key: cord-0864687-8cxu6ixq
authors: Bangaru, Sandhya; Antanasijevic, Aleksandar; Kose, Nurgun; Sewall, Leigh M.; Jackson, Abigail M.; Suryadevara, Naveenchandra; Zhan, Xiaoyan; Torres, Jonathan L.; Copps, Jeffrey; Torrents de la Peña, Alba; Crowe, James E.; Ward, Andrew B.
title: Structural mapping of antibody landscapes to human betacoronavirus spike proteins
date: 2021-10-01
journal: bioRxiv
DOI: 10.1101/2021.09.30.462459
sha: 056a5a7f02cfe270a5457fdf4ff64988af2d2cbc
doc_id: 864687
cord_uid: 8cxu6ixq

Preexisting immunity against seasonal coronaviruses (CoV) represents an important variable in predicting antibody responses and disease severity to Severe Acute Respiratory Syndrome CoV-2 (SARS-2) infections. We used electron microscopy based polyclonal epitope mapping (EMPEM) to characterize the antibody specificities against β-CoV spike proteins in sera from healthy donors (HDs) or SARS-2 convalescent donors (CDs). We observed that most HDs possessed antibodies specific to seasonal human CoVs (HCoVs) OC43 and HKU1 spike proteins while the CDs showed reactivity across all human β-CoVs. Detailed molecular mapping of spike-antibody complexes revealed epitopes that were differentially targeted by antibodies in preexisting and convalescent serum. Our studies provide an antigenic landscape to β-HCoV spikes in the general population serving as a basis for cross-reactive epitope analyses in SARS-2 -infected individuals. One-Sentence summary We present the epitope mapping of polyclonal antibodies against beta-coronavirus spike proteins in human sera.

Four human coronaviruses (HCoVs) of genus α (HCoV-229E and HCoV-NL63) or β (HCoV-OC43 and HCoV-HKU1) are endemic in the human population contributing up to a third of the common cold infections (1, 2). While the infection rate and prevalence of these HCoVs vary based on the region, primary infections occur early in life with a majority of the population infected before 15 years of age (2) (3) (4) (5) . Most individuals possess antibodies to HCoVs targeting the trimeric spike glycoprotein and the nucleocapsid protein (N) though antibodies wane over time permitting reinfection even within a year (6) (7) (8) (9) . In addition to HCoVs OC43 and HKU1, the β-CoV genus also contains three highly pathogenic CoVs associated with human disease: Middle East Respiratory Syndrome CoV (MERS), Severe Acute Respiratory Syndrome CoV (SARS) and the novel SARS-2, the causative agent of the ongoing coronavirus disease 2019 (COVID- 19) pandemic (10, 11) .

