The Ethical Committee (The Ethical Committee at the University of Tartu) approval (#323T-21) was received for the studies. The participants provided their written informed consent to participate in the study. The SARS-CoV-2 antibody-positive individuals were identified by KoroSero-EST-1 cross-sectional seroepidemiological study conducted from May 8 to July 31, 2020, in two general practitioners at Tallinn and Saaremaa island and is presented in detail elsewhere (20). The population in Saaremaa, in particular, had an outbreak of COVID-19 in March 2020 with a seroprevalence of 6.3%.

The 56 antibody-positive individuals were reanalyzed for the longitudinal samples in November 2020, which occurred on average 7-8 months after their initial infection in March 2020 i.e. 6 months after the seropositivity analysis ( Table 1 ). From these, 44 individuals had the asymptomatic infection as they did not report acute respiratory disease or other COVID-19 symptoms since March 2020 but were seropositive for SARS-CoV-2 antibodies. Twelve antibody-positive individuals reported symptoms of respiratory disease. As negative controls, we studied 115 age and sex-matched uninfected individuals from Saaremaa and Tallinn populations. Samples for serum analysis were kept +4°C during the transportation to the laboratory, where stored at -20°C until testing in SYNLAB Estonia Central Laboratory or at the research laboratories of the University of Tartu, Estonia.

The samples were tested by chemiluminescent microparticle immunoassay (CLIA) for detection of IgG against SARS-CoV-2 nucleoprotein (N) (Abbott Architect SARS-CoV-2 IgG with ARCHITECT i2000SR analyzer; Abbott Laboratories, USA) according to the manufacturer’s instructions and as reported earlier (20, 21). Luciferase-based immunoprecipitation assays (LIPS) with S-RBD and N proteins were performed as reported (22). Results are expressed as percent of positive control's luciferase activity values (the discrimination level was 1.6 for S-RBD, and 1.3 for N antigens).

An in-house neutralization assay was performed for positive samples as reported (20). The neutralization assay was done in duplicates, the serum samples were two-fold serially diluted in Virus Growth Media (VGM – DMEM/0.2% BSA/Pen-Strep) starting from 1:4 to 1:4096 in a volume of 50 μl per well. The viral stock was diluted in VGM to contain 100 pfu (plaque-forming units) per 50 μl. Virus (a local Estonian SARS-CoV-2 isolate from a COVID-19-positive patient’s nasal swab; from SYNLAB) was added to the wells containing sera dilutions, and the mixtures were incubated for 1 hour at 37°C. After this, 4x10⁴ Vero E6 cell suspension in 100 μl of VGM per well was added to the wells. Plates were incubated at 37°C and 5% CO₂ for up to 5 days. Neutralization titer was calculated by microscopically evaluating cytopathic effect (CPE) in infected wells (round/floating cells compared to an untouched monolayer in non-infected wells). The dilution of sera that did not show CPE was considered a neutralization titer.

PBMC were isolated from CPT tubes (BD Biosciences) according to the manufacturer’s instructions and cryopreserved in CTL-CryoTM ABC Media Kit (ImmunoSpot). Before T cell stimulation the PBMC vials were thawed, washed, counted, and suspended in an X-Vivo15 serum-free medium (Lonza). 1-2 million of PBMCs were used per condition. Overlapping peptide pools of SARS-CoV2 spike, nucleocapsid, and membrane proteins were purchased from Miltenyi Biotec and used at a final concentration of 1ug/ml of each peptide. The stimulations (mock, S, M, and N peptide pools) were carried out for 20 hours in the presence of anti-CD28 and anti-CD49d. CEFX peptide pool (JPT Peptides) was used as a positive control. After the stimulation T cells were stained with monoclonal antibodies: CD3 Brilliant Violet 650, CD4 Alexa Fluor 700, CD8 Brilliant Violet 605, CCR7 PE-Dazzle, CD45RA APC, CD69 Brilliant Violet 510 (all from Biolegend), and CD137 PE (from Miltenyi Biotech). Before the acquisition with LSRFortessa (BD Biosciences), 7AAD was added for the discrimination of dead cells. Data were analyzed using FCS Express 7 (DeNovo Software).

A panel of 92 inflammation-related biomarkers was measured in serum by the Proximity Extension Assay technique by Olink Proteomics. The assay uses two oligonucleotide-conjugated antibodies that bind to protein targets. Upon binding to the protein epitope, the paired oligonucleotide sequences are amplified through a quantitative real-time PCR reaction. Data is then generated using normalized protein expression (NPX) values on a log2 scale whereby a higher NPX correlates with higher protein expression. Proteins containing NPX values >50% below the assay’s limit of detection (LOD) were excluded from the analysis. The data were pre-processed by Olink^® using NPX Manager software.

The Wilcoxon Kruskal-Wallis tests with Dunn multiple correction and Spearman correlation calculations were performed using GraphPad version 9. The significance symbols on violin and dot plots are marked: * = adj. P ≤ 0.05, ** = adj. P ≤ 0.01, *** = adj. P ≤ 0.001 and **** = adj. P ≤ 0.0001.

