This study was approved by the Internal Review Board of Hannover Medical School (MHH, approval number 3639_2017, 9001_BO-K) and includes 206 donors, who gave written informed consent for their peripheral blood to be collected and used for research purposes at Hannover Medical School (MHH). All donors are recovered COVID-19 patients, who were sampled between 14 and 433 days after symptom onset. This study population consists of 98 female and 108 male donors, being 18 to 69 (Median = 46.25) years old (Table 1). For PBMC isolation, either whole blood samples or residual blood samples from platelet and plasma disposables were used. From each donor, two to thirteen samples were collected between April 2020 and May 2021.

All donors had been infected with SARS-CoV-2, confirmed by reverse transcriptase polymerase chain reaction (RT-PCR). The dates of symptom onset (ranging from February 2020 to October 2020) and offset of symptoms (ranging from March 2020 to November 2020) were documented. Consequentially, symptomatic periods ranged from zero (“symptom-free”) to 73 days (Median = 12 days). Disease severity was categorized by the WHO clinical progression scale (22). In this study population, the WHO score ranged from one (ambulatory) to seven (severe disease with mechanical ventilation required).

Serology for SARS-CoV-2 was performed by ELISA (anti-SARS-CoV-2 S1 spike protein domain IgG, QuantiVac, #EI 2606-9601-10 G and anti-SARS-CoV-2 NCP IgG; Euroimmun, Lübeck, Germany, #EI 2606-9601-2 G) according to the manufacturer's instructions (plasma dilution 1:500). Antibody levels are expressed as IgG RU/ml (RU= Relative units) or IgG Ratio (optical density divided by calibrator < 0.8, negative; 0.8–1.1, intermediate; >1.1, positive).

Flow cytometry was performed on the FACSCanto 10c system (BD Biosciences, Heidelberg, Germany) using the BD FACSDiva Software version 8.0.1. The following markers for innate leukocytes as well as T and B cells were applied to determine the cellular immune status (Panel 1) and B cell phenotype (Panel 2): anti-CD3 fluorescein isothiocyanate (FITC, BD Biosciences, #345764), anti-CD4 Peridinin-Chlorophyll-Protein (PerCP, BD Biosciences, #345770), anti-CD8 allophycocyanin (APC, BD Biosciences, #345775), anti-CD14 Brilliant Violet 510 (BV510, BD Biosciences, #563079), anti-CD19 AF-700 (BD Biosciences, #557921) or BV510 (BD Biosciences, #302242), anti-CD20 APC-cyanine 7 (APC-Cy7, BioLegend, #302314), anti-CD24 PerCP (BioLegend, #311114), anti-CD27 BV421 (BioLegend, 356418), anti-CD38 APC (BioLegend, #356606), anti-CD45 APC-H7 (BD Biosciences, #641417) or AlexaFluor 700 (AF-700, BioLegend, #368514), anti-CD56 phycoerythrin (PE, BD Biosciences, #345812), anti-CD45RA BV605 (BioLegend, #304134), anti-CD62L BV421 (BioLegend, #304828) or BV510 (BioLegend, #304844), anti-IgM FITC (BioLegend, #314506), and anti-IgD PE (BioLegend, #348204). Cells were stained at room temperature and in the dark for 30 minutes either before (Panel 1) or after (Panel 2) lysis of erythrocytes according to the manufacturer's instructions (Panel 1: BD Biosciences, #349202, Panel 2: Beckman Coulter, Brea, CA, USA, #A07799). For the analysis of total cell numbers, whole blood samples were measured on a single platform involving TruCount™ tubes (BD Biosciences, #340334).

