Strong and Working Immune Memory Lasting Up to 9 Months After COVID-19 Infection in Southeast Asia

Sixty-four individuals with confirmed SARS-CoV-2 were included and re-assessed 6-9 months after infection. SARS-CoV-2 infection was confirmed by positive molecular diagnosis as part of the national surveillance system. Since Cambodia had minimal detection of SARS-CoV-2 during the follow-up period, the probability of re-exposure to SARS-CoV-2 was minimal (37) in 2020. For 33 individuals, we obtained a blood sample 2-9 days after laboratory confirmed infection (Figure S1A). For all 64 study participants, between 1 to 15 follow-up nasopharyngeal/oropharyngeal (NP/OP) swab samplings assessed the duration of SARS-CoV-2 RNA shedding during the acute phase of infection via RT-PCR (38). Based on the duration of SARS-CoV-2 RNA shedding, 53% of individuals were considered "long shedders" with detection of viral RNA in NP/OP swabs for ≥10 days (Figure S1B). Overall, 70% of the patients displayed mild or moderate symptoms, and 30% remained asymptomatic (Table S1). For all assays, samples were selected based on availability and quality.

The presence of S-binding antibodies was measured using the S-Flow assay, which sensitively and quantitatively measures anti-S IgG, IgA, and IgM by flow cytometry (21, 39) (Figure 1A). The National Institute for Biological Standards and Control (NIBSC) references were utilized to validate the assays and pre-pandemic samples obtained from nineteen individuals were measured to set the cutoff for each assay (Figure S2). Anti-S IgM, IgG, and IgA titers decreased significantly between acute phase and late convalescence (p=0.02, p<0.0001, p<0.0001, respectively) (Figure 1B). Within the total S-binding antibodies, the percentage of anti-S IgM and IgA decreased whereas anti-S IgG increased over time (p=0.03, Figure 1C). The detection of neutralizing antibodies was achieved by foci reduction neutralization test using a Cambodian SARS-CoV-2 isolate. There was no difference in the titers of SARS-CoV-2 neutralizing antibodies between the acute and convalescent phase, even though titers tended to decrease over time (Figure 1D). Over time, the percentage of individuals positive for anti-S IgM (p<0.0001) and anti-S IgA (p<0.0001) decreased (Figure 1E). In the acute phase, 91% of individuals were positive for anti-S IgG, and only 70% of the individuals were positive for neutralizing antibody titers. Up to nine months post infection, the frequency of individuals positive for anti-S IgG remained stable (88%) whereas the frequency of individuals with neutralizing titers decreased to 56% (p=0.055) (Figure 1E). Analyzing only individuals with paired samples available, revealed similar results as the whole cohort (Figures S3A, B). Taken together, these data show that despite decreases in antibody titers over time, the percentage of individuals positive for anti-S IgG remains stable.

Besides neutralization, antibodies can mediate Fc-effector functions, such as complement activation, killing of virus-infected cells and phagocytosis of viral particles (20). To further define the humoral response in these individuals, we assessed antibody effector functions in vitro. The NIBSC references were utilized to validate the assays and nineteen pre-pandemic samples were measured to set the cutoff for each assay. Antibody-dependent cellular phagocytosis (ADCP) assay measures the engulfment of neutravidin beads coated with SARS-CoV-2 derived S1 by THP-1 cells (Figures 2A, S4). A decrease in ADCP can be observed between the acute and late convalescent phase (p=0.005, Figures 2B, C). The percentage of subjects with ADCP activity decreased from 73% to 55% over time. However, when calculating the proportion of ADCP within the total anti-S IgG, we observed a significant increase of the proportion of ADCP over time (p=0.003, Figure 2D).

Next, to evaluate the contribution of anti-S antibodies to complement dependent cytotoxicity (CDC), we assessed cell death in Raji cells engineered to express S protein in the presence of normal human serum as source of complement (Figure 2E, S5) (21). No differences in CDC activity was observed between the acute and late convalescent phase, where 60% and 56% of the subjects showed CDC activity, respectively (Figures 2F, G). The proportion of CDC-mediating antibodies within the total anti-S IgG fraction significantly increased between acute and late convalescence (p=0.0002, Figure 2H).

