key: cord-0808391-n0actmsc
authors: Ng, Kevin W.; Faulkner, Nikhil; Cornish, Georgina H.; Rosa, Annachiara; Harvey, Ruth; Hussain, Saira; Ulferts, Rachel; Earl, Christopher; Wrobel, Antoni; Benton, Donald; Roustan, Chloe; Bolland, William; Thompson, Rachael; Agua-Doce, Ana; Hobson, Philip; Heaney, Judith; Rickman, Hannah; Paraskevopoulou, Stavroula; Houlihan, Catherine F.; Thomson, Kirsty; Sanchez, Emilie; Brealey, David; Shin, Gee Yen; Spyer, Moira J.; Joshi, Dhira; O’Reilly, Nicola; Walker, Philip A.; Kjaer, Svend; Riddell, Andrew; Moore, Catherine; Jebson, Bethany R.; Wilkinson, Meredyth G.Ll.; Marshall, Lucy R.; Rosser, Elizabeth C.; Radziszewska, Anna; Peckham, Hannah; Ciurtin, Coziana; Wedderburn, Lucy R.; Beale, Rupert; Swanton, Charles; Gandhi, Sonia; Stockinger, Brigitta; McCauley, John; Gamblin, Steve; McCoy, Laura E.; Cherepanov, Peter; Nastouli, Eleni; Kassiotis, George
title: Pre-existing and de novo humoral immunity to SARS-CoV-2 in humans
date: 2020-07-23
journal: bioRxiv
DOI: 10.1101/2020.05.14.095414
sha: b9863901386ce78a9ad15d549d362f967073e938
doc_id: 808391
cord_uid: n0actmsc

Several related human coronaviruses (HCoVs) are endemic in the human population, causing mild respiratory infections1. Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), the etiologic agent of Coronavirus disease 2019 (COVID-19), is a recent zoonotic infection that has quickly reached pandemic proportions2,3. Zoonotic introduction of novel coronaviruses is thought to occur in the absence of pre-existing immunity in the target human population. Using diverse assays for detection of antibodies reactive with the SARS-CoV-2 spike (S) glycoprotein, we demonstrate the presence of pre-existing humoral immunity in uninfected and unexposed humans to the new coronavirus. SARS-CoV-2 S-reactive antibodies were readily detectable by a sensitive flow cytometry-based method in SARS-CoV-2-uninfected individuals and were particularly prevalent in children and adolescents. These were predominantly of the IgG class and targeted the S2 subunit. In contrast, SARS-CoV-2 infection induced higher titres of SARS-CoV-2 S-reactive IgG antibodies, targeting both the S1 and S2 subunits, as well as concomitant IgM and IgA antibodies, lasting throughout the observation period of 6 weeks since symptoms onset. SARS-CoV-2-uninfected donor sera also variably reacted with SARS-CoV-2 S and nucleoprotein (N), but not with the S1 subunit or the receptor binding domain (RBD) of S on standard enzyme immunoassays. Notably, SARS-CoV-2-uninfected donor sera exhibited specific neutralising activity against SARS-CoV-2 and SARS-CoV-2 S pseudotypes, according to levels of SARS-CoV-2 S-binding IgG and with efficiencies comparable to those of COVID-19 patient sera. Distinguishing pre-existing and de novo antibody responses to SARS-CoV-2 will be critical for our understanding of susceptibility to and the natural course of SARS-CoV-2 infection.