The spike protein is an important determinant of host range and cell tropism as it mediates virus attachment and entry into the host cells, making it a major target for neutralizing antibodies and a key component for vaccine development (12) (13) (14) (15) (16) . While the SARS-2 spike shares high structure and sequence homology with the SARS (69.2%) spike, it is less conserved across other β-CoVs, with as little as 27.2% sequence homology between SARS-2 and OC43 (17) . Despite the low sequence conservation, preexisting immunity against seasonal CoV spike proteins has been associated with COVID-19 disease outcome as a consequence of either back-boost or induction of cross-reactive antibodies following SARS-2 infection (8, 9, (18) (19) (20) (21) (22) . Of interest, COVID-19 donors with high SARS-2 antibody titers also possessed increased levels of antibodies against β-HCoVs (8, 23) . It is not clear if infection triggers a recall of preexisting HCoV-specific antibodies or preferentially elicits cross-reactive β-CoV antibodies targeting conserved epitopes. Here, we elucidate the β-HCoV spike epitopes targeted by pre-existing serum antibodies and compare them to those elicited following SARS-2 infection using EMPEM methodology (24, 25) Soluble ectodomains of spike proteins for β-CoVs, HKU1, OC43, SARS, MERS and SARS-2 (4 stabilized constructs were used for SARS-2) were generated and characterized by negative stain electron microscopy (ns-EM) and shown to be homogeneous in their prefusion conformation (Fig. S1A) . To determine the baseline serum antibody titers to β-CoV spikes in the general population, either sera or plasma (based on availability) from eight HDs, collected prior to the COVID-19 pandemic, with unknown HCoV infection history were screened for spike antibodies by enzyme-linked immunosorbent assay (ELISA). All 8 donors exhibited reactivity to the OC43 spike, with half maximal effective concentration (EC50) serum dilution values ranging from 0.0007 to 0.02, while HKU1 antibody titers were lower in general (serum dilution EC50 of 0.001 to 0.06) ( Fig. 1A and Fig. S1B ). This finding is consistent with OC43 being the most commonly encountered HCoV globally while HKU1 is less prevalent (4, 5) . Reactivity against SARS-2 spike was not detected in any of the HD sera, and only 1 out of 8 donors (D1124) exhibited low-level reactivity against SARS and MERS spikes. For comparison, we then assessed β-CoV spike reactivity in 3 sera samples from COVID-19 CDs (~day 56 post-infection), all of whom exhibited high antibody titers to SARS-2 spike. Notably, the 3 donors also showed reactivity against other β-CoV spikes including SARS and MERS ( Fig. 1A and fig. S1B ). Given that the donors were immunologically naïve to these pathogenic CoVs, the data indicate that SARS-2 infection can elicit some level of cross-reactivity against the β-CoV spikes. While OC43 reactivity was high in both HD and CD sera, HKU1 spike antibody titers appeared enhanced in CDs ( Fig.   1A and Fig. S1B ). To investigate if serum antibody reactivity translated to inhibitory activity, we performed neutralization assays with both HD and CD serum against the OC43 virus and VSV-pseudotyped SARS and SARS-2 viruses. Overall, the serum inhibitory titers against the OC43 virus correlated well with their binding titers ( Fig. 1A and Fig. S1C ). While none of the HD sera neutralized SARS or SARS-2 virus, the CD sera exhibited neutralizing activity (Serum dilution IC50 of 0.002 to 0.007) against the SARS-2 virus and some weak activity against the SARS virus ( Fig. 1A and Fig. S1C ).

We next employed ns-EMPEM to determine the epitope specificities of spike antibodies in the HD sera. Structural analysis of polyclonal Fabs complexed with spike proteins from OC43, HKU1, SARS or MERS revealed OC43-reactive antibodies in all 8 donors and HKU1-reactive antibodies in 1 donor. We did not detect antibodies to either SARS or MERS spikes (Figs. 1B-D).

Published cryo-EM structures of β-CoV spikes show the cleavable S1 and S2 subunits comprising an N-terminal domain (NTD), a C-terminal domain (CTD), subdomains 1 and 2 (SD1 and SD2), the fusion peptide (FP) and heptad repeats 1 and 2 (HR1 and HR2) (26) (27) (28) (29) . OC43 NTD-reactive 1B-C). The prevalence of NTD-site 1 Fabs that can sterically block receptor engagement correlated well with the OC43 inhibitory titers observed across HDs. While neither the CTD nor the SD1 of OC43 spike is associated with any known function, antibodies to CTD were seen in at least 3 donors, to SD1 in 2 donors and to S1 inter-protomeric interfaces in 2 donors. A single S2-reactive antibody from donor 1051 displayed a broad footprint with potential interactions with residues 800 to 807, 1013 to 1031, and 1062 to 1068 (Fig. S1D ). Of interest, donor 1412 with a relatively low OC43 neutralization titer displayed the greatest diversity of Fab specificities, targeting 6 distinct S1 epitopes including the inter-protomeric interfaces. This individual was also the only donor with detectable Fab responses to the HKU1 spike targeting the CTD, the SD1 and the SD2 (Fig. 1D ).

