Recent infection with HCoV-OC43 may be associated with protection against SARS-CoV-2 infection

Summary Antibodies against seasonal human coronaviruses (HCoVs) are known to cross-react with SARS-CoV-2, but data on cross-protective effects of prior HCoV infections are conflicting. In a prospective cohort of healthcare workers (HCWs), we studied the association between seasonal HCoV (OC43, HKU1, 229E and NL63) nucleocapsid protein IgG and SARS-CoV-2 infection during the first pandemic wave in the Netherlands (March 2020 – June 2020), by 4-weekly serum sampling. HCW with HCoV-OC43 antibody levels in the highest quartile, were less likely to become SARS-CoV-2 seropositive when compared with those with lower levels (6/32, 18.8%, versus 42/97, 43.3%, respectively: p = 0.019; HR 0.37, 95% CI 0.16–0.88). We found no significant association with HCoV-OC43 spike protein IgG, or with antibodies against other HCoVs. Our results indicate that the high levels of HCoV-OC43-nucleocapsid antibodies, as an indicator of a recent infection, are associated with protection against SARS-CoV-2 infection; this supports and informs efforts to develop pancoronavirus vaccines.


INTRODUCTION
The ongoing SARS-CoV-2 pandemic is characterized by a large individual variability in the risk of contracting infection and subsequent disease severity (Hu et al., 2021;Liu, 2021). Vaccination efforts have been successful in protecting individuals against symptomatic infection and especially severe disease, but sustaining long term protection remains a problem, especially in the light of emerging immune-evasive variants (Hoffmann et al., 2022;Lin et al., 2022b). The potential of cross-protection against SARS-CoV-2 infection elicited by previous infections with seasonal human coronaviruses (HCoVs) is therefore of great interest, but studies have yielded conflicting results (Anderson et al., 2021;Dugas et al., 2021;Ladner et al., 2021;Lin et al., 2022a;Ortega et al., 2021;Sagar et al., 2021;Song et al., 2021).
In this study we prospectively followed a cohort of health care workers (HCW) with different levels of exposure to SARS-CoV-2, and assessed the association between levels of pre-existing HCoV antibodies, incidence of SARS-CoV-2 infection over time, disease severity and SARS-CoV-2 neutralizing immunity in those that became infected. Higher baseline HCoV-OC43 nucleocapsid protein IgG concentrations are associated with markedly lower incidence of SARS-CoV-2 infection. Future interventions against coronaviruses could take advantage of this cross-protective effect, e.g., by incorporating conserved coronavirus antigens to generate pancoronavirus vaccines.

RESULTS
High HCoV-OC43 nucleocapsid IgG levels are associated with lower SARS-CoV-2 incidence Serum IgG antibodies against the C-terminal domain of nucleocapsid protein (NCt) of seasonal HCoVs OC43, HKU1, 229E, NL63, and total Ig antibodies against S1-RBD of SARS-CoV-2, were measured every 4 weeks during the first COVID-19 wave in the Netherlands (March 2020 -June 2020) in a cohort of 150 HCW (see Table 1 for characteristics). IgG concentrations against all seasonal HCoVs remained relatively stable during the study period ( Figures 1A-1H). We hypothesized that if there was any cross-protection by HCoV immunity, this would most likely affect HCW with the most recent seasonal HCoV infection, and therefore those with the highest IgG levels. Plotting the HCoV anti-NCt IgG levels against the probability of contracting a SARS-CoV-2 infection revealed that these potential associations were likely non-linear (Figures 2A-2D). We therefore used baseline seasonal HCoV antibody concentration as a dichotomous determinant throughout the study (highest quartile versus lower concentrations; see Tables S1 and S2). During follow-up, 18.8% (6/32) of participants with anti-NCt IgG concentrations against HCoV-OC43 in the highest quartile at baseline became SARS-CoV-2 seropositive, compared with 43.3% (42/97) of those with lower antibody concentrations (p = 0.019; HR 0.37, 95% CI 0.16-0.88; Figure 3A and Table 2). To correct for possible confounding effects by work-related bedside exposure to COVID-19 patients, we performed a multivariable Cox regression analysis, which showed a consistent result (HR 0.41, 95% CI 0.18-0.97, Table 2). We did not find an association between SARS-CoV-2 infection and anti-NCt IgG levels against HCoV-HKU1, HCoV-229E and HCoV-NL63 (Figures 3B-3D and Table 2). To justify the use of baseline HCoV anti-NCt IgG levels, rather than the antibody levels at each measurement, we performed a sensitivity analysis by using a time-varying determinant in the Cox regression analysis, which results mirrored the earlier found association between HCoV-OC43 IgG concentration and SARS-CoV-2 incidence (HR 0.48, 95% CI 0.23-1.00; Table 2). We did not find an association between SARS-CoV-2 infection and HCoV-OC43 anti-NCt IgA levels (Table 2 and Figures S1A-S1D). Serum-IgA is regarded as one of the earliest markers of infection, yet is only moderately elevated during the first weeks following infection, and therefore a less sensitive marker for recent infection than serum-IgG ( Figure 4) (Callow et al., 1990).