SARS-CoV-2-specific T lymphocytes were detected by Interferon-gamma (IFN-γ) Enzyme-linked Immunospot (EliSpot) assay. In short, after PBMC isolation from whole blood samples using discontinuous density gradient centrifugation and resuspension in culture medium, which was comprised of RPMI1640 (Lonza, Vervies, Belgium, #BE12-702F) with 10% of human AB serum (C.C.pro, Oberdorla, Germany, #S-41-M), at a concentration of 1 x 10⁷ cells/ml, the cells were plated in 24-well plates for overnight resting. On the next day, the rested PBMCs were transferred to anti-IFN-γ pre-coated EliSpot plates (Lophius Biosciences, Regensburg, Germany, #12100010) in co-culture with specific antigens of interest for 16–18 h at a density of 2.5 x 10⁵ cells/well. Overlapping peptide pools against the following peptides were used for stimulation at a final concentration of 1 μg of each peptide/ml peptide pool: SARS-CoV-2 membrane protein (M, #130-126-703), nucleocapsid protein (N, #130-126-699), and CMV phosphoprotein 65 (pp65, #130-093-435) (all Miltenyi Biotec, Bergisch Gladbach, Germany) as well as SARS-CoV-2 proteins S1 and S2 (PM-WCPV-S), huCoV epitopes (strains 229E, #PM-229E-S-1 and OC43, #PM-OC43-S-1; S1 and S2), and antigens derived from human RSV (nucleoprotein, NP, #PM-HRSVB-NCPN) and IAV (matrix protein 1, MP1, PM-INFA-MP1-H1N1) (all JPT, Berlin, Germany). A positive control (PC) consisted of cells stimulated with staphylococcal enterotoxin B (1 μg/ml, SEB, Merck, Taufkirchen, Germany, #S4881) while PBMCs in culture medium alone served as a negative control (NC). The detection of IFN-γ was achieved using streptavidin-alkaline phosphatase (Mabtech Stockholm, Sweden, #12360002-S) and 5-13 bromo-4-chloro-3-indolyl phosphate/nitroblue tetrazolium (BCIP/NBT Liquid Substrate, SERVA, Heidelberg, Germany, #15246.01). Emerging IFN-γ spots were counted on an AID iSpot spectrum reader system using the AID EliSpot 8.0 Software (both from AID, Strassberg, Germany). For duplicate wells, means were calculated and for each specific antigen, the result was expressed as the number of spots per well (spw). For determining non-responders to this assay, a threshold was set at ≥3 spw or >2xNC. Additionally, spots/10,000 CD3⁺ T cells were calculated based on staining of PBMCs with anti-CD3 FITC (BD Biosciences, #345764), anti-CD4 PerCP (BD Biosciences, #345770), anti-CD8 APC (BD Biosciences, 345775), and anti-CD45 APC-H7 (BD BioSciences, #641417) followed by flow cytometric analysis.

Intracellular cytokine staining for characterization of antiviral memory T-cell subsets was performed and analyzed as previously described (23). Briefly, 1 x 10⁶ isolated PBMCs were rested overnight in TexMACS culture medium (Miltenyi Biotec, #170-076-307) and then were stimulated with overlapping peptide pools derived from SARS-CoV-2 proteins M, N, S1 and S2, huCoV S proteins (strains 229E and OC43) as well as RSV NP, IAV MP1, and CMV pp65 (Miltenyi Biotec and JPT, catalog numbers see above) at a final concentration of 1 μg of each peptide/ml peptide pool. Stimulation with phorbol 12-myristate 13-acetate (PMA; 10 ng/ml, #P1585) and ionomycin (500 ng/ml, Sigma Aldrich, #I9657) as well as CytoStim (Miltenyi Biotec, #130-092-172) served as a positive control whereas unstimulated PBMCs were used as a negative control. After one hour of stimulation, 5 μg/ml of Brefeldin A (BioLegend, #420601) were added to each well, incubating for another 4 h afterwards. Following the total 5 h of stimulation period, the protocol included harvesting and extracellular staining with anti-CD4 PerCP (BioLegend, #, anti-CD8 PE-Cy7 (BioLegend, #344712), anti-CD45 Pacific Blue (BioLegend, #304022), anti-CD45RA BV605 (BioLegend, #304134), and anti-CD62L FITC (BioLegend, #304804). Thereafter, cells were fixed and permeabilized using the IntraPrep Kit according to the Manufacturer's instructions (Beckman Coulter, #A07803), followed by intracellular staining with anti-TNF-α (tumor necrosis factor alpha) APC (BioLegend, #502912) and anti-IFN-γ PE (BioLegend, #502509). Flow cytometric measurements were performed on a FACSCanto 10c system (BD Biosciences), acquiring at least 50,000 events in the CD45⁺ lymphocyte gate for each analysis.