patients with or without RT-qPCR-confirmed HCoV infection, sampled before or during the early spread of SARS-CoV-2 in the UK (Table S1 ). Although the presence of SARS-CoV-2 S-reactive antibodies of all three Ig classes effectively distinguished COVID-19 patients, remarkably, a proportion of SARS-CoV-2-uninfected patients also had SARS-CoV-2 S-specific antibodies (Fig. 1a) . Again, sera from the latter group contained predominantly lower levels of S-reactive IgG and no IgM or IgA antibodies (Fig. 1a) , suggesting the presence of cross-reactive immunological memory. The S2 subunit exhibits a higher degree of homology among coronaviruses than S1 (Extended data Fig. 5) and was likely the main target of cross-reactive antibodies 6 . Indeed, whereas the addition of recombinant soluble S1 completely abolished binding of the S1-reactive CR3022 antibody 19 to SARS-CoV-2 S-expressing cells, it did not affect the binding of HCoV patient sera (Fig. 1b,c) . Binding of COVID-19 patient sera was reduced by approximately 30% by soluble S1, indicating recognition of both S1 and S2 (Fig. 1b,c) . Consistent with these observations, sera from both COVID-19 and HCoV patients reacted comparably in a SARS-CoV-2 S-coated ELISA, whereas only those from COVID-19 patients reacted in a SARS-CoV-2 S1-coated ELISA (Fig. 1d) . These data highlighted cross-reactivity of HCoV patient sera with SARS-CoV-2 S, and specific recognition of the S1 subunit by COVID-19 patient sera. To confirm and determine the extent of such cross-reactivity using an independent assay, we developed ELISAs with a number of SARS-CoV-2 recombinant proteins (Methods). These included the stabilised trimeric S ectodomain, the S1 subunit, the receptor-binding domain (RBD) and the nucleoprotein (N). In a cohort of 135 COVID-19 patients (Table S1 ), the sensitivity of ELISAs was lower than that of the FACS-based assay, with 20 COVID-19 samples testing negative on ELISA coated with the S1 subunit, 8 of which also tested negative on S-coated or N-coated ELISAs (Extended data Fig. 6 ). Overall, the S1-coated ELISA for IgG detection demonstrated 92.1% sensitivity and 99.6% specificity of seropositivity detection for all samples taken after day 16 post symptoms onset (Extended data Fig. 7a-c) . Rates of IgG seropositivity determined by SARS-CoV-2 S1-coated ELISA were congruent with, but overall lower than those determined by FACS, underscoring the higher sensitivity of the FACS-based assay (Extended data Fig. 7d ,e). We next tested our initial cohort of COVID-19 patients and SARS-CoV-2-uninfected individuals (SARS-CoV-2 -), which included earlier samples of unknown HCoV status and more recent samples of known HCoV status (Table S1 ), in the different assays. SARS-CoV-2 S-reactive IgG antibodies were detected by FACS in 5 of 34 SARS-CoV-2-uninfected individuals with recent HCoV infection, as well as in 1 of 31 individuals without recent HCoV infection (Fig. 2a) . This suggested that these antibodies might have persisted from earlier HCoV infections, rather than having been induced by the most recent one. The 3 SARS-CoV-2-uninfected individuals with the highest cross-recognition of S on FACS, and a further 4 individuals, also had IgG antibodies detectable on ELISA coated with SARS-CoV-2 S ectodomain, as well as with SARS-CoV-2 N (Fig. 2b-e) , also highly conserved among coronaviruses (CoVs). In contrast, none of the control samples had IgG antibodies detectable on ELISAs coated with the less conserved SARS-CoV-2 S1 or RBD (Fig. 2c,e) . Collectively, these results suggested that a total of 12 of 95 control samples exhibited IgG antibodies cross-reactive with conserved epitopes in SARS-CoV-2 proteins (S2 and N). Moreover, a sample from an 81-year-old COVID-19 patient in this cohort collected on day 16 post mild COVID-19 symptoms exhibited only IgG reactivity to SARS-CoV-2 S, but not to S1, which would be more characteristic of pre-existing antibody memory to HCoVs, than a de novo response to SARS-CoV-2 ( Fig. 2a-e) . Although the data presented here and emerging in the recent and unrefereed literature 14, 20, 21 suggest extensive antibody cross-reactivity between SARS-CoV-2 and HCoVs, the potential consequences of pre-existing immunity for the course of SARS-CoV-2 infection or susceptibility to COVID-19 have not been fully considered. One of the ways in which HCoV-elicited antibodies might protect against SARS-CoV-2 infection is inhibition of entry into the target cell. SARS-CoV-2 binding to one identified cellular receptor, angiotensin-converting enzyme 2 (ACE2) 2, 6, 7, 22 , is mediated by the RBD. As HCoV-induced antibodies cross-react with S2, but not the RBD, they might not be expected to block ACE2-RBD interactions. However, SARS-CoV-2-neutralising antibodies that do not block 5 RBD-ACE2 interaction have been reported 23,24 , as have neutralising antibodies targeting the S2 subunit of SARS-CoV 25, 26 or of SARS-CoV-2 24 . It is therefore plausible that HCoV patient sera targeting the S2 also neutralise, either by indirectly affecting ACE2-mediated viral entry or by interfering with alternative routes of entry. Indeed, entry of SARS-CoV-2 can also be facilitated by the alternative receptor CD147, also known as basigin (BSG) 27 , neuropilin 1 28, 29 , and possibly also by receptorindependent mechanisms, as has been described for other CoVs 30, 31 . To examine the ability of HCoV patient sera to inhibit viral entry, we used HEK293T cells as targets. These cells lack ACE2 expression, but are nevertheless permissive to entry of lentiviral particles pseudotyped with SARS-CoV-2 S (Extended data Fig. 8) . Moreover, transduction efficiency of HEK293T cells by SARS-CoV-2 S pseudotypes was not further increased by ACE2 overexpression (Extended data Fig. 8) , highlighting ACE2-independent entry. In contrast, HEK293T cells expressed high levels of BSG and NRP1, encoding CD147 and neuropilin 1, respectively (Extended data Fig. 8 ). Sera from seroconverted (Ab +) COVID-19 patients efficiently neutralised SARS-CoV-2 S pseudotypes, consistent with prior reports 10, 15, 17, 32, 33 , whereas those from COVID-19 patients that had not yet produced binding antibodies (Ab -) did not (Fig. 3a) . Surprisingly, sera from SARS-CoV-2-uninfected patients that contained SARS-CoV-2 S-reactive antibodies neutralised these pseudotypes, with efficiencies comparable with sera from seroconverted COVID-19 patients (Fig. 3a) . None of the sera neutralised lentiviral particles pseudotyped with the control glycoprotein of vesicular stomatitis virus (VSV) (Fig. 3a) . Titres of neutralising antibodies in COVID-19 patient sera correlated well with titres of SARS-CoV-2 RBD-binding antibodies, determined by ELISA and with titres of SARS-CoV-2 S-binding antibodies, determined by FACS (Fig. 3b ). Despite the low numbers, the correlation between neutralising antibodies in SARS-CoV-2-uninfected patient sera with SARS-CoV-2 S-binding IgG antibodies determined by FACS was even stronger (Fig. 3c) . Collectively, these data indicated that HCoV-elicited cross-reactive antibodies can interfere with at least one mode of SARS-CoV-2 entry into target cells. To confirm these findings in different experimental setting, we used infection of Vero E6 cells with authentic SARS-CoV-2 and assessed the ability of patient sera to reduce the number and size of plaques formed. All COVID-19 patient sera tested had measurable neutralising activity in this assay (Fig. 3d) . Importantly, 5 of 6 sera from SARS-CoV-2-uninfected patients with FACS-detectable SARS-CoV-2 S-reactive antibodies also neutralised SARS-CoV-2 in this assay, albeit on average less potently than COVID-19 patient sera, whereas none of the sera from SARS-CoV-2uninfected patients from the same cohort, but without FACS-detectable SARS-CoV-2 S-reactive antibodies did (Fig. 3d) , supporting the functional relevance of FACS-detectable SARS-CoV-2 Sbinding IgG antibodies. Together, our results from four independent assays demonstrated the presence of pre-existing immunity to SARS-CoV-2 in uninfected individuals. Given their potential importance in shaping the natural course of SARS-CoV-2 infection, we next examined the prevalence of such cross-reactive antibodies in healthy donors. We further tested samples from a cohort of 50 SARS-CoV-2-uninfected pregnant women (Table S1 ), none of which had SARS-CoV-2 S-reactive antibodies of all three Ig classes (IgG, IgM and IgA) concurrently (Extended data Fig. 9 ). However, at least 5 of these samples (all collected in May 2018) showed evidence for lower levels of SARS-CoV-2 S-reactive IgG antibodies alone, which could not have been elicited by SARS-CoV-2 infection (Extended data Fig. 9 ). An additional 2 samples reacted with N, but not S or S1 on ELISA (Extended data Fig. 9 ). These data suggested that a minimum of 10% of this cohort had pre-existing antibodies cross-reactive with SARS-CoV-2 proteins. In a separate cohort of 101 SARS-CoV-2-uninfected donors (all sampled in May 2019), 3 had SARS-CoV-2 S-reactive IgG antibodies (Extended data Fig. 10 ), which did not correlate with antibodies to diverse viruses and bacteria also present in several of these samples (Table S1) . SARS-CoV-2 S-reactive IgM and IgA were also detected in 2 of these donors (aged 23 and 33), albeit at considerably lower levels than those in COVID-19 patients (Extended data