The ongoing COMMUNITY cohort study investigates immune responses to SARS-CoV-2 infection and vaccination. 2149 HCW were enrolled at Danderyd Hospital, Stockholm, Sweden, in April 2020. Study participants have since enrollment been followed-up every four months. Serum, saliva and nasal secretion samples are collected at each follow-up. Vaccination data (type and time) is obtained from the Swedish vaccination register (VAL Vaccinera). SARS-CoV-2 infection is defined as either seroconversion to the SARS-CoV-2 spike antigen at any of the follow-ups prior to vaccination, seroconversion to the SARS-CoV-2 nucleocapsid antigen, a record of a positive qPCR test in the national communicable diseases register or a positive Rapid Diagnostic Test reported at any of the follow-ups within the study.

For this sub study, serum, saliva and nasal secretion samples were obtained from 72 HCW in November 2022 ( Table 1 ; Figure 1 ), and from 976 HCW in February 2023 ( Table 2 ; Figure 1 ). Participants were stratified according to prior SARS-CoV-2 infection. Written informed consent was obtained from all study participants and the study protocol was approved by the Swedish Ethical Review Authority (dnr 2020-01653).

The collection and processing of saliva samples followed standardized protocols and the procedure started with the methods that least interfered with saliva production. First, participants were instructed to passively drool in a clean cup for five minutes. Saliva was transferred using a pipette into a 2mL tube. A buccal swab (the soft end of a FLOQSwab, cat nr. CP501CS01, COPAN Diagnostics Inc., United States) was then immersed in saliva briefly accumulated on the tongue, with no/minimal brushing of the oral mucosa. The swab was transferred into a 15 mL Falcon tube with 0.5 mL PBS. The swab was gently pressed against the tube wall to extract saliva and tubes were vortexed for 10 seconds. Next, participants were instructed to place a Salivette^® absorbent roll (cat nr: 51.1534, Sarstedt AG & Co. KG, Germany) between the cheek and gum and hold it still for one minute. The roll was then returned to the Salivette^® tube, which was centrifuged for 5 minutes at 1000g, leaving saliva at the bottom of the tube. A new absorbent roll was then kept in the mouth for three minutes while saliva production was stimulated by chewing on the cotton roll. The roll was then returned to the Salivette^® tube, which was centrifuged for 5 minutes at 1000g, leaving saliva at the bottom of the tube. To investigate the effect of centrifugation on passive drool samples and buccal swab samples, the samples were first kept at 4°C over the day, and then centrifuged at 400g for one minute at the same temperature to segregate any debris. The supernatant was transferred into 50 μL aliquots.

A nasal swab (the soft end of a FLOQSwab, cat nr. CP501CS01, COPAN Diagnostics Inc., United States) was gently inserted 2 cm into a nostril, twirled for five seconds, and subsequently placed into a separate 15 mL Falcon tube with 1 mL of PBS. The tubes were vortexed for 10 seconds and the solutions aliquoted.

Blood samples were obtained from participants by venipuncture into serum separator tubes (SST II, BD Vacutainer®). All samples were allowed to clot for at least 30 min at room temperature and then centrifuged at 2000g for 10 minutes. The supernatant was collected and stored at -80°C within 1-2 hours for later analyses.

