The CoNAN study (COVID-19 Outbreak in Neustadt-am-Rennsteig) was a prospective longitudinal population-based cohort study in Neustadt am Rennsteig a village in the Ilm-district in central Thuringia, Germany with 883 inhabitants in which a SARS-CoV-2 outbreak had occurred in Spring 2020. Due to the isolated location of the village, the extensive testing of the population, and the clear and controlled outbreak, Neustadt am Rennsteig is well suited to study the seroprevalence and potential development of immunity of SARS-CoV-2 infections. On March 22, 11 confirmed COVID-19 cases had been diagnosed in the district of which six (55%) were Neustadt residents with further 69 residents classified as contact persons. As a consequence, local public health authorities declared 14-day quarantine on the entire village. With support of the local family physician, an outbreak containment team of the public health department conducted a mandatory mass screening using nasopharyngeal swabs starting on April 2, in which 865 SARS-CoV-2 PCR tests were performed. Health authorities reported 51 SARS-CoV-2 infections and three SARS-CoV-2–associated deaths in the community during the outbreak. All persons with positive PCR results were defined as COVID-19 cases. The initiated containment measured controlled the outbreak; the spread to neighboring villages was prevented. Quarantine on the village was lifted on 5 April 2020. For the CoNAN study, samples were taken at three defined time points. The first sampling was performed from 13 to 16 May 2020. A total of 626 participants were included. The results of the antibody testing have been published (8). The second sampling was performed from 7 to 9 October 2020 and included the participants of the first round who had shown antibodies in at least two different IgG antibody assays and a control group matched after sex, age, and comorbidities, 145 participants in total. The third sampling was performed from 13 to 15 April 2021, with the participants of round 2 along with some new participants, 224 in total ( Figure 1 ). Participation in the study was voluntary and could be withdrawn at any time. Refusal to participate had no consequences. Participants were enrolled at a central study site that was set up in the villages’ town hall or in rare cases if requested by home visits. After informed consent, questionnaires and blood samples were directly taken at the study site. All inhabitants of the community of Neustadt am Rennsteig regardless of age, gender, or infection status were eligible for participation in the first phase. Individuals who do not reside in Neustadt am Rennsteig or who live in the adjacent community of Kahlert were not eligible for inclusion. Informed consent was provided by the participants or the parents/legal representatives. In the second and third phase of the study, participants who had a proven infection with SARS either by SARS-CoV-2 PCR or antibody positivity in the first phase were invited along with an age, sex, and comorbidity matched control group. Inhabitants of the village who were not invited, however, could also perform antibody testing.

Participants completed a pseudonymized questionnaire directly at the study site during all three rounds. After re-assessing the original case report forms on paper, obvious errors were corrected, and duplicated entries were deleted. Plausibility checks of demographic data were performed. Symptoms were noted if reported. Strength and duration of symptoms was not weighted in the analysis of this manuscript. Self-reported information on a positive SARS-CoV-2 PCR test at the time point of the outbreak/quarantine initiation was double checked with the information by the health department of the Ilm-district if the participants gave their permission on the consent form.

Five serological tests were performed in all three rounds. Characteristics of the tests are provided in Supplementary Table S1 . Two of the tests detected antibodies recognizing the S-antigen, one recognized the 2019-nCoV recombinant antigen and two tests recognized the N-antigen of SARS-CoV-2. Detection of SARS-CoV-2 IgG antibodies was performed with five different quantification methods, of which two were enzyme-linked immunosorbent assays (ELISA) and three were chemiluminescence-based immunoassays (CLIA/CMIA). All tests were carried out according to manufacturers’ instructions. For detailed information on assay characteristics and instruments used, see Supplementary Table S1 . Sensitivities and specificities are shown as provided by the manufacturer. The following assays were used; EDI Novel Coronavirus SARS-CoV-2 IgG ELISA kit (Epitope Diagnostics Inc., San Diego, CA, USA), SARS-CoV-2 IgG ELISA kit (Euroimmun, Lübeck, Germany), SARS-CoV-2 S1/S2 IgG CLIA kit (DiaSorin, Saluggia, Italy), 2019-nCoV IgG kit (Snibe Co., Ltd., Shenzhen, China) and Elecsys Anti–SARS-CoV-2 kit (Roche, Basel, Switzerland).