Fig. 10 ), suggestive of recent or ongoing response. The last two cohorts of SARS-CoV-2-uninfected individuals were sampled in late spring, a season when HCoV spread typically declines, and it was therefore less likely that they had experienced very 6 recent HCoV infection. To examine if recent infection with a particular type of HCoV correlated with the presence of SARS-CoV-2 S-reactive antibodies, we tested an additional 13 donors who had each been infected with one of the 4 HCoV types between 12 and 55 days prior to sampling (Table S1 ). Of these 13 donors, only 1, recently infected with HCoV-OC43, had SARS-CoV-2 S-reactive IgG antibodies, and these at very low levels (Extended data Fig. 11 ), suggesting that their emergence is not simply a common transient event following each HCoV infection in this age group (median age 51 years; Table S1 ). Instead, given that HCoV-reactive antibodies are present in virtually all adults [34] [35] [36] , the emergence of SARS-CoV-2 S-cross-reactive antibodies in a relatively small proportion of adults (15 of 262; 5.72%), indicates additional requirements, such as random B cell receptor repertoire focusing or context of initial exposure or frequency of HCoV infection, than time since the most recent HCoV infection. Indeed, the frequency of infection with each HCoV type displays a characteristic age distribution with children and adolescents experiencing the highest overall infection frequency 1, [35] [36] [37] [38] [39] . We therefore examined a cohort of younger SARS-CoV-2-uninfected healthy donors (aged 1-16 years; Table S1 ). Remarkably, at least 21 of these 48 donors had readily detectable levels of SARS-CoV-2 S-reactive IgG antibodies (Fig. 4a,b) . The prevalence of SARS-CoV-2 S-reactive IgG antibodies increased with age, peaking at 62% after 6 years of age (Fig. 4c) , when HCoV seroconversion in this age group also peaks 34, 35, 37, 38 , and was significantly higher than in adults (p<0.00001, Fisher's exact test). Moreover, sera from SARS-CoV-2-uninfected children and adolescents efficiently neutralised SARS-CoV-2 S pseudotypes, but not VSVg pseudotypes in vitro (Fig. 4d) . Sera from two SARS-CoV-2-uninfected adolescents, from which sufficient sample was available for testing, also neutralised authentic SARS-CoV-2 with comparable efficiency (Fig. 4e) . These observations support a model whereby exposure to HCoVs elicits humoral immunity that cross-reacts with conserved protein domains in other coronaviruses, including SARS-CoV-2. Infection with SARS-CoV-2 additionally induces de novo antibody responses to variable domains unique to this virus, specifically S1. Consistent with this model, SARS-CoV-2 S-reactive class-switched antibodies were detected by FACS concurrently with IgM antibodies in our COVID-19 patients, kinetics typically associated with anamnestic responses. A similar flow assay was recently used to demonstrate crossreactivity of COVID-19 patient sera with the S proteins, but not the RBDs of the other two zoonotic CoVs, SARS-CoV and MERS 10 . Sera from SARS-CoV-recovered patients have been suggested to neutralise, to a degree, SARS-CoV-2 S pseudotypes 7 . Moreover, infection with SARS-CoV-2 has recently been shown to increase IgG seroreactivity to HCoVs 33 . SARS-CoV-2 S-reactive T helper cells, a prerequisite for class-switched antibody responses, have been detected in the majority of COVID-19 patients, as well as between one-third and two-thirds of healthy individuals, albeit at lower clonal frequencies 20, 40 . In agreement with the specificity of the antibody response we describe here, T helper cells from COVID-19 patients targeted both subunits of SARS-CoV-2 S, whereas those from healthy individuals targeted the S2 subunit 20 . Collectively, these studies highlight the existence of potentially functionally relevant antigenic epitopes in the S2 subunit, which is conserved among HCoVs and SARS-CoV-2. They echo findings with antigenic epitopes in the F subdomain of the influenza A virus hemagglutinin, which is also highly conserved among all influenza A virus 16 subtypes and the target of broadly neutralising antibodies 41, 42 . Over its entire length, SARS-CoV-2 S exhibits closer homology with the S proteins of the two betacoronaviruses, HCoV-OC43 and HCoV-HKU1, than of the alphacoronaviruses HCoV-NL63 and HCoV-229E (Extended data Fig. 12a ). Infections with HCoV-OC43 are more prevalent than with HCoV-HKU1, also in the UK and in children and adolescents in particular 38 . Although these considerations would incriminate HCoV-OC43 as the likely inducer of SARS-CoV-2 S-reactive antibodies, other HCoVs are sufficiently homologous with SARS-CoV-2 to elicit cross-reactive humoral immune responses. To probe potential shared epitopes targeted in SARS-CoV-2 S, we constructed overlapping peptide arrays spanning the last 743 amino acids of SARS-CoV-2 S and tested the reactivity of adult and adolescent SARS-CoV-2-uninfected donor sera that had crossreactive antibodies detectable by FACS (SARS-CoV-2 -Ab +) (Extended data Fig. 12b ). As controls, we 7 also used sera from seroconverted COVID-19 patients (SARS-CoV-2 + Ab +) and sera from adult SARS-CoV-2-uninfected donors without cross-reactive antibodies detectable by FACS (SARS-CoV-2 -Ab -). Multiple peptides in the array were recognised by the sera, identifying several putative epitopes in S2 (Fig. 5a ,b; Extended data Fig. 12b ). However, there were at least 5 epitopes recognised more strongly by sera with cross-reactive antibodies, particularly from adolescent donors, than by control sera (Fig. 5a,b) . These included epitopes S901-906 (QMAYRF) and S810-816 (SKPSKRS) (Fig. 5 ), which were embedded in immunodominant regions previously experimentally defined in SARS-CoV and computationally predicted in SARS-CoV-2, respectively 43 . Both these epitopes were conserved between SARS-CoV and SARS-CoV-2, as well as among all four HCoVs (Fig. 5a,b) . We further identified epitopes S851-856 (CAQKFN), S1040-1044 (VDFCG) and S1205-1212 (KYEQYIKW), all of which were highly conserved among HCoVs (Fig. 5a,b) . Epitopes CAQKFN and KYEQYIKW mapped to a surface loop and the membrane-proximal region of S2, respectively, and were therefore missing from prior structures. However, the loop containing the CAQKFN epitope is present in the most recently reported structure 44 , where, similarly to other identified epitopes, it was located on the surface of S2 and accessible by antibodies (Fig. 5c) . Lastly, epitope S997-1002 (ITGRLQ), also targeted more strongly by cross-reactive than control sera, was not located on the surface of pre-fusion S conformation (Fig.  5c ), suggesting that additional epitopes might be accessible in alternative conformations, such as after the loss of the S1 subunit or after premature acquisition of the post-fusion conformation, as it was recently proposed 45, 46 . Cross-reactivity with the identified epitopes was examined by ELISAs coated with synthetic peptides, which provided evidence for cross-reactivity (Extended data Fig. 13 ). Indeed, sera from seroconverted COVID-19 patients reacted with the selected peptides, as did those from SARS-CoV-2-uninfected adults or adolescents, in some cases more strongly than the former (Extended data Fig. 13 ). Reactivity with at least two peptides (containing epitopes CAQKFN and FIEDLLFN) appear restricted to SARS-CoV-2-uninfected donor samples with FACS-detectable crossreactive antibodies and was not seen in control samples without FACS-detectable cross-reactive antibodies (Extended data Fig. 13 ). These findings further supported the notion of pre-existing and de novo responses targeting conserved and unique SARS-CoV-2 S epitopes, respectively. However, the sequence conservation of the identified epitopes among the 4 HCoVs did not indicate if one was more likely than the others to have induced cross-reactive antibodies against these epitopes. We therefore examined reactivity of the sera against the S proteins of the 4 HCoVs separately using our FACS-based assay (Extended data Fig. 14) . As expected, all sera exhibited some level of reactivity with one or more HCoVs, with HCoV-229E being the most frequent ( Fig. 5d and Extended data Fig. 13 ). However, levels specifically of IgG antibodies against each of the 4 HCoV types, determined by this assay, were higher in sera from COVID-19 patients and SARS-CoV-2-infected donors with FACS-detectable cross-reactive antibodies to SARS-CoV-2 S, than the respective control sera (Fig. 5d) . Indeed, IgG and IgA reactivity against HCoVs was higher in SARS-CoV-2-infected adults with, than without SARS-CoV-2-reactive IgG (p=1.4×10 -6 for IgG and p=0.017 for IgA, Student's t-test), and in SARS-CoV-2-infected children or adolescents with, than without SARS-CoV-2-reactive IgG (p=0.010 for IgG and p=0.021 for IgA, Student's t-test), supporting a direct link between the two. Accordingly, IgG reactivity against each HCoV type was independently correlated with the presence of cross-reactive antibodies against SARS-CoV-2 (Fig. 5d) . Weak correlation was also seen for IgA reactivity specifically against HCoV-OC43, but not the other HCoV or for IgM reactivity against any HCoV (Fig. 5d) . Consistent with sequence conservation of the putative targeted epitopes, these data suggest the presence of IgG memory reactive with HCoVs, as well as SARS-CoV-2.