Influenza-and RSV-antibodies are not associated with SARS-CoV-2 incidence
To support the conclusion of a HCoV-OC43-or betacoronavirus-specific protective effect we examined the presence of cross-immunity induced by non-coronavirus respiratory viruses; seasonal influenza virus and   HCoV nucleocapsid IgG did not affect severity of SARS-CoV-2 infection or neutralizing capacity None of the participants in our cohort reported severe COVID-19 disease requiring hospital admission. Asymptomatic SARS-CoV-2-infection was reported by 18 (37.5%) of the 48 seropositive participants, and 29 (60.4%) reported symptoms varying from minimal to moderate. For one participant the severity of symptoms was not reported. We found no clear association between baseline high or low HCoV-OC43 anti-NCt IgG concentrations and presence of any symptoms during SARS-CoV-2 infection (OR 0.68, 95% CI 0.17-2.76; Table 3), even after adjustment for possible confounding by age and sex (OR 0.28, 95% CI 0.05-1.38; Table 3). We also found no association between HCoV-HKU1, HCoV-229E, and HCoV-NL63 anti-NCt IgG concentrations, and COVID-19 disease severity (Table 3).
Seasonal HCoV IgG levels against spike protein are not associated with SARS-CoV-2 incidence To explore why previous studies using seasonal HCoV spike (S) IgG found no evidence for cross-immunity, we replicated our analysis with anti-S (rather than anti-NCt) IgG. There was substantial variation in HCoV anti-S IgG compared with HCoV anti-NCt IgG over time. This was comparable for those with and without SARS-CoV-2 infection during the study period ( Figures S3A-S3H). We found no significant difference in the incidence of  Table S4). We considered whether cross-reactivity of HCoV anti-S IgG boosted by SARS-CoV-2 infection could mask a potential relation between HCoV anti-S IgG and SARS-CoV-2 infection. Therefore, we repeated the analysis including only the samples in which HCoV anti-S IgG were measured 4 weeks before   Table S4). Similarly, no significant difference in the incidence of SARS-CoV-2 infection was detected in individuals with highest quartile versus lower anti-S IgG concentrations of HCoV-HKU1, HCoV-229E and HCoV-NL63 with either analysis method (see Table S4).