Peripheral blood mononuclear cells (PBMCs) from 56 inhabitants of the village Neustadt am Rennsteig were analyzed for S-Protein-specific T_H cell response. DMSO as solvent of S-peptide mixes was used as control. S-peptide mixes 1 and 2 represent the S-Protein N-terminal part and C-terminal part, respectively. General reactivity was controlled by SEB/TSST1 stimulation. There were no non-responders. PBMCs were isolated by gradient density centrifugation on Biocoll solution (Bio&SELL GmbH, Feucht, Germany) at 800g at room temperature (RT) for 20 min without brakes. PBMCs were washed with PBS and subsequently cryoconserved in medium containing penicillin/streptomycin, 10% DMSO and 50% FCS (all Sigma-Aldrich, St. Louis, MO, USA). For analysis, PBMCs were thawed and immediately washed with cell culture medium (supplemented with 10% human AB serum [PAN Biotech, Aidenbach, Germany], penicillin/streptomycin). Upon recovery at 37°C for 2h, a maximum number of 5 × 10⁶ PBMCs were restimulated in cell culture medium containing 1 µg/ml recombinant anti-human CD28 antibody (clone CD28.2, BioLegend, San Diego, CA, USA, RRID : AB_314303) and either 0.2% DMSO (negative control), SARS-CoV-2 Spike glycoprotein PepMix 1 (S1, N-terminal coverage) or 2 (S2, C-terminal coverage) (both JPT Peptide Technologies GmbH, Berlin, Germany). As high controls 10⁶ PBMCs were restimulated with 1 µg/ml TSST1 and 1 µg/ml SEB (both Sigma-Aldrich) in presence of 1 µg/ml recombinant anti-human CD28, or with anti-human CD3/CD28 beads (Gibco/Thermo Fisher Scientific, Lithuania) at a ratio of 1 bead/PBMC. After stimulation for 2h, Brefeldin A (BioLegend, San Diego, CA, USA) was added for another 14h of incubation. Cells were shortly incubated with 1 mg/ml beriglobin followed by staining with anti-human CD3 Pacific Blue (clone UCHT1, BioLegend, RRID : AB_1595437) and anti-human CD4 Brilliant Violet 605 (clone RPA-T4, BioLegend RRID : AB_2564391). After 5 min at 4°C in the dark, Zombie Aqua fixable dead cells stain (BioLegend) was added and samples were mixed and incubated for another 10 min at 4°C in the dark. Incubation was stopped with PBA/E and the cells were fixed in 2% Formaldehyde/PBS at RT for 20 min, blocked with 1 mg/ml beriglobin/0.5% Saponin and intracellularly stained with anti-human CD154 APC (clone 24-31, BioLegend, RRID : AB_314832), anti-human CD137 PE/Cy7 (clone 4B4-1, BioLegend, RRID : AB_2207741), anti-human IFNγ APC/Cy7 (clone 4S.B3, BioLegend, RRID: AB_10663412), anti-human TNFα PerCP/Cy5.5 (clone MAb11, BioLegend, RRID: AB_2204081), anti-human IL-4 PE (clone MP4-25D2, BioLegend, RRID: AB_315129), and anti-human IL-17A FITC (clone BL168, BioLegend, RRID: AB_961390) in 0.5% Saponin (Sigma-Aldrich) in PBA/E at 4°C for 20 min. Samples were analyzed on a FACS-Canto-Plus (BD), and data were analyzed with FlowJo V10.7 (BD, Ashland, OR, USA). S-protein–specific CD3⁺CD4⁺ T helper (TH) cells are depicted as CD137⁺CD154⁺ among living CD4⁺CD3⁺. Representative FACS plots are shown in ( Supplementary Figure S2 ).

The Uniform Manifold Approximation and Projection (UMAP) algorithm and the FlowSOM algorithm were used for unsupervised high dimensional flow cytometric analyses of the entire dataset with FlowJo version 10.8.1. Proportionally downsampled single cells/live/CD3⁺/CD4⁺ populations for each FCS file were concatenated in one single FCS file. UMAP was used for dimensionality reduction by using Euclidean as distance function with 15 Nearest Neighbors and 0.5 minimum distance. The following markers were used for building UMAP maps: 4-1BB, CD154, TNF, IFNγ, IL-4, and IL-17a. Resulting UMAP maps were fed into the FlowSOM (34). To identify clusters, heatmaps were built with median fluorescence intensity (MFI) values from each marker.

