Immunogenicity and efficacy of heterologous ChAdOx1–BNT162b2 vaccination

, , Following severe adverse reactions to the AstraZeneca ChAdOx1-S-nCoV-19 vaccine 1,2 , European health authorities recommended that patients under the age of 55 years who received one dose of ChAdOx1-S-nCoV-19 receive a second dose of the Pfizer BNT162b2 vaccine as a booster. However, the effectiveness and the immunogenicity of this vaccination regimen have not been formally tested. Here we show that the heterologous ChAdOx1-S-nCoV-19 and BNT162b2 combination confers better protection against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection than the homologous BNT162b2 and BNT162b2 combination in a real-world observational study of healthcare workers ( n = 13,121). To understand the underlying mechanism, we conducted a longitudinal survey of the anti-spike immunity conferred by each vaccine combination. Both combinations induced strong anti-spike antibody responses, but sera from heterologous vaccinated individuals displayed a stronger neutralizing activity regardless of the a possible confounder) and the vaccination regimen. Results showed that age is not associated with infection status ( P = 0.4514). In particular, patients aged 60 years or above (715 out of 13,121) were not more infected (3 out of 715; 0.42%) than patients aged 60 years or below (88 out of 12,406 (0.71%); two-sided Fisher’s exact test, P = 0.4890).

The ChAdOx1-S-nCoV-19 and BNT162b2 vaccines (hereafter referred to as ChAd and BNT, respectively) have been shown to confer strong protection against SARS-CoV-2 infection and to induce robust immune responses directed against the spike protein of SARS-CoV-2 when administered in a homologous setting [3][4][5] . However, the effectiveness of a ChAd-BNT combination as a prime-boost regimen has not been formally tested in vaccine trials. To compare the risk of SARS-CoV-2 infection following heterologous ChAd-BNT or homologous BNT-BNT vaccination schedules, we extracted data from the occupational medicine database of the University Hospital of Lyon (Hospices Civils de Lyon), France. Specifically, we analysed the number of individuals who received each vaccination regimen (which started in January 2021 in both groups) and the number of SARS-CoV-2 infections (documented by a positive PCR with reverse transcription (RT-PCR) result) that occurred at least 2 weeks after the booster dose in each group (Table 1). Ten infections were identified out of 2,512 individuals (0.40%) in the heterologous vaccination group compared with 81 infections out of 10,609 individuals (0.76%) in the homologous vaccination group (Fig. 1). As the age of the individual and the vaccination regimen are known covariates, we used a multiple logistic regression model to model the probability of being infected. Our predictors were age (set as a continuous variable, considered as a possible confounder) and the vaccination regimen. Results showed that age is not associated with infection status (P = 0.4514). In particular, patients aged 60 years or above (715 out of 13,121) were not more infected (3 out of 715; 0.42%) than patients aged 60 years or below (88 out of 12,406 (0.71%); two-sided Fisher's exact test, P = 0.4890).

Article
However, the vaccination regimen was significantly associated with the probability of being infected after vaccination (P = 0.0384). Individuals vaccinated with BNT-BNT were twice as likely to be infected than those vaccinated with ChAd-BNT (relative risk of 2.03).

