Participants for this study were recruited at the “Centre Hospitalier Universitaire de Québec – Université Laval (CHUL)” in Quebec City, Canada, as part of a longitudinal study of the humoral and cellular responses to SARS-CoV-2 (41, 42). The cohort consisted of 304 adult volunteers over 18 years of age who were food and retail workers (bar/restaurant, grocery, or hardware store) from the greater Quebec City area at the time of recruitment. Participants were enrolled after giving written informed consent. They came for three visits at 12-week intervals between April 2021 and May 2022, during which they were surveyed about demographic, socioeconomic, behavioural, and occupational characteristics. Whole blood for immune response analysis was collected at visit one (V1) and visit three (V3) ( Figure 1A ). The study was approved by the “Comité d’éthique de la recherche du CHU de Québec – Université Laval” (registration number 2021-5744).

Peripheral blood from venipuncture was collected in EDTA-containing tubes and processed within two hours. PMNs were isolated as previously described (43). In brief, 2% dextran (Sigma, Cat. 31392-50G) sedimentation of erythrocytes was first carried out, followed by Lymphocyte Separation Medium (Wisent, Cat. 305-010-CL) density gradient centrifugation. After removing the supernatant, remaining contaminating erythrocytes were eliminated through hypotonic lysis. Resulting purified granulocytes were then used for R848 (Resiquimod) (InvivoGen, Cat. vac-r848) stimulation.

Peripheral blood was first centrifuged to remove plasma. PBMCs were isolated as previously described (44). Briefly, they were isolated using Lymphocyte Separation Medium (Wisent, Cat. 305-010-CL) density gradient centrifugation. The lymphocyte ring was collected and washed. PBMCs were then used for R848 (InvivoGen, Cat. vac-r848) stimulation.

After isolation from blood samples, PBMCs and PMNs were resuspended at a concentration of 2 x 10⁶ cells/mL in the high-performance TheraPEAK™ X-VIVO™-15 Serum-free Hematopoietic Cell Medium (Lonza, Cat. BEBP04-744Q). They were plated in round-bottom 96-well plates at 1 x 10⁵ cells/well and 2 x 10⁵ cells/well and stimulated with the viral analogue R848 (InvivoGen, Cat. vac-r848). Four R848 conditions (0, 0.5, 1 and 2 µg/mL) were analyzed in triplicate. They were then incubated for 24 h at 37°C, 5% CO₂. The next day, plates were centrifuged 5 min at 400 x g and the supernatant was transferred to a flat-bottom 96-well plate and frozen at -20°C until use.

The human IL-8/CXCL8 Duoset ELISA kit from R&D Systems (Cat. DY208) was used to measure IL-8 concentrations in the supernatants of R848-stimulated cells. Ninety-six-well plates were coated with 4 µg/mL of diluted mouse anti-human capture antibody overnight at room temperature. The next morning, plates were rinsed three times with wash buffer (0.05% Tween-20 in phosphate-buffered saline (PBS)) with the HydroControl washer (Tecan, Salzburg, Austria) and incubated for 1 h with block buffer (1% bovine serum albumin (BSA) in PBS, pH 7.2-7.4, 0.2 µm filtered, R&D Systems, Cat. DY995) at room temperature. After rinsing, 100 µL of samples, standards (31.3-2000 µg/mL) and controls were added and incubated for 2 h at room temperature. Plates were washed, then 10 ng/mL of biotinylated goat anti-human detection antibody was added to the wells for 2 h at room temperature. After washing, streptavidin-HRP diluted 40-fold was added and left to incubate for 20 min at room temperature in the dark. Plates were washed again and prepared for the detection of bound antibodies by adding the substrate solution (1:1 mixture of H₂O₂ and tetramethylbenzidine, R&D Systems, Cat. DY999) for 20 min at room temperature in the dark. The stop solution (2N H₂SO₄, R&D Systems, Cat. DY994) was added to the wells before reading at 450 and 540 nm with the SpectraMax 190 microplate reader (Molecular Devices, San Jose, CA, USA).

All analyses were performed with GraphPad Prism 10.2.0 (GraphPad Inc, San Diego, CA, USA). For all data, initial tests of normality and log normality showed that the data did not fit both distributions. Therefore, all analyses were conducted using nonparametric tests. Where applicable, Friedman or Kruskal-Wallis tests were performed followed by Dunn’s multiple comparisons test for paired or unpaired data, respectively. All tests were two-sided and P-values < 0.05 were considered statistically significant. Data are shown as mean ± standard error of the mean (SEM).