Killing of virus-infected cells can also be mediated by activated NK cells, after binding of immunocomplexes to CD16 (20). Therefore, antibody-dependent cellular cytotoxicity (ADCC) activity was measured using S-expressing 293T cells as target cells with degranulation measured by CD107a staining in primary NK cells as a readout for ADCC (Figures 2I, S6). ADCC activity did not change between the acute and late convalescent phase (Figures 2J, K). At both time points, 59% - 66% of individuals showed anti-S mediated ADCC activity. However, similar to ADCP and CDC, the proportion of ADCC-mediating antibodies within the fraction of anti-S IgG increased significantly over time (p<0.0001, Figure 2L). Analyzing only individuals with paired samples available, revealed similar results as the cohort as a whole (Figures S3C–H). Overall, these data show that antibody effector functions mediated by S-specific antibodies remain stable over time and that the proportion of the functional antibody response within the total anti-S antibodies increases over time.

Upon re-infection, memory B cells are rapidly activated to differentiate into antibody-producing plasmablasts and/or re-initiate germinal centers in the case of secondary heterologous infection with antigenically similar pathogens (40). Therefore, they may play an important role in long-term immune memory to SARS-CoV-2 and their evolving variants. We assessed the phenotype and frequency of antigen-specific memory B cells from 40 SARS-CoV-2 infected individuals in the cohort by staining with site-specific biotinylated recombinant S1 and N protein (Figures 3A, S7A, B). At late convalescence, 0.10% of median of the total CD27⁺ B cells are S1-specific, whereas 0.66% of median are N-specific (p<0.0001, Figure 3B). We observed the highest frequencies of S1-specific and N-specific B cells in the activated memory B cell compartment (CD27⁺CD38⁺) compared to plasma blast (CD27⁺CD38^hi) and resting memory compartments (CD27⁺CD38^–) (Figures 3C, D) (41, 42). The proportion of CD27⁺CD38⁺ S1-specific B cells (median=75%, IQR=30%) is significantly increased compared to the proportion of CD27⁺CD38⁺ N-specific B cells (median=39%, IQR=26%, Mann-Whitney Test, p<0.0001) (Figure 3E). Moreover, the proportion of S1- versus N-specific B cells varies within each CD27⁺ B cell subset (p<0.0001, Figure 3F). We next analyzed S1- and N-specific B cells within the unswitched (IgD^–IgM⁺) and switched (IgD^–IgG⁺ and IgD^–IgA⁺) B cell compartments (Figure S7A). S1-specific B cells were mainly IgD^–IgG⁺, whereas N-specific B cells were either IgD^–IgM⁺ or IgD^–IgG⁺ (Figures 3G, H). The proportion of IgD^–IgG⁺ S1-specific B cells (median=75%, IQR=24%) was significantly increased compared to the proportion of IgD^–IgG⁺ N-specific B cells (median=37%, IQR=17%) (p<0.0001) (Figure 3I). Therefore, within each switched B cell subset, the proportion of S1- versus N-specific B cells was different (p<0.0001) (Figure 3J). Taken together, SARS-CoV-2 infection induces a robust memory B cell response targeting both S and N.

