Characteristics of study groups
To assess humoral and cellular immune responses in the lung mucosa and blood following SARS-CoV-2 vaccination and hybrid immunity, we collected bronchoalveolar lavage (BAL) fluid and paired blood samples from 7 vaccinees with no history or evidence of SARS-CoV-2 infection (naïve, vaccinated group) and 15 vaccinees who had serologically-confirmed asymptomatic infection or experienced symptomatic infection between 2 to 11 months (56 – 333 days) prior to receiving SARS-CoV-2 vaccination. Vaccinated individuals with asymptomatic and symptomatic infection were combined in one ‘hybrid immunity’ group, due to the lack of obvious difference in immune responses to SARS-CoV-2 antigens. All vaccinated individuals received two doses of either mRNA or adenoviral vector vaccine. We also included a pre-pandemic group (n=11) of unexposed, unvaccinated individuals as controls (Figure 1A). Table 1 summarizes the demographic characteristics of the three study groups and the time of sample collection in relation to infection or last vaccination.
Airway antibody responses following hybrid immunity or vaccination alone
Levels of circulating and mucosal antibodies against spike (S), receptor binding domain (RBD) and nucleocapsid (N) protein were measured in serum and BAL samples in all three study groups. Antibody responses to N protein (non-vaccine protein) were used to confirm absence of past infection in the vaccinated cohort and classify the groups. Limit of sensitivity (LOS) was set as median + 2 x standard deviation (SD) of the results in unexposed (pre-pandemic) donors. As expected, anti-N IgG was below or near the LOS in the naïve vaccinated group, whereas in the infected vaccinated group, it was detected in all individuals (15/15) in serum and in 47% (7/15) in the BAL fluid (Extended Data Figure 1A and B).
SARS-CoV-2 vaccination elicited robust systemic IgG responses to both S and RBD protein, with levels being more pronounced in the infected vaccinated group (3.4-fold and 3.3-fold median increase of anti-S and anti-RBD IgG compared to naïve vaccinated group, respectively) (p=0.039) and p=0.11, respectively) (Figure 1B and C). Such systemic antibody differences as a result of hybrid immunity have been extensively demonstrated in large cohort vaccination studies 32, 33. High levels of anti-S and -RBD IgG were also detected in the BAL fluid of SARS-CoV-2 vaccinees. Importantly, anti-S and anti-RBD IgG levels in the lung were also significantly elevated in the hybrid immunity group compared to the naïve vaccinated group (8.2-fold and 9.4-fold increase for S and RBD, respectively) (p=0.024 and p=0.014, respectively) (Figure 1D and E).
As IgA plays a crucial role in the antiviral immune defence in mucosal surfaces 34, 35, IgA responses against SARS-CoV-2 proteins were also assessed in BAL samples. In the naïve vaccinated group, mucosal IgA levels against S, RBD and N did not differ from the control group. However, the infected vaccinated group had significantly greater mucosal anti-S IgA (2.5-fold increase from control, p=0.014) and a trend to higher anti-RBD IgA (Figure 1F and G), whereas the majority had non- detectable anti-N IgA levels in BAL (Extended Data Figure 1C).
Vaccine-induced antibody responses to S protein demonstrated a strong correlation between serum and BAL for IgG and slightly less so for IgA (Extended Data figure 1D and E).
Presence of SARS-CoV-2 specific memory B cells in the lung following hybrid immunity
Memory B cells are critical for long-term humoral immunity. To identify SARS-CoV-2 specific memory B cells (MBCs), fluorescently labelled S, RBD and N proteins were used to assay PBMCs and lung leukocytes (Figure 2A) (see gating strategy, Extended Data Figure 2A). As expected, and in line with the antibody responses, only vacinees who had previously been exposed to viral nucleoprotein through infection had detectable N-specific MBCs in the blood. By contrast, both naïve vaccinated and infected vaccinated individuals had circulating S- and RBD-specific MBCs above the background staining threshold (set as median + 2 x SD of pre-pandemic levels).