DISCUSSION
In a prospective cohort of HCWs followed during the first wave of the SARS-CoV-2 pandemic, we found that for individuals with high baseline HCoV-OC43 NCt-IgG levels, the probability of SARS-CoV-2 infection was substantially reduced. IgG antibody concentrations steadily decline over time, therefore high antibody levels suggest a recent HCoV infection (Callow et al., 1990;Edridge et al., 2020). The effect was robust for adjustment for Our findings complement the growing body of evidence that pre-existing immunity to seasonal HCoV can protect against SARS-CoV-2 (Braun et al., 2020;Kundu et al., 2022;Ladner et al., 2021;Mateus et al., 2020;Ortega et al., 2021;Sagar et al., 2021;Schulien et al., 2021;Song et al., 2021), although results have not been consistent throughout previous studies. Higher levels of nucleocapsid-antibodies against HCoV-OC43 and HCoV-HKU1 were associated with a less severe course of COVID-19, and lower levels with a higher rate of intensive care admissions (Dugas et al., 2021). In line with our findings, a previous study observed a trend towards higher HCoV N-antibody levels at baseline in HCW who subsequently did not become infected with SARS-CoV-2 (Ortega et al., 2021). On the contrary, two previous studies examining the correlation between baseline HCoV antibodies and protection against SARS-CoV-2 infection in longitudinally sampled populations, did not find the protective effect we describe (Anderson et al., 2021;Lin et al., 2022a). Authors concluded pre-existing betacoronavirus antibodies may actually negatively impact protection, because higher magnitudes correlate with more SARS-CoV-2 antibodies following infection, as a proxy for greater disease severity (Lin et al., 2022a). The use of different antibody targets (i.e. spike or nucleocapsid protein) may explain these discordant results. The C-terminal domain of nucleocapsid protein we used in our ELISA, is carefully chosen because this part of the viral protein is well preserved within, but less conserved between HCoV species Edridge et al., 2020;Leach et al., 2021). The specificity of this test is 100% and sensitivity 97%, therefore, the detected antibodies against NCt are unlikely to be cross-reactive . In contrast, when one uses the fulllength version of the S protein in a serological test the specificity is likely reduced, because antibodies against epitopes located in the S2-subunit are known to be more reactive across species Ladner et al., 2021). The SARS-CoV-2 infection-induced antibodies that cross-react with the S for seasonal coronaviruses, may mask the fact that those not infected by SARS-CoV-2 had higher antibodies recognizing HCoV-OC43 S to start with. Furthermore, HCoV immunity is not long-lasting ; differences in interval between baseline HCoV antibody sampling and SARS-CoV-2 exposure observed between previous studies and this one, may further contribute to different outcomes.
Sera of SARS-CoV-2 uninfected individuals that contained cross-reacting antibodies were found to have neutralizing potential which could contribute to protection against SARS-CoV-2 (Galipeau et al., 2021;  Results of linear regression analysis for the association between highest quartile versus lower quartiles of antibody levels against HCoV C-terminal of nucleocapsid protein (NCt) and SARS-CoV-2 neutralizing capacity (measured in June 2020) in log ID 50 in SARS-CoV-2 seropositive participants. Multivariable analysis is corrected for time to infection with SARS-CoV-2.
We did not demonstrate an association between baseline HCoV-OC43 anti-NCt IgG and decreased SARS-CoV-2 severity, nor did we detect signs of antibody dependent enhancement of disease, as suggested by others (Arvin et al., 2020;Huang et al., 2020;Lin et al., 2022a). In addition, we did not find an association between baseline seasonal HCoV anti-NCt IgG and neutralization capacity after SARS-CoV-2 infection or SARS-CoV-2 S-or RBD-antibody concentration (as a surrogate for potential neutralization capacity) in SARS-CoV-2 uninfected individuals. We also did not find a protective association for anti-NCt IgG concentrations of betacoronavirus HCoV-HKU1 and alphacoronaviruses HCoV-NL63 and HCoV-229E. The latter might be explained by the decreased homology between alpha-and betacoronaviruses including SARS-CoV-2 (Huang et al., 2020;Liu, 2021;Song et al., 2021). The less robust antibody response following HCoV-HKU1 infection, with difficulty to recognize recent infection by this virus, may explain the lack of detection of a protective association with HCoV-HKU1 anti-NCt IgG Sechan et al., 2022). iScience Article As described above, humoral cross-immunity may mediate protection in itself. Alternatively, protection against SARS-CoV-2 could be explained by cross-reactive cellular immunity (Braun et al., 2020;Grifoni et al., 2020;Le Bert et al., 2020;Lineburg et al., 2021;Mateus et al., 2020;Schulien et al., 2021), with HCoV antibody levels being merely a marker for recent infection; and both explanations are not mutually exclusive. Kundu et al. recently reported on household contacts recruited shortly after exposure to COVID-19 patients (Kundu et al., 2022). Baseline N-targeting antibodies against seasonal HCoVs were associated with both higher frequencies of cross-reactive T cells, and not contracting SARS-CoV-2 infection during follow-up. Contacts who remained SARS-CoV-2 negative during follow-up showed significantly higher frequencies of specific IL-2 secreting memory T cells that cross-react with HCoV, compared to contacts who became SARS-CoV-2 positive. Cross-reactive T cells were depleted from the bloodstream within days to weeks after exposure, suggesting migration from the circulation to the affected respiratory mucosa (Kundu et al., 2022). Similarly, another study demonstrated that closely (NAAT and serologically) monitored HCW who did not contract SARS-CoV-2 infection despite exposure, had stronger, more multispecific memory T cells, compared with an unexposed pre-pandemic cohort, with expansion of T cells able to cross recognize shared HCoV epitopes (Swadling et al., 2021).
An important strength of this study is the prospectively collected, longitudinal data on seasonal HCoV immunity with detailed surveillance of SARS-CoV-2 incidence. The study period comprises the very first pandemic months of the Netherlands, with fresh immunologic memory of endemic HCoV infections, which is not yet hampered by SARS-CoV-2 preventive measures. The 4-week interval of serum sampling allows us to interpret dynamics of HCoV antibody levels, rather than solely depend on cross-sectional antibody concentrations. Our study also benefits from the use of a highly specific HCoV anti-NCt assay. iScience Article In conclusion, this study found that HCW with high IgG antibodies against HCoV-OC43 NCt were less frequently infected with SARS-CoV-2. We corroborate that immunity induced by one HCoV infection can confer short-lived protection against another HCoV. Downscaling strict SARS-CoV-2 preventive measures will likely be accompanied by recurrence of endemic HCoV infections. Cross-protection derived from HCoV might, at least partially, contribute in controlling the SARS-CoV-2 pandemic, and vaccine development may benefit from incorporating antigens which are conserved between coronavirus species to attempt to generate pancoronavirus immunity.

