key: cord-0747627-i5k5o3c6
authors: Ortega, N.; Ribes, M.; Vidal, M.; Rubio, R.; Aguilar, R.; Williams, S.; Barrios, D.; Alonso, S.; Hernandez-Luis, P.; Mitchell, R. A.; Jairoce, C.; Cruz, A. M.; Jimenez, A.; Santano, R.; Mendez, S.; Lamoglia, M.; Rosell, N.; Llupia, A.; Puyol, L.; Chi, J.; Rodrigo, N.; Parras, D.; Serra, P.; Mayor, A.; Barroso, S.; Varela, P.; Vilella, A.; Trilla, A.; Santamaria, P.; Carolis, C.; Tortajada, M.; Izquierdo, L.; Angulo, A.; Engel, P.; Garcia-Basteiro, A.; Moncunill, G.; Dobano, C.
title: Seven-month kinetics of SARS-CoV-2 antibodies and protective role of pre-existing antibodies to seasonal human coronaviruses on COVID-19
date: 2021-02-23
journal: nan
DOI: 10.1101/2021.02.22.21252150
sha: 81be0706d73d0f49bafe62282e10a4b0c4ac1124
doc_id: 747627
cord_uid: i5k5o3c6

Unraveling long-term kinetics of antibodies to SARS-CoV-2 and the factors influencing its course, like prior antibody levels to human coronaviruses causing common cold (HCoVs), is essential to understand protective immunity and effective surveillance strategies. Antibody levels against six SARS-CoV-2 and four HCoV antigens were quantified by Luminex, and antibody neutralization capacity was assessed by flow cytometry in a cohort of health care workers followed-up for 6 months. Seroprevalence increased over time from 13.5% (month 0) and 15.6% (month 1), to 16.4% (month 6). Levels of antibodies were stable over time, except IgG against nucleocapsid antigen and IgM levels that waned. After the peak response, anti-spike antibody levels increased from ~150 days post-symptom onset in all individuals (73% for IgG), in the absence of any evidence of re-exposure. The neutralizing capacity of antibodies was maintained. Pre-existing antibodies to alpha-HCoV were lower in individuals who subsequently seroconverted for SARS-CoV-2. IgG and IgA to HCoV were significantly higher in asymptomatic than symptomatic seropositive individuals. Thus, pre-existing cross-reactive HCoVs antibodies could have a protective effect against SARS-CoV-2 infection and COVID-19 disease.

Coronavirus disease 2019 , caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), has already caused a loss of 2.3 M lives globally 1 . Since its emergence, a key priority has been the understanding of the kinetics and protective role of the immune response in the population, to assess the degree of exposure in serosurveys and to understand immunity to the virus. This knowledge guides vaccine development, selection of donors for hyperimmune serum-transfusion therapies, and combining antigens with the highest immunogenic and neutralizing capacity to improve surveillance interventions.

Longitudinal studies assessing SARS-CoV-2 antibody kinetics have found that IgA and IgM peak between week 3 and 4 post symptoms onset (PSO) and wane thereafter, with IgA persisting longer than IgM 2-7 . IgA and IgM seroreversion was estimated between days 71 and 49, respectively 8 , but IgA has also been found to remain detectable 6 months post infection and to be less affected by the decay than IgM 9, 10 . Several studies have observed relatively stable levels of IgG to the spike (S) protein after three 11, 6 , four 12, 13 and six to eight months 2, 9, [14] [15] [16] . However, others reported that IgG only lasted around 3-4 months PSO 17, 18 .

Many studies consistently observe that IgG to the nucleocapsid (N) protein, found inside the virus or infected cells, decay faster than IgG to S, being a marker of a most recent infection but less sensitive for assessing population seroprevalence 2,13-20 . While antibodies targeting N protein are unlikely to directly neutralize SARS-CoV-2, those targeting S, responsible for the interaction with the ACE2 receptor in the host cells, are considered the main neutralizers 21 .

Studies up-to-date point that neutralizing antibodies (nAbs) strongly correlate with antibody titers to S 16, [19] [20] [21] [22] and that disease severity is associated with higher neutralizing antibody capacities [23] [24] [25] [26] .

