Abstract
Advanced age is a main risk factor for severe COVID-19 and thus elderly were often prioritized for vaccination. However, low vaccination efficacy and accelerated waning immunity have been reported in this age group. To elucidate age-related differences in immunogenicity, we analysed cellular, serological and salivary SARS-CoV-2 spike glycoprotein-specific immune responses to BNT162b2 COVID-19 vaccine in old (69-92 years) and middle-aged (24-57 years) vaccinees compared to natural infection (COVID-19 convalescents of 21-55 years). Serological humoral responses to vaccination exceeded those of convalescents but salivary anti-spike subunit 1 (S1) IgA and neutralizing capacity were less durable in vaccinees. In old vaccinees, we observed that pre-existing spike-specific CD4+ T cells correlated with efficient induction of serological anti-S1 IgG and neutralizing capacity after vaccination. Our results highlight the role of pre-existing cross-reactive CD4+ T cells with respect to SARS-CoV-2 vaccination particularly in old individuals, in whom their presence predicted efficient COVID-19-vaccine-induced humoral immune responses.
Main
Global efforts have been mounted to develop efficient vaccines against coronavirus disease 2019 (COVID-19) caused by the pandemic severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)1. As severe COVID-19 mainly affects older individuals, many vaccination campaigns have prioritized the elderly population2. However, vaccination efficacy is known to be decreased in this age group as compared to younger individuals, particularly for primary vaccination3,4,5. For COVID-19 vaccination - given the distinct homology of certain antigen target regions of SARS-CoV-2 to common cold coronaviruses (HCoV) - one possible explanation could be an age-related reduced number of pre-existing cross-reactive CD4+ T cells in old individuals6,7. To assess the immunogenicity of the COVID-19 vaccine in this particularly vulnerable age group and identify possible relations to pre-existing SARS-CoV-2-specific cross-reactivities, we examined systemic cellular, serological and salivary humoral SARS-CoV-2-specific immunity before and during the course of COVID-19 vaccination with BNT162b2 mRNA vaccine (Tozinameran™, Comirnaty™8,9) in old and comorbid nursing home residents (n=18; mean age 83±6) and their middle-aged caregivers (n=14; mean age 47±10) at baseline (prior to first vaccination), at day 28 (d28, 7 days after booster vaccination) and at day 49 (d49, 28 days after booster vaccination). For comparison with naturally acquired immunity, we additionally analysed COVID-19 convalescents (age-matched to the middle-aged cohort; mean age 36±11) after mild natural SARS-CoV-2 infection at ∼d28 (n=10), ∼d49 (n=16) or ∼d94 (n=11) after symptom onset (Table 1).
We first analysed anti-SARS-CoV-2 spike glycoprotein subunit 1 (S1) immunoglobulin G (IgG) and anti-S1 immunoglobulin A (IgA) antibody levels in serum (Fig. 1a and b). At d28, eight old donors (44%) failed to seroconvert for anti-S1 IgA and five of them (28%) also for anti-S1 IgG. In contrast, all middle-aged donors exhibited high anti-S1 IgG and anti-S1 IgA levels at d28 indicating a fast and robust systemic humoral response to vaccination. Anti-S1 IgG and anti-S1 IgA levels were more heterogeneous and significantly lower in the old compared to the middle-aged cohort at d28 (Fig. 1a and b). At d49, anti-S1 IgG levels were still significantly lower in the old vaccinees (Fig. 1a), but no significant differences were present in anti-S1 IgA at d49 comparing old and middle-aged vaccinees (Fig. 1b). Assessing S1-specific functional neutralization capacity in serum using a surrogate viral neutralization assay (sVNT), all middle-aged vaccinees demonstrated a prompt and uniform increase of virus neutralization at d28, which remained stable at d49 (Fig. 1c). In the old vaccinees, this response was significantly delayed and remained lower than in middle-aged vaccinees at d49 with one non-responder (inhibition < 30%). Compared to COVID-19 convalescents, anti-S1 IgG levels and median S1 neutralizing capacity in serum were more homogenously distributed and, moreover, significantly higher in age-matched middle-aged and in old vaccinees at both time points (Fig. 1a and c).