Antibodies and virus neutralization
To understand the immunological basis of this difference, we assembled a cohort of healthcare workers (HCWs) without comorbidity (Covid-Ser-Vac). Half of this HCW cohort received the homologous BNT-BNT combination within the recommended 4-week interval between the two doses, whereas the second half received the BNT booster after receiving the first ChAd dose approximately 12 weeks before. For both groups, the mean age was 41 years and 34 years, respectively, and about 70% of participants were female ( Table 2). Three blood samples were drawn: before vaccination, before the second dose and 4 weeks after the second dose. Immunological analyses were performed longitudinally (Extended Data Fig. 1), and a linear regression model was used to compare immune parameters between groups to adjust for age differences.
As previously shown 6,7 , the BNT vaccine induced a stronger level of SARS-CoV-2 receptor binding domain (RBD) or spike S1-specific IgG than the ChAd vaccine after the first dose. However, these IgG titres were no longer statistically different between groups after the second dose (Fig. 2a, b). Moreover, the positivity rate was 100% for both heterologous and homologous vaccinations. Serum IgA levels measured after the booster vaccination tended to be greater in the homologous than in the heterologous setting (median (interquartile range (IQR)) of 46.7 ng m -1 equivalent (Eq) (36.30-78.70 ng m -1 ng Eq) versus 37.4 ng ml -1 Eq (25.40-59.80 ng ml -1 Eq), respectively; equivalence was to the standard used (Methods)) ( Fig. 2c).
We then measured the ability of serum samples to neutralize SARS-CoV-2 spike-pseudotyped viral particles 8 . None of the sera displayed a detectable neutralizing activity before vaccination or after the first dose. However, most of the sera displayed a neutralizing activity after the second dose, and those from the ChAd-BNT vaccinated individuals displayed a higher neutralizing efficacy than those from the BNT-BNT vaccinated individuals (median (IQR) of 62% (34-93%) versus 99% (89-100%) of neutralization, respectively; P < 0.001) (Fig. 2d).
To validate these results, we tested the ability of the sera to prevent Vero E6 cell infection by SARS-CoV-2 isolates belonging to various clades, including 19A (B38 lineage), Alpha (B.   sequenced and confirmed to contain the characteristic mutations of its viral clade. The ChAd-BNT vaccinated individuals exhibited 2.3-fold to 3.6-fold higher serum neutralizing antibody titres against the different variants than BNT-BNT vaccinated individuals (Fig. 2e). Moreover, in the latter group, the neutralizing responses observed against the Beta, Gamma and Delta variants were significantly reduced compared with that against the 19A clade. By contrast, the neutralizing activity detected in sera from ChAd-BNT vaccinees was similarly high against all strains except the Beta clade (Fig. 2e, f). These data demonstrate that the neutralizing potential of antibodies generated by heterologous vaccination is less affected by spike mutations appearing in variants of concern than that of antibodies induced by homologous vaccination.

Vaccine-induced memory cells
We then studied the memory B cell (mBC) compartments using fluorescently labelled tetrameric RBDs to track RBD-specific mBCs and to analyse their phenotype (the gating analysis is presented in Extended Data Fig. 2a). As shown in Fig. 3a and Extended Data Fig. 2b, which correspond to concatenated flow cytometry plots, the frequency of RBD-binding mBCs was comparable for both vaccination groups after the prime dose but was on average twofold higher for the ChAd-BNT vaccination schedule after the booster dose. Moreover, switched (IgD -CD27 + ) mBCs accounted for about 67% and 47% of the RBD-specific mBC pool in the ChAd-BNT group and the BNT-BNT group, respectively (Fig. 3b, c, Extended Data Fig. 2c). The expression pattern of IgM and IgG isotypes was next used to compare the immunoglobulin isotype distribution among the switched mBCs. The results showed that the homologous vaccination schedule promoted more IgG-switched mBCs (62% versus 48%) than the heterologous vaccination schedule (Fig. 3d, Extended Data Fig. 2d). Finally, to address the activation status of mBCs, we analysed their expression of CD21 and CD11c. It is generally considered that the concurrent loss of CD21 and the upregulation of CD11c are phenotypic features associated with B cell activation independent of the developmental stage 10 .
Resting mBCs were therefore defined as CD21 + CD11ccells, whereas CD21 + CD11c + , CD21 -CD11c + and CD21 -CD11ccells were globally considered as activated mBCs. The frequencies of activated RBD-specific mBCs were significantly enhanced in the ChAd-BNT group compared with the BNT-BNT group (Fig. 3e, Extended Data Fig. 2e). In particular, the proportions of the CD21 -CD11c + subset were almost three times higher in the ChAd-BNT group. Altogether, our results indicate that compared with the homologous vaccination schedule, the ChAd-BNT combination induces significantly more RBD-specific B cells, increases the frequencies of post-switch mBCs and induces a more active mBC generation process. In all panels, box-and-whiskers plots are shown (see Methods for details), and the median is represented by the magenta middle line. A linear regression model was used to compare values between groups, and this model was corrected for age.
Exact P values are shown for the indicated comparisons when significant or nearly significant. f, Comparison of serum neutralizing activity against the reference lineage 19A and against the variants of concern for each group. P values are shown and calculated using the linear regression model described in e.