Understanding the extent of antibody cross-reactivity with other human coronaviruses (HCoV) is important to elucidate the impact of such pre-existing antibodies on COVID-19 immunity. Four low-pathogenic HCoV causing common cold have circulated among humans for at least 100 years: the alphacoronaviruses 229E and NL63, and the betacoronaviruses OC43 and HKU1. They account for about 10% of all acute respiratory tract infections, and . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. ; https://doi.org/10.1101/2021.02.22.21252150 doi: medRxiv preprint thus, a substantial proportion of the global population is expected to carry antibodies against them 27, 28 . Previous studies found some cell-mediated 29, 30 and antibody cross-reactivity of HCoV immune responses with SARS-CoV-2 [31] [32] [33] . Regions within N and S antigens with high amino acid homology between SARS-CoV-2 and HCoV are potential targets of crossreactive antibodies [32] [33] [34] , and could exert cross-protective effects against SARS-CoV-2 infection and/or disease. Prior studies have not found protection against infection, as participants with recent documented infection with an endemic HCoV had similar rates of SARS-CoV-2 acquisition than those without recent HCoV infection [35] [36] [37] . Regarding antidisease protection, COVID-19 patients with a recent HCoV diagnosis had statistically significant lower odds for COVID-19 intensive care unit admission and death 37 , but other studies did not find any association between confirmed prior history of seasonal HCoVs and COVID-19 severity 35, 36 . Some recent studies have suggested that this pre-existing immunity would not confer cross-protection but, rather, be responsible for an immunological imprinting or 'original antigenic sin', a phenomenon well studied for influenza virus infections. This suggests that the immune system privileges recall of existing memory responses -in this case of HCoV-, in detriment of stimulating de novo responses -here to SARS-CoV-2-leading to poor outcomes or severe disease 30, 31 . The possibility of antibodies to HCoVs acting as antibody-derived enhancement (ADE) has also been reviewed and the most recent evidence shows no clinical, in vitro or animal evidence 38, 39 . Disentangling the role of pre-existing HCoVs antibodies on anti-SARS-CoV-2 responses may have implications in the deployment of potentially effective vaccines, as well as for the interpretation of serological studies.

At the beginning of the pandemic, healthcare workers (HCW) were considered to be at a higher risk of SARS-CoV-2 infection than the general population, although there is now evidence that seroprevalence is similar when using adequate personal protective equipment.

We previously observed 9.3% (95% CI, 7.1-12.0) SARS-CoV-2 seroprevalence in a random cohort of 578 HCW from Hospital Clínic in Barcelona (HCB) between March-April 2020 40 , and of 14.9% after a month follow-up 4 , based on the detection of antibodies to one antigen . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint In the present study, we aimed to characterize the antibody kinetics and neutralization capacity between March and October 2020 at four cross-sectional surveys and estimate the seroprevalence in the same cohort of HCW. For this analysis, we measured IgM, IgG and IgA isotypes against an expanded panel of six SARS-CoV-2 antigens and tested crossreactivity with the four endemic HCoVs (HKU1, 229E, OC43 and NL63) to assess its potential impact on COVID-19 protection.

From the initial cohort, 507 individuals participated in a fourth visit (M6) six months after baseline (12.3% lost to follow-up). Mean age was 42.7 (SD: 11.2) and 72% were female. Full demographic characteristics at baseline (M0), one (M1) and three (M3) month follow up visits were as described 4, 40 . is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Table 2 ). Age, sex and other variables were not found to be associated with SARS-CoV-2 infection. Sixty-nine percent of the infections were symptomatic and a single participant required hospitalization in our cohort.

Levels of SARS-CoV-2 antigen specific isotypes (IgM, IgA, IgG) were plotted against time with up to four observations with a maximum 7.7 months PSO, in a total of 235 samples from 76 symptomatic participants (Figure 1 ). Kinetic curves were very similar when plotted against days since positive rRT-PCR in participants who were asymptomatic or symptomatic (Supplementary Figure 1) .

IgA or IgM peaked within the first month PSO, while IgG peaked around day 50 PSO. SARS-CoV-2 IgG levels were generally steady for S antigens (S, S1, S2 and RBD) and for IgA up to 230 days PSO (71% and 69% of the participants remained seropositive six months PSO, respectively), and waned at a clearly slower rate than IgM (34% of the participants remained seropositive) and IgG to N-related antigens (26% of the participants remained seropositive).