To evaluate salivary humoral responses as potential correlate of first-line adaptive protection against SARS-CoV-2 infection, we furthermore analysed the presence of anti-S1 secretory IgA (sIgA) in the saliva. Following vaccination, we observed increased levels of anti-S1 sIgA in both the middle-aged and old cohort (Fig. 1d). At d28, anti-S1 sIgA levels were significantly higher in the middle-aged donors as compared to the old individuals. All 14 middle-aged donors showed anti-S1 sIgA levels above their age group’s maximum pre-vaccination level (0.45 optical density (OD) ratio) while only six of the old vaccinees (60%) reached above their age group’s maximum pre-vaccination level (0.31 OD ratio). At d49, we observed anti-S1 sIgA levels comparable to pre-vaccination in all vaccinees (Fig. 1d). By contrast, COVID-19 convalescents exhibited significantly higher anti-S1 sIgA levels at ∼d49 and even ∼d94 compared to age-matched middle-aged vaccinees at d49 (Fig. 1d). S1 neutralizing activity in saliva developed more heterogeneously after vaccination (Fig. 1e). An increase in S1 neutralizing activity was detected in most vaccinees at d28 but dropped back to pre-vaccination levels at d49 mirroring anti-S1 sIgA kinetics (Fig. 1d and e). In some responsive old donors, a delayed anti-S1 sIgA response was reflected by a similarly delayed increase of salivary S1 neutralizing activity, which caused some old donors to show increased anti-S1 sIgA and S1 neutralizing activity at d49 (Fig. 1d und e). COVID-19 convalescents demonstrated higher S1 neutralizing capacity in the saliva around ∼d49 compared to age-matched vaccinees at d49. In contrast to anti-S1 sIgA, S1 salivary neutralizing capacity did not remain significantly increased in convalescents at ∼d94 compared to d49 in age-matched middle-aged vaccinees indicating that anti-S1 sIgA does not correspond to S1 neutralizing activity following infection at this time point (Fig. 1d und e).
Finally, we analysed frequencies of antigen-reactive CD40L+ 4-1BB+ CD4+ T cells after in vitro stimulation with the N-terminal part (S1, covered by peptide mix S-I) and the C-terminal part (S2, peptide mix S-II) of the spike glycoprotein (Fig. 2a and b, Suppl. Fig. 2). We have recently demonstrated that, in contrast to the N-terminal part, the C-terminal part of the spike glycoprotein contains highly conserved domains and triggers CD4+ T cell cross-reactivity to SARS-CoV-210. One week after the second vaccination (d28), all middle-aged and 89% (S-I; n=16/18) and 94% (S-II; n=17/18) respectively of the old donors acquired T cell reactivity to S-I and S-II. However, S-I- and S-II-reactive T cell frequencies increased more quickly and peaked at a higher level in the middle-aged than in the old cohort. At d49, S-I- and S-II-reactive CD4+ T cell frequencies of the middle-aged vaccinees dropped slightly so that both age groups reached comparable levels. In convalescents, we observed a significantly lower T cell reactivity to S-I and S-II at ∼d49 after symptom onset compared to both vaccinated cohorts (Fig. 2a and b). Pre-existing T cell reactivity to S-II at baseline was significantly lower in old individuals than in the middle-aged cohort (Fig. 2b and c). Remarkably though, in these old individuals, high levels of S-II-specific, but not S-I-specific, CD4+ T cells at baseline correlated with higher anti-S1 IgG and consistently with elevated S1 neutralizing capacity in serum at d28 (Fig. 2d and e, Suppl. Fig. 1). These correlations did not exist for anti-S1 sIgA or salivary S1 neutralizing capacity (Fig. 2f and g). Hence, in old vaccinees, increased pre-existing frequencies of S-II-specific CD4+ T cells predicted the efficiency of neutralizing and anti-S1 IgG humoral vaccination responses.
In this study, we investigated systemic and mucosal immune responses to the COVID-19 mRNA vaccine BNT162b2 in middle-aged and old vaccinees compared to COVID-19 convalescents (age-matched to middle-aged vaccinees). While the middle-aged cohort demonstrated a remarkably homogeneous and prompt induction of both cellular and humoral immune responses, a significantly decreased anti-S1 IgG response was observed in the old vaccinees after two doses of BNT162b2, which was intra-individually associated with an equally reduced anti-S1 IgA response (Suppl. Fig. 3). However, compared to COVID-19 convalescents with a mild disease course, COVID-19 vaccination induced more pronounced systemic immunity with significantly higher levels of spike-reactive CD4+ T cells, S1-specific antibodies and serum S1 neutralizing capacity regardless of age.