Cross-reactivity between seasonal HCoVs and the pandemic SARS-CoV-2 needs to be carefully considered in the development and interpretation of assays for precise detection of SARS-CoV-2specific antibodies. The FACS-based method we employed demonstrated the highest degree of specificity and sensitivity of the assays used here. This assay allows the distinction between pre-8 existing and de novo antibody responses to SARS-CoV-2, based on the levels of SARS-CoV-2 Sreactive IgG and parallel detection of IgM and IgA. These distinguishing features were maintained and, indeed, appear to increase during the first 43 days post onset of COVID-19 symptoms (the latest time-point we have examined). It is possible that these features are maintained for longer periods post SARS-CoV-2 infection, but it is also expected that the discriminating ability will be lost over time. Within each assay, increasing specificity by using higher positivity thresholds or omitting conserved protein domains may be at expense of sensitivity. This is evident in the use of less conserved SARS-CoV-2 protein domains such as S1 and RBD, which exhibited approximately 78% sensitivity during the first 43 days post onset of COVID-19 symptoms in our study, and which is likely to drop further with waning antibody titres over longer periods. The ability of HCoV patient sera to neutralise authentic SARS-CoV-2 and SARS-CoV-2 S pseudotypes raise similar concerns regarding the specificity of neutralisation assays. For example, 5 of 1,000 samples from healthy Scottish blood donors collected in March 2020 neutralised SARS-CoV-2 S pseudotypes and 1 of 100 samples collected in 2019 also had neutralising activity in the absence of a strong ELISA signal 32 . Notably, these samples were described to have low or no SARS-CoV-2 Sreactive IgM antibodies 32 , a feature we would associate with immune memory of HCoVs. In addition to its implications for serology assay development and interpretation or for the design of vaccination studies, potential cross-reactivity between seasonal HCoVs and the pandemic SARS-CoV-2 has important ramifications for natural infection. Thorough epidemiological studies of HCoV transmission suggest that cross-protective immunity is unlikely to be sterilising or long-lasting 39 , which is also supported by repeated reinfection of all age groups 4 , sometimes even with homologous HCoVs 47 . Nevertheless, prior immunity induced by one HCoV has also been reported to reduce the transmission of homologous and, importantly, heterologous HCoVs, and to ameliorate the symptoms where transmission is not prevented 1,4,5 . A possible modification of COVID-19 severity by prior HCoV infection might account for the age distribution of COVID-19 susceptibility, where higher HCoV infection rates in children than in adults 5, 35, 37 , correlates with relative protection from COVID-19 48 , and might also shape seasonal and geographical patterns of transmission. Public health measures intended to prevent the spread of SARS-CoV-2 will also prevent the spread of and, consequently, maintenance of herd immunity to HCoVs, particularly in children. It is, therefore, imperative that any effect, positive or negative, of pre-existing HCoV-elicited immunity on the natural course of SARS-CoV-2 infection is fully delineated. 9 