Antibody levels were observed to increase from ~150 days PSO onwards (Figure 1) . To further explore this, we grouped participants based on their antibody levels at M6 compared to the previous visit (M1 or M3). We only considered participants who had already shown a decrease in antibodies after the peak response. We therefore calculated an "antibody is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. ; https://doi.org/10.1101/2021.02.22.21252150 doi: medRxiv preprint increase index" between both visits for each antigen-isotype combination and labelled the individuals as "decayers'' when the ratio of antibody levels between both visits was <1, and as "sustainers/increasers'' when the ratio was ≥ 1, in line with the methodology by Chen et al. 17 . Increased levels were observed in all antigen-isotype combinations (Figure 1) . Most sustainers/increasers had a boost for more than one antigen-isotype pair, as assessed by a Venn diagram (Supplementary Figure 2) . Levels at seroconversion visit were higher in decayers than sustainers/increasers, being statistically significant for N IgG, S2 IgG and S1

IgM (Supplementary Figure 3a) . There was no association of the antibody increase index or being a sustainer/increaser or a decayer with age. We observed a trend towards having a higher antibody increase index, mainly for IgG, in participants who reported current or past symptoms at M6 since last visit, several months after COVID-19 disease recovery (Supplementary Figure 3b) . Of note, none of these individuals reported new infections and we did not find any association between having reported a contact with a COVID-19 case at M6 and the antibody increase index (data not shown). We also identified a trend towards higher antibody increase index in participants with shorter duration of symptoms (<10 days) compared to those who had symptoms for >10 days (Supplementary Figure 3c) . We explored all the variables in univariable and multivariable logistic regression models and none were robustly associated with being a sustainer/increaser or a decayer (data now shown).

Plasma neutralizing capacity measured as RBD-ACE2-binding inhibition generally increased between the day of onset of symptoms until day 80 and remained stable thereafter up to 250 days PSO (Figure 2) . We correlated the antibody neutralizing capacity and levels at the different study visits. At the first cross-sectional visit (M0, mean days PSO=20) levels of all three Ig isotypes against RBD and S antigens correlated with neutralization capacity (rs=0.19-0.32, p<0.05), while the correlation between antibody levels against N and RBD-ACE2 neutralization did not reach statistical significance (Figure 3a) . At the fourth cross- is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. ; https://doi.org/10.1101/2021.02.22.21252150 doi: medRxiv preprint sectional visit (M6, mean days PSO=200), IgM levels to any antigen did not correlate with neutralization percentage, whilst IgG and IgA levels against all six antigens showed moderate to strong correlations (rs=0.24-0.76, p<0.05), with higher correlations for S antigens (Figure 3b ). We performed a PCA for all antigen-isotype pairs (Supplementary Figure 4) and the first 5 components, explaining 75.12% of the variance were included as predictors in a model with neutralizing capacity as an outcome (P<0.05, adjusted R 2 0.575). Component 1 and 5 were significantly associated with neutralizing capacity (Supplementary Table 3 ). In these components, S and S1 IgG, and S2 IgM, contributed to an increase in the neutralization activity, whilst N C-terminal IgG negatively influenced it (P<0.001). We observed that antibodies to S antigens were highly contributing to the prediction of the neutralization percentage (component 1, longer vectors).

We did not find any significant difference in neutralizing capacity between sustainers/increasers and decayers. The neutralizing capacity was also not associated with the antibody increase index, except for IgM increase index that inversely correlated with the neutralization percentage at M0 and after six months PSO (Supplementary Figure 5) .