Samples from 3 donors (269, 1051 and 1412) were chosen for high resolution cryo-EMPEM studies with OC43 spike as they represented individuals with antibodies against all the unique epitopes observed (Table S1 ). We reconstructed 10 high-resolution maps of unique spike-Fab complexes ( Fig. 2A, Figs . S2-5, Table S2 ). High-resolution analysis of immune complexes from donor 269 revealed Fabs bound to the CTD, CTD-NTD interface and to NTD-site 1; NTDsite 1 Fab was not reconstructed during the ns-EMPEM studies. For donor 1051, cryo-EM analysis enabled differentiation of polyclonal Fabs targeting the CTD that were originally observed as a single species by ns-EM. We were unable to obtain reconstructions of either the SD1 or S2 antibody despite multiple attempts at focused classification, likely owing to low Fab abundance or dissociation of the complex during the cryogenic sample preparation process. For donor 1412, we reconstructed 5 of the 6 specificities seen in ns-EMPEM, targeting NTD-site 1, NTD-site 2, SD and inter-protomeric S1 interfaces. In all reconstructed maps we observed an additional non-spike density buried within a hydrophobic pocket in the CTD; the location and size resembling linoleic acid in SARS-2 spike (30, 31) . The MW of 254 g/mol obtained by mass-spectrometry analysis of the OC43 spike and the corresponding density in the OC43 map are however consistent with sapienic acid (6Z-hexadecenoic acid; Fig S7C ). Lastly, Fab10 (5 Å) targets the SD1 with primary interactions with Asp624, Glu646, Arg676, and glycans at Asn648 and Asn678 (Fig. S7D) . Importantly, Fab9 and Fab10 both make extensive contacts to the Asn675 glycan which represents an important immunogenic determinant within the SD1 epitope. An epitope summary of commonly elicited β-CoV spike antibodies in healthy human serum is shown in Fig. 3H and CDR lengths for Fabs1-7 determined by structural homology are summarized in table S3

Next, we sought to investigate the nature of spike antibodies in serum following SARS-2 infection. Sera from 3 CDs were screened for antibodies to SARS-2 spike by ns-EMPEM. Analysis of EM data (2D and 3D) revealed both NTD and RBD (or CTD) antibodies though the latter were relatively fewer in number and more difficult to reconstruct owing to the flexible RBD (Fig. 4A) .

While RBD antibodies have been well documented to provide protection against SARS-2 infection, NTD responses are being recognized as an important component of the neutralizing response to SARS-2, particularly those targeting the supersite comprising of residues 14 to 20, 140 to 158 and 245 to 264 (32) (33) (34) (35) (36) (37) . Collectively, these donors possessed several polyclonal antibodies targeting this supersite along with antibodies to other previously described sites (32) . Interestingly, we also observed antibody-pairs in both donors 1988 and 1989 that appeared to partially bind each other while also recognizing some part of the spike NTD (Fig. 4A) . It is unclear why these antibodies are triggered in SARS-2 donors, and the implications of this finding for understanding COVID pathology needs further investigation.

To obtain detailed molecular information on immunodominant epitopes within the SARS- Our ELISA data demonstrated that there is an increase in antibody binding titers to non-SARS-2 β-CoV spikes following an infection with SARS-2 virus ("convalescent donors") ( Fig   1A) . This finding indicates either a back-boost of pre-existing responses or elicitation of novel cross-reactive antibodies to conserved epitopes. Structural mapping of SARS-2 spike residues that are either identical to or have a conserved substitution in at least 3 of the 4 other β-CoVs, OC43, HKU1, SARS and MERS, revealed several conserved patches in the S2 subunit that could potentially elicit cross-reactive responses (Fig. S10A) . Several recent studies have found S2 as a target for cross-reactivity across β-CoVs (8, 9, 18, 19, 22) . Two individual studies also revealed SARS-2 spike residues in and around 560 to 572, 819 to 824, and 1,150 to 1,156 and their homologous regions on other HCoV spikes as being recognized with higher frequency in COVID-19 patients as compared to pre-COVID controls (Fig. S10A) (18, 20) . To determine if these epitopes are targeted following SARS-2 infection, we performed ns-EMPEM on CD sera with OC43, HKU1, SARS and MERS. As with HDs, the SARS-2 CDs had serum antibodies to the OC43 spike protein; antibodies to NTD-site 1 were seen in 2 donors, antibodies to interface in 2 donors and an S2 antibody was observed in 1 donor (Donor 1989) (Fig. 4A) . While we are uncertain if the S2 antibody was induced by SARS-2 infection, the antibody appears to target the helix 1014 to 1030 that is highly conserved across the β-CoV spikes (Fig. S10B) . Notably, donors who possessed high levels of OC43 antibodies also had some SARS-2-reactive antibodies prepandemic that did not correlate with protection against SARS-2 (9). When complexed with the HKU1 spike, we were able to detect antibodies in all 3 CD samples, which was higher than seen for HD (3D reconstructions were possible only for 1 of the 8 HD sera) suggesting an increase in HKU1 antibody titers following SARS-2 infection (Fig. 4A) . Of interest, Song et al. observed higher HKU1 spike antibody titers in post-COVID sera compared to pre-pandemic sera, whereas titers remained comparable for other HCoV spikes (8) . Whereas donors 1988 and 1989 had antibodies to the HKU1 CTD and/or the NTD, donor 1992 sera contained an S2 antibody binding to the base of the trimer. The epitope is analogous to that of the β-CoV cross-reactive spike monoclonal antibody (mAb) CC40.8 isolated from a COVID donor (Fig. S10C) ; CC40.8 binds strongly to SARS-2 and HKU1 spikes while also exhibiting some reactivity to SARS and OC43 spikes (8) . We also reconstructed a MERS spike antibody in donor 1989 that partly overlaps with the known MERS mAb G4 targeting the S2 connector domain near the trimer base ( Fig.   S10D ) (27) . The presence of a MERS-reactive antibody in a MERS-naive donor illustrates induction of cross-reactive responses following SARS-2 infection. We were not able to reconstruct any antibodies to SARS even though the CD sera had detectable titers against the spike. An overall comparison of antibody specificities between the HD and CD sera revealed antibody classes that were present in both the groups primarily targeting the S1 subunit while antibodies to the more conserved S2 subunit were enriched in the COVID donors. Collectively, these results suggest that SARS-2 infection triggers induction of cross-reactive antibodies to conserved β-CoV spike epitopes while some HCoV spike-specific antibodies may be back-boosted. This cross-boosting while associated in COVID-19 pathogenesis may also have long-lasting implications for immunity to seasonal CoVs as much of the population will be vaccinated and/or infected with SARS-2. 