In addition to humoral immune memory, the generation and maintenance of virus-specific cellular immune responses is critical to help prevent reinfection. Long-term maintenance and phenotypes of SARS-CoV-2-specific memory T cell responses are still under investigation (4, 43, 44). SARS-CoV-2-specific CD4⁺ and CD8⁺ T cells were assessed in 33 individuals at late convalescence by incubating PBMCs with peptide pools covering immunodominant sequences of the viral S1, M and N protein (Figure 4A). Post incubation, activation induced marker (AIM) assays identified CD4⁺ antigen-specific cells using OX40⁺CD137⁺ combined with phenotypic markers to measure different memory and T helper (Th) subsets (Figures S8A–D). Percentages of both S1- and M-specific CD4⁺ T cells were significantly increased compared to the percentage of N-specific cells (p<0.0001, p=0.0002), Figure 4B). Phenotypically, 42% of virus-specific T cells displayed an effector memory phenotype (CD45RA^–CCR7⁺) and 87% of the cells showed a Th1-skewed phenotype (CXCR3⁺CCR6^–) (Figures 4C, D). Comparing the memory phenotype of S1-, M- and N-specific cells, we observed that a lower proportion of S1-specific cells displayed an effector memory phenotype (23%) compared to M-specific cells (41%, p=0.0457) and N-specific cells (58%, p<0.0001) (Figure S9A). Moreover, 97% of M-specific cells showed a Th1-skewed phenotype compared to only 65% (p<0.0001) of the S1-specific cells and 71% (p=0.0130) of the N-specific cells (Figure S9B). In eight individuals, sufficient cell numbers were available to assess functionality by cytokine production after peptide stimulation using a multi-parameter ex vivo intracellular cytokine staining (ICS) assay (Figure S8E). SARS-CoV-2-specific CD4⁺ T cells produced Interleukin (IL)-2 (36%) or IL-6 (28%) after peptide stimulation, and were polyfunctional (Figures 4E, F). Percentages of IL-2⁺ and IL-17⁺ cells were significantly higher after S1 stimulation compared to M stimulation (p=0.046, p=0.017) (Figure S9C).

Next, we assessed the frequency and phenotype of cytotoxic CD8⁺ T cells by AIM assay using CD69⁺CD137⁺ to identify antigen-specific CD8⁺ T cells. Frequency of total SARS-CoV-2-specific CD8⁺ cells is 0.44% (median) (Figure 5A) with 61% of these SARS-CoV-2-specific CD8⁺ T cells being terminally differentiated effector memory cells (TEMRA, CD45RA⁺CCR7^–) (Figures 5B, S9D). No differences were observed between S1-, M- and N-specific CD8⁺ T cells. Similar to antigen-specific CD4⁺ T cells, SARS-CoV-2-specific CD8⁺ T cells produced either IL-2 (56%) or IL-6 (16%) after peptide stimulation, and were polyfunctional (Figures 5C, D), (Figure S9E). Interestingly, 2/33 (6%) individuals displayed no CD4⁺ T cell reactivity, and 6/33 (19%) individuals lacked a CD8⁺ T cell response after stimulation. In summary, sustained and functional CD4⁺ and CD8⁺ T cell responses are detected in the study participants, even after experiencing only mild or asymptomatic SARS-CoV-2 infection. These data suggest that SARS-CoV-2 can induce a long-lived cellular immune response, which could confer protection after reinfection or could be reactivated with vaccination.

In order to assess if symptomatic disease is associated to altered immune memory formation, we compared the functional immune response between asymptomatic and symptomatic patients with mild/moderate clinical presentation. Overall, no differences occurred in the titers of anti-S IgM, IgA, IgG, neutralization or antibody-effector functions assessed in the acute phase of infection (Figures 6A, S10). At late convalescence, symptomatic disease resulted in increased ADCC activity compared to asymptomatic individuals (p=0.0034) (Figure 6A). Other Fc-mediated effector functions, such as ADCP and CDC and neutralizing titers were not different between the patients (Figure S10). Percentages of N-specific CD27⁺ B cells, but not S1-specific, were increased in asymptomatic individuals versus patients who were symptomatic (p=0.051) (Figure 6B). Symptomatic disease resulted in increased percentage of SARS-CoV-2-specific CD4⁺ T cells (p=0.0018) with a central memory phenotype (p=0.0498) (Figures 6C, D), but no differences were observed in the CD8⁺ T cell compartment (Figure S10).

These data suggest that the outcome of acute infection has an imprint on the memory immune response with implications for response to subsequent infection or vaccination.

In order to assess the relation between antibody titers, functional humoral immune memory, and the cellular T and B cell compartment we performed extensive correlation analysis (Figure 7A). Age correlated to anti-S antibody titers and S1-specific CD19⁺IgD^–IgG⁺ and CD19⁺CD27⁺B cells. In the acute phase of infection, anti-S IgG, IgM and IgA titers, functionality, measured by seroneutralization, and effector functions correlated. Seroneutralization, anti-S IgA, and ADCC correlated over time, albeit not very strong. Of note, viral shedding did not correlate with anti-SARS-CoV-2 immunity (Figure 7A), and subdividing individuals into short or long viral RNA shedding using an arbitrary cut-off of 10 days did not yield any differences in SARS-CoV-2 specific immune responses (Figure S11).