Pre-pandemic plasma samples had some antibody reactivity against SARS-CoV-2 antigens, particularly against N protein, and levels of antibodies against N from SARS-CoV-2 positively correlated with antibodies to HCoV N antigens (to a lesser extent for IgM), indicating crossreactivity between them (Supplementary Figure 6) . Therefore, we analyzed the antibody levels against HCoV N antigens prior and after SARS-CoV-2 infection in the 33 participants who seroconverted during the study period. While some participants showed stable anti-HCoV N antibody levels, a general upward trend was observed. IgG to 229E and NL63, and IgM to 229E and HKU1, significantly increased after SARS-CoV-2 seroconversion. Considering that not all seroconverters had an increase in levels supports a back-boost of N HCoV beyond cross-reactivity (Supplementary Figure 7) . is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint We investigated whether having higher baseline anti-HCoV N antibody levels could be protective against SARS-CoV-2 infection. Overall, we observed a consistent trend towards higher baseline IgG levels to alpha-HCoV 229 (p=0.06) and NL63 (p=0.15) in participants who did not seroconvert compared to seroconverters, although these differences did not reach statistical significance (Figure 4a) . We assessed whether having higher anti-HCoV N antibody levels prior to infection could confer protection against COVID-19 symptoms in participants who seroconverted during the study period. Although statistical significance was only reached for IgA against OC43, we observed a common trend towards higher levels of anti-HCoV N IgA and IgG in asymptomatic than symptomatic SARS-CoV-2 seropositive participants ( Figure 4b) . Consistently, levels of all three isotypes against alpha HCoVs and of IgA to OC43 experienced a higher fold-increase after SARS-CoV-2 infection in asymptomatic than symptomatic seroconverters (p<0.05) (Figure 4c ), suggesting that a back boost -beyond cross-reactivity-in anti-HCoV antibody levels could confer disease-protective immunity. In line with this finding, seropositive asymptomatic participants had significantly higher IgG levels against all four HCoVs than symptomatic participants in the first visit after SARS-CoV-2 positivity (Figure 4d ). In contrast, anti-SARS-CoV-2 N antibody levels were higher in symptomatic seropositive participants (p<0.05) (Figure 4e ).

Finally, we tested whether baseline anti-HCoV antibody levels impacted de novo production of antibodies to SARS-CoV-2. To test this hypothesis we correlated the increase in anti-N SARS-CoV-2 antibody levels from baseline to seroconversion for the three isotypes against is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint 

To our knowledge, this is the first longitudinal study to assess the antibody response to such a wide panel of antigens from SARS-CoV-2 and HCoV, up to 7.7 months after infection, and the first to show evidence of COVID-19 protection by pre-existing HCoV antibodies. This is important to track the evolution of the immunity in asymptomatic and mild/moderate cases, particularly in an indispensable population like HCW, and to understand why some people may be less affected by COVID-19. A strength of the present study is the availability of sequential sampling within a random cohort including asymptomatic and symptomatic subjects.

Importantly, we observed a trend towards higher levels of antibodies against HCoVs N proteins at baseline in those participants who did not become infected with SARS-CoV-2, suggesting some level of cross-protection against infection. Moreover, asymptomatic SARS-CoV-2 seropositive participants tended to have higher anti-HCoV N IgA and IgG levels prior to seroconversion than symptomatic participants, suggesting cross-protection against disease. In addition, asymptomatic seropositive participants had significantly higher anti-HCoV N IgG levels after infection than symptomatics, pointing towards a disease-protective back-boost of anti-HCoV antibodies. Combined with the observation that higher baseline anti-HCoV N antibody levels correlated with less de novo anti-SARS-CoV-2 N antibody production, we propose a protective effect of previous exposure to HCoVs, which could be the result of a diminished exposure (decreased viral load) due to the suggested protective role of anti-HCoV antibodies. Other studies have reported a lack of anti-disease crossprotection [35] [36] [37] ; and some studies have associated severe COVID-19 with a back-boosting of antibodies against S2 from betacoronaviruses 31 , and N and S from OC43 30 . However, these studies included only hospitalized patients, as opposed to our cohort that included mainly asymptomatic and participants with mild/moderate symptoms.

We show a cumulative prevalence of SARS-CoV-2 infection of 19.6% (95% CI 16.4-23.0%) after six months of follow-up (October 2020). The cumulative prevalence around May 2020, corresponding to our second visit (M1), recalculated here with a wider antigen panel, was . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint [41] [42] [43] . Around 28% of the total infections detected throughout the follow up were newly diagnosed after the first visit (M0), which would reveal that infections in the hospital setting mostly happened within the first pandemic wave. No re-infections were reported in our cohort and this could be related to the induction and maintenance of robust neutralizing antibodies along the study period, in contrast with another study in a cohort of 173 primary HCW in which 4 reinfections were reported 44 . Surprisingly, only 56% of participants with evidence of infection by serology had a positive rRT-PCR, highlighting that almost half of the infections went under-detected, mainly during the baseline visit (only 49% had a previous positive rRT-PCR) and going up to 73% of rRT-PCR detection rate in the following visits. We observed a high seroreversion rate for IgA and IgM at visits M1 and M3, decreasing at visit M6. This finding reinforces the rapid decay below the seropositivity threshold of these two isotypes compared to IgG, for which only 9 participants seroreverted between M1 and M6 visits. Although some reports have pointed to a higher antibody decay in HCW with mild symptoms 45 , our results show that IgG levels are maintained up to 7.7 months PSO, in line with other studies 2, 9, [14] [15] [16] . Interestingly, IgA levels were maintained in those individuals who did not serorevert during the first 3 months PSO. Furthermore, IgG to N C-terminal rapidly decreased below the positivity threshold, as seen in other studies 2,13-20 .