Serum or plasma samples were obtained from University College London Hospitals (UCLH) COVID-19 patients testing positive for SARS-CoV-2 infection by RT-qPCR and sampled between March 2020 and April 2020. An initial cohort of 35 patients (31 annotated) and an extended cohort of 135 patients were tested between 2 and 43 days after the onset of COVID-19 symptoms (Table S1 ). A total of 262 samples form SARS-CoV-2-uninfected adults and 48 SARS-CoV-2-uninfected children and adolescents were also used (described in Table S1 ). Samples from adults were obtained from UCLH and Public Health Wales, University Hospital of Wales, and samples from children and adolescents were obtained from the Centre for Adolescent Rheumatology Versus Arthritis at University College London (UCL), UCLH, and Great Ormond Street Hospitals (GOSH), and Great Ormond Street Institute for Child Health (ICH) with ethical approval (refs 11/LO/0330 and 95RU04). All patient sera and sera remaining after antenatal screening of healthy pregnant women were from residual samples prior to discarding, in accordance with Royal College Pathologists guidelines and the UCLH Clinical Governance for assay development and GOSH and ICH regulations. All serum or plasma samples were heat-treated at 56 o C for 30 min prior to testing.

SARS-CoV-2 nucleic acids were detected at Health Services Laboratories, London, UK, by an RT-qPCR method as recently described 49 . HCoV nucleic acids were detected by RT-qPCR, as part of a diagnostic panel for respiratory viruses, run by Health Services Laboratories (HSL), London, UK.

HEK293T cells were obtained from the Cell Services facility at The Francis Crick Institute, verified as mycoplasma-free and validated by DNA fingerprinting. Vero-E6 cells were from the National Institute for Biological Standards and Control, UK. HEK293T cells overexpressing ACE2 were generated by transfection, using GeneJuice (EMD Millipore), with a plasmid containing the complete human ACE2 transcript variant 1 cDNA sequence (NM_001371415.1) cloned into the mammalian expression vector pcDNA3.1-C' FLAG by Genscript. Cells were grown in Iscove's Modified Dulbecco's Medium (Sigma Aldrich) supplemented with 5% fetal bovine serum (Thermo Fisher Scientific), Lglutamine (2 mmol/L, Thermo Fisher Scientific), penicillin (100 U/ml, Thermo Fisher Scientific), and streptomycin (0.1 mg/ml, Thermo Fisher Scientific). The SARS-CoV-2 isolate hCoV-19/England/02/2020 was obtained from the Respiratory Virus Unit, Public Health England, UK, (GISAID EpiCov TM accession EPI_ISL_407073) and propagated in Vero E6 cells. analysis software. Transfection efficiencies were determined by staining with a fixed concentration of the S1-reactive CR3022 antibody (100 ng/ml) (Absolute Antibodies) and control convalescent sera (1:50 dilution), followed by BV421 anti-IgG antibody. Staining with convalescent sera, detecting epitopes over the entire S ectodomain, was consistently higher than staining with CR3022, detecting only the S1 epitope, and was taken as the maximum transfection efficiency. This varied between 68% and 95% between experiments.

The SARS-CoV-2 RBD and S1 constructs, spanning SARS-CoV-2 S (UniProt ID: P0DTC2) residues 319-541 (RVQPT…KCVNF) and 1-530 (MFVFL…GPKKS), respectively, were produced with C-terminal twin Strep tags. To this end, the corresponding codon-optimised DNA fragments were cloned into mammalian expression vector pQ-3C-2xStrep 50 . A signal peptide from immunoglobulin kappa gene product (METDTLLLWVLLLWVPGSTGD) was used to direct secretion of the RBD construct. Stabilised ectodomain of the SARS-CoV-2 S glycoprotein (residues 1-1208) with inactivated furin cleavage site (RRAR, residues 682-685 mutated to GSAS) and a double proline substitution (K986P/V987P) 51,52 was produced with a C-terminal T4 fibritin trimerization domain and a hexahistidine (His6) tag from the pcDNA3 vector. Expi293F cells growing at 37 o C in 5% CO2 atmosphere in shake flasks in FreeStyle 293 medium were transfected with the corresponding plasmids using ExpiFectamine reagent (Thermo Fisher Scientific). Conditioned medium containing secreted proteins was harvested twice, 3-4 and 6-8 days post-transfection. Twin Strep-and His6-tagged proteins were captured on Streptactin XT (IBA LifeSciences) or Talon (Takara) affinity resin, respectively, and purified to homogeneity by size exclusion chromatography through Superdex 200 (GE Healthcare) in 150 mM NaCl, 1 mM EDTA, 20 mM Tris-HCl, pH 8.0. Full-length SARS CoV2 N gene product was produced with an N-terminal His6 tag from pOPTH-1124 plasmid (kindly provided by Jakub Luptak and Leo James, Laboratory for Molecular Biology, Cambridge, UK). Escherichia coli C43(DE3) cells (Lucigen) transformed with pOPTH-1124 were grown in terrific broth medium, and expression was induced by addition of 1 mM Isopropyl β-D-1-thiogalactopyranoside at 37 o C. Bacteria, harvested 4 h post-induction, were disrupted by sonication in core buffer (1M NaCl, 10 mM imidazole, 20 mM HEPES-NaOH, pH 8.0) supplemented with BaseMuncher nuclease (Expedion; 1 ml per 40 ml cell suspension) and Complete EDTA-free protease inhibitor mix (Roche). The extract was precleared by centrifugation, and His6tagged protein was captured on NiNTA agarose (Qiagen). Following extensive washes with core buffer supplemented with 20 mM imidazole, the protein was eluted with 500 mM imidazole. SARS CoV-2 N, was further purified by cation exchange and heparin affinity chromatography prior polishing by gel filtration through a Superdex 200 16/40 column (GE Healthcare), which was operated in 300 mM NaCl, 20 mM HEPES-NaOH, pH 8.0. Purified SARS CoV2 antigens, concentrated to 1-5 mg/ml by ultrafiltration using appropriate VivaSpin devices (Sartorius), were snap-frozen in liquid nitrogen in small aliquots and stored at -80 o C.

