The participants studied were part of a larger prospective observational SARS-CoV-2 vaccination cohort recruiting participants up to 60 years of age by approaching people from the general population via several Personal Records Database (BRP) drawings (IIVAC study; NL76440.041.21, EudraCT: 2021-001357-31) (12). In addition, children 5-11 years of age were solicited to participate in the IIVAC study through recruitment via the Long COVID KIDS study (16) and were largely healthy controls from that study invited via BRP drawings and social media campaigns. Before inclusion, children were asked whether they experienced post-COVID-19 complaints. Of note, no such COVID complaints were reported by 5–11-year-old children before inclusion in this study. Children with cancer, transplants, chronic kidney disease, Down syndrome, HIV, auto immune diseases and/or any immune deficiency through disease or medication were excluded from participation in the study. Participant characteristics can be found in Table 1 .

Children received 10 µg/dose (5-11 years) or 30 µg/dose (12-17 years) of BNT162b2 (Comirnaty, Pfizer) by Public Health Services (GGD). Children were offered two vaccine doses, but some elected to only receive one vaccine dose. Blood samples (via finger prick or venipuncture) and questionnaires were taken prior to COVID-19 vaccination and at fixed intervals after vaccination. Questionnaires included demographic and SARS-CoV-2 testing information. Participants who reported a positive SARS-CoV-2 test within 28 days after receiving a vaccine and those who showed evidence of a new infection in this period based on the presence of nucleoprotein (NP) antibodies were excluded from the current analysis. In addition, participants were only included in the antibody analysis if at least two consecutive samples (one before and one after the first or the second vaccine dose) were available. For cellular assays, sample selection was limited by availability of PBMC samples. Within this availability, PBMC samples were selected to include participants with or without a prior infection, participants who received one or two vaccine doses and, if possible, those within each of these categories from both age groups (aged 5-11 and 12-17 years). We aimed for equal distribution of samples among these categories.

Serum and PBMC samples were collected before the first vaccination and 28 days after the first (± 2 days) and/or second (- 7 days till + 21 days) vaccination of the primary vaccination series. Serum samples were either self-collected by finger prick in microtubes and returned by mail, as described previously (17), or via venipuncture by a research nurse during a home visit. Once received by the laboratory at the National Institute for Public Health and the Environment (RIVM), the Netherlands, serum from microtubes and from blood sample tubes were spun down and frozen at -80°C as previously described (12). Heparin tubes were used to collect PBMCs (subgroup only), which were isolated with Lymphoprep as described previously (18).

SARS-CoV-2 multiplex immunoassays were carried out to determine serum antibody levels towards SARS-CoV-2 Spike S1 (Sino Biological, 40591-V08H) and Spike S1 Omicron B1.1.529 (Sino Biological, 40591_V08H41) as described previously (19). In short, the serum was incubated with antigen-coupled beads for 45 minutes in the dark, followed by a 30-minute incubation with Goat-anti-human IgG. Samples were incubated in SM01 (Surmodics) and washing steps after each incubation were carried out with PBS. Antibody binding to antigen-coupled beads was determined with an FM3D (Luminex) and antibody levels were expressed as binding antibody units (BAU)/ml for S1 and as arbitrary units (AU)/ml for S1 Omicron BA.1. Antibody data for the 12–17-year-old participants in this study were published previously (12). The threshold for seropositivity was set at 10.1 BAU/mL for Spike S1 Wuhan and 14.3 BAU/mL for Nucleoprotein, as previously standardized for the Wuhan (vaccine) strain against the NIBSC/WHO COVID-19 reference serum 20/136. No cut-off has been established for Spike S1 Omicron antibodies.