However, the vast majority of participants with a previous infection remained seropositive for S-related antigens. This finding is of special relevance because RBD and S IgG antibody levels have been shown to correlate with neutralizing activity and S is the main target of currently deployed COVID-19 vaccines and most products under development.

Remarkably, we noticed a pronounced increase in S-related IgG levels from day 150 PSO onwards in 34/46 (73.9%) participants. Previous studies that reached 150 days of follow-up have not highlighted this phenomenon 6, 14, 16 , but it was observed in Figueiredo-Campos et al. 9 . Chen et al. assessed a subset of individuals with stable or increasing antibody levels at day ~100 17 . In our study, nearly all increasers showed the boost in levels for more than one antigen-isotype pair, in line with the results observed by Chen et al. 17 . We also found a . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. ; https://doi.org/10.1101/2021.02.22.21252150 doi: medRxiv preprint consistent tendency pointing to shorter duration of symptoms in participants with higher increase indices, labeled as quick healers, independently from their age. In contrast with their work, we found statistically significant differences in SARS-CoV-2 antibody levels at seroconversion, with decayers showing higher levels compared to sustainers-increasers for N IgG, S2 IgG and S1 IgM. The increase in antibody levels in recovered participants could be related to a natural boost after a re-exposure, although we do not have any evidence of reinfection, and sustainers/increasers did not report more contacts with positive cases than decayers. A similar late increase in antibody levels has been reported in a study describing immunity to Ebola virus, showing a pattern of decay-stimulation of antibody production in survivors who had been neither re-exposed nor vaccinated, and had been asymptomatic is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint The main limitations of this study are that our cohort had few participants with severe disease, and that we only assessed the impact of anti-HCoV N antibodies on SARS-CoV-2 response, while anti-N antibodies are not expected to have neutralizing capacity. However, it is likely that sera with high levels of N HCoV antibodies would also have high levels of antibodies targeting S antigens and B and T cells specific to HCoV, which could explain the potential association with a protective effect. All together, further studies will be needed to 

We measured the levels of antibodies to SARS-CoV-2 antigens in blood samples of 578 randomly selected HCW from HCB followed up at four visits: baseline -hereby termed "M0"-(month 0, March 28 th to April 9 th 2020, n = 578), "M1" (month 1, April 27 th to May 6 th 2020, n = 566), "M3" (month 3, July 28 th to August 6 th 2020) when only participants with previous evidence of infection were invited (n=70), and "M6" (month 6, Sept 29 th to Oct 20 th 2020, n = 507) (12.3% lost to follow-up). We collected retrospective data on symptoms in order to set is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. ; https://doi.org/10.1101/2021.02.22.21252150 doi: medRxiv preprint the beginning of the disease, and the longest period since symptoms onset was 231 days (7.7 months).

The study population included HCW who delivered care and services directly or indirectly to patients, as described 4, 40 . We collected nasopharyngeal swabs for SARS-CoV-2 rRT-PCR at M0 and M1 and a blood sample for antibody and immunological assessments at all visits.

SARS-CoV-2 detection by rRT-PCR followed the CDC-006-00019 CDC/DDID/NCIRD/ Division of Viral Diseases protocol, as previously described 4, 40 . Participants isolated at home due to a COVID-19 diagnosis or on quarantine, were visited at their households for sample and questionnaires collection.

Written informed consent was obtained from all study participants prior to study initiation. The study was approved by the Ethics Committee at HCB (Ref number: HCB/2020/0336). Data for each participant were collected in a standardized electronic questionnaire as described 40 .