All statistical analyses were performed in the analysis population as indicated and stratified by age, PCR status, symptomatic disease, sex or sero-status from the serological assays if applicable. Descriptive analyses included the calculation of mean with standard deviation (SD) and medians with minimum and maximum or interquartile range (IQR) values for continuous variables, and absolute counts (n, with percentages) for categorical variables. Owing to the great data completeness, we performed no data imputations. All reported p-values are unadjusted and two sided. Time course experiments were analyzed with the Friedman test and Dunn’s post-hoc test for multiple comparisons. Antibody ratios were compared with the Student’s t-test with Welch correction. Non-parametric estimation of Spearman’s rank correlation was performed with the following strength cutoffs ≥ 0.8 as very strong; ≥ 0.6–0.8 as moderately strong; ≥ 0.3–0.6 as fair and < 0.3 as poor (adapted from (35)). For comparison of two paired samples, the Wilcoxon-matched-pairs signed-rank test as implemented in GraphPAD PRISM 9 was used. When more than two paired groups were compared, non-parametric Kruskal–Wallis with the Dunns post-hoc test was used.

The study was conducted according to the current version of the Declaration of Helsinki and has been approved by the institutional ethics committees of the Jena University Hospital and the respective data protection commissioner (approval number 2020-1776) and the ethics committee of the Thuringian chamber of physicians. All data were collected with unique pseudonyms on paper case report forms. These identifiers were later used to merge the questionnaire information with the laboratory information in an electronic study database. The study is registered at the German Clinical Trials Register: DRKS00022416.

The study flow chart of all three visits of the CoNAN study is shown in Figure 1 . A total of 626 of the 883 community inhabitants (71%) participated in the 1^st round of the study in April 2020. Of those, individuals with a prior positive SARS-CoV-2 PCR and/or antibody titer (“infected”) and age and sex matched non-infected controls were invited via mail to participate in the 2^nd and 3^rd visit. Villagers that were not invited but nevertheless wanted to take part additionally in the 2^nd and 3^rd visit, for example, to be informed about their antibody status or to contribute to the scientific project, could also partake. Thus, of the initial 626 participants, 146 individuals took part in the 2^nd visit in October 2020 and 224 in the 3^rd visit in April 2021. There were 132 individuals that participated in all three rounds. Antibody levels were determined in all of these participants. T-cell analysis was performed in all previously infected participants as well as in randomly chosen previously non-infected individuals. The detailed characteristics of the subjects are given in Tables 1 and 2 and Supplementary Material Table S1 .

A matched analysis of antibody concentrations in 40 participants was conducted from all three time points. These were defined as being “seropositive” if at least two of five performed serological tests. In a matched analysis of T_H cell immunity, we investigated 30 previously infected participants and 16 non-infected controls and excluded 10 participants that had been vaccinated or infected during the survey period. A comparison of the test performance between the five serological IgG assays in the participants is shown in Supplementary Table S1 .

In the population-wide CoNAN 1 study—in this manuscript referred to as 1.5-month time point—52 participants were anti–SARS-CoV-2 antibody seropositive (AB+) (reported in 8). Of these, 44 individuals participated in the 2^nd and 46 individuals in the 3^rd round of the study, respectively. Four participants had been vaccinated during the course of the study and were excluded. The remaining 40 participants were assessed for antibody course and are referred to as the “infected” group.

From the participants of the 3^rd visit, 161 had initially been tested negative (AB-) of which 40 (24, 8%) participants became AB+. Of these, 18 persons had been vaccinated against SARS-CoV-2. For one participant, the information on vaccination history was missing. Nine participants had a PCR-confirmed SARS-CoV-2 infection between the 2^nd and 3^rd visit. Furthermore, 21 participants had not been vaccinated against SARS-CoV-2.

The remaining 40 participants (median age 60.5 years [range 5–83, IQR 51,75–71], male n = 24 [57, 43%], female 18 [42, 86%]) were antibody positive in the first round and had no re-infection or vaccination against SARS-CoV-2. These were assessed in the longitudinal serology study. Participants’ characteristics are provided in Table 1 .