Interferon gamma (IFN-γ) ELISpot was adapted from a protocol described earlier (20). In brief, ethanol-activated Multiscreen HTS IP filter plates (Millipore; MSIPS4510) were coated with 5 μg/ml anti-human IFN-γ monoclonal antibodies (1-D1K, MabTech, 3420-3-1000) in PBS, overnight at 4°C. After washing with PBS, the plates were blocked using AIM-V medium (Gibco BRL, 12055-083) supplemented with 2% heat-inactivated human serum (Sigma, H6914) (AIM-V + 2%HS) for 1 hour. Then, PBMCs that were thawed using RPMI medium (Gibco, 52400-025) supplemented with 10% Fetal Bovine Serum (FBS; GE Life Sciences, SH30071.03) and 1% Penicillin/Streptomycin (Lonza, DE17-602E), were plated at 2x10⁵ cells/well (or 5x10⁴ cells/well for the positive control) in AIM-V + 2%HS. Cells were stimulated in duplicate or triplicate with peptide pool containing 15-mer overlapping peptides (11 amino acids overlap) spanning the entire Spike protein from the Wuhan strain or VOCs or the nucleoprotein from the Wuhan strain of SARS-CoV-2 (0.65 µg/ml; JPT), DMSO (negative control), or PHA (positive control; 1 µg/ml), for 22-24 hours, at 37°C, 5% CO₂. Next, plates were washed with PBS + 0.05% Tween-20 and incubated with 1 μg/ml anti-human IFN-γ biotinylated detection antibody (7-B6–1, MabTech, 3420-6-1000) in PBS + 0.5% FBS for 1 hour. After washing with PBS + 0.05% Tween20, ELISpot plates were incubated with Extravidin Alkaline Phosphatase (1 μg/ml; Sigma, E2636) in PBS+0.5% FBS for 1 hour. Next, the plates were washed with PBS +0.05% Tween20, then rinsed with PBS, and developed with 5-bromo-4-chloro-3-indolyl phosphate (BCIP)/nitro blue tetrazolium (NBT) (SigmaFast; Sigma, B5655). The reaction was stopped after approximately 7 minutes by H₂0 and the plates were dried in the dark. Spots were analyzed with an Immunospot C.T.L. reader and software. The number of spot-forming units (SFU) from DMSO controls was subtracted from total SFU counted in cultures with antigen. Samples that showed low response to PHA (< 33 SFU/5x10⁴ PBMC) compared to other samples from the same donor were excluded from the analysis due to technical issues. Finally, counted spots are represented as IFN-γ spot-forming units (SFU).

SARS-CoV-2 B cell ELISpots were carried out as described previously (18). In short, PBMC samples were cultured for 5 days in AIM-V medium with 10% heat-inactivated FBS, 50uM β mercaptoethanol, 3µg/ml CpG, 10ng/ml recombinant Interleukin 2 and 10ng/ml recombinant interleukin 10. ELISpot filtration plates (Millipore, MSIPS4510) were pre-wetted with 70% ethanol and coated with 10ug/ml of SARS-CoV-2 S1 (Sino biological, 40591-V08H), S1 Omicron BA.1 (Sino biological), MERS S1 (Sino Biological, 40069-V08H), or goat anti-human IgG (MP biomedicals, 855071). After overnight incubation of previously cultured cells on the pre-coated plates, antibody binding was detected with 0.2µm filtered 50 µl/well AP-conjugated goat-anti-human IgG (Seracare, 5220-0462), diluted in 1:5000 in PBS/0.01% Tween20/1% goat serum (ICN Biomedicals, 191356). Spots were made visible with BCIP (Sigma, B5655 or KPL, Seracare, 5420-0038) and subsequently imaged and counted using the same settings throughout each experiment (Immunospot C.T.L, Sample sensitivity = 185, Spot separation = 5, gating = 0.0133 to 0.3339, diffuse processing = large). No background subtraction was carried out and spot numbers were normalized to numbers of specific memory B-cells per 100.000 PBMCs.