Reduced humoral and cellular vaccination responses in old individuals have been described for vaccines against influenza, yellow fever, pneumococcal disease, and tetanus as well as for COVID-1912-15. In the light of the increased risk of the elderly for severe COVID-19 and the current discussions on their need of booster vaccinations, it is essential to identify and evaluate possible predictors of low vaccination efficiency particularly in this age group. Several studies have supported the notion of beneficial effects of pre-exposure SARS-CoV-2 cross-reactivity6, 16, 17. In the elderly, the frequency of these pre-existing S-II-specific CD4+ T cells was reduced (Fig. 2b and c). However, whether this phenomenon has a direct effect on BNT162b2 immunogenicity in the elderly was unclear. Our findings here highlight the relevance of pre-existing CD4+ T cell cross-reactivity for an efficient immune response to COVID-19 vaccination particularly in this age group. Possible explanations for this age-related reduction of cross-reactivity could be cellular senescence or impaired (oronasal) mucosal immunity11, 18-20.
We additionally investigated anti-S1 sIgA in the saliva as a potential correlate of local mucosal protection from SARS-CoV-2 infection following COVID-19 vaccination as compared to natural infection. We assume that the presence of S1-specific antibodies and S1 neutralizing capacity in the saliva may contribute to protection against SARS-CoV-2 infection and reduce local replication21, 22. However, within four weeks (d49) after booster vaccination, anti-S1 sIgA levels and S1 neutralizing activity in the saliva returned to low levels. By contrast, COVID-19 convalescents exhibited significantly higher salivary anti-S1 sIgA levels and higher salivary S1 neutralizing activity at ∼d49 after symptom onset compared to age-matched vaccinees. We found a correlation between anti-S1 sIgA levels and salivary S1 neutralization in convalescents at ∼d28 and at ∼d49, which was not observed in vaccinees (Suppl. Fig. 4). This indicates that neutralizing capacity in the saliva following vaccination may not only rely on anti-S1 sIgA but possibly anti-S1 IgG, which is consistent with reports on detectable anti-S1 IgG in the saliva of vaccinated individuals23, 24. In COVID-19 convalescents, anti-S1 sIgA secretion in salivary glands (and salivary S1 neutralizing activity) is likely induced by locally primed B and T cells in nasopharyngeal lymph nodes and/or tonsils25. In contrast, the more transient presence of anti-S1 sIgA in the saliva of vaccinated individuals could be the result of transfusion of serum-derived anti-S1 IgA through the endothelium into the oral mucosa26. An alternative explanation would be mucosal homing of peripheral anti-S1 IgA plasmablasts although intra-muscular injection does not reliably induce mucosal homing27.
Taken together, the presence of anti-S1 sIgA and S1 neutralizing capacity in the saliva after vaccination is of shorter duration and lower magnitude than after natural infection, which points out the need to determine the role of mucosal immunity, e.g., in the form of sIgA in saliva, for evaluation of SARS-CoV-2 immunity and its transmission. Currently, vaccines for intranasal application are in development, which may fill the gap in mucosal immunity observed here28. Our findings demonstrate that BNT162b2 induces strong immune responses in middle-aged as well as most old and comorbid individuals. However, for some old individuals, the serological response to vaccination is hampered and may leave these individuals at higher risk of infection and severe disease courses, thus promoting recommendations for regular immune status check-ups and further vaccination boosts. Importantly, we show here that pre-existing SARS-CoV-2 spike glycoprotein cross-reactive memory T cells predict vaccination efficiency in the elderly and may generally contribute to the high responsiveness to COVID-19 vaccines.