The following software was used for data analysis: Microsoft Excel 2010 (Microsoft Corporation, Redmond), FlowJo™ v10 (FlowJoTM LLC, BD Biosciences), and BD FACSDiva v8.0.1 (BD Biosciences). Graphic visualization and statistical analysis were performed using Prism Version 8.2.0 (GraphPad Software, San Diego, California, USA), which also included Linear Regression and Kruskal-Wallis or Friedman Test followed by multiple comparison correction. Significance levels were calculated and expressed as p-values (^*p < 0.05, ^**p < 0.01, ^***p < 0.001, ^****p < 0.0001).

Sampling times were divided into three time periods depending on the time between symptom onset and sample collection: the first time period was defined as 0–26 weeks (n = 193 donors), the second time period was defined as 27–52 weeks (n = 82 donors), and lastly, the third time period was defined as 53 weeks and more (n = 24 donors) after symptom onset (Table 1). Figure 1A shows the distribution of sampling times in relation to time of symptom onset. The column's subdivisions reflect the three time periods described above. Notably, the distribution of symptom onset times is comparable to the general incidence of SARS-CoV-2 infections in Germany during that time (24). Most of the donors included in this study had a mild course of the disease (WHO clinical progression scale 2). The distribution of disease severity according to the time after symptom onset is evenly distributed amongst the three time periods. Due to the fact that from some donors, several samples were collected within one time period, we uniformly selected the first donation in each time period.

Whole blood samples were analyzed via flow cytometry to generate a cellular immune profile over the course of the study. The results are shown either as percentages or as cells per microliter blood (Figures 1B–D). CD4⁺ T cell phenotypes were stable while there were rises in CD8⁺ T central memory (TCM) cell counts as well as decreases in T effector memory re-expressing CD45RA cells (TEMRA) respectively (Figure 1B) from the first to the second time period. Additionally, CD8⁺ T effector memory (TEM) cells increased similarly (Figure 1B). Some innate and adaptive immune cell types, namely γδ T cells, granulocytes, and monocytes (Figure 1C) as well as CD19⁺ cells (Figure 1D), showed a slight progression toward median values that were observed for healthy, SARS-CoV-2-naïve donors in a previous study (23), indicating a longer retention time in the blood after the initial immune response. Interestingly, marginal zone B cell counts decreased significantly from the second to the third time period. To conclude, while there were minor changes in the CD8⁺ T cell compartment, most immune cell subsets remained stable over time.

Serum and plasma samples were analyzed for levels of anti-S and anti-Nucleocapsid protein (N or NCP) IgG antibodies. Results are shown as anti-Spike IgG RU/ml and IgG NCP Ratio. Median antibody levels against SARS-CoV-2 S (Figure 1E) as well as NCP (Figure 1F) significantly decreased toward the second time period and then remained on a stable level from the second to the third time period. Notably, highest measured values were consistently receding over time for S and NCP. In summary, levels of anti-S and anti-NCP IgG antibodies appeared to decrease toward a lower but stable level over time with anti-NCP IgG levels presenting an overall stronger decline.