In a first step, we assessed the course of the serum-antibody concentrations over 1 year after SARS-CoV-2 infection. Three of five tests revealed discrete results normalized to a standard for all three time points (1). EDI, recognizing anti-nucleocapsid antibodies; (2). Liason Diasorin, recognizing anti-spike antibodies and (3). Maglumi Snibe, recognizing anti-spike and anti-nucleocapsid antibodies), whereas one test (Euroimmune; recognizing anti-spike antibodies) provided an OD and semi-quantitative data and one test (Roche, recognizing anti-nucleocapsid antibodies) provided qualitative results only. As the missing standardization of the Roche test is a possible bias for tests performed at different time points, the course of the Roche test was not assessed further. During the 1-year observation period, the four quantitive tests showed a significant decline of the serum antibody concentrations ( Figure 2A ). The extent of the decline varied between the individual tests and time points (shown in detail in Figure 2A and Supplementary Figure S1A ). Other authors had shown that anti-nucleocapsid antibodies become undetectable as early as already 8 months infection (36), suggestive of a shorter half-life of this antibody subset (37, 38). However, in our study, the opposite was the case. When comparing the decline of antibody concentrations between the EDI (N) test and the Euroimmune (S) test between 1.5 and 6 months, the decrease in anti-nucleocapsid antibodies was less pronounced (mean_EDI = 0.84; SD_EDI = 0.37 vs. mean_EU = 0.21; SD_EU = 0.10; t-test p < 0.001). This effect persisted after 12 months (mean_EDI = 0.31; SD_EDI = 0.11 vs. mean_EU = 0.21; SD_EU = 0.12; t-test p < 0.001) ( Supplementary Figure S2A ). For the early time points, this was confirmed in the comparison of the EDI test with the Diasorin (S1/S2) test (mean_DS = 0.65; SD_EDI = 0.46; t-test p = 0.05). Yet, after 12 months, the Diasorin assay had a less pronounced decline (mean_DS = 0.76; SD_EDI = 0.62; t-test p < 0.001). The data suggest a rapid waning of serum antibodies detected in some but not all tests during the first 6 months and a less rapid waning and preservation of antibodies within one year after SARS-CoV-2 infection.

In a subgroup analysis, we also stratified the serum antibody concentrations with respect to age, PCR-positive versus PCR-negative participants, the presence or absence of symptoms or the sex of the individuals. For all groups, the time course results remained unaffected by this stratification ( Figures 2B–D ; Supplementary Figures S1B and C ). Notably, except three, all of the participants that initially had been tested seropositive—defined as at least two different positive assays—remained seropositive after one year.

To assess SARS-CoV-2 T cell–mediated immunity, we then analyzed S-protein–specific CD154⁺4-1BB⁺ cells among peripheral blood CD3⁺CD4⁺ T helper (TH) cells. This T cell–mediated immunity was determined in a matched cohort of 46 study individuals that participated in the first and third round of the study ( Figure 1 ). Of these, 16 individuals were non-infected and considered as control cohort. Thirty individuals constituted the infected cohort. Of which, six were PCR-positive only, 10 were PCR-positive and antibody-positive after 1.5 months, and 14 were antibody-positive only.

Characteristics of the participants in the longitudinal T-cell study are provided in Table 2 . To identify the spike-reactive T_H cells, we compared T_H cells restimulated with mixes of peptides covering the N-terminal part (S.Pep1 [N]) or the C-terminal part (S.Pep2 [C]) of the SARS-CoV-2 spike protein to T_H cells responding in presence of DMSO alone ( Figure 3 ). As shown in Figures 3A, B , we could detect the presence of spike-specific CD154⁺4-1BB⁺ T_H cells among all TH cells at 1.5 months and still 12 months after infection ( Figures 3A, B ). Notably, there was a slight, but significant reduction in the frequency of spike-specific T_H cells in the previously infected cohort over time ( Figures 3C, D ). In only two participants (6.7%), the spike-specific T_H cell response had vanished after 1 year ( Figures 3C, D ). When compared with SARS-CoV-2–negative participants, previously SARS-CoV-2–infected participants clearly showed an overall higher frequency of SARS-CoV-2 specific T cells at all time points that were investigated ( Figures 3E, F ). In addition, when assessing the subgroup of antibody-positive (and PCR-negative) participants only, this trend persisted ( Figures 3G, H ). Despite a persistent presence of spike-reactive T_H cells 12 months after infection in this group, a slight decrease in the frequency of these antigen-specific T_H cells among all T_H cells was also detectable ( Figure 3H ). Interestingly, a significantly higher frequency of spike-specific T_H cells was detected in this antibody-only group, when compared with individuals with an initially positive PCR status without measurable antibody titers ( Supplementary Figure S3 ). Notably, several healthy subjects showed a T_H cell response against SARS-CoV-2 already at the beginning of the study, which was maintained over time.