Article
We then longitudinally monitored the T cell response of vaccinees against RBD peptides using a whole-blood interferon-γ (IFNγ) release assay. The T cell response was higher after ChAd immunization than after BNT priming and was similarly increased by the BNT booster dose in both groups (median of 0.33 UI ml -1 versus 0.43 UI ml -1 ) (Fig. 3f). To confirm and consolidate these results, we measured the spikespecific CD4 and CD8 T cell response at the single-cell level by flow cytometry after stimulating peripheral blood mononuclear cells (PBMCs) from vaccinees with a pool of commercial peptides spanning the entire spike protein and then staining for intracellular IFNγ (the gating strategy is presented in Extended Data Fig. 3). Both vaccination regimens were able to induce a progressive increase in the frequency of S-reactive CD4 and CD8 T cells from the pre-vaccine to the post-booster phase (Fig. 3g, h). However, the heterologous combination resulted in a stronger CD4 T cell response, both after priming and boosting, and in an increasing trend in the CD8 T cell response after the booster dose.
Finally, to understand interrelations between immune parameters, we performed a matrix analysis (Extended Data Fig. 4). This confirmed that the neutralizing antibody activity correlated with the titres of spike-specific IgG, regardless of the variant analysed, and not correlated with IgA levels. Moreover, the neutralization activity correlated with the percentage of switched RBD + mBCs and with the percentage of activated RBD + mBCs. This provides further support that better neutralizing antibodies are produced as a result of sustained B cell activation.