At late convalescence, anti-S IgG correlated with all three effector functions, but not with neutralizing capacity. Within the B cell compartment, N-specific IgD^–IgG⁺, IgD^–IgA⁺ and CD27⁺ B cells correlated to one another, as did S1-specific IgD^–IgG⁺, IgD^–IgA⁺ and CD27⁺ B cells. No correlation was identified between S1- and N-specific B cells. Anti-S IgG titers, ADCP, and CDC correlated with S1-specific IgD^–IgG⁺ and CD27⁺ B cells. The S-specific CD4⁺ T cell responses correlated with S-specific CD8⁺ T cell responses, but did not correlate to antibody titers nor to effector functions or to S1-specific B cells.

Next, we assessed in more detail the relationship between the different effector functions at late convalescence. We included 56 individuals with measurable anti-S IgG above the cutoff based on the pre-pandemic samples (Figures 1, 7B). We found that plasma from each individual could induce neutralization or at least one antibody-effector function (ADCP, CDC and ADCC). In 11 out of the 56 individuals (19.6%), both neutralization titers and all antibody-effector functions could be measured. Eighteen out of the 56 individuals (32,1%) had measurable antibody-effector functions, but no neutralizing titers, while in 4 out of the 56 (7.1%) individuals only neutralizing titers were detected (Figure 7B).

We further detailed if individuals showed measurable immune responses in one or multiple immune compartments at 6-9 months post infection. In twenty individuals we assessed in parallell anti-S IgG, S1-specific B cells and S-specific T cells. We set an arbitrary cutoff for positive B and T cell responses at 0.1% of S-specific CD27⁺ B, CD4⁺ T and CD8⁺ T cells. Four out of the 20 individuals (20.0%) had responses in all compartments. Three individuals (15.0%) showed only measurable anti-S IgG titers. Moreover, 7 individuals (35.0%) had anti-S IgG titers and detectable S1-specific CD27⁺ B cells but no S-specific T cell responses. In contrast, 9 (45.0%) individuals had measurable S-specific T cell responses, but no S1-specific B cell responses or anti-S antibody titers (Figure 7C).

Overall, different aspects of a functional immune memory response do not fully correlate with one another and require separate evaluation when considering long-term immune memory to SARS-CoV-2.

Ethical approval for the study was obtained from the National Ethics Committee of Health Research of Cambodia. Written informed consent was obtained from all participants prior to inclusion in the study. Pre-pandemic blood samples were obtained from clinically healthy individuals included in the dengue vaccine initiative study in 2015-2016. Clinically healthy adult volunteers who presented at the International Vaccination Centre, Institut Pasteur du Cambodge before the onset of the pandemic were included to validate the antigen-specific B and T cell staining. Acute SARS-CoV-2 infected patients were identified via screening of imported cases in Cambodia between 6^th March to 12^th August 2020. All laboratory confirmed cases are quarantined and monitored for symptoms. Moreover, 1-15 follow-up nasopharyngeal/oropharyngeal swab samplings for SARS-CoV-2 detection were conducted to assess SARS-CoV-2 RNA shedding. Patients were only discharged after two consecutive negative RT-PCR tests within 48h. Symptomatic patients displayed mild/moderate symptoms such as running nose, cough, fever and difficult to breath. In total, we included 64 individuals for follow up. In 33 individuals, 2-9 days after laboratory confirmation, a blood sample was obtained. A second blood sample was obtained 6-9 months later from all 64 study participants. Participant characteristics and clinical signs are summarized in Table S1. Plasma was collected and stored at -80°C, The PBMCs were isolated via Ficoll-Paque separation, cryopreserved in 10% DMSO/FBS and stored in liquid nitrogen until analysis. The National Institute for Biological Standards and Control (NIBSC) 20/130 (research reagent) and 20/118 (reference panel) have been obtained from WHO Solidarity II, the global serologic study for COVID-19.

