. When comparing the neutralizing activity of antibodies against wildtype and omicron from these children, we found that neutralizing activity to omicron was consistently and significantly lower than to the wildtype strain ( Supplementary Figure 5C ). Overall, the data showed that, unlike neutralizing antibodies, CD4 and CD8 T cell recognition of the omicron spike protein was largely preserved.

To investigate the effect of preexisting immune memory on vaccine responses, we measured antibody titers generated with BNT162b2 or Ad26.COV2-S vaccines in a subgroup of recovered (n=10) and seronegative children (n=10). The samples were collected 1.1 [IQR 0.9] and 1.3 months [IQR 0.6] after receipt of the second dose of BNT162b2 (n=8 recovered; n=10 seronegative) or one dose of Ad26.COV2-S (n=2 recovered). For the comparison of neutralizing activity, we tested pre- and post-vaccination serum samples using live-virus neutralization assays with wildtype and omicron BA.1 and BA.2 variants. Vaccination yielded increased neutralizing antibody titers in both groups ( Figures 3A, B ). However, median titers to SARS-CoV-2 wildtype, omicron BA.1 or BA.2 were 3.4-fold, 6.7-fold or 7-fold higher in recovered than in seronegative children, respectively ( Figure 3C ). Notably, this difference was even more pronounced when we compared specific neutralizing activities, i.e. the ratios of neutralizing antibodies to SARS-CoV-2 IgG. Sera from vaccinated children with pre-existing immunity yielded 3.9-fold, 9.9-fold and 14-fold higher neutralization activities against wildtype, omicron BA.1 and BA.2 than those from seronegative children, respectively ( Figure 3D ). The comparison of individual neutralization profiles between the two groups showed that post-vaccination sera from children with pre-existing immunity efficiently neutralized BA.1 and BA.2, exhibiting 2.7-fold and 3.4-fold differences to the wildtype strain. In contrast, the differences between wildtype and BA.1 or BA.2 were 5.6-fold and 9.3-fold in sera from seronegative children, respectively ( Figures 3E, F ). These results indicate that vaccination of children with preexisting immunity resulted not only in higher neutralizing antibody titers, but also in broader neutralization of VOCs compared to seronegative vaccinees. Analysis of neutralizing antibody titers in non-vaccinated children revealed no differences between samples obtained at V3 (12 months [IQR 0.8]) or 15 months [IQR 1.3] after serodiagnosis ( Supplementary Figures 6A, B ).