Limitations of the study
The study is limited by the relatively small sample size. The width of several 95% confidence intervals, e.g. for the association between HCoV-OC43 anti-NCt IgG concentration and SARS-CoV-2 severity (OR 0.68, 95% CI 0.17-2.76, or 0.28, 95% CI 0.05-1.38 after correction for sex and age), suggests the study may have been underpowered to detect clinically relevant associations. Also, none of the relatively young and healthy participants suffered from severe COVID-19. The generalizability of our findings to the current era of emerging variants of concern is uncertain.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

ACKNOWLEDGMENTS
We thank the Amsterdam UMC health care workers that took time to participate, and everyone assisting with sampling and otherwise enabling this study during those busy first pandemic months. We also thank Alex Schuurman, Tom Reijnders, Justin de Brabander, Adinda Pijpers, Esmee Das, Nikita Borstlap, and Lisa Urlings for their essential help in collecting the study data and Birgit Lissenberg for advice on statistical analysis. We acknowledge the GRACE network (LSHM-CT-2005-518226), especially the Work Package 9 Study on the etiology, diagnosis, and prognosis of lower respiratory tract infections in primary care. We thank the principal investigators, the study participants, and other contributors of the GRACE study, with special credits for Herman Goossens, Katherine Loens, Margareta Ieven

Materials availability
This study did not generate new unique reagents.
Data and code availability d The data reported in this study cannot be deposited in a public repository because privacy restrictions may apply. To request access to data, contact Marije Bomers (m.bomers@amsterdamumc.nl). In addition, summary statistics describing these data have been deposited in tables and figures of this manuscript and the supplemental material and are publicly available as of the date of publication.
d This article does not report original code.
d Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request.

Study design and participants
In March 2020, the first month of the SARS-CoV-2 pandemic in the Netherlands, we started a prospective serologic surveillance cohort study among hospital HCWs in the Amsterdam University Medical Center (UMC), consisting of two tertiary care hospitals (S3 study) . Participants underwent phlebotomies combined with surveys regarding exposure to COVID-19 patients, presence of COVID-19 related symptoms, and results of nucleic acid amplification testing (NAAT). All participants were assumed to be seronegative on the day the first SARS-CoV-2 infection was established in the Netherlands (February 27, 2020). Follow-up visits were scheduled every 4 weeks over 18 weeks during the first wave (started March 23, 2020 and finished on June 25, 2020). The first patient with confirmed COVID-19 was admitted to the Amsterdam UMC on March 9, 2020.
Within this cohort we compared the 60 HCW that contracted SARS-CoV-2 infection during follow-up to a group of 90 seronegative HCW. The latter were selected on work-related COVID-19 exposure and highest attendance to follow-up visits. Work-related exposure was defined as working in direct patient care with COVID-19 patients (e.g., intensive care unit, emergency department or a dedicated COVID-19 ward) versus not working in patient care.
The study was approved by the medical ethics review committee of both hospitals, and written informed consent was obtained from each participant. More comprehensive details about the original S3 cohort have been published previously .