Figures S1 to S10 

Epidemiology, genetic recombination, and pathogenesis of coronaviruses

Epidemiology and clinical presentations of the four human coronaviruses 229E, HKU1, NL63, and OC43 detected over 3 years using a novel multiplex real-time PCR method

Immunological characteristics govern the transition of COVID-19 to endemicity

Epidemiology characteristics of human coronaviruses in patients with respiratory infection symptoms and phylogenetic analysis of OC43 during 2010-2015 in Guangzhou

Human coronavirus circulation in the United States 2014-2017

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

Seasonal coronavirus protective immunity is short-lasting

Cross-reactive serum and memory B-cell responses to spike protein in SARS-2 and endemic coronavirus infection

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

Origin and evolution of pathogenic coronaviruses

A novel coronavirus from patients with pneumonia in China

Mechanisms of coronavirus cell entry mediated by the viral spike protein

Structure, function, and evolution of coronavirus spike proteins

The spike protein of SARS--a target for vaccine and therapeutic development

MERS spike protein: a key target for antivirals

SARS-2 SPIKE PROTEIN: an optimal immunological target for vaccines

Kinetics of viral load and antibody response in relation to COVID-19 severity

Viral epitope profiling of COVID-19 patients reveals cross-reactivity and correlates of severity

Severe COVID-19 patients display a back boost of seasonal coronavirus-specific antibodies

Epitope-resolved profiling of the SARS-2 antibody response identifies cross-reactivity with endemic human coronaviruses

Serologic cross-reactivity of SARS-2 with endemic and seasonal betacoronaviruses

Preexisting and de novo humoral immunity to SARS-2 in humans

Immunological imprinting of the antibody response in COVID-19 patients

Polyclonal antibody responses to HIV Env immunogens resolved using cryoEM

Electron-microscopy-based epitope mapping defines specificities of polyclonal antibodies elicited during HIV-1 BG505 envelope trimer immunization

Pre-fusion structure of a human coronavirus spike protein

Immunogenicity and structures of a rationally designed prefusion MERS spike antigen

Cryo-electron microscopy structures of the SARS spike glycoprotein reveal a prerequisite conformational state for receptor binding

Structural basis for human coronavirus attachment to sialic acid receptors

Structural analysis of full-length SARS-2 spike protein from an advanced vaccine candidate

Free fatty acid binding pocket in the locked structure of SARS-2 spike protein

N-terminal domain antigenic mapping reveals a site of vulnerability for SARS-2

Prevalent, protective, and convergent IgG recognition of SARS-2 non-RBD spike epitopes

Potent SARS-2 neutralizing antibodies directed against spike N-terminal domain target a single supersite

The receptor binding domain of the viral spike protein is an immunodominant and highly specific target of antibodies in SARS-2 patients

Potent neutralizing antibodies against multiple epitopes on SARS-2 spike

Potent neutralizing antibodies from COVID-19 patients define multiple targets of vulnerability