Molecular detection of SARS-CoV-2 in combined nasopharyngeal/oropharyngeal swabs was performed as previously described (38). Briefly, RNA was extracted with the QIAamp Viral RNA Mini Kit (Qiagen) and real-time RT-PCR assays for SARS-CoV-2 RNA detection were performed in using primers/probes from Charité Virologie [Berlin, Germany (80)] to detect both E and RdRp genes.

The detection of neutralizing antibodies was achieved by foci reduction neutralization test (FRNT) similar as described before (81) and adapted to SARS-CoV-2. Briefly, serial diluted, heat-treated plasma samples were incubated with a Cambodian SARS-CoV-2 isolate (ancestral strain; GISAID: EPI_ISL_956384 (38);) for 30min at 37°C and 5% CO₂. The mixtures were distributed on African green monkey kidney cells (VeroE6; ATCC CRL-1586) and incubated again for 30min 37°C and 5% CO₂. Afterwards, the mixtures were replaced by an overlay medium containing 2% carboxymethyl cellulose (Sigma-Aldrich) in Dulbecco's modified Eagle medium (DMEM; Sigma-Aldrich) supplemented with 3% FBS (Gibco) and 100 U/mL penicillin-streptomycin (Gibco). Infection was visualized 16-18h after inoculation by staining of infected cells with a SARS-CoV-2-specific antibody (rabbit, antibodies-online GmbH), targeting the S2 subunit of the viral spike protein, and afterwards with antibody anti-rabbit IgG HRP conjugate (goat; antibodies-online GmbH). Finally, cells were incubated with TrueBlue TMB substrate (KPL), and infection events appear as stained foci and were counted with an ELISPOT reader (AID Autoimmune Diagnostika GmbH, Strassberg, Germany). The amount of neutralizing antibodies is expressed as the reciprocal serum dilution that induces 50% reduction of infection (FRNT50) compared to the positive control (virus only) and is calculated by log probit regression analysis (SPSS for Windows, Version 16.0, SPSS Inc., Chicago, IL, USA). FRNT50 titers below 10 are considered negative.

Transfected cell lines, Raji (ATCC^® CCL‐86™) and 293T (ATCC^® CRL‐3216™), with SARS-Cov-2 spike plasmid or a control plasmid using Lipofectamine 2000 (Life technologies) are kind gifts from Olivier Schwartz, Institut Pasteur, Paris, France (21). Spike-expressing Raji cells and Raji control cells were cultured at 37°C, 5% CO2 in RPMI medium while 293T-spike cells and 293T control cells were cultured in DMEM medium. All media were completed with 10% FBS (Gibco, MT, USA), 1% L glutamine (Gibco), 1% penicillin/streptomycin and puromycin (1 μg/mL, Gibo™) for cell selection during the culture.

The S-Flow assay was performed as previously described (39). Briefly, plasma samples were diluted (1:200) in 1xPBS with 2mM EDTA and 0.5% BSA (PBS/BSA/EDTA) and incubated with 293T-spike cells (80000 cells/100µl) for 30 minutes on ice. The cells were washed with PBS/BSA/EDTA and stained either with anti-IgM PE (dilution 1:100, Biolegend) and anti-IgG Alexa Fluor™ 647 (dilution 1:600, Thermo Fisher) or anti-IgA Alexa Fluor 647 (dilution 1:800, Jakson ImmunoResearch) for 30 minutes on ice. The cells were washed with 1xPBS and fixed using buffer of the True-Nuclear Transcription Factor Staining kit (Biolegend). After fixing, the cells were washed and resuspended in 1xPBS. The results were acquired using FACS Canto II, BD Biosciences. The gating strategy for anti-IgM, anti-IgG or anti-IgA positive cells was based on the 293T control cells incubated with negative SARS-CoV-2 reference plasma. The data were reported as percentage of positive cells for anti-IgM, anti-IgG or anti-IgA. The NIBSC Research Reagent (20/130) and panel (20/118) (WHO Solidarity II) was utilized to set the cutoff for positivity based on the background staining of the negative SARS-CoV-2 plasma and calculated following formula: cut-off= % positive cells + 2x standard deviation.