Using population-based serological screening we identified 56 SARS-CoV-2 antibody-positive individuals (20), from which 44 were seropositive without any sign of infection since March 2020 (asymptomatic) and 12 individuals who self-reported infection and were tested positive for the SARS-CoV-2 PCR test (symptomatic). We employed an automated CLIA and in-house LIPS method to detect SARS-CoV-2 antibodies. In the LIPS approach, we measured separately antibodies to N protein and S-RBD, the latter correlating with the antibody neutralization activity (23). Overall the CLIA method, targeting SARS-CoV-2 N protein, correlated well with LIPS N protein assay (r= 0.89, p<0.0001, Supplementary Figure 1A ), slightly less with LIPS S-RBD (r= 0.59, p=0.0001, Supplementary Figure 1B ), whereas there was a strong correlation between LIPS S-RBD and N findings (r=0.72, p<0.0001, Supplementary Figure 1C ) among positive samples.

The antibody levels between asymptomatic and symptomatic patients were not different at 7-8 months but were significantly higher compared to negative controls as measured by CLIA ( Figure 1A ) and both LIPS assays ( Figures 1B, C ). As the humoral response in SARS-CoV-2 infected individuals has been reported to decline with time, we studied the persistence of SARS-CoV-2 antibody levels at 7-8 months compared to the time point of seropositivity finding. Indeed, we found a significant decrease in SARS-CoV-2 N protein antibody levels in asymptomatic individuals by CLIA ( Figure 2A ) and LIPS N ( Figure 2B ) assays as well as in symptomatic individuals by CLIA ( Supplementary Figure 2 ). In contrast, the antibody levels to S-RBD were not significantly changed ( Figure 2C ). Specifically, antibodies to N protein decreased in 82% (36 out of 44), whereas antibodies to S-RBD were lower in 54% (24 out of 44) of asymptomatic individuals. Moreover, S-RBD antibodies at 7-8 months correlated with the virus neutralization titers (r=0.421; p<0.005) ( Supplementary Figure 3A ), and similarly to S-RBD antibody levels, the neutralization titers were stable without significant change 7-8 months after the infections ( Supplementary Figure 3B ). The antibody levels among the sexes and age groups were similar. These results show that antibody levels to SARS-CoV-2 remain slightly decreased or stable 7-8 months after the infection. Although the antibodies to N protein decline with time, the S-RBD antibodies remain stable and, importantly, correlate with the virus neutralization capacity.

We next analyzed the SARS-CoV-2 -specific memory T cell responses in asymptomatic and symptomatic antibody-positive individuals 7-8 months after their infection time. To this end, we tested the antigen-specific CD4⁺ T cell activation based on the upregulation of CD137 and CD69 activation markers on memory cells (single- and double-negative cells for CCR7⁺ and CD45RA^-) using three SARS-CoV-2 antigen peptide mixes of S, N, and M proteins ( Figure 3A ) in 45 antibody-positive (34 asymptomatic and 11 symptomatic) and 41 antibody-negative individuals. The CD4⁺ T cells responded the strongest to M (mean % of activated CD4⁺ memory T cells 0.11; range 0%-0.47%) and S (0.10%; 0%-0.40%) derived peptides and had a slightly weaker response to N peptides (0.096%; 0%-0.82%). Calculating the sum of the three antigen responses among memory T cells showed a significantly higher proportion of antigen-specific activated CD4⁺ T cells among asymptomatic antibody-positive individuals (0.31%; 0%-1.44%) compared to negative controls (0.08%; 0%-0.23%) ( Figure 3B ), and though the symptomatic group (0.57%; 0.20%-1.24%) tended to have a higher number of positive T cells, this was not significant when compared to asymptomatic group. Further correlation with corresponding SARS-CoV-2 antibody levels showed CD4⁺ T cell responses to have a moderate but significant correlation (Spearman ρ= 0.54, p<0.001) with anti-RBD antibodies ( Figure 3C ) but not with anti-N antibodies as tested by CLIA and LIPS assays.

In search of signs of long-term inflammation in asymptomatic and symptomatic individuals, we applied Olink technology and profiled their serum samples for inflammation proteins. The screening of 50 antibody-positive (39 asymptomatic and 11 symptomatic) and 54 antibody-negative individuals showed overall similar levels in the majority of markers 7-8 months after the infection. Nevertheless, four serum proteins, S100A12, TGF-alpha, IL18, and OSM, all associated with activated macrophage-monocytic cells, were elevated in both the asymptomatic and symptomatic groups ( Figures 4A–D ). The four markers were thus more likely associated with the long-term effect of SARS-CoV-2 infection and not with symptoms of the disease. Within this panel, we searched for other inflammation markers associated with the four markers and found four other protein biomarkers (CASP8, HGF, TNFSF14, and IL8) which were in moderate correlation with elevated markers identified in seropositive individuals ( Figure 4E ). We did not see the elevated inflammation markers to correlate with antibody levels, T cell responses, or sex and age.