Discussion
Heterologous prime-boost vaccinations have been reported to be more immunogenic than homologous ones in experimental settings 11 . This has also been proven for some human vaccines 12,13 . Moreover, studies using mice have demonstrated the strong immunogenicity of the ChAd-BNT combination 14 . Here we report that heterologous ChAd-BNT vaccination for intracellular IFNγ expression among CD4 (g) or CD8 (h) T cells after PBMC stimulation with commercial spike peptides. In all panels, box-and-whiskers plots (see Methods for details) of n = 29 and n = 31 participants in heterologous and homologous vaccination groups, respectively, are shown. The median value is shown in each group as a purple line. A high cellular mortality rate at the thawing step led to the exclusion of a few samples in each panel; each individual dot corresponds to one participant, and each measurement was performed once. In all panels, a linear regression model was used to compare values between groups, and this model was corrected for age. Exact P values are shown for the indicated comparisons when significant or nearly significant.
confers better protection against infection-which is associated with more switched mBCs and higher virus neutralizing antibody titres-irrespective of the variant analysed. This finding is of particular importance considering the global increase in the SARS-CoV-2 Delta variant 15 . Other studies reported better neutralizing potential of sera from ChAd-BNT vaccinated individuals compared with BNT-BNT vaccinated individuals, but these studies only used surrogate 16,17 or pseudovirus 18 neutralization assays. Moreover, the ability of such assays to predict neutralizing activity against authentic clinical virus isolates is still debated 19 . Heterologous vaccination induced antibodies with a stronger neutralization potential than homologous vaccination, but the spike-specific IgG antibody titres were comparable. This result suggests that it is the quality of the antibodies that underlies the neutralization potential of the antibody response induced by heterologous vaccination. Assuming that the secondary antibody response is primarily derived from the mBC clones generated by primo-immunization, we can postulate that ChAd and BNT formulations elicit different mBC compartments. mBCs generated by the ChAd formulation could, for example, bear antigen receptors displaying a larger breadth of epitope recognition or a better fit for the SARS-CoV-2 spike protein. This could be linked to the different conformations of the spike protein, as that of the BNT vaccine bears a mutation stabilizing the protein in its pre-fusion conformation 20 . Our data showed that mBCs still exhibit features of activation 4 weeks after the booster dose in the heterologous regimen, which suggests that the post-boost process of mBC differentiation is more active in this schedule. This could result from a more efficient or long-lasting germinal centre reaction that facilitates an extended cycle of somatic hypermutations and possibly output of B cell clones with higher affinity antigen receptors. An enlarged pool of CD4 + T helper cells could contribute to an amplified or prolonged germinal centre reaction. In keeping with this notion, we report here a higher frequency of spike-specific CD4 T cells after priming with the ChAd vaccine, which has been observed in other studies 6,7 . Moreover, stronger T cell responses may contribute to better protection independently of B cell responses 21 .
The better neutralization potential of sera from ChAd-BNT vaccinated individuals could also be linked to their relatively lower level of S1-specific IgA levels. Indeed, plasma IgA monomers specific to SARS-CoV-2 proteins are twofold less potent than IgG equivalents 22 . Furthermore, IgA levels persist better than IgG after infection, which coincides with a reduction in the serum neutralizing potential 23 . Following a possible competition between serum IgA and IgG, the decrease in IgA levels in heterologous vaccination conditions may lead to better neutralization effects. This phenomenon has already been observed in the context of the RV144 vaccination trial against HIV-1, in which the level of envelope-specific IgA correlated with the risk of infection 24 . In this context, IgA was also found to inhibit antibody-dependent cell cytotoxicity by competing with IgG 25 .
Our data do not allow the discarding of the possibility that the later timing of administration of the booster dose contributed to the high efficacy of the heterologous regimen. Still, it is noteworthy that the size of the RBD + mBC pool after the prime dose was comparable for both groups of vaccinees, despite the disparity in the analysis time point after the prime vaccination (4 versus 12 weeks). This observation indicates that the extended interval between prime and boost in the ChAd-BNT cohort has not been translated into a greater size of the RBD-specific mBC compartment. It has been recently demonstrated for both the ChAd vaccine 26 and the BNT vaccine 27 that long injection time intervals (12 weeks or more) provide higher binding and neutralizing antibody titres than shorter intervals (less than 6 weeks). For a comparable long vaccination interval (more than 9 weeks), the heterologous ChAd-BNT vaccination schedule induced better humoral immunity (titres and neutralization) than the homologous ChAd-ChAd combination 28 . To our knowledge, there is only one study in which ChAd-BNT and BNT-BNT vaccination schedules were compared with the same prime-boost interval (4 weeks) 6 . That study 6 concluded that for short intervals, the humoral responses induced by both regimens were comparable. However, a longer interval (that is, 12 weeks) between prime and boost could be needed to allow the synergy between heterologous vaccines.
Together, we present a real-world observational study of HCWs showing that the heterologous ChAd-BNT vaccination regimen confers stronger protective immunity than the homologous BNT-BNT prime-boost schedule. As hidden confounding factors might be present in our study (for example, different levels of exposure to the virus), confirmatory studies and a longer follow-up of vaccinated participants are warranted. However, our data suggest that the heterologous combination could be particularly suitable for immunocompromised individuals.

Online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41586-021-04120-y.

Ethical statement and cohort description
The use and analysis of data from the occupational health medical file were authorized after a regulatory declaration to the National Commission for Information Technology and Civil Liberties according to the reference methodology (declaration MR004 number 20-121 of 30 April 2020). The declaration of SARS-CoV-2 infection is compulsory for all staff to obtain daily allowances without loss of salary during the imposed quarantine.
For the Covid-Ser-Vac study, clinical data were recorded by a trained clinical research associate using Clinsight software (v.Csonline 7.5.720.1). Blood samples were processed and stored at the Centre de Ressource Biologique Neurobiotec, 69500 Bron. Sixty naive HCWs for COVID-19 and vaccinated with BNT and/or ChAd vaccines were included in a prospective longitudinal cohort study conducted at the Hospices Civils de Lyon. Blood sampling was performed before vaccination, before the second dose of vaccine and 4 weeks after the end of vaccination schedule. The absence of previous SARS-CoV-2 infection was confirmed using the Wantai SARS-CoV-2 Ab total assay in the pre-vaccine sample. Demographic characteristics and delays between doses are depicted in Table 2 and Extended Data Fig. 1. Written informed consent was obtained from all participants. Ethics approval was obtained from the national review board for biomedical research in April 2020 (Comité de Protection des Personnes Sud Méditerranée I, Marseille, France; ID RCB 2020-A00932-37), and the study was registered at ClinicalTrials.gov (NCT04341142).