Implementing IFN-γ EliSpot, we investigated T cell frequencies against overlapping peptide pools derived from SARS-CoV-2 and huCoV strains OC43 and 229E as well as RSV, IAV, and CMV. Results are shown as spots per 10,000 CD3⁺ T cells. We found SARS-CoV-2 M- and N-specific T cells to be stable on the long-term despite decreased frequencies in the second time period (Figures 2A,B). When evaluating T cell frequencies for SARS-CoV-2 S, S1 subunit-specific T cell frequencies appeared to be stable over the period of observation while S2 subunit-specific T cell frequencies decreased initially and then remained on a stable level toward the third time period (Figure 2C), similarly to SARS-CoV-2 N-specific T cells. Furthermore, we investigated the relation between the progression of T cell frequencies against SARS-CoV-2 and huCoV strains OC43 and 229E S1 and S2 subunits (Figure 2C). Within one virus strain, values for S2-specific T cells were generally higher than for S1, with generally highest values observed for SARS-CoV-2. Over time, T cell frequencies were stable for huCoV 229E S1 as well as for 229E S2 and OC43 S1, despite decreased levels in the second time period. The huCoV OC43 S2-specific T cell response was initially higher and then decreased to a stable level 6 months post symptom onset. By correlating the values for T cell frequencies against the S1 with the S2 subunit we found a strong level of correlation for SARS-CoV-2 (Figure 2D) and lower levels of correlation for the huCoV strains OC43 and 229E (data not shown). Lastly, we also analyzed T cell frequencies against RSV, IAV (Figure 2E), and CMV (Figure 2F). Lowest values were observed for RSV with a pattern similar to the one observed for SARS-CoV-2 S2. For IAV, T cell frequencies were comparable to those seen against SARS-CoV-2. Here, we saw a progression similar to that of SARS-CoV-2 M. Finally, largely stable T cell frequencies against CMV were detectable in CMV-seropositive donors. In summary, T cell responses against SARS-CoV-2 epitopes were detectable over the entire course of the study, with the most stable frequencies having been measured for SARS-CoV-2 S1 while values for SARS-CoV-2 S2 initially decreased and remained stable afterwards.

We analyzed samples obtained from 18 donors individually in order to assess their individual progressions of SARS-CoV-2-specific T cell frequencies and antibody levels. This cohort consists of nine female and nine male donors, ranging from 24 to 68 years of age (Median = 49 years) and having shown mostly mild symptoms during infection [WHO score 2 (n = 17) and 4 (n = 1, hospitalized)] (Table 1). The results are expressed as spots per 10,000 CD3⁺ T cells with the T cell frequencies shown for SARS-CoV-2 M, N, S1, and S2 separately (Figures 3A–D) as well as their cumulative values (Figure 3E). For the analysis of each peptide pool, the respective non-responders to that peptide pool were excluded whereas in Figure 3E all donors are shown, since every donor responded to at least one peptide pool in each time period. We observed that donors with high frequencies in the beginning presented lower levels over the course of time. Individuals who showed equally low levels of T cell frequencies at the beginning presented consistent levels one year later. Interestingly, in two donors we detected a slight rise of SARS-CoV-2-specific T cell frequencies throughout time of convalescence. Furthermore, initial T cell frequencies were higher against SARS-CoV-2 M and S1 than against the other viral proteins. To characterize the specific long-term antiviral T cell populations (53–60 weeks post symptom onset) in more detail, we performed intracellular staining for IFN-γ and TNF-α after stimulation with the same peptide pools used in the EliSpot assays. Consistent with the findings described above, we were able to detect specific T cells against all SARS-CoV-2 derived peptide pools (Figure 3F). Furthermore, the results showed a pattern largely comparable to our previous work (23). There, highest specific T cell frequencies were detected within the subset of T effector memory (TEM) cells, as is the case here. Interestingly, CD4⁺ T cell responses against SARS-CoV-2 S shifted from SARS-CoV-2 S2 toward SARS-CoV-2 S1 subunits, which is in line with the results from the EliSpot assay (Figure 2C). In summary, individual T cell frequencies appeared to converge from variously high levels to a stable level over the course of 1 year after infection, which resembles our findings in the large cohort.

For the cohort described above, results of serological analyses are presented individually. Antibody levels against S (Figure 3G) and NCP (Figure 3H) were both initially declining toward the second time period. Subsequently, progressions vary from stable to mildly decreasing toward the third time period with apparently more consistent values for anti-S IgG. To conclude, we also observed a similar progression of this individual cohort's antibody levels in comparison with the large cohort.