IgM, IgG and IgA antibodies to the full length SARS-CoV-2 S protein, its subregions S1 and S2, RBD that lies within the S1 region, the N full length protein and its specific C-terminal region, and the full length N protein of the HCoVs HKU1, 229E, OC43 and NL63, were measured by Luminex (Supplementary Information) based on a previously described protocol 49 . Sequential plasma samples from the same individual were tested together. Assay positivity cutoffs specific for each isotype and analyte were calculated as 10 to the mean plus 3 standard deviations (SD) of log10-transformed mean fluorescence intensity (MFI) of 129 pre-pandemic controls. Results were defined as undetermined when the MFI levels for a given isotype-analyte were between the positivity threshold and an upper limit at 10 to the mean plus 4.5 SD of the log10-transformed MFIs of pre-pandemic samples, and no other isotype-antigen combination was above the positivity cutoff and the participant did not have any previous evidence of seropositivity or rRT-PCR positivity.

. CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Percentage of inhibition of RBD binding to ACE2 by plasma was analyzed through a flow cytometric-based in vitro assay as detailed in the Supplementary Information. Briefly, a murine stable cell line expressing the ACE2 receptor was incubated with RBD-mFc fusion protein, composed of RBD fused to the Fc region of murine IgG1, previously exposed to the different plasma samples at a dilution 1/50. Cells were stained with anti-mouse IgG-PE, washed, and analyzed by flow cytometry using standard procedures. One hundred and one samples were tested along positive and negative pre-pandemic controls, in duplicates.

Prevalence of SARS-CoV-2 antibodies or SARS-CoV-2 infection confirmed by rRT-PCR, and cumulative prevalence of past or current infection (positive SARS-CoV-2 rRT-PCR and/or antibody seropositivity at any time point) were calculated as proportions with 95% CI.

We tested the association between variables with the Chi-square or Fisher's exact test for categorical variables, and with the Wilcoxon Sum Rank test for continuous variables. Paired Samples Wilcoxon Test was used for paired continuous data. We assessed the relationships between continuous variables using linear regression models and Spearman's rank correlation test. Locally estimated scatterplot smoothing (LOESS) was used to visualize trends in antibody levels over days PSO or post rRT-PCR diagnosis.

A Venn diagram was created to illustrate the overlap between anti-N full-length protein, anti-N C-term, anti-RBD, anti-S, anti-S1, anti-S2 in the Sustainer/Increaser groups 50 .

Univariable and multivariable linear regression models were run to assess factors associated with SARS-CoV-2 antibody levels and prevalence. We additionally explored the association between the SARS-CoV-2 antibody levels and the percentage of neutralization of RBD at month 6 in a Principal Components Analysis (PCA) that included all isotype/antigen pairs. Before the PCA, we confirmed the adequacy of the analysis by testing the colinearity of the variables with the Kaiser-Meyer-Olkin analysis (>0.5) and the Bartlett's sphericity test (p < 0.001). The number of factors chosen was based on eigenvalues >1 that explained >75% of the total variance. To investigate the relationships between HCoV levels and a subset of is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. ; https://doi.org/10.1101/2021.02.22.21252150 doi: medRxiv preprint variables with clinical outcomes and SARS-CoV-2 antibody levels, we built multivariable logistic and linear models, respectively, for those participants for whom we had a sample prior to seroconversion.

A P-value of ≤ 0.05 was considered statistically significant and 95% and CIs were calculated for all estimates. We performed the statistical analysis in R version 4.0.3 (packages tidyverse and corrplot). is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. 