The cohort and study timeline have been described in detail by Santerre et al. (42). Briefly, 304 food and retail workers from the Quebec City metropolitan area, Canada, were enrolled in order to study innate and adaptive immunity to SARS-CoV-2 ( Table 1 ). Their occupational distribution was as follows: 149 bar/restaurant workers (49%), 112 grocery store employees (37%), and 43 hardware store workers (14%). Men accounted for 42% of participants (n=128) and women for 58% (n=176). As for age, 139 participants were between 18 and 39 years old (46%), 118 were between 40 and 59 (39%), and 47 were 60 years or older (15%). The median age was 41 years old (interquartile range (IQR) 26-56). Three visits were planned for each participant at 12-week intervals. The first visit (V1) took place between April and October 2021, the second visit (V2) between July 2021 and January 2022, and the third visit (V3) between October 2021 and May 2022 ( Figure 1B ). The first vaccination campaign had already begun prior to recruitment. Thus, before V1, most participants had received at least one dose of vaccine (79%) and 63% were already fully vaccinated (two or more vaccine doses). Many participants (38%) also received a third dose before their third visit, which mostly coincided with the fifth wave (emergence of the Omicron variant) of the pandemic.

In order to assess the innate response, IL-8 production by PMNs ( Figure 2 ) and PBMCs ( Figure 3 ) was first measured following 24 h of stimulation with R848. Figure 2A shows a 30-day rolling average of IL-8 production by PMNs to illustrate the global fluctuation over the course of the study. The graph shows that the production increases in a dose-response manner with increasing R848 concentrations. Since there were two time points for each participant, results between both visits were first compared ( Figure 2B ). No significant differences were seen in IL-8 production between V1 and V3 for all conditions. Further analyses were performed using clinical and demographic data such as sex ( Figures 2C, G ), age ( Figures 2D, H ), body mass index (BMI) ( Figures 2E, I ), and the presence of comorbidities ( Figures 2F, J ) at both V1 and V3 separately. Again, results for all of these parameters did not show any significant difference for both visits.

The 30-day rolling average of IL-8 production by PBMCs after R848 stimulation shows some fluctuations throughout the year, yet, unlike PMNs, they do not follow a dose-response pattern ( Figure 3A ). There were also significant differences between both visits for all conditions ( Figure 3B ). Indeed, all values increased at V3 compared to V1, without stimulation and following activation with R848. Again, differences were assessed for sex ( Figures 3C, G ), age ( Figures 3D, H ), BMI ( Figures 3E, I ), and comorbidities ( Figures 3F, J ). There were no significant differences in relation to all those parameters neither at V1 nor at V3.

Given the information about the vaccine regimen of participants, different parameters were used to analyze IL-8 such as the number of doses received, the type of vaccine and the number of days between the visit and the last dose of vaccine received. Moreover, since previous results showed significant differences between visits, they were analyzed separately for the above-mentioned factors. Regarding the number of vaccine doses received, no differences were found for the PMN response at V1 ( Figure 4A ), nor at V3 ( Figure 4B ). However, results for PBMCs differed between subjects who were fully vaccinated compared to those who were not. Indeed, IL-8 production after R848 stimulation was significantly lower at V1 for fully vaccinated participants compared to unvaccinated and partially vaccinated participants ( Figure 4C ). At V3, subjects who received 3 vaccine doses had a significantly decreased IL-8 response compared to those who got only 2 doses ( Figure 4D ). It is notable that only nine participants remained unvaccinated, and that all other participants were fully vaccinated.

Depending on the type of vaccine received, participants were separated into five different groups. Group 1 consists of subjects who were not vaccinated. In groups 2 and 3, subjects received the BNT162b2 Pfizer-BioNTech vaccine (1-3 doses) or the mRNA-1273 Moderna vaccine (1-3 doses), respectively. Group 4 includes subjects who received the ChAdOx1-S AstraZeneca vaccine as a first dose, followed by either a second AstraZeneca dose, and/or one or two doses of an mRNA vaccine (Pfizer-BioNTech or Moderna). Finally, group 5 consists of participants who received a mix of mRNA vaccines (Pfizer-BioNTech and Moderna). Comparison of IL-8 production between those groups shows no effect of the type of vaccine regimen received on PMNs at V1, nor at V3 ( Figures 5A, B ). Likewise, no effect was seen for PBMC response at both visits ( Figures 5C, D ).

Considering the variation in time between the last vaccine dose and the visits of each participant, additional analyses were performed to investigate if it influenced the IL-8 response. Groups were separated as follows: unvaccinated participants, last vaccine received less than 14 days, between 15 and 30 days, between 31 and 60 days, between 61 and 90 days, between 91 and 120 days, and more than 120 days prior to the visit. No effect was seen on the IL-8 response by PMNs at V1 and at V3 ( Figures 6A, B ). For the PBMC response, there was also no effect at V1 ( Figure 6C ). However, at V3 ( Figure 6D ) there was a significant increase in the IL-8 response following R848 stimulation (2 µg/mL) between subjects who received their last vaccine dose 31-60 days prior to their visit and those who received it 91-120 days prior to their visit.