Methods
Participants and ethics
The study was approved by the ethics committee of Charité – Universitätsmedizin Berlin (EA/152/20) and was conducted in accordance with the Declaration of Helsinki. A written informed consent was obtained from all participants. The 39 participants (22 nursing home residents (old vaccinees), 17 caregivers (middle-aged vaccinees)) analysed for this study were recruited at three different nursing homes in Berlin between September and November 2020 and were available for follow-up visits 28 days and 49 days after their first COVID-19 vaccination in January and February 2021. Furthermore, we collected saliva and blood samples of a total of 36 COVID-19 convalescents with mild disease course (WHO II) at ∼28 (n=10), ∼49 (n=16) or ∼94 days (n=11) post symptom onset. Baseline data of vaccinees and data of convalescents had been collected and partially analysed as part of the Charité Corona Cross (CCC) study for a project investigating T cell cross-reactivity to SARS-CoV-26. Visits included nasopharyngeal swabs, blood and saliva sampling at all time points. 4 older and 3 middle-aged donors with signs of previous SARS-CoV-2 infection (either positive anti-S1 IgG levels or a S-I T cell stimulation index > 3.0 at baseline) were excluded from analysis.
SARS-CoV-2 RT-qPCR
Nasopharyngeal swabs for RT-qPCR were taken from all participants at all three time points. Vaccinees who tested positive by RT-PCR in the study period (n=2) were excluded from all analyses. Swab material was suspended in 4.3 ml Cobas PCR Media. RNA was extracted using the MagNA Pure 96 system (Roche). The viral RNA extraction was performed using 200 μl swab dilution eluted in 100 μl of extraction buffer. SARS-CoV-2 detection was based on two genomic targets (E- and N gene, TIB Molbiol) using 5 μl of the RNA eluate. Quantification of SARS-CoV-2 copy numbers was achieved using calibration curves with serial diluted photometrically quantified in-vitro transcribed RNA as described before20,21. RT-PCR was performed using the LightCycler 480 II (Roche).
Blood sampling, serum preparation and PBMC isolation
Whole blood was collected in lithium heparin tubes for peripheral blood mononuclear cell (PBMC) isolation and SST™II advance (all Vacutainer®, BD) tubes for serology. SST™II advance tubes were centrifuged at 1000 x g, 10 min and serum supernatant aliquots frozen at −80 °C until further use. PBMCs were isolated by gradient density centrifugation according to the manufacturer’s instructions (Leucosep tubes, Greiner; Biocoll, Bio&SELL).
Ex vivo T cell stimulation
Freshly isolated PBMC were cultivated at 5×106 PBMC in AB-medium containing RPMI 1640 medium (Gibco) supplemented with 10% heat inactivated AB serum (Pan Biotech), 100 U/ml penicillin (Biochrom), 0.1 mg/ml streptomycin (Biochrom). Stimulations were conducted with 11aa overlapping 15-mer PepMix™ SARS-CoV-2 spike glycoprotein peptide pool 1 or 2 (termed here S-I and S-II; JPT) at concentrations of 1 μg/ml per peptide respectively. Stimulation controls were performed with equal concentrations of DMSO in PBS (unstimulated control) and 1 μg/ml per peptide of CEFX Ultra SuperStim pool (positive control; JPT). All approaches contained 1 μg/ml purified anti-CD28 (clone CD28.2, BD Biosciences). Incubation was performed at 37 °C, 5% CO2 for 16 h in the presence of 10 μg/ml brefeldin A (Sigma-Aldrich) during the last 14 h.
Flow Cytometry
Stimulations were stopped by incubation in 20 mM EDTA for 5 min. Surface staining was performed for 15 min in the presence of 1 mg/ml beriglobin (CSL Behring) with the following fluorochrome-conjugated antibodies titrated to their optimal concentrations: CD3-FITC (REA613, Miltenyi), CD4-VioGreen (REA623, Miltenyi), CD8-VioBlue (REA734, Miltenyi), CD38-APC (REA671, Miltenyi), HLA-DR-PerCpVio700 (REA805, Miltenyi). During the last 10 min of incubation, Zombie Yellow fixable viability staining (Biolegend) was added. Fixation and permeabilization were performed with eBioscience™ FoxP3 fixation and PermBuffer (Invitrogen) according to the manufacturer’s protocol. Intracellular staining was carried out for 30 min in the dark at room temperature with 4-1BB-PE (REA765, Miltenyi) and CD40L-PE-Vio770 (REA238, Miltenyi). All samples were measured on a MACSQuant® Analyzer 16 (Miltenyi) according to the gating strategy illustrated in Suppl. Fig. 2. Instrument performance was monitored prior to every measurement with Rainbow Calibration Particles (BD).
