Seventy-seven healthcare workers undergoing COVID-19 vaccination were recruited at Hospital Universitario 12 de Octubre (Madrid, Spain). Participants received the two 30 µg dose, 21 days apart, BNT162b2 vaccination schedule in early 2021, and after 10 months they received the 50-µg mRNA-1273 boost, in December 2021. Samples were longitudinally collected at 7 time points: pre-vaccine baseline, 21 days after the first BNT162b2-dose, 15 days, 1, 3 and 6 months after the second BNT162b2-dose and 30 days after mRNA-1273 third boost dose. The subjects were further stratified on the basis of SARS-CoV-2 infection prior to vaccination confirmed by either positive RT-PCR or S1 domain of the Spike glycoprotein (S1), Nucleocapsid (N) and/or Membrane (M) SARS-CoV-2-specific T cells above the positivity threshold by fluorospot (60 SARS-CoV-2-naïve; 17 SARS-CoV-2-recovered) ( Figure 1A ). Vaccination side effects after the three vaccine doses were collected using a standardized questionnaire. The Institutional Review Board approved the study (21/039 and 21/056).

Freshly isolated peripheral blood mononuclear cells (PBMCs) were seeded in duplicate at 300,000 cells/well in IFN-γ IL-2 FluoroSpot™ plates (MabTech). Cells were supplemented with 15-mer overlapping peptides covering the S1 domain of the Spike glycoprotein (SARS-CoV-2 S1 scanning pool, MabTech), the Nucleocapsid protein (Epitope Mapping Peptide Set [EMPS] SARS-CoV-2 NCAP-1, JPT), and the Membrane protein (EMPS SARS-CoV-2 VME1, JPT) at a final concentration of 1 µg/mL. Assays were incubated for 16-18 hours at 37°C. Spots were counted using automated IRIS™ FluoroSpot Reader System (MabTech). Results were expressed as IFN-γ- or IL-2-producing spot forming units (sfu) per 10⁶ PBMCs. Reponses were considered positive if the results were at least three times higher than the mean of the negative control and >25 IFN-γ sfu/10⁶ PBMCs for S1, >14 IFN-γ sfu/10⁶ PBMCs for N, and >21 IFN-γ sfu/10⁶ PBMCs for M proteins, as previously published (19, 35). Apart from the sfu, which correspond to the frequency of antigen-specific T cell clones, the relative spot volume (RSV), which represented the amount of secreted IFN-γ per spot, was also obtained.

Serum SARS-CoV-2 IgG antibodies targeting the S1 protein were detected with the Euroimmun Anti-SARS-CoV-2 ELISA (Euroimmun AG, Lübeck, Germany) according to manufacturer’s instructions. Results were evaluated semi-quantitatively by calculating the ratio of the OD value of the sample over the OD value of the calibrator (relative OD), with the following cut-off values: OD ratio <1.1: negative; and OD_ratio ≥1.1: positive.

SARS-CoV-2-pseudotyped rVSV-luc were produced as previously published (11, 36, 37). Serum samples were tested at dilutions 1:80, 240, 720, 2160, 6480. SARS-CoV-2-pseudotyped rVSV-luc was normalized for infectivity to a MOI of 0.5 and incubated with the serum samples at 37° C for 1 h. Next, Vero E6 cells were seeded onto the virus-serum mixture and incubated at 37°C for 24h. Cells were then lysed and assayed for luciferase expression. Neutralizing titer 50 (NT50) was calculated using a nonlinear regression model fit with settings for log agonist versus normalized response curve. Neutralization potency of serum samples were calibrated using the WHO International Standard 20/136, which was assigned 813 International Units per ml [IU/ml]. Calibrated NT50 in IU/ml for each serum sample was calculated as the observed NT50 titers multiplied by the calibration factor (assigned as 0.756), which is estimated as 813 IU/ml divided by NT50 (tested as 1:1075) of the 20/136 standard (11, 36). The cut-off titer for neutralization positivity was 45 IU/ml (38).

Quantitative data were shown as the median with IQR, and qualitative variables were expressed as absolute and relative frequencies. Non-parametric Mann-Whitney U or Wilcoxon signed-rank tests were applied for comparison within two groups, when necessary. Kruskal-Wallis test was used to compare three or more unmatched groups. Correlations between continuous variables were evaluated using Spearman’s rank test. Grouping of individuals was done according to mean SFU and neutralization titer. Differences were considered statistically significant when p<0.05. Statistical analysis was performed using GraphPad Prism version 8.0 software (GraphPad Software Inc, LaJolla, CA) and R software v4.1.1.