Serum (dilution 1:50000), nasal secretion (dilution 1:1000) and saliva (dilution 1:100) spike-specific IgA and IgG were quantified by V-PLEX SARS-CoV-2 panel 31 according to the manufacturer´s instructions (Meso Scale Diagnostics (MSD), Maryland, USA). Total IgA concentrations were quantified using Isotyping Panel 1 Human/NHP Kit (MSD) according to the manufacturer´s instructions (Saliva: IgA, 1:10000; IgG, 1:1000. Nasal secretion samples: IgA 1:5000; IgG 1:500). To correct for possible differences in sampling efficacy, ratios between spike-specific Ig in saliva and nasal secretion samples and total IgA concentration in the same sample were calculated. The ratios were multiplied by 10^7 for graphical purposes. Cut-off values for detectable WT spike-specific IgG and IgA in serum were provided by the manufacturer. To set cut-off-values for detectable nasal secretion and saliva spike-specific IgA and IgG, the mean + 3SD of fourteen pre-pandemic nasal secretion samples, and six pre-pandemic saliva samples were used and set at 3.6 AU/mL for nasal IgA, 5.9 AU/mL for saliva IgA, 3.7 AU/mL for nasal IgG and 22.9 AU/mL for saliva IgG. When analysing and comparing Ig-levels between previously SARS-CoV-2 infected and infection naïve participants, nasal secretion and saliva Ig-levels below cut-off were set to cut-off/√2 for graphical and statistical purposes. Only samples with values above Limit of Detection (LoD, set by MSD Diluent 100 only + 2.5 SD, plate specific) in both spike specific IgA and total IgA were included when analysing correlation coefficients.

SIgA was measured as previously described (19). Briefly, nasal secretion and saliva samples were analysed for SIgA using a V-PLEX SARS-CoV-2 panel according to standard protocol with the following modifications: After sample incubation (dilution 1:50, time 2h) and 3x wash in 150 μL/well of 1X MSD Wash buffer, the supplied detection antibody was switched to a monoclonal mouse anti-human SIgA antibody (HP6141, Calbiochem, Sigma-Aldrich) at a final concentration of 5 µg/mL. After 1h incubation, the plates were washed x3 in 150 μL/well of 1X MSD Wash buffer and a SULFO-TAG conjugated goat anti-mouse antibody (#R32AC-5 ska, MSD) was added at a final concentration of 0.5 µg/mL. The final steps were then performed as specified by the manufacturer. A set of samples with expected high SIgA levels were initially analysed and the three samples with the highest IgA signal were pooled and used as calibrator for the standard curve. A four-fold serial dilution was used with two replicates at each of the seven calibrator levels and buffer only added as an additional zero calibrator blank. The calibration curve (plate specific) was established by fitting the signals from the calibrators to a 4-parameter logistic (or sigmoidal dose-response) model with a 1/Y2 weighting in the DISCOVERY WORKBENCH 4.0 Analysis Software (MSD). By correcting for dilution, the final antibody concentrations in undiluted samples were obtained. As for the IgA measurements, ratios between mucosal spike-specific secretory IgA and total mucosal IgA concentration were then calculated. The ratios were multiplied by 10^7 for graphical reasons. All plates were analysed on a MESO SECTOR S600 instrument (MSD), which is an electrochemiluminescence reader measuring the light emitted from the SULFO-TAG. Results for the antibody assays are reported in arbitrary units (AU)/mL.

Continuous variables (antibody levels) are presented as medians and compared using the Mann-Whitney U test. Categorial variables are presented as percentages and compared with the Chi-square test. The significance of correlation was analysed with Spearman’s rank correlation. Difference between two dependent correlation coefficients were analysed with Williams´s test. Coefficient of variation (CV) was calculated using the formula: CV = (standard deviation/mean) * 100. To investigate variant cross-binding capacity of SARS-CoV-2 specific Ig, a ratio between SARS-CoV-2 BA.5 and WT antibody levels was determined. To assess correlations between variant cross-binding capacity and antibody levels, a spearman rank correlation between SARS-CoV-2 BA.5/WT ratio and SARS-CoV-2 WT antibody levels was determined. All statistical analyses were performed in GraphPad Prism version 9.2.0 (GraphPad Software, San Diego, California, USA) or R version 4.3.2. Within R, the following packages were used: tidyverse, psych, table1 and readxl.