ELISA 96-well MaxiSorp plates (Thermo Fisher Scientific) were coated overnight at 4 o C with purified protein in PBS (3 µg/ml per well in 50 µL) and blocked for 1 hr in blocking buffer (PBS, 5% milk, 0.05% Tween 20, 0.01% sodium azide). Sera was diluted in blocking buffer (1:50) and 50 µL added to the plate then incubated for 1 hr at room temperature. After washing 4 times with PBS-T (PBS, 0.05% Tween 20), plates were incubated with alkaline phosphatase-conjugated goat anti-human IgG (1:1000, Jackson ImmunoResearch) for 1 hr. Plates were developed by adding 50 µL of alkaline phosphatase substrate (Sigma Aldrich) for 15-30 min after 6 washes with PBS-T. Optical densities were measured at 405 nm on a microplate reader (Tecan). CR3022 (Absolute Antibodies) was used as a positive control for ELISAs coated with S, S1, and RBD. For ELISAs with synthetic peptides, 96well Nunc Immobilizer Amino Plates (Thermo Fisher Scientific) were used for coating overnight at 4 o C with peptides (10 µg/ml per well in 50 µL)

Lentiviral particles pseudotyped with either SARS-CoV-2 S or Vesicular Stomatitis Virus glycoprotein (VSVg) were produced by co-transfection of HEK293T cells with plasmids encoding either of these glycoproteins together with a plasmid encoding the SIVmac Gag-Pol polyprotein and a plasmid expressing an HIV-2 backbone with a GFP encoding gene, using GeneJuice (EMD Millipore). Viruscontaining supernatants were collected 48 hr post-transfection and stored at -80 o C until further use. For neutralisation assays, lentiviral pseudotypes were incubated with serial dilutions of patient sera at 37 o C for 30 minutes and were subsequently added to HEK293T cells seeded in 96-well plates (3,000 cells/well). Polybrene (4 ug/mL, Sigma Aldrich) was also added to the cells and plates were spun at 1,200 rpm for 45 min. The percentage of transduced (GFP+) cells was assessed by flow cytometry 72 hours later. The inverse serum dilution leading to 50% reduction of GFP+ cells was taken as the neutralising titre.

Confluent monolayers of Vero E6 cells were incubated with 10-20 plaque-forming units (PFU) of SARS-CoV-2 strain hCoV-19/England/2/2020 and two-fold serial dilutions of human sera (previously heat-treated at 56C for 30 min) starting at 1:40 dilution, for 3 hours at 37C, 5% CO2, in triplicate per condition. The inoculum was then removed and cells were overlaid with virus growth medium containing avicel at a final concentration of 1.2%. Cells were incubated at 37C, 5% CO2. At 24 hours post-infection, cells were fixed in 4% paraformaldehyde and permeabilised with 0.2% Triton-X-100/PBS and virus plaques were visualised by immunostaining, as described previously for the neutralisation of influenza viruses 53 , except using a rabbit polyclonal anti-NSP8 antibody and antirabbit-HRP conjugate and detected by action of HRP on a tetra methyl benzidine (TMB) based substrate. Virus plaques were quantified and IC50 for sera was calculated using LabView software as described previously 53 .

Peptide arrays were synthesised on an Intavis ResPepSL Automated Peptide Synthesiser (Intavis Bioanalytical Instruments, Germany) on a cellulose membrane by cycles of N(a)-Fmoc amino acids coupling via activation of carboxylic acid groups with diisopropylcarbodiimide (DIC) in the presence of Ethyl cyano(hydroxyimino)acetate (Oxyma pure) followed by removal of the temporary α-amino protecting group by piperidine treatment. Subsequent to chain assembly, side chain protection groups were removed by treatment of membranes with a deprotection cocktail (20 mls 95% trifluoroacetic acid, 3% triisopropylsilane 2% water 4 hours at room temperature) then washing (4 × DCM, 4 × EtOH 2 × H20, 1 × EtOH) prior to being air dried. Membranes were blocked for 1 hour in blocking buffer (PBS, 5% BSA, 0.05% Tween 20, 0.05% sodium azide), then stained with pooled sera (1:100 in blocking buffer) for 2 hours at room temperature. Membranes were washed 3 times in PBS-T, then stained with IRDye 800CW Goat anti-Human IgG (Licor; 1:15,000 in blocking buffer) for 1 hour at room temperature in the dark. Membranes were washed 3 times in PBS-T and once in PBS before imaging on an Odyssey CLx Infrared scanner (Licor). Scanned images were analysed in Image Studio v5.2 (Licor).

Individual peptides were synthesised on an Intavis ResPepSLi Automated Peptide Synthesiser (Intavis Bioanalytical Instruments, Germany) on Rink amide resin (0.26 mmole/g, 0.1 mmol) using N(a)-Fmoc amino acids and HATU as the coupling reagent. Following amino acid chain assembly, peptides were cleaved from the resin by addition to cleavage cocktail (10 ml, 92.5% TFA, 2.5% H2O, 2.5% EDT, 2.5% 12 TIS) for 2 hours. Following resin removal, peptide precipitation and extensive washing with ether, the peptides were analysed by LC-MS on an Agilent 1100 LC-MSD.