The authors declare no competing interests. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. Dashed line represents the positivity threshold. Participants were grouped based on their antibody levels at M6 compared to the previous visit, individuals were labelled for each isotype-antigen pair as "Decayers" (pink) when the ratio of antibody levels between both visits was <1 and as "Sustainers/Increasers" (light blue) when the ratio was ≥ 1 and grey when the classification was not applicable. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Figure 2 . Venn Diagram illustrating the overlap between antigen-specific IgGs in the "Sustainers/Increasers'' group. Participants were grouped based on their antibody levels at M6 compared to the previous visit, individuals were labelled for each isotypeantigen pair as "Decayers" when the ratio of antibody levels between both visits was <1 and as "Sustainers/Increasers" when the ratio was ≥ 1. Here, we only represent participants who classify as "Sustainers/Increasers'' for IgG against each of the studied antigens (n=34). Between parentheses are the number of sustainers/Increasers seropositive for IgG against the indicated antigen. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Figure 3 . Comparisons of serological and clinical characteristics between sustainers/increasers and decayers. a) Differences in antibody levels at seroconversion between sustainers/increasers and decayers (median fluorescence intensity, MFI) of IgA, IgG and IgM against each antigen (Nucleocapsid (N), and its C-terminal domain, the Receptor Binding Domain (RBD), full S protein and its subregions S1 and S2)). b) Differences in antibody increase index (represented in log scale) between seropositive participants who reported symptoms and those who did not in month 6 (M6) after recovering from a previous SARS-CoV-2 infection. c) Differences in antibody increase index (represented in log scale) between seropositive symptomatic participants who reported less than 10 days of symptom duration and those who reported more than 10 days. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. ; https://doi.org/10.1101/2021.02.22.21252150 doi: medRxiv preprint Supplementary Figure 5 . a) Sample collection timeline. b) Correlations between antibody increase index and neutralization capacity. Spearman's rank correlation test between antibody increase index (MFI increase between M6 and previous visit) of IgA, IgG and IgM against each study antigen (Nucleocapsid full length protein (N), and its C-terminal domain, the Receptor Binding Domain (RBD), full S protein and its subregions S1 and S2), and the plasma neutralization capacity at M0 (as a percentage of RBD-ACE2 binding inhibition). Pvalues and rs correlation coefficients are color-coded for each antigen/isotype pair. Colored lines represent the fitted curve calculated using the linear model method. Shaded areas represent 95% confidence intervals. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. ; https://doi.org/10.1101/2021.02.22.21252150 doi: medRxiv preprint Supplementary Figure 8 . Linear regression models showing the relation between antibody levels against the four HCoV N proteins (three isotypes added) and the anti-SARS-CoV-2 N antibody ratio of seroconversion (seroconversion levels/baseline levels) for all three isotypes. P-values and rs correlation coefficients are given for each isotype. Black lines represent the fitted curve calculated using the linear model method. Shaded areas represent 95% confidence intervals. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 23, 2021. ; https://doi.org/10.1101/2021.02.22.21252150 doi: medRxiv preprint

COVID-19) Dashboard. World Health Organization

Immunological memory to SARS-CoV-2 assessed for up to 8 months after infection

Kinetics and Isotype Assessment of Antibodies Targeting the Spike Protein Receptor Binding Domain of SARS-CoV-2 In COVID-19 Patients as a function of Age and Biological Sex

SARS-CoV-2 Seroprevalence and Antibody Kinetics Among Health Care Workers in a Spanish Hospital After 3 Months of Follow-up

IgA-Ab response to spike glycoprotein of SARS-CoV-2 in patients with COVID-19: A longitudinal study

Longitudinal evaluation and decline of antibody responses in SARS-CoV-2 infection

Durability of neutralizing antibodies and T-cell response post SARS-CoV-2 infection

Seroprevalence of anti-SARS-CoV-2 antibodies in COVID-19 patients and healthy volunteers up to 6 months post disease onset

Evolution of Antibody Immunity to SARS-CoV-2

Persistence of serum and saliva antibody responses to SARS-CoV-2 spike antigens in COVID-19 patients

SARS-CoV-2-specific antibody detection for sero-epidemiology: a multiplex analysis approach accounting for accurate seroprevalence

Kinetics of SARS-CoV-2 specific antibodies (IgM, IgA, IgG) in nonhospitalized patients four months following infection

Stable neutralizing antibody levels six months after mild and severe COVID-19 episode

Orthogonal SARS-CoV-2 Serological Assays Enable Surveillance of Low-Prevalence Communities and Reveal Durable Humoral Immunity

Comprehensive analysis of SARS-CoV-2 antibody dynamics in New Zealand

Quick COVID-19 Healers Sustain Anti-SARS-CoV-2 Antibody Production

SARS-CoV-2 induces a durable and antigen specific humoral immunity after asymptomatic to mild COVID-19 infection

Homologous and heterologous antibodies to coronavirus 229E, NL63, OC43, HKU1, SARS, MERS and SARS-CoV-2 antigens in an age stratified crosssectional serosurvey in a large tertiary hospital in The Netherlands