Anti-SARS-CoV-2 S1 enzyme-linked immunosorbent assay (ELISA) in serum and saliva
Humoral responses to vaccination were determined by anti-SARS-CoV-2 S1 IgG and IgA testing. Anti-SARS-CoV-2 IgG and IgA tests in sera were performed using a commercially available ELISA kit (Euroimmun) as previously described22. Test results for sera were considered positive above an optical density (OD) ratio of 1.1. The same ELISA kit was used for anti-SARS-CoV-2 IgA tests in saliva. The assay was performed with 1:100 diluted serum and 1:10 diluted saliva samples. Values were capped at an OD ratio of 10. No positivity thresholds have yet been validated for saliva.
Surrogate virus neutralization assay (sVNT)
A competition ELISA-based surrogate virus neutralization assay (sVNT; medac) mimicking the SARS-CoV-2 receptor (ACE2) binding process was used to identify neutralizing anti-SARS-CoV-2 antibodies in participant serum and saliva23,24. The assay was performed with 1:10 diluted serum and 1:5 diluted saliva samples respectively following the manufacturer’s instructions. Inhibition activity above 30 % was considered positive in serum.
Data processing and statistical analysis
Study data were collected and managed using REDCap electronic data capture tools hosted at Charité25,26. Flow cytometry data were analysed using FlowJo 10 (BD). Prism 9 (GraphPad) was used for data plotting. For statistical group comparisons between antibody OD ratios and T cell frequencies, non-parametric testing (Mann-Whitney U-test) and ROUT outlier tests were performed. Statistical significance was reported as follows: *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. Correlation coefficients were reported as r. CD4+ T cell activation was plotted as stimulation index (StimIndex), i.e., ratio of % of CD40L+4-1BB+ CD4+ T cells in stimulated samples and % of CD40L+4-1BB+ CD4+ T cells in unstimulated controls (zero background values were set to a minimum of 0.001 before division). Lines in figures displaying T cell stimulation results indicate stimulation indices of 1.5 (above background; dashed line) and 3.0 (positivity threshold; dotted line)6.
Data Availability
The datasets generated during the current study are available from the corresponding author on reasonable request.
Author Contributions
Conceptualization: LMA, JB, VMC, AT. Data Analysis: LMA, TS, JB. Funding acquisition: CD, VMC, AT. Resources: LMA, TS, LL, LH, BK, MD, KG, ZUA, MAM, LES, ID, RL, JB. Writing: LMA, CGT, JB, VMC, AT.
Competing Interests
VMC and MAM are named together with Euroimmun on a patent application filed recently regarding detection of antibodies against SARS-CoV-2.
Data and materials availability
The datasets generated during the current study are available from the corresponding author on reasonable request.
Supplementary Figures and Figure Legends
Acknowledgment and Funding
We thank the CCC Study Group E. Baysal, T. Panne, F. Legler, M. Girod, B. Bohnen, T. Nguyen, J. Schmitz, R. Klemencic, L. Hintze, N. Avinc, P. Resch, A. Maraj, A. Farghaly, P. Schulz, K. Du, N. Matuschewski, K. Gutwenger, I. Katsianas, Z. Cheng, A. Braginets for their contributions to donor recruitment, sample processing and measurement. Furthermore, we thank T. Bleicker, J. Tesch, P. Tscheak, M. L. Schmidt, P. Mackeldanz and J. Riege for their support in conducting ELISA and PCR analyses. We also thank Andreas Hetey for support with maintaining the REDCap database and Nicola Brindle for critical reading of the manuscript. This work was funded by the German Federal Ministery of Health to A.T. (Charité Corona Cross (CCC) and Charité Corona Cross 2.0 (CCC 2.0)) and to C.D. and V.M.C (Konsiliarlabor für Coronaviren) by the German Ministry of Research through VARIPath (01KI2021) to V.M.C, by the German Research Foundation through KFO339 to J.B. and in part by the German Federal Ministry of Education and Research (Bundesministerium für Bildung und Forschung, BMBF) to NaFoUniMedCovid19 – COVIM, FKZ: 01KX2021.