We analyzed the induction of activation markers OX40 (TNFRSF4, CD134) and CD154 (CD40L) by CD4⁺ T cells and activation markers OX40 (TNFRSF4, CD134) and CD137 (4-1BB) by CD8⁺ T cells in response to Spike peptides using flowcytometry. To this end, PBMCs that had been cultured with Spike peptides or with DMSO as unstimulated control for IFNγ ELISpot were collected after finishing the culture step of the ELISpot protocol. PBMCs were washed and stained using a cocktail containing a Live/dead dye (FVS780, BD Biosciences) and monoclonal antibodies to stain outer membrane markers using CD3-BUV496, CD4-BUV805, CD8-BUV395, OX40-BB700 (all BD Biosciences), CD154-PE-Cy7, and CD137-BV605 (both Biolegend) directly after collection of the cells. After subsequent washing, the cells were analyzed on a Symphony A3 flow cytometer (BD Biosciences). Data were analyzed with FlowJo software Version 10 (BD Biosciences) to determine frequencies of cells showing dual expression of activation markers among the live CD3⁺CD4⁺ lymphocytes and the live CD3⁺CD8⁺ lymphocytes ( Supplementary Figure S4 ). Responsiveness to Spike antigen per participant was calculated by subtracting the frequency of positive cells in the DMSO control from the frequency of positive cells after culture with Spike antigen.

Data were analyzed in R (version 4.3.0) or Graphpad Prism (version 9.5.1). Appropriate tests to investigate statistical significance are described in the respective figure legends.

Participants were invited to receive their first vaccine dose as part of the primary vaccination series from August 2021 through August 2022. Thirty out of 64 (47%) participants were classified as infected based on a self-reported positive COVID-19 test or the presence of antibodies in serum to Spike S1 or nucleoprotein before receiving the first vaccination. ELISpot data indicate that the sum of T-cell responsiveness to Spike + NP prior to vaccination confirms that children included in our cellular assays and classified as not infected based on antibodies and COVID-19 tests were naive to SARS-CoV-2 ( Supplementary Table S1 ). In samples of all twelve children classified as not previously infected and of which ELISpot data were complete, the sum of spot counts to Spike + NP (0-7 SFU/2x10⁵ PBMC) was lower than this sum of spot counts in samples from all children classified as previously infected (9-55 SFU/2x10⁵ PBMC). SARS-CoV-2-variants of concern Delta (Summer 2021 to Winter 2021/2022) and Omicron (Winter 2021/2022 to Summer 2022 and beyond) were dominant in the Netherlands before and during the vaccination period. The majority of vaccinated participants 5-11 years of age for whom vaccination became available from January 2022 were classified as infected before vaccination (21 of 23 children; 91%). For participants 12-17 years of age, vaccinated from July 2021 onwards, 9 of 41 participants (22%) had been infected before vaccination. All uninfected participants received two vaccine doses, while 18 of the 30 infected participants (60%) received only one dose ( Table 1 ).

To investigate to what extent vaccination induces antibody responses to SARS-CoV-2, levels of Spike S1-specific antibodies were determined before vaccination, after the first vaccine dose and after the second vaccine dose. Participants without a previous infection showed robust antibody responses after each vaccine dose ( Figure 1A , p = 0.003 for the first dose, p < 0.0001 for the second dose) with median antibody levels increasing from 0.6 BAU/ml (n = 14) at baseline, to 618.4 BAU/ml (n = 34) after the first vaccine dose and 3853.2 BAU/ml (n = 34) after the second vaccine dose. Participants who had been infected prior to vaccination also showed an increased antibody response after the first vaccine dose ( Figure 1B , p < 0.0001), with antibody levels increasing from 156.8 BAU/ml (n = 30) to 3699.305 BAU/ml (n = 30), indicating that vaccination of these previously infected children further enhances specific antibody responses to the Spike S1 antigen. Antibody levels, however, did not significantly increase further after a second vaccine dose ( Figure 1B , p = 1, 5058.6 BAU/ml, n = 9). Notably, antibody levels after one vaccine dose were higher for participants who were previously infected compared to those who were not ( Figure 1C , p < 0.0001). Furthermore, after one dose, previously infected participants already had antibody levels comparable to uninfected participants who received two vaccine doses ( Figure 1C , p = 1), indicating that a second vaccine dose as part of the primary immunization series does not contribute significantly to the antibody response in previously infected participants. Within the infected group, no differences were observed between age groups after one vaccine dose ( Figure 1 , p = 1). Nevertheless, the first vaccine dose induces robust levels of SARS-CoV-2-specific antibodies in the current population, and these levels are higher in children who previously experienced an infection compared to naive children.