Measurement of IgG titres
Sera were immediately stored at -80 °C after blood sampling. S1-specific IgG and RBD-specific IgG were measured using Siemens Atellica IM SARS-CoV-2 IgG (sCOVG) and bioMérieux Vidas SARS-CoV-2 IgG diagnosis kits, respectively, according to the manufacturers' recommendations. For standardization of these assays to the first World Health Organization international standard, the concentrations were transformed into binding antibody units per ml (BAU ml -1 ) using the conversion factors provided by the manufacturers.

Measurement of spike-specific IgA
ELISAs to evaluate IgA binding to the SARS-CoV-2 spike protein were performed as previously described 29 . High-binding 96-half-well plates (2310M, Nunc) were coated with 100 μl per well of a spike protein solution (1 μg ml -1 ; 40591-V08H spike S1-RBD, Sino Biologicals) in PBS overnight at 4 °C. Plates were washed with washing buffer containing 1× PBS with 0.05% Tween 20 (Sigma-Aldrich) and incubated with 170 μl of blocking buffer per well containing 1× PBS with 3% fat milk powder and 0.05% Tween 20 (Sigma-Aldrich) for 1 h at room temperature. Immediately after blocking, recombinant anti-RBD IgA (B Cell Design, IB3C4 PV) or serum samples diluted in PBS were added and incubated for 1 h at 37 °C. Plasma samples were assayed at a 1:100 starting dilution and 7 additional 3-fold serial dilutions. Recombinant human anti-RBD IgA was used to perform a calibration curve starting at 1.5 μg ml -1 . Plates were washed and then incubated with anti-human IgA (A0295, Sigma-Aldrich) secondary antibody conjugated to horseradish peroxidase in blocking buffer at 1:10,000. Plates were developed by the addition of the horseradish peroxidase substrate 3,3′,5,5′-tetramethylbenzidine (TMB; 34021, Thermo Fisher Scientific) for 10 min, and the developing reaction was stopped by adding 50 μl of 1 M HCl. Optical density units were measured at 450 nm in a microplate reader (Tecan). For serum samples, a positive control (serum pool from critically ill patients with COVID-19, diluted 200-fold in PBS) and a negative control (pool of historical serum samples) were added in duplicate to each run. After deduction of the background, a relative content in IgA equivalent (ng ml -1 Eq) was calculated using the calibration curve. The limit of detection of the assay was 0.1 ng ml -1 Eq.

Live-virus neutralization experiments
A plaque reduction neutralization test (PRNT) was used for the detection and titration of neutralizing antibodies. A tenfold dilution of each serum specimen in culture medium was first heated for 30 min at 56 °C to avoid complement-linked reduction of viral activity. Serial twofold dilutions (tested in duplicate) of the serum specimens in culture medium were mixed in equal volume with the live SARS-CoV-2 virus. After gentle shaking and a contact of 30 min at room temperature in plastic microplates, 150 μl of the mix was transferred into 96-well microplates covered with Vero E6 cells (American Type Culture Collection (ATCC), CRL-1586, not authenticated but regularly tested for mycoplasma contamination). The plates were incubated at 37 °C in a 5% CO 2 atmosphere. Infection efficiency was evaluated by microscopy 5 days later when the cytopathic effect of the virus control reached 100-500 TCID 50 (median culture infectious dose) per 150 μl. Neutralization was recorded if more than 50% of the cells present in the well were preserved. The neutralizing titre was expressed as the inverse of the higher serum dilution that exhibited neutralizing activity; a threshold of 20 was used (PRNT 50 titre ≥ 20). All experiments were performed in a biosafety level 3 laboratory. The different viral strains that were used were sequenced and deposited at GISAID

Monitoring of T cell responses using whole-blood IFNγ release assay
Fresh blood collected in heparinized tubes was stimulated for 22 h at 37 °C under 5% CO 2 with SARS-CoV-2 peptide pools (derived from the prototype Wuhan strain NC_045512.2) targeting RBD (46 peptides, [C] = 8 μg ml -1 ) (bioMérieux) diluted in IGRA solution (bioMérieux). The peptides (15-mer) encompassed the entire protein sequence and overlapped by five residues. The IGRA solution was used as a negative control and a mitogen as a positive control. The concentration of IFNγ in the supernatant was measured using a VIDAS automated platform (VIDAS IFNγ RUO, bioMérieux). The measuring range was 0.08-8 IU ml -1 and WB IGRA positivity thresholds were defined at 0.08 IU ml -1 . The IFNγ response was defined as detectable when the IFNγ concentration of the test was above threshold and the negative control was below threshold or when the IFNγ concentration of the test minus IFNγ concentration of the negative control was above threshold.