B cells are an underrepresented cell population in the lung mucosa; their presence in the lung is usually associated with infection or chronic inflammation36. Although data on anti-viral B cell immunity in human respiratory mucosa are scarce, murine model studies of influenza infection demonstrated the generation of flu-specific memory B cells in the lung following influenza infection that were able to produce more antibodies with enhanced potential to recognise viral variants4, 5, 6. In this study, the small number of B cells in the BAL samples allowed the assessment of SARS-CoV-2 specific MBCs only in the hybrid immunity group. The frequencies of S-, RBD- and N-specific MBCs were greater in the lung mucosa of infected vaccinated individuals compared to pre-pandemic controls (median 5.5% vs 0.08% for S, 2.88% vs 0.06% for RBD and 1.69% vs 0.06% for N-specific responses (p=0.0004, p=0.016 and p=0.016, respectively) (Figure 2E-2G). Paired sample comparison of the frequencies of circulating and mucosally detected anti-viral MBCs in the infected vaccinated group revealed enrichment of S- and RBD-specific MBCs in the lung mucosa. The median frequencies of S- and RBD-specific MBCs were 2.1-fold (p=0.0078) and 3.8-fold (p=0.062) higher in the BAL compared to PBMC sample from the same donors (Figure 2H). We detected that in the lung mucosa memory B cells were mainly class-switched MBCs, whereas paired blood samples had a significantly increased proportion of unswitched MBCs cells (Figure 2I).
We also stratified the infected vaccinated group based on the vaccine they received. Despite the small sample size, mRNA vaccinated individuals exhibited 1.8-fold higher frequency of circulating S-specific MBCs compared to ChAdOX1_S recipients (p=0.037), and a trend to higher RBD-specific B cells (Extended Data Figure 2B). In lung mucosa, a similar trend was observed for S-specific MBCs levels (Extended Data Figure 2C), but low cell yields hindered a robust analysis.
Robust T cells responses in the lung mucosa after infection and vaccination but not vaccination alone
Circulating and tissue resident memory (TRM) T cells are important in constraining viral spread and protect against severe disease when neutralising antibodies fail to confer sterilising immunity37, 38, 39. Moreover, we showed that T cells targeting the early expressed replication transcription complex (RTC: NSP7,12,13) are selectively associated with infection being aborted before detection by PCR or seroconversion and can be detected in pre-pandemic blood and lung samples14, 40. Therefore, we examined T cell responses in blood and BAL samples following vaccination alone or infection and vaccination in blood and paired BAL samples.
The frequencies of circulating and lower airway CD4+ and CD8+ T cells were measured based on the expression of activation-induced markers (AIM assay) after stimulation with SARS-CoV-2 peptides (for full gating strategy see Extended Data figure 3A) and were compared to pre-existing cross-reactive responses detectable using the same assays in cryopreserved pre-pandemic BAL samples. BAL samples were further divided by the expression of prototypic tissue residency markers (CD69/CD103 co-expression for CD8 and CD69/CD49a expression for CD4/CD49a into TRM and recirculating T cells. As reported by others41, 42, 43, SARS-CoV-2 vaccination alone induced notable S-specific CD4+ and CD8+ T cell responses in the circulation when compared to pre-pandemic controls (Figure 3B and 3C). In the infected vaccinated group, the frequency of circulating S-specific CD4+ and CD8+ T cells tended to be higher than the naïve vaccinated group (1.8-fold and 4.8-fold increase, respectively). Despite the induction of T cell immunity systemically, vaccination alone did not elicit S-specific T cell responses that were significantly greater than those in pre-pandemic samples within the global (Figure 3D and 3E) or TRM lung compartment (Figure 3F and 3G).
As opposed to vaccination alone, BAL samples from those who acquired hybrid immunity exhibited greater anti-Spike T cell responses than either the pre-pandemic or naïve vaccinated group (Figure 3D-3G). Within the global T cell population, the frequency of S-specific CD4+ and CD8+ T cells increased by 2.8-fold and 5.3-fold higher in the infected, vaccinated group compared to the naïve, vaccinated group (p=0.048 and p=0.012, respectively) (Figure 3D and 3E). A similar profile was observed in the TRM T cell compartment, with S-specific CD4+ and CD8+ T cell frequencies being 2.8-fold and 4.8-fold greater, respectively, in the hybrid immunity group compared to naïve vaccinated group (p=0.05 and p=0.017, respectively) (Figure 3F and 3G). In addition, within the global T cell population, the frequencies of S-specific CD4+ and CD8+ T cells were substantially higher in BAL than in paired blood from infected vaccinated individuals (median 2.45% vs 0.62% of S-specific CD4+ T cells and median 1.84% vs 0.24% of S-specific CD8+ T in BAL and paired blood of infected, vaccinated individual, respectively) (p=0.0005 and p=0.0002, respectively) (Figure 3H and 3I). In the naïve, vaccinated group the frequency of S-specific CD4+ but not CD8+ T cells was slightly higher in BAL than PBMCs (median 0.89% vs 0.35% in BAL and paired blood, respectively).