Cellular and humoral SARS-CoV-2-specific immune responses were prospectively analysed in 77 healthcare workers undergoing vaccination, comprising a total of 539 longitudinal samples studied ( Figure 1A ). The median age of the cohort was 39 years old (interquartile range [IQR] 22-64 years) and 57/77 (74%) participants were females, with no relevant comorbidities. Based on SARS-CoV-2 infection prior to vaccination, the cohort included 60 SARS-CoV-2 naïve and 17 recovered individuals. All recovered subjects had been infected by the ancestral Wuhan SARS-CoV-2 strain during the first pandemic wave in Spain. They had been either asymptomatic or suffered mild COVID-19 (WHO ordinal scale of 0-2). The median time from symptom onset to administration of the first BNT162b2-dose was 285 days ([IQR] 272-298 days). There were no differences in sex or age between naïve and recovered individuals. Vaccination side effects were more frequent among subjects with prior SARS-CoV-2 infection compared to naïve subjects ( Figure S1 ), and increased their frequency with subsequent vaccine doses (median [IQR] number of side effects after the first, second and third doses were 2 [2-3], 4 [2-6] and 5 [3-7], respectively).

We studied the development and maintenance of SARS-CoV-2-specific T cells, IgG, and neutralizing antibodies after vaccination in SARS-CoV-2 naïve and recovered individuals. Among naïve subjects, the peak S1-specific cellular response was achieved 15 days after the second vaccine dose, then it decreased steadily and finally reached a plateau at month 3, remaining stable up to month 6 post-vaccination ( Figure 1B ). All vaccinees except one remained above the positivity threshold for cellular response during this period. The administration of the third booster dose significantly increased the number of S1-specific T cells compared to 6 months post-vaccination ( Figure 1B ), although, the magnitude of this response did not reach peak levels. Regarding the humoral response, peak levels of S1-specific IgG antibodies were also detected 15 days after the second dose, and then they gradually declined over the next 6 months although remained detectable in all subjects ( Figure 1C ). The third vaccine dose boosted anti-S1 IgG to levels comparable to the peak response ( Figure 1C ). Similarly, the maximum neutralizing response against SARS-CoV-2 elicited by 2 vaccine doses was observed 15 days after the second dose and it experienced a remarkable reduction over the course of the next 6 months ( Figure 1D ). All vaccinees except one remained positive for neutralization, although the levels of neutralization 6 months after the second dose were only slightly above the 45 IU/ml positivity threshold. Among naïve individuals the third vaccine dose elicited the highest levels of neutralizing antibodies, with a 20.9-fold increase compared to the previous measurement ( Figure 1D ). Anti-S1 IgG positively correlated with the neutralizing activity of serum samples at all time points ( Figure S2 ). The development of IL-2-producing and IFN-γ+IL-2 double positive S1-specific T cells followed a dynamic similar to that of IFN-γ-producing T cells ( Figures S3A, B ), although 3 and 6 months after the second dose the number of IL-2-producing clones significantly exceeded that of IFN-γ ( Figure S3C ).

Similar to naïve individuals, in SARS-CoV-2 recovered subjects the maximum cellular response to vaccination was reached 2 weeks after the second dose, it declined up to month 3 and then remained stable up to month 6 ( Figures 1E , S3D, E ). During this period, there were more IFN-γ-producing T cells than IL-2 ( Figure S3F ). The third vaccine dose significantly enhanced the S1-specific T cell response which reached peak levels ( Figure 1E ). Similarly, anti-S1 IgG levels reached their peak 2 weeks after the second dose and then experienced a progressive decline up to month 6, although all subjects remained S1-IgG-positive ( Figure 1F ). After the third dose the anti-S1 IgG antibodies were the highest. The neutralizing activity followed a similar profile to that of total antibodies, and despite the decline over time, neutralizing antibodies remained well above the positivity threshold 6 months after vaccination ( Figure 1G ). The third vaccine administration boosted the neutralizing activity with a 2.3-fold increase, which nevertheless did not reach the maximal level observed 15 days post-second dose. As in naïve individuals, anti-S1 IgG positively correlated with the neutralizing activity of serum samples at all time points ( Figure S2 ).