SARS-CoV-2 pseudoparticle preparation and neutralization
SARS-CoV-2 spike-pseudotyped murine leukaemia virus retrovirus particles were produced as described for SARS-CoV 30 . In brief, HEK293T Article cells (ATCC, CRL-1573, not authenticated but regularly tested for mycoplasma contamination) were transfected with constructs expressing murine leukaemia virus Gag-Pol, the green fluorescent protein (GFP) reporter and the SARS-CoV-2 spike protein (a gift from D. Lavillette (CAS Key Laboratory of Molecular Virology & Immunology, Institut Pasteur of Shanghai Chinese Academy of Sciences, Pasteurien College, Soochow University, Jiangsu, China)). Control particles pseudotyped with the unrelated RD114 virus surface glycoprotein (from a cat endogenous virus) were generated as previously described 31 . For neutralization assays, a sample of approximately 1 × 10 3 pseudoparticles was incubated with a 100-fold dilution of sera or control antibodies for 1 h at 37 °C, spinoculated for 2 h at 2,500g before infection of Vero E6 cells. After 72 h of infection, the percentage of GFP-positive cells was determined by flow cytometry. As a control, the same procedure was performed using RD114 pseudoparticles. Anti-spike SARS-CoV-2 RBD (Sino Biological, 40150-V08B2) and anti-gp70 RD114 (ViroMed Biosafety Labs) antibodies were used with a 100-fold dilution as positive and negative control, respectively.

Statistical analyses
Statistical modelling for epidemiological data. To model the probability of being infected knowing the subject age and the used vaccination regimen, we used the generalized linear model (glm) function in R to perform a multiple logistic regression model with the logit link function. The outcome variable was the infection status, whereby infected = 1 and not infected = 0. Our predictors were the age, as a continuous variable (considered as a possible confounder), and the vaccination regimen, whereby Chad-BNT = 0 and BNT-BNT = 1. On the basis of the model coefficients for the vaccination regimen (B0 = -5.306 and B1 = 0.712), we also assessed the relative risk (RR) as follows: RR = (1 + exp(-B0))/(1 + exp(-B0 - B1)).

Statistical analysis for immunological and virological analyses.
ChAd-BNT and BNT-BNT populations showed a significant difference in the average ages. To test for the consequence of the vaccination regimen on the different immune parameters, we fitted a multiple linear regression model, which allows for the correction of age (as a possible confounding factor) by including age and vaccination regimen simultaneously as predictors in the right explanatory side of the lm function in R (one model per immune parameter). Results of statistical significance displayed in Figs. 2 and 3 are from these adjusted models, for which age showed no effect. For box and whiskers plots, the upper and lower bounds of the box are the 75th and the 25th percentile (the third Q3 and first Q1 quartile), respectively. The IQR = Q3 - Q1; an observation was considered an outlier if it was above Q3 + 1.5 × IQR or below Q1 - 1.5 × IQR. Upper and lower whiskers represent the maximum and the minimum values, respectively, without taking into account the outliers.

Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this paper.    All studies must disclose on these points even when the disclosure is negative.

Sample size
We did not perform sample size calculations as we did not know what to expect in terms of difference for the different measurements. We tried to analyze 30 patients in each group for all immunological measurements, as, based on our experience, it was sufficient to reach good statistics for most of the measurements, given that we minimize variability by selecting patients without comorbidity with similar age range, to avoid batch effects patients from each group (ie homologous vs heterologous) were included in each experiment and the number of experiments performed to test all samples was reduced as much as technically feasible.
Data exclusions In the T cell experiments in Figure 3G-H we had to exclude several samples in both groups that were all analyzed on the same day.