This cohort also includes subjects who were not vaccinated against COVID-19 at the time of their first visit, who became fully vaccinated before the third visit. Hence, their IL-8 response over time was analyzed to see if vaccination had an impact or not. The PMN response remained unchanged between both visits ( Figure 7A ). However, the PBMC response without stimulation was significantly stronger at V3 compared to pre-vaccination ( Figure 7B ).

It is well known that the circadian rhythm affects the immune response (45, 46). Hence, the daily time of sampling was analyzed to investigate if it could influence our results ( Supplementary Figure S1 ). Neither the PMN, nor the PBMC response to R848 stimulation was affected by the time at which samples were collected. As the study spanned over a whole year, and since the rolling averages showed some fluctuations, it seemed pertinent to evaluate seasonality on IL-8 responses. Each condition of stimulation was first separated and compared for the IL-8 response at each season of the study. For the PMN response, the resting condition ( Figure 8A ) is slightly increased in summer. Activation conditions ( Figures 8B–D ) display significant differences between spring 2021 and the following seasons. It is important to note that most participants who came during spring 2021 were not yet vaccinated against SARS-CoV-2. There is also a significantly higher IL-8 response after 2 µg/mL R848 stimulation in fall compared to winter. Responses during other seasons were similar. The PBMC response also shows strong variations between the different seasons of the study. Contrary to PMNs, the resting response remained stable throughout the course of the study ( Figure 8E ). The response after R848 stimulation also showed differences for PBMCs. Indeed, responses in spring 2021 were significantly stronger than in other seasons, as well as IL-8 responses in fall ( Figures 8F–H ). Moreover, differences remain marked at a high concentration of R848, contrary to PMNs. Variations in the responses separated by seasons do not concord with those following vaccination. Therefore, differences seen in PBMC response between seasons might be independent to the status of vaccination.

Since these results did not seem conclusive, seasonal differences were further investigated. PMN and PBMC activity was thus analyzed by comparing paired responses between visits 1 and 3 of participants, separating them by the seasons during which those visits took place. For PMNs ( Figure 9 ), the strongest differences occur between summer and fall ( Figure 9B ), where there is an increase in the IL-8 response following stimulation in fall. Interestingly, PMN activation also increases in fall compared to spring ( Figure 9A ). Here, the majority of participants were not vaccinated in spring for their first visit, but received their full vaccine before their third visit. This result differs from those seen earlier in Figure 7A , suggesting that the season might have a greater impact than vaccination on PMN responses. No difference was observed for other seasons ( Figures 9C–E ). As no participants came first in spring and then in summer for their third visit, that comparison could not be made. For PBMCs ( Figure 10 ), stronger responses were seen between summer and winter, as well as between fall and winter and between fall and spring. Surprisingly, all significant differences were an increase in the PBMC response at V3, no matter the season, suggesting that the changes seen between seasons might be due to other factors. There was an increase of IL-8 production for PBMCs at rest between spring and fall ( Figure 10A ), but no differences between summer and fall ( Figure 10B ). In subsequent analyses, IL-8 measurements were separated by month. It was found that there were many fluctuations during the year. For the PMN response, there seem to be peaks of activity across summer and fall following R848 stimulation ( Supplementary Figure S2 ). The PBMC response shows strong peaks at the end of spring and at the end of fall ( Supplementary Figure S3 ). The first peak coincides with the first wave of vaccination, as seen in the timeline ( Figure 1B ). However, the second peak comes before peak vaccination and the burst of infections observed during the fifth wave, which indicates that it must be independent of vaccination or infection.

To attempt to discriminate further between seasons and vaccination, vaccinated and non-vaccinated participants were separated in the seasonal analysis. For PMNs, there were no differences in IL-8 between seasons for both non-vaccinated and vaccinated participants ( Figures 11A, B ). It is notable that, except for spring 2021, the number of non-vaccinated subjects who came during following seasons was rather small (summer n=15, fall n=8, winter n=7, spring 2022 n=1); hence, it is difficult to draw conclusions for those groups. The results for PBMCs were quite different. Indeed, there was only a difference between spring and summer 2021 in the IL-8 response following activation with 1 and 2 µg/mL R848 for the non-vaccinated participants ( Figure 11C ). In contrast, vaccinated participants showed differences for all conditions of activation ( Figure 11D ). Finally, analyses were done to compare IL-8 production in COVID-19 infected and uninfected participants ( Figure 12 ) and no differences were seen between both groups.