mRNA vaccines induced robust circulating cellular and antibody responses specific to the S1 SARS-CoV-2 protein. However, these responses showed different dynamics depending on whether or not individuals had COVID-19 prior to vaccination. Although the peak T cell response elicited 2 weeks after the second dose was similar in both groups ( Figures 1H , S3G, H ), the maintenance of IFN-γ-producing T cell responses thereafter was significantly higher in SARS-CoV-2 recovered than in naïve individuals, both the number of specific T cells and the amount of IFN-γ released by each T cell clone ( Figures 1H , S3I ). After the third dose, the T cell response developed in subjects with prior COVID-19 was significantly higher both in the number of S1-specific clones ( Figure 1H ) and in the amount of IFN-γ secreted per clone ( Figure S3I ) compared to naïve individuals.

Humoral response paralleled T cell behaviour, since anti-S1 IgG peak levels were similar at 2 weeks post-second vaccine dose but the decline in antibody levels was significantly lower in SARS-CoV-2 recovered compared to naïve individuals over the course of the next 6 months ( Figure 1I ). The administration of the third vaccine dose elicited similar anti-S1 IgG levels in both groups ( Figure 1I ). Two-dose vaccination in SARS-CoV-2 recovered individuals induced a much higher level of neutralizing antibodies as compared with naïve individuals up to 6 months post-vaccination ( Figure 1J ). Three vaccine doses were required to develop a similar neutralizing activity in both cohorts ( Figure 1J ). The boost in neutralizing capacity triggered by the third dose was much higher in naïve than in recovered individuals (20.9- versus 2.3-fold increase).

We found no association between age and strength of the cellular or humoral immune response as measured at month 6 after 2-dose vaccination, either in SARS-CoV-2 naïve or recovered individuals ( Figures S4A–C ). Finally, significant positive correlations between the number of side effects and the magnitude of both T cell and antibody responses were exclusively found in SARS-CoV-2 naïve individuals after the second vaccine dose ( Figures S4D–I ).

Five naïve individuals had a SARS-CoV-2 infection 3 to 6 months after completing the standard 2-dose vaccination and another 12 naïve subjects suffered a breakthrough infection approximately 1 month after the boost, coinciding with the Omicron pandemic wave ( Figure 2A ). All breakthrough infections in naïve individuals ranged from asymptomatic to mild COVID-19 (WHO ordinal scale of 0-2). In contrast, no breakthrough infection was recorded in the COVID-19 recovered individuals, neither after the second or the third dose ( Figure 2B ). As expected, none of the individuals who remained naïve developed M- or N-specific T cells during follow-up ( Figure 2C ), while the absence of breakthrough infections in recovered individuals could be confirmed by the lack of any boost in the specific T cell response against M or N proteins ( Figure 2D ). Of note, all SARS-CoV-2 recovered individuals maintained cellular response against M and/or N above the positivity threshold during follow-up ( Figure 2D ), which meant recovered subjects exhibited detectable virus-specific memory T cells up to 22 months after infection, coinciding with the post-boost sample. Recovered subjects were protected against breakthrough infection compared to naïve (0/17 versus 17/60, Fisher’s test, p=0.017). Demographic and immunological characteristics of subjects infected after vaccination are shown in Table 1 . None of the analysed immune parameters were consistently low in infected compared to uninfected individuals. Interestingly, a cluster analysis in naïve individuals based on the mean peak cellular and neutralizing response after vaccination showed that subjects who developed both high T cell (>700 S1 IFN-γ sfu/10⁶ PBMCs) and neutralizing (>1/1206 IU/ml) response, were protected against breakthrough infection ( Figure 2E ). These analyses suggest that individuals with a lower-than-average cellular or neutralizing response may benefit from the administration of a third booster dose.

Finally, regarding the effect of breakthrough infections on immunity, SARS-CoV-2 infections after the second or the third vaccine dose increased the S1-specific T cell response compared to individuals who remained naïve throughout the study ( Figures 3A–C , S5A–D ), and, as expected, also induced the development of IFN-γ T cells against SARS-CoV-2 M and N proteins ( Figures S5E, F ). Both total ( Figures 3D–F ) and neutralizing ( Figures 3G–I ) antibodies were also boosted after exposure to the virus.