Data were analysed and plotted in GraphPad Prism v8 (GraphPad Software) or SigmaPlot v14.0 (Systat Software). Sequence alignments were performed with Vector NTI v11.5 (Thermo Fisher Scientific). Levels of IgM are indicated by a heatmap. b, Ability of soluble S1 to inhibit binding to SARS-CoV-2 S of the S1-specific CR3022 antibody or antibodies in the sera of SARS-CoV-2-infected (SARS-CoV-2 +) or HCoV-infected (SARS-CoV-2 -HCoV +) patients. One representative of two patients is shown. c, Quantitation of the inhibitory effect of soluble S1 on binding to SARS-CoV-2 S of sera from patients described in b. Each dot is an individual patient serum tested at 1:50 and 1:200 dilution. d, Optical densities from ELISAs coated with S or S1 of sera from patients described in b. Each dot is an individual patient serum tested at 1:50 dilution. The dashed lines represents the optical densities obtained with the CR3022 antibody. Boxed peaks indicate the regions that were recognised differentially by sera with and without SARS-CoV-2 S-cross-reactive antibodies, as determined by FACS. b, Alignment of the amino acid sequences of SARS-CoV-2 S and those of the 4 HCoV types. Similarity and homology is indicated with different shading and the consensus sequence is also shown. Boxes indicate the predicted core epitopes in each of the selected regions. c, Depiction of the linear predicted epitopes targeted by SARS-CoV-2 S-cross-reactive antibodies on the most complete structure of the trimeric SARS-CoV-2 spike (PDB ID: 6ZGE). One monomer is shown in colour, with S1 subunit coloured in blue and S2 in pale red, and the remaining two in grey. Peptides corresponding to the epitopes are shown for one monomer only; those present in the structure are depicted as crimson surfaces and the one derived from the membrane proximal region not present in the structure is represented as a circled dashed line. The patient experienced mild COVID-19 symptoms that did not require hospitalisation, but remained positive for SARS-CoV-2 infection, with RT-qPCR confirmation in late April 2020. None of the serial serum samples had significant neutralising activity against SARS-CoV-2 pseudotypes.

Extended data Fig. 6 . Comparison of antibody detection methods in an extended cohort of COVID-19 patients.

Samples from a total of 135 confirmed COVID-19 patients were tested (Table S1 ). a, Frequency of cells that stained with all three antibody classes (IgM+IgG+IgA+) or only with IgG (IgG+) in each of these samples, ranked by their IgM+IgG+IgA+ frequency. b-d, Optical densities from ELISAs coated with S (b), S1 or RBD (c) or N (d) of the same samples. Dashed lines in a-d denote the assay sensitivity cut-offs. e, Summary of the results from a-d, represented as a heatmap of the quartile values. Extended data Fig. 7 . Performance of the S1 ELISA assay for SARS-CoV-2-reactive antibody detection. a, Optical densities (ODs) from S1-coated ELISAs on sera from 170 confirmed COVID-19 cases and 262 controls. The dashed line represents the calculate cut-off for positivity. b-c, Sensitivity and specificity for the assay calculated for all the sample collected at any time-point post symptoms onset (b) or only for samples collected 16 days after the onset of COVID-19 symptoms (c). The receiver operating characteristics (ROC) curve areas and 95% confidence intervals (CI) are also shown for each group. d, Kinetics of seroconversion in COVID-19 patients determined by the S1-coated ELISA. ODs are plotted over time of sample collection since symptoms onset. Seropositivity was calculated for each consecutive week since symptoms onset. Only COVID-19 patients with known date of symptoms onset are included. e, Correlation of seropositivity determined by the S1-coated ELISA and the FACSbased assay. Each symbol represents and individual COVID-19 patient sample. 

Seasonality and immunity to laboratory-confirmed seasonal coronaviruses (HCoV-NL63, HCoV-OC43, and HCoV-229E): results from the Flu Watch cohort study

A pneumonia outbreak associated with a new coronavirus of probable bat origin

A Novel Coronavirus from Patients with Pneumonia in China

The time course of the immune response to experimental coronavirus infection of man

The dominance of human coronavirus OC43 and NL63 infections in infants

Structure, Function, and Antigenicity of the SARS-CoV-2 Spike Glycoprotein

SARS-CoV-2 Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically Proven Protease Inhibitor

SARS-CoV-2 serological analysis of COVID-19 hospitalized patients, paucisymptomatic individuals and blood donors. medRxiv

Profiling Early Humoral Response to Diagnose Novel Coronavirus Disease (COVID-19)

Potent human neutralizing antibodies elicited by SARS-CoV-2 infection. bioRxiv

Evaluation of nine commercial SARS-CoV-2 immunoassays. medRxiv

Antibody responses to SARS-CoV-2 in patients with COVID-19

Serology characteristics of SARS-CoV-2 infection since the exposure and post symptoms onset. medRxiv

Severe Acute Respiratory Syndrome Coronavirus 2-Specific Antibody Responses in Coronavirus Disease

Rapid generation of neutralizing antibody responses in COVID-19 patients. medRxiv

Humoral immune response and prolonged PCR positivity in a cohort of 1343 SARS-CoV 2 patients in the New York City region. medRxiv

Neutralizing antibody responses to SARS-CoV-2 in a COVID-19 recovered patient cohort and their implications. medRxiv

Antibody responses to SARS-CoV-2 in patients of novel coronavirus disease 2019

Human monoclonal antibody combination against SARS coronavirus: synergy and coverage of escape mutants

Presence of SARS-CoV-2 reactive T cells in COVID-19 patients and healthy donors. medRxiv

Distinct systems serology features in children, elderly and COVID patients. medRxiv

Characterization of spike glycoprotein of SARS-CoV-2 on virus entry and its immune cross-reactivity with SARS-CoV

A human monoclonal antibody blocking SARS-CoV-2 infection. bioRxiv

A neutralizing human antibody binds to the N-terminal domain of the Spike protein of SARS-CoV-2

A human SARS-CoV neutralizing antibody against epitope on S2 protein

Human monoclonal antibodies against highly conserved HR1 and HR2 domains of the SARS-CoV spike protein are more broadly neutralizing

SARS-CoV-2 invades host cells via a novel route: CD147-spike protein. bioRxiv

Neuropilin-1 facilitates SARS-CoV-2 cell entry and provides a possible pathway into the central nervous system. bioRxiv