We first set out to compare total and spike-specific IgA and IgG in paired nasal secretion, saliva and serum samples from 72 HCW. Samples were collected in November 2022, and the majority of participants had received at least three SARS-CoV-2 vaccine doses (96%) and also experienced at least one prior infection (85%) ( Table 1 ). Nasal secretions were collected by swabbing the lining of the anterior nostrils and saliva was collected by passive drooling. Substantially higher levels of total IgA as compared to total IgG were detected in both saliva (nine-fold higher with a median of 1.2 x 10⁸ vs. 1.4 x 10⁷ pg/mL, p<0.0001) and nasal secretions (six-fold higher with a median of 1.2 x 10⁷ vs. 2.1 x 10⁶ pg/mL, p<0.0001) ( Figures 2A, B ). Spike-specific IgA correlated well in nasal secretions and saliva (r=0.79, p<0.0001) ( Figure 2C ). However, spike-specific IgA levels were three-fold higher in nasal secretions as compared to in saliva (median of 36.7 vs. 12.9 AU/mL) ( Figure 2D ). A threshold for detectable spike-specific IgA was set in both nasal secretions and saliva using pre-pandemic nasal secretion and saliva samples. The median spike-specific IgA level was ten-fold above the threshold of 3.6 AU/mL in nasal secretions while only two-fold above the threshold of 5.9 AU/mL in saliva, and there was a trend towards a larger proportion of participants with detectable spike-specific IgA in nasal secretions as compared to in saliva (83% vs. 77%, p = 0.33) ( Figure 2D ).

Considering the possibility of a monomeric IgA transudation from the circulation to mucosal secretions, we determined the proportion of spike-specific IgA bound to the secretory component (secretory IgA; SIgA) in nasal secretions and in saliva. As expected, there were strong correlations between the total population of spike-specific IgA and spike-specific SIgA in both compartments (nasal secretions r=0.96, p<0.001; saliva 0.77, p<0.001), but the correlation was significantly stronger in nasal secretions as compared to in saliva (p<0.0001), suggesting that there may be a larger transudation of monomeric IgA to saliva than to nasal secretions ( Figures 2E, F ). Interestingly, spike-specific IgA in serum correlated stronger (p<0.0001) to the total population of spike-specific IgA in saliva as compared to the total population of spike-specific IgA in nasal secretions (r=0.73, p<0.001 vs r=0.54, p<0.001) ( Figures 2G, H ). Collectively, these findings imply that, although spike-specific IgA correlated well between nasals secretions and saliva, the levels are higher and likely holds a larger proportion of polymeric SIgA in nasal secretions as opposed to slightly lower levels with a larger proportion of monomeric IgA transudate from the circulation, in saliva.

Prior studies have demonstrated the superior neutralizing capacity of dimeric IgA over monomeric IgA and IgG (12, 13), potentially due to the polymeric conformation. We therefore proceeded to compare the variant cross-binding capacity of spike-specific SIgA and the total population of spike-specific IgA in nasal secretions and in saliva. The variant cross-binding capacity was assessed by determining the ratio between SARS-CoV-2 BA.5 and wild-type (WT) spike-specific binding for SIgA, IgA, and IgG across compartments. The median ratios between BA.5 and WT spike-specific binding were higher for SIgA in both nasal secretions and saliva (1.17 and 1.19 respectively) as compared to IgA in the same samples (ratio 0.66 and 0.59 respectively), p<0.0001. IgA in serum displayed even lower variant cross-binding capacity (ratio 0.47), p<0.0001, albeit significantly higher than that for IgG in nasal secretion, saliva and serum (ratio 0.29, 0.26 and 0.26 respectively), p<0.0001 ( Figure 3 ).

There were only modest correlations between WT spike-specific titers and variant cross-binding capacity for SIgA in nasal secretions (r=0.29, p=0.01) and for IgG in serum (r=0.35, p=0.002), but no other significant correlations were observed between the titers of respective antibody and its variant cross-binding capacity in either of the compartments (all p > 0.05). There were very minor or no differences in the variant cross-binding capacity of each antibody isotype between nasal secretions and saliva, ( Figure 3 ), implying that the variant cross-binding capacity is isotype-dependent.