THP-1 cells (ATCC^® TIB-202™) were used as phagocytic cells. For this, 1 µg of biotinylated S1 protein (Genscripts) was used to saturate the binding sites on 1 µl of FluoroSphere neutravidin beads (Thermo Fisher) overnight at 4°C. Excess protein was removed by washing the pelleted beads. The protein-coated beads were incubated with 40 µl heated-inactivated plasma diluted in complete RPMI (1:40) for 15 minutes at room temperature. Then, 5x10⁴ THP-1 cells suspended in 50 µl complete RPMI were added to the complex and incubated for 16 hours at 37°C, 5% CO₂ After incubation, the cells were washed with 1xPBS and fixed using buffer in True-Nuclear Transcription Factor Staining kit (Biolegend). After fixing, the cells were washed and resuspended in 1xPBS. The samples were analyzed using FACS Canto II, BD Biosciences. Phagocytosis activity was scored by the integrate mean fluorescence intensity (iMFI) value (% positive fluorescence THP-1 cells x MFI of the positive fluorescence THP-1 cells).

The assay used spike-expressing Raji cells as target cells, pooled serum (4 healthy donors) as complement source and heated-inactivated patient plasma as antibody source. In short, 50 µl of heated-inactivated plasma (1:50) were incubated with Raji-spike cells for 30 minutes at 37°C, 5% CO₂. Afterward, 50 µl of complete RPMI containing 15% of pooled serum was added into the cells and incubated at 37°C, 5% CO₂ for 14 hours. The cells were washed with PBS and stained with Zombie Aqua viability dye (BioLegend) for 20 minutes on ice and then stained anti-APC C3/C3b/iC3b antibody (Cedarlane) for 30 minutes on ice. The cells were fixed with fixation buffer in True-Nuclear Transcription Factor Staining kit (Biolegend) for 20 minutes on ice. After fixing, the cells were washed and resuspended in 1xPBS. The samples were acquired using FACS Canto II, BD Biosciences. The results were reported as percentage of cell death and MFI of C3 deposition on the cells.

The assay used 293T-spike cells as a target cell and purified NK cells from healthy donor PMBCs as effector cells. First, 293T-spike cells were incubated with heated-inactivated patient plasma diluted in complete DMEM medium (1:50) at 37°C, 5% CO₂ for 30 minutes. The NK cells were enriched by magnetic negative selection (Miltenyi) according to manufactor's instruction. The 293T-spike cells were washed five times with complete RPMI medium. The NK cells were mixed with 293T-spike cells at a ratio 1:1 at final volume of 100 µl complete RPMI. Anti-CD107a and Monensin (Biolegend) 1:1000 dilution were added to the suspension and incubated at 37°C, 5% CO₂ for 6 hours. The cells were washed with 1xPBS and stained with Zombie Aqua viability dye (BioLegend) for 20 minutes on ice. Then the cells were stained with anti-CD3 and anti-CD56 for 30 minutes on ice. The cells were washed and fixed/permeabilized using True-Nuclear Transcription Factor Staining kit (Biolegend) for 20 minutes on ice. After staining, the cells were washed and resuspended in 1xPBS. The samples were acquired using FACS Canto II, BD Biosciences.

Biotinylated SARS-CoV-2 S1 protein and biotinylated SARS-CoV-2 N protein were purchased from GenScript. The biotinylated proteins were combined with different streptavidin (SA) fluorophore conjugates, BUV496 (BD Biosciences) and PE (Biolegend), respectively, at 1:1 molar ratio. Briefly, each SA was added gradually (3 times, every 20 minutes) to 20 µl of each biotinylated protein (1 µM) on ice. The reaction was quenched with D-biotin (GeneCopeia) at 50:1 molar ratio to SA for a total probe volume of 30 µl for 30 minutes on ice. Probes were then used immediately for staining. Each staining used 5 µl of probe. Shortly, patient PBMCs was washed with 1xPBS and stained with Zombie Aqua viability dye (BioLegend) for 10 minutes on ice. The cells were stained with the probes. Then the cells were washed and stained with anti-IgG antibody, for 30 minutes on ice. After that, the cells were washed and stained with master mix containing of anti-CD3, anti-CD19, anti-CD27, anti-CD38, anti-IgD, anti-IgM and anti-IgA antibodies for 30 minutes on ice Antibodies are listed in. After staining, the cells were washed and resuspended in 1xPBS with 2% FBS. The samples were analyzed using FACS Aria, BD Biosciences. The flow cytometry gating strategy to classify memory B cell subsets and switched B cells is shown in. Overall, 40 samples were of sufficient quality and were included in the analysis. 1 1