In conclusion, individuals with initially detectable antibody levels showed a higher T_H cell response after 12 months than individuals with an antibody titer below detection threshold despite a PCR-confirmed infection. When we assessed intracellular cytokine expression of INFγ, TNF, IL-4, and IL-17A ( Figures 3I–L ; Supplementary Figure S3 ), we observed that significantly elevated initial levels of INFγ expressing SARS-CoV-2 specific T_H cells in inpreviously infected patients remained elevated up to one year after infection ( Figure 3I and Supplementary Figure S3 ), whereas the expression of TNF, IL-4, or IL-17A was not increased at any time point ( Figures 3J–L and Supplementary Figure S3 ). Overall, despite the slight decrease in particular subgroups, the data suggest that a T_H cell–mediated immunity after SARS-CoV-2 infection prevails for at least 1 year and contains a robust and maintained specific T_H1 cell immunity.

To gain a further unbiased perspective on the immunity toward the spike proteins as represented by the S.Pep.1 (N) or S.Pep.2 (C)–specific T_H response, we performed multi-dimensional flow cytometry analyses. Representative Uniform Manifold Approximation and Projection (UMAP) maps were color coded according to the resulting clusters using the FlowSOM algorithm (34) ( Figure 4 and Supplementary Figure S4 ). With the limitation that only small populations could be analyzed, the global high-dimensional analyses revealed diverse T-cell activation status of SARS-CoV2–infected individuals in time ( Figures 4A, B ). For both S1(N) and S2(C), we observed populations that resembles different phenotypes: (i) resting cells (cluster 1: negative for activation markers and cytokines), (ii) 4-1BB⁺ activated CD4⁺ T cells (cluster 2: 4-1BB⁺IFNγ⁺TNF^low), and (iii) CD154⁺4-1BB⁺ activated CD4⁺ T cells (cluster 3 for S.Pep1 and cluster 4 for S.Pep2: CD154⁺4-1BB⁺IFNγ^highTNF^low). Moreover, a S2(C)-specific population depicting a CD154^low4-1BB^high phenotype was observed (CD154^low4-1BB^highIFNγ^highTNF^high) ( Figure 4B ). For both S1(N) and S2(C), we only observed a marginal contribution of IL-4 and IL-17A. Despite obtaining these clusters, we did not detect any significant differences in infected individuals between 1.5 months and 12 months. This unbiased result supports our notion that the T_H-specific response is maintained over time after SARS-CoV-2 infection.

We then asked whether the 12-month antibody levels could be predicted by the T cell immune response mounted after 1.5 months and vice versa. Therefore, we correlated antibody concentrations and T cell immunity at 1.5 months and 12 months after infection. We assessed the Spearman correlation for the quantitative Diasorin, the Snibe, the EDI as well as for the semiquantitive Euroimmune assays with the percentage of the S1(N)-specific CD154⁺4-1BB⁺ T_H cells ( Figures 5A–D ) and the S2(C)-specific CD154⁺4-1BB⁺ T_H cells ( Supplementary Figures S5A–D ). None of the comparisons revealed a strong correlation. There was a moderate correlation (Spearman r = 0.6–0.8) of the serological tests at 1.5 months to T_H cell responses at 1.5 months and 12 months after infection ( Figures 5A–D ; Supplementary Table S3 ). When including all participants with T-cell data regardless of the antibody status, this prevailed for the 1.5 months correlation. However, S2(C)-specific T_H cells moderately correlated with antibody concentrations for the EDI test at 1.5. months (r_EDI = 0.61; p < 0.0001) ( Supplementary Figure S5A ) and the IgG index of the Euroimmune assay (r_EU = 0.62; p < 0.0001) ( Supplementary Figure S5D ). The correlation to S1(N)-specific T_H cells was only weak (r_EDI = 0.39, p = 0.05; r_EU = 0.48, p = 0.0079 ( Figures 5A, D ).