Neuropilin-1 is a host factor for SARS-CoV-2 infection. bioRxiv

Coronaviruses induce entry-independent, continuous macropinocytosis

MHVR-independent cell-cell spread of mouse hepatitis virus infection requires neutral pH fusion

Neutralising antibodies to SARS coronavirus 2 in Scottish blood donorsa pilot study of the value of serology to determine population exposure. medRxiv

Virological assessment of hospitalized patients with COVID-2019

Development of a nucleocapsid-based human coronavirus immunoassay and estimates of individuals exposed to coronavirus in a U.S. metropolitan population

Human coronavirus NL63 and 229E seroconversion in children

A systematic review of antibody mediated immunity to coronaviruses: antibody kinetics, correlates of protection, and association of antibody responses with severity of disease

Human Coronavirus Infections in Israel: Epidemiology, Clinical Symptoms and Summer Seasonality of HCoV-HKU1

Epidemiology of seasonal coronaviruses: Establishing the context for COVID-19 emergence

Coronavirus occurrence and transmission over 8 years in the HIVE cohort of households in Michigan

Targets of T cell responses to SARS-CoV-2 coronavirus in humans with COVID-19 disease and unexposed individuals

A neutralizing antibody selected from plasma cells that binds to group 1 and group 2 influenza A hemagglutinins

Structure and Function Analysis of an Antibody Recognizing All Influenza A Subtypes

A Sequence Homology and Bioinformatic Approach Can Predict Candidate Targets for Immune Responses to SARS-CoV-2

Evolution of SARS-CoV-2 spike glycoprotein&nbsp

Distinct conformational states of SARS-CoV-2 spike protein. bioRxiv

SARS-CoV-2 structure and replication characterized by in situ cryo-electron tomography. bioRxiv

Human Coronavirus NL63 Molecular Epidemiology and Evolutionary Patterns in Rural Coastal Kenya

Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Infection in Children and Adolescents: A Systematic Review

Extraction-free COVID-19 (SARS-CoV-2) diagnosis by RT-PCR to increase capacity for national testing programmes during a pandemic. bioRxiv

A bipartite structural organization defines the SERINC family of HIV-1 restriction factors

Immunogenicity and structures of a rationally designed prefusion MERS-CoV spike antigen

Cryo-EM structure of the 2019-nCoV spike in the prefusion conformation

Optimisation of a micro-neutralisation assay and its application in antigenic characterisation of influenza viruses

We thank L. James and J. Luptak (Laboratory for Molecular Biology, Cambridge, UK) for a generous gift of SARV CoV2 N expression construct and M. Pizzato (University of Trento, Italy) for providing codon-optimised SARS CoV2 S cDNA. We also thank the entire CRICK COVID-19 Consortium. We are grateful for assistance from the Cell Services and High Throughput Screening facilities at the Francis Crick Institute and UCLH Biochemistry (A. Goyale and C. Wilson) and to Mr Michael Bennet and Mr Simon Caidan for training and support in the high-containment laboratory. This work was supported by a Centre of Excellence Centre for Adolescent Rheumatology Versus Arthritis grant, 21593, as well as support from the Great Ormond Street Childrens Charity, CureJM Foundation and the NIHR Biomedical Research Centres at GOSH and UCLH. This work was supported by the Francis Crick Institute, which receives its core funding from Cancer Research UK, the UK Medical Research Council, and the Wellcome Trust. Fig. 5 . Conservation of SARS-CoV-2 S subunits. Structure of the SARS-CoV-2 S protomer with each amino acid residue coloured according to conservation among 24 animal and human CoVs. The alignment and the figure were generated using the ConSurf algorithm (https://consurf.tau.ac.il) using a single chain of the SARS-CoV-2 spike protein in the closed state (PDB ID 6VXX) as a reference.

13 52 (21 -75) 1/2019 -4/2020 n/a 18 (12 -55) Patients at University Hospital of Wales. A total of 16 samples were taken from these 13 patients. One haematology patient, persistently infected with NL63 was sampled 4 separate times and all other donors were sampled once. They include patients infected with OC43 (5), NL63 (3), 229E (2) and HKU1 (3). Extended data Fig. 9 . SARS-CoV-2-reactive antibody detection in an additional control cohort. Samples from a total of 50 SARS-CoV-2-uninfected individuals collected in 2018 were tested. All 50 were pregnant healthy women visiting antenatal clinics (Table S1 ). a, Frequency of cells that stained with all three antibody classes (IgM+IgG+IgA+) or only with IgG (IgG+) in each of these samples, ranked by their IgM+IgG+IgA+ frequency. b-d, Optical densities from ELISAs coated with S (b), S1 or RBD (c) or N (d) of the same samples. (Table S1 ). a, Frequency of cells that stained with all three antibody classes (IgM+IgG+IgA+) or only with IgG (IgG+) in each of these samples, ranked by their IgM+IgG+IgA+ frequency. Arrows indicate the 3 samples with SARS-CoV-2-cross-reactive antibodies. b, Optical densities from S1-coated ELISA of the same samples. Dashed lines in (a, b) denote the assay sensitivity cut-offs. c, FACS profiles of the 3 samples that were positive for SARS-CoV-2-cross-reactive antibodies. Fig. 11 . SARS-CoV-2-reactive antibody detection in an additional control cohort. A total of 16 samples from 13 individuals with recent HCoV infection (Table S1 ) were tested. One haematology patient, persistently infected with NL63 was sampled 4 separate times and all other donors were sampled once. a, Frequency of cells that stained with all three antibody classes (IgM+IgG+IgA+) or only with IgG (IgG+) in each of these samples, ranked by their IgM+IgG+IgA+ frequency. The arrow indicate sample with SARS-CoV-2-crossreactive antibodies. b, Optical densities from S1-coated ELISA of the same samples. Dashed lines in (a, b) denote the assay sensitivity cut-offs. c, FACS profile of the sample with SARS-CoV-2-cross-reactive antibodies from 66year-old donor infected with OC43 18 days prior to sampling.