In agreement with large vaccination studies44, 45, we observed that adenoviral vector vaccine tended to induce increased frequency of S-specific T cells in the periphery compared to mRNA vaccines. The tendency of the adenoviral vector vaccine to induce stronger T cell immunity was also observed in the lower airways, with 3.3-fold and 2.5-fold higher S-specific CD4+ and CD8+ T cell responses when compared to RNA vaccination (Extended Data Figure 3B-C).
We also examined T cell specificities for non-vaccine included SARS-CoV-2 structural proteins (N and membrane [M]) and non-structural proteins (NSP-7, NSP-12 and NSP-13 pool, representative of the core replication-transcription complex [RTC]) in blood and BAL (Extended Data Figure 3D-E and Figure 4). As expected, the frequencies of circulating N- and M-specific CD4+ and CD8+ T cells were significantly higher than in pre-pandemics only in the hybrid immunity group, as those vaccinees had a past SARS-CoV-2 infection (Figure 4A and 4B). In the case of RTC-specific T cells, their frequency did not differ amongst groups, as SARS-CoV-2 cross-reactive CD4+ and CD8+ T cell responses were detected systemically in 3 out 8 pre-pandemic controls, in line with previous studies14, 40, 46. In BAL samples, the frequency of the aforementioned T cell specificities was tested in a subset of pre-pandemic and infected, vaccinated individuals based on cell number availability. Interestingly, the hybrid immunity group had, or tended to have, higher N- and M- and RCT-specific T cell responses within the global and TRM T cell compartment in BAL samples compared to levels detected in pre-pandemic controls (Figure 4C-4F). In addition, these SARS-CoV-2 specific CD4+ and CD8+ T cell responses were enriched in the lower airways compared to the periphery (Figure 4G and 4H).
The hierarchy of SARS-CoV-2 antigen recognition by circulating and lower airway T cells of each distinct peptide pool (S, N, M and RTC) was analysed in a subset of 8 infected vaccinated individuals (Extended Data Figure 3D-E). The antigen recognition profile differed between systemic and airway localised T cells, and between T cell subsets. SARS-CoV-2 specific CD4+ T cells were largely dominated by S-specific CD4+ T cells in the periphery and lung mucosa, however in lower airway they were enriched with additional T cell specificities (Extended Data figure 3D). In the case of SARS-CoV-2 CD8+ T cells, their antigen recognition profile was more diverse in both sites, with S-specific CD8+ T cells being apparent but not dominant (Extended Data figure 3E).
Longevity of antibody and T cells responses in lung mucosa following vaccination alone or hybrid immunity
To assess the longevity of vaccine-induced SARS-CoV-2 immune memory in the lung mucosa following vaccination and hybrid immunity, antibody and T cell responses assessed in BAL and paired blood of naïve, vaccinated and infected vaccinated individuals were plotted in association with time post the 2nd vaccine dose (which was 2-11 months after any known infection dates). Levels of circulating anti-S and anti-RBD IgG were negatively correlated with time post-vaccination in the naïve vaccinated but not the infected vaccinated group, implying quicker antibody decay in the naïve vaccinated donors (Figure 5A). In the lung mucosa, anti-S and anti-RBD IgG levels exhibited similar rates of decay between the two vaccinated groups (Figure 5B). On the other hand, levels of anti-S and RBD IgA in BAL were detectable only following hybrid immunity but they quickly reached pre-pandemic levels (at 5-months post-vaccination, Figure 4C). This result is in agreement with previous studies in convalescent patients that reported short-lived IgA-mediated immunity at mucosal sites21, 47.
Circulating and lower-airway S-specific T cell frequencies were also plotted in association with time post-vaccination in both vaccinated groups. In blood, numbers of S-specific CD4+ and CD8+ T cells declined over time in the infected vaccinated group and at 5-months post-vaccination reached the frequencies induced by vaccination alone (Figure 6A). On the other hand, S-specific T cell responses were better sustained in the lung mucosa following hybrid immunity. Despite a trend of negative association with time post-vaccination, S-specific CD4+ T cell responses were detectable from the lung mucosa of infected vaccinated individuals for over 5-months post vaccination. Lower-airway S-specific CD8+ T cells did not associate negatively with time, suggesting they remained at stable levels throughout the period of 5-months post vaccination (Figure 6B and 6C). The human lung also retained partial immune memory to SARS-CoV-2 over a year post infection. Despite decline over time, T cell specificities not affected by SARS-CoV-2 vaccination, such as M-, N- and RTC-specific CD4+ and CD8+ T cells, were detectable in various frequencies between 6 to 18 months post infection (Figure 6D), rendering these conserved SARS-CoV-2 proteins as potential vaccine targets.