We recently demonstrated that SARS-CoV-2 infection elicits high and durable levels of spike-specific IgA in nasal secretions (19), but that current systemically administered SARS-CoV-2 vaccines do not trigger a mucosal spike-specific IgA response (18). As expected, and in line with these findings, spike-specific IgA levels were significantly higher in both nasal secretions and saliva from previously infected participants compared to from SARS-CoV-2 infection naïve participants (all p<0.01) ( Figures 4A, B ). Notably, however, these differences were substantially larger in nasal secretions as opposed to in saliva. Spike-specific IgA levels were eighteen-fold higher in nasal secretions from participants with prior infection as compared to SARS-CoV-2 infection naïve participants (median 46.0 vs. 2.5 AU/mL, p<0.001), while corresponding difference was only three-fold in saliva (13.8 vs. 4.2 AU/mL, p<0.01) and two-fold in serum (14 400 vs. 7 300 AU/mL) from the same individuals ( Figures 4A–C ). Spike-specific IgG levels were not higher in participants with prior infection in any of the mucosal compartments nor in serum ( Figures 4D–F ).

We next proceeded to validate our findings in a larger cohort of 976 participants sampled in February 2023 ( Table 2 ). Due to the time-consuming nature of passive drooling and the relatively large sample size, we first compared this method with the relatively easier saliva collection methods using a cotton roll (Salivette®) with or without chewing or a simple buccal swab in a subset of 18 participants, all with prior infection. Paired analyses revealed strong correlations and comparable spike-specific IgA levels in saliva samples collected by all four methods ( Figures 5A, B ). Furthermore, the coefficients of variation (CV) of spike-specific IgA were low between the collection methods, with the lowest CV between passive drooling and buccal swabbing (%CV for passive drooling and buccal swabbing was 9.1; passive drooling and cotton roll with chewing 12.7; passive drooling and cotton roll without chewing 21.0; buccal swabbing and cotton roll with chewing 14.1; buccal swabbing and cotton roll without chewing 20.5, and cotton roll with and without chewing 25.1) ( Figure 5C ). We next investigated the effect of centrifugation prior to analyses of saliva samples collected with buccal swabbing and by passive drooling. Centrifugation had a smaller effect on saliva spike-specific IgA levels in samples collected by buccal swabbing as opposed to by passive drooling (r=0.85, p=0.006, %CV: 7.2 vs. r=0.71, p=0.08, %CV: 18.8). As the cotton roll protocol stipulates a centrifugation step, we did not examine analyses of spike-specific IgA in saliva collected by cotton rolls without centrifugation. These experiments ensured us that any of these saliva collection methods can be used to generate comparable results, and, for simplicity, we therefore proceeded to sample saliva by buccal swabbing without centrifugation.

In line with our results from the smaller cohort collected in November 2022, spike-specific IgA levels correlated in saliva and nasal secretions (r = 0.66, p < 0.0001) ( Figure 6A ). Median spike-specific IgA levels were six-fold higher in nasal secretions as compared to in saliva (43.8 vs. 6.8 AU/mL, p<0.0001), a significantly larger proportion of participants had detectable spike-specific IgA in nasal secretions than in paired saliva samples (85% vs. 55%, p<0.0001), and median spike-specific IgA levels were twelve-fold higher than the threshold in nasal secretions (43.8 vs. 3.6 AU/mL) compared to just above cut-off in saliva (6.8 vs 5.9 AU/mL) ( Figure 6B ). Consistent with the findings from the smaller cohort, spike-specific IgA was detected predominantly in nasal secretions and saliva from individuals who had experienced a prior infection, with a larger difference in nasal secretions as opposed to in saliva (median difference between convalescents and naïve was nineteen-fold in nasal secretions, 48.7 vs. 2.5 AU/mL, and two-fold, 7.3 vs. 4.2, AU/mL in saliva) ( Figures 6C, D ). Prior infections had a negligible impact on serum spike-specific IgG levels ( Figure 6E ).