Antibody Immunological Imprinting on COVID-19 Patients

Analysis of Serologic Cross-Reactivity Between Common Human Coronaviruses and SARS-CoV-2 Using Coronavirus Antigen Microarray

Pre-existing and de novo humoral immunity to SARS-CoV-2 in humans

Immunogenicity and crossreactivity of antibodies to SARS-CoV-2 nucleocapsid protein

Seasonal human coronavirus antibodies are boosted upon SARS-CoV-2 infection but not associated with protection

SARS-CoV-2 infection and COVID-19 severity in individuals with prior seasonal coronavirus infection

Recent endemic coronavirus infection is associated with less-severe COVID-19

A perspective on potential antibody-dependent enhancement of SARS-CoV-2

Antibody-dependent enhancement and SARS-CoV-2 vaccines and therapies

IgA (%)

IgG (%)

the S1 (aa 1-681, expressed in Expi293 and His tag-purified) and S2 (purchased from SinoBiologicals), the receptor binding domain (RBD) (fused with C-terminal 6xHis and StrepTag purification sequences and purified from supernatant of lentiviral-transduced CHO-S cells cultured under a fed-batch system), the nucleocapsid full protein (N) and the specific N C-terminal region, and the four HCoV N full length proteins (expressed in E. coli and His tag-purified). Assay performance was previously established as 100% specificity and 95.78% COOH-microspheres

Equal amounts of each antigen-coupled microspheres were multiplexed and stored at 2000 microspheres/μL at 4°C, protected from light

Twenty five μL of goat anti-human IgG-phycoerythrin (PE) (GTIG-001, Moss Bio) diluted 1:400, goat anti-human IgA-PE (GTIA-001, Moss Bio) 1:200, or goat anti-human IgM-PE (GTIM-001, Moss Bio) 1:200 in Luminex buffer were added to each well and incubated for 30 min. Plates were washed and microspheres resuspended with 80 μL of Luminex Buffer, covered with an adhesive film and sonicated 20 seconds on sonicator bath platform, before acquisition on the Flexmap 3D reader. At least 50 microspheres per analyte per well were acquired, and median fluorescence intensity (MFI) was reported for each analyte. Neutralizing antibodies The stable cell line 300.19-ACE2 was obtained by transfecting 300.19 cells with a plasmid encoding human ACE2 cDNA (SinoBiological) with an Amaxa cell line Nucleofector kit V, followed by hygromycin selection and subsequent subcloning. RBD-mFc fusion protein, containing RBD fused to the Fc region of murine IgG1 was obtained by cloning RBD amplified from the pcDNA3-SARS-CoV-2-S-RBD-Fc (Addgene) into the PFUSE-mIGg1-Fc1 (InvivoGen). HEK-293T cells were transiently transfected with the RBD-mFc

19-ACE2 cells per well in a 96-well plate were incubated for 30 min at 4°C with 4 mg/mL of RBD-mFc fusion protein previously exposed to diluted plasma (1:50) for 30 min at 4°C. Samples were stained with anti-mouse IgG-PE (Jackson ImmunoResearch), washed, and analyzed by Flow cytometry using standard procedures. Samples were acquired with a FACSCanto II (BD Biosciences) and analyzed with FlowJo

Supplementary References

Highly sensitive and specific multiplex antibody assays to quantify immunoglobulins M, A and G against SARS-CoV-2 antigens

Suppression of non-specific binding in serological Luminex assays

Heterophile antibody interference in a multiplexed fluorescent microsphere immunoassay for quantitation of cytokines in human serum

Subversion of natural killer cell responses by a cytomegalovirus-encoded soluble CD48 decoy receptor

We thank the participation of health care workers who are committed with this study as well as with their professional activity that has a positive impact in the society, even more relevant during the pandemic. We are grateful to Eugénia Chóliz, Pau Cisteró, Antía Figueroa-Romero, Silvia Folchs, Jochen Hecht, Mikel J. Martínez, Núria Pey, Patricia Sotomayor and Sara Torres who participated in the field and/or laboratory work during previous visits. We

*Only participants with previous evidence of SARS-CoV-2 infection were invited to visit M3, thus, no seroprevalence or seroconversion data are presented for this visit.