Via IFN-γ EliSpot, we examined 37 samples from 34 donors for specific T cells against peptide pools derived from the S1 and S2 subunits of three SARS-CoV-2 variants, namely the wild type (previously presented), alpha (B.1.1.7), and beta (B.1.351) VOCs. This cohort consists of 14 females and 20 males, ranging from 25 to 69 years of age (Median = 46 years), who have shown mostly mild symptoms during infection [WHO score 1 (n = 1), 2 (n = 29), and 4 (n = 4, hospitalized)]. Samples were collected 21 to 61 weeks after symptom onset (Median 55.2 weeks) with the last sample collected in May 2021. Results are shown as spots per 10,000 CD3⁺ T cells. Considerably, no significant differences in T cell frequencies against each of the three variants were found (Figure 4A). When comparing S1 to S2, the T cell responses against the alpha and beta variants appeared to be lower for S2, as was the case for the wild type virus. We went on to correlate the cumulative T cell frequencies for each variant with each other. In this analysis, the alpha and beta variants showed the strongest correlation (R² = 0.7612, P-value < 0.0001) while the correlations of alpha and beta variants to wild type SARS-CoV-2 both showed a similar but lower correlation level (alpha: R² = 0.1605, P-value = 0.014; beta: R² = 0.1351, P-value = 0.0298) (Figures 4B–D). Of note, all donors included in this study have most likely been infected with SARS-CoV-2 wild type virus due to the fact that the first case of a SARS-CoV-2 VOC infection in Germany occurred in late December of 2020 and the latest infection in this study was dated in October 2020 (25). In summary, T cell responses to S-derived peptide pools from wild type, alpha, and beta SARS-CoV-2 showed no significant differences and various degrees of correlation, of which the strongest correlation was observed between the alpha and beta variants.

During the time of this study, one donor (male, >50 years old) became reinfected with SARS-CoV-2 almost 1 year after the first infection. Both infections were symptomatic with similar symptom durations. However, while during the first infection strong limb and internal pain was reported, symptoms during the second infection were milder (cough and elevated body temperature). We analyzed three samples from this donor, two of which were collected before the reinfection. Throughout this course of time, only small fluctuations for various immune cells were measured, indicating a rise directly after infection and a decrease over time (Figure 5A). We observed stable to slightly increasing counts of CD19⁺ cells, naïve, marginal zone, non-switched memory, and switched-memory B cells (Figure 5B). Notably, cell counts for transitional B cells and plasmablasts strongly decreased in the second sample, then rose again after reinfection with a more than two-fold increase in plasmablast levels compared to the first sample (Figure 5B). Strikingly, we saw a drastic increase in SARS-CoV-2-specific antibody levels, both against S and NCP, after reinfection (Figure 5C), which is consistent with the high cell count of plasmablasts. Moving on to the T cell frequencies (Figure 5D), it is remarkable that the majority of T cell responses after reinfection were lower than in the first donation. Interestingly, SARS-CoV-2 N-specific T cells appeared to be partially re-emerging after reinfection in this donor, whereas values for SARS-CoV-2 M resemble those of the second donation. Furthermore, frequencies for wild type SARS-CoV-2 S1 and S2 showed a strong decrease in the second donation, which is consistent in the third. Comparing those with the endemic coronaviruses, we also saw high initial T cell responses to OC43 S2 and 229E S1 decreasing to a stable level over time. On the contrary, OC43 S1 and 229E S2-specific T cells appeared more frequently after the reinfection with SARS-CoV-2. Lastly, in compliance with the general progression of T cell frequencies in this donor, RSV- and IAV-specific T cells were highest in the first sample, considerably lower in the second and increased again in the third, while T cell responses against CMV remained stable. Moreover, at the time of the third sampling, we compared the T cell responses to SARS-CoV-2 VOCs alpha and beta. As shown in Figure 5E, cumulative values for S1 and S2 were considerably low for the wild type virus and highest for SARS-CoV-2 alpha. Of note, the virus variant that caused the reinfection was not determined. In summary, in this case of reinfection with SARS-CoV-2, symptoms were milder but similar in duration. Re-exposure to SARS-CoV-2 was also followed by particularly high cell counts for plasmablasts and transitional B cells as well as distinctly increased antibody levels. The frequency of N-specific T cells increased upon reinfection, while it remained stable for the remaining tested antigens. Responses to VOCs were considerably higher than to wild type virus.