Antigen-specific CD4⁺ and CD8⁺ T cells, as well as memory T cells and T helper subsets were assessed by Activation-Induced Marker (AIM) assay (4, 8). Cells were cultured at 37°C, 5% CO₂, in the presence of SARS-CoV-2-specific S1, M and N protein pools [1 µg/mL] (PepTivator^® SARS-CoV-2 regents; Miltenyi Biotec) in 96-well U-bottom plates at 0,5-1x10⁶ PBMCs per well. After 24 hours, cells were washed in 1xPBS supplemented with 0.5% bovine serum albumin (BSA) and 2 mM EDTA (FACS buffer) and stained with Zombie Aqua Fixable Viability kit (Biolegend) and incubated for 20 min at 4°C followed by surface staining for 30 min at 4°C. Stained cells were washed and resuspended in FACS buffer and analyzed using a FACSAria Fusion (BD Biosciences). Antibodies are listed in Table S2. Negative controls without peptide stimulation were included for each donor. Antigen-specific CD4⁺ and CD8⁺ T cells were measured subtracting the background (unstimulated control) from the peptide-stimulated sample. Negative results were set to zero. Data were analyzed with FlowJo software version 10.7.1 (FlowJo LLC). Overall, 33 samples were of sufficient quality and were included in the analysis.

Functional SARS-CoV-2-specific CD4⁺ and CD8⁺ T cells were assessed by surface and intracellular staining in a subset of individuals if sufficient amount of PBMCs were obtained (n=8). Cells were cultured at 37°C, 5% CO2, in the presence of SARS-CoV-2-specific S1, M and N protein pools separately [1 µg/mL each] (PepTivator^® SARS-CoV-2 reagents; Miltenyi Biotec), Monensin (Biolegend) 1:1000 dilution and anti-Human CD28/CD49d purified [100 µg/mL] (BD Bioscience) in 96-well U-bottom plates at 0,5-1x10⁶ PBMCs per well. After 6 hours, cells were washed in FACS buffer and stained using a Zombie Aqua Fixable Viability kit (Biolegend) and incubated for 20 minutes at 4°C. Cells were then washed in PBS and fixed/permeabilized with True-Nuclear™ Transcription Factor Buffer Set (Biolegend). Surface (CD3, CD4 and CD8) and intracellular markers (IFN-γ, IL-2, IL-4, IL-6 and IL-17) were detected via the subsequent addition of directly conjugated antibodies incubating for 30 minutes at 4°C. Antibodies are listed in Table S2. Stained cells were finally washed and resuspended in FACS buffer and analyzed using a FACSAria Fusion (BD Biosciences). Antigen-specific CD4⁺ and CD8⁺ T cells were measured subtracting the background (unstimulated control) from the peptide-stimulated sample. Negative results were set to zero. Data were analyzed with FlowJo software version 10.7.1 (FlowJo LLC).

Calculations, figures and statistics were made using Prism 9 (GraphPad Software) or RStudio (Version 1.2.1335). The data were tested for statistical normality before applying the appropriate statistical tests. All information about sample sizes and statistical tests performed were shown in the figure legends. Spearman correlation plot was calculated and visualized with the following packages: FactoMineR, factoextra (https://cran.r-project.org/web/packages/factoextra/index.html↗) and corrplot (https://github.com/taiyun/corrplot↗) in R (Version 3.6.1) and RStudio (Version 1.2.1335).