Data collection
For analyses of infections (Figure1), Health Care Workers at the Lyon University hospital (HCL) who received each vaccination regimen were monitored for infections starting in January 2021 in both groups. Data in Figure 1 show the infections that occurred after the 14-days postboost period, up to the end of the recording (08/15/2021) For immunological analyses (Figures 2-3), clinical data were collected using the Clinsight software, during January-April 2021. . Blood samples were processed and stored at the Centre de Ressource Biologique Neurobiotec, 69500 Bron. Serological and immunological analyses were performed at the Lyon-Sud hospital or at the Centre International de recherche en infectiologie (CIRI) in Lyon.

Outcomes
For figure 1 (analyses of infections) the primary outcome was the infection status ie not infected vs infected. This was assessed by performing RT-PCR for SARS-CoV-2 upon contact with confirmed cases or upon symptoms onset. For figures 2 and 3, (covid-ser study), the primary outcome was the positivity of the SARS-Cov-2 serological test at different time points, and the secondary outcome was the serum level of IgM and IgG titers and the serum neutralization capacity.

Flow Cytometry
Plots Confirm that: The axis labels state the marker and fluorochrome used (e.g. CD4-FITC).
The axis scales are clearly visible. Include numbers along axes only for bottom left plot of group (a 'group' is an analysis of identical markers).
All plots are contour plots with outliers or pseudocolor plots.
A numerical value for number of cells or percentage (with statistics) is provided.

Methodology
Sample preparation For T cell stimulation and staining: Cryopreserved PBMCs were thawed and rested overnight in RPMI 1640 medium supplemented with 10% FBS in 5% CO2 at 37°C. On the second day, cytokine-producing T cells were identified by intracellular cytokine staining (ICS): Briefly, the PBMCs were stimulated with individual peptides for 1 h in the presence of 1 μg/mL monoclonal antibodies against human CD28 (Biolegend) and CD49d (Biolegend) then for an additional 5h with GolgiPlug (brefeldin A, BD Biosciences), GolgiStop (monensin, BD Biosciences). Then a standard antibody staining was carried out: dead cells were first labelled with LIVE/ DEAD™ Fixable eF780 dye (ThermoFisher) and then followed by surface antibody staining. Subsequently, Cytofix/Cytoperm kit (BD Biosciences) was used for permeabillizing the cells before staining the cells with antibodies against molecules expressed intracellularly. For SARS-CoV-2 RBD-specific B cells analysis : PBMC staining were clearly described in the methods section of the manuscript. Briefly, cryopreserved PBMCs were centrifuged and resuspended in PEB Buffer (PBS + 0.5% BSA and 2 mM EDTA) and incubated with Fc receptor block (Miltenyi 130-059) for 15 min at 4°C. Next, cells were washed in PEB and stained for 30 min in brilliant stain buffer at 4°C in the dark using surface antibodies with both the PE-and APC-conjugated recombinant RBD tetramers. Then, cells were washed in PEB, and resuspended in a PEB dilution of the fixable viability dye eFluor 780 (ebiosciences 65-0865-18). They were next washed and fixed in 4% paraformaldehyde (PFA) for 20 min at 4°C in the dark before a final wash and resuspension for flow cytometry analysis. Gating strategy Peptide-specific T cell subsets were identified via the following gating strategy: Viable lymphocytes were addressed by successive gating in SSC-A/FSC-A plot followed by excluding DUMP positive cells (dead cells, CD14, CD19, CD56) in a FSC-A/ DUMP plot. Then, singlets were gated in a FSC-A/FSC-H plot and CD3+ T cells were gated in CD3/FSC-A plot. From CD3+ T cells, CD4+ and CD8+ T cells were gated in CD4/CD8 plots. Next, peptide-specific CD4 T cells were gated by plotting CD4/IFNg and peptide-specific CD8 T cells were gated by plotting CD8/IFNg.
Gating strategy for RBD-specific B cell analysis is provided in the supplementary information file and cell population are defined. Briefly, successive gating were applied for single cells and viable cells. Then, B cells were gated as CD19+/CD3-cells.