As memory B cells are important for antibody production upon repeated antigen contact, SARS-CoV-2 Spike-specific memory B cells were quantified by ELISpot in a subset of participants to detect their presence and their dynamics in response to vaccination. Memory B-cell data are available from a subset of participants with a previous infection before and after vaccination (n = 7). For participants without a previous infection, memory B cell data are available after vaccination only (n=7) ( Supplementary Table S2 ). All previously infected participants showed an increase in memory B-cell responses after vaccination ( Figure 2A , p = 0.03). In a direct comparison to participants who were not previously infected, a previous infection resulted in higher numbers of Spike S1-specific memory B cells after vaccination ( Figure 2B , p = 0.009. No differences were noted in memory B cell responses between different age groups or between participants who received one or two vaccine doses ( Supplementary Figures S1A , S2A ). These data show a limited induction or maintenance of memory B cells in peripheral blood upon infection. However, upon vaccination, significant boosting of spike specific memory B cells was observed.

Spike-specific T-cell numbers measured by ELISpot increased significantly after vaccinations, both in not previously infected ( Supplementary Table S3 , Figure 3A , p = 0.001) and in previously infected children ( Figure 3B (p = 0.004). Compared to non-infected vaccinees, exposure to SARS-CoV-2 before vaccination resulted in a higher T-cell response to primary vaccination ( Figure 3C , p = 0.048). Differences between the two age groups ( Supplementary Figure S1B ) or between previously infected children who received one or two vaccine doses ( Supplementary Figure S2B ) were not statistically significant. Altogether, these data show that vaccination significantly contributes to Spike-specific T-cell responses in children, and that these T-cell responses become even stronger in those who had been infected with SARS-CoV-2 prior to vaccination.

To gain insight in the type of T cells that respond to vaccination, we analyzed the induction of activation markers expressed by CD4⁺ and CD8⁺ T cells in response to Spike peptides by flow cytometry. Naive and SARS-CoV-2-infected children showed comparable frequencies of responding CD4⁺ T cells prior to vaccination ( Figure 4A , p = 0.8). In naive children, frequencies of responding CD4⁺ T cells were significantly elevated after primary vaccination ( Figure 4B , p = 0.02), whereas in previously SARS-CoV-2-infected children a slight but unsignificant elevation of the frequency of responding CD4⁺ T cells was observed after vaccination ( Figure 4C , p = 0.08). Remarkably, previous infection with SARS-CoV-2 induced significantly higher frequencies of responding CD8⁺ T cells prior to vaccination as compared to naive participants ( Figure 4D , p = 0.03). In contrast to CD4⁺ T cells, no significant vaccine-induced changes of responsive CD8⁺ T cells were detected, both in non-infected ( Figure 4E , p = 0.7) and previously infected children ( Figure 4F , p = 0.1). Together, these data imply that primary mRNA COVID-19 vaccination may drive T-cell responsiveness towards the CD4⁺ subset, whereas a priming SARS-CoV-2 infection may drive an increase in T-cell responsiveness more profoundly in the CD8⁺ subset.

Finally, we assessed whether T-cell and B-cell responses of children generated by the original strain Spike antigen from the BNT162b2 mRNA vaccine would also engraft recognition of Spike proteins from later SARS-CoV-2 variants of concern. Of all dominant strains circulating at the time of sample collection, Omicron was antigenically most diverted from the Original Strain in the evolution of SARS-CoV-2. Vaccination with the original Spike antigen resulted in an increase in antibodies targeting the Spike S1 from Omicron B1.1.529 ( Figure 5A , p < 0.001). Furthermore, no significant differences were observed in vaccine-induced responses between the original and the Omicron variant of Spike for both B cells ( Figures 5B–D ) and T cells ( Figures 5E–G ), indicating a substantial cross-reactive cellular response against the Omicron variant is induced in children by vaccination with BNT162b2. Moreover, in both naive and previously SARS-CoV-2-exposed children, vaccine-induced T-cell responses to original strain Spike strongly correlated with T-cell responses observed to Omicron ( Supplementary Figure S3 ). These findings suggest that T cells addressed by Original Strain Spike antigen are likely to respond to a large extent to currently occurring new variants of concern. Combined, these observations indicate that primary vaccination with Original Spike antigen generates similar cellular response to Spike protein of the more distant Omicron variant of concern in children.