key: cord-0843426-iwd6femw authors: Camerini, D.; Randall, A. Z.; Trappl-Kimmons, K.; Oberai, A.; Hung, C.; Edgar, J.; Shandling, A.; Huynh, V.; Teng, A. A.; Hermanson, G.; Pablo, J. V.; Stumpf, M. M.; Lester, S. N.; Harcourt, J.; Tamin, A.; Rasheed, M.; Thornburg, N. J.; Satheshkumar, P. S.; Liang, X.; Kennedy, R. B.; Yee, A.; Townsend, M.; Campo, J. J. title: Mapping SARS-CoV-2 Antibody Epitopes in COVID-19 Patients with a Multi-Coronavirus Protein Microarray date: 2021-01-15 journal: nan DOI: 10.1101/2021.01.14.21249690 sha: 9bf971d57de1b6c6585163a51895ec9182c42907 doc_id: 843426 cord_uid: iwd6femw The emergence and rapid worldwide spread of SARS-CoV-2 has accelerated research and development for controlling the pandemic. A multi-coronavirus protein microarray was created containing full-length proteins, overlapping protein fragments of varying lengths and peptide libraries from SARS-CoV-2 and four other human coronaviruses. Sera from confirmed COVID-19 patients as well as unexposed individuals were applied to multi-coronavirus arrays to identify specific antibody reactivity. High level IgG, IgM and IgA reactivity to structural proteins S, M and N, as well as accessory proteins, of SARS-CoV-2 were observed that was specific to COVID-19 patients. Overlapping 100, 50 and 30 amino acid fragments of SARS-CoV-2 proteins identified antigenic regions. Numerous proteins of SARS-CoV, MERS-CoV and the endemic human coronaviruses, HCoV-NL63 and HCoV-OC43 were also more reactive with IgG, IgM and IgA in COVID-19 patient sera than in unexposed control sera, providing further evidence of immunologic cross-reactivity between these viruses. The multi-coronavirus protein microarray is a useful tool for mapping antibody reactivity in COVID-19 patients. A novel human coronavirus, which causes severe acute respiratory syndrome, now known as Previous studies have shown that COVID-19 patients rapidly seroconvert to SARS-CoV-2 and 39 produce IgM, IgG and IgA antibodies directed to several viral proteins (5) (6) (7) (8) . Reinfection challenge 40 studies in rhesus macaques showed that the humoral and cellular immune response to SARS- CoV-2 infection was effective in blocking reinfection (25, 26) . Nevertheless, it is not clear whether 42 all antibody responses are beneficial or whether some antibody responses to SARS-CoV-2 lead 43 to a less favorable course of disease (9, 10) . Moreover, enhancement of infection by antibodies 44 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.14.21249690 doi: medRxiv preprint NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. has been reported for severe acute respiratory syndrome coronavirus (SARS-CoV), which is 45 closely related to SARS-CoV-2 (11) (12) (13) 24 ). We have created and used a multi-coronavirus protein microarray, containing over one-thousand coronavirus proteins, protein fragments and peptides to map IgG, IgA and IgM antibody epitopes 48 in sera from COVID-19 patients. Our approach localizes the antibody reactivity of COVID-19 49 patients within SARS-CoV-2 proteins and allows us to map the antigenic regions bound. Furthermore, we can similarly measure the antibody reactivity of COVID-19 patients and healthy 51 controls with endemic human coronaviruses and with the two previous epidemic coronaviruses, 52 SARS-CoV and Middle East Respiratory Syndrome coronavirus (MERS-CoV). Our findings and 53 the multi-coronavirus protein microarray we created will be useful in discerning which de novo 54 and cross-reactive antibody responses to SARS-CoV-2 are protective and which may be less 55 useful in preventing disease or may even be detrimental. In addition, if high levels of antibody to 56 specific epitopes are found to be especially protective, the array could be used to screen 57 convalescent plasma for therapeutic potential and vaccine recipient sera as a preliminary (Table 1) . The multi-coronavirus array was incubated with sera from two sets of patient samples and 69 associated negative controls collected in different regions of the USA. The first set of sera from 70 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.14.21249690 doi: medRxiv preprint CoV nucleocapsid (N), S and membrane (M) proteins compared to healthy control sera (Fig. 1) . The receptor binding domain of the SARS-CoV-2 S protein (RBD) had overall weaker antibody . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.14.21249690 doi: medRxiv preprint differences between IgG binding of the negative and positive groups, and the differences are the is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.14.21249690 doi: medRxiv preprint 6 of each of these nine SARS-CoV-2 proteins and to produce the structural proteins and some 124 accessory proteins of HCoV-NL63, HCoV-OC43 and MERS-CoV. Using aa start and end 125 positions of each fragment within the protein, differential reactivity between the COVID-19 and 126 healthy donor groups was mapped in a circular heatmap for the structural proteins (Fig. 2) . This 127 analysis allowed us to identify antigenic regions in each SARS-CoV-2 structural protein. The 128 SARS-CoV-2 N protein showed the strongest reactivity in its carboxy terminal 100 aa fragment, 129 as well as in 50 aa fragments covering the same region. This region was recognized by IgG, IgA 130 and IgM with significant differential reactivity between COVID-19 patients and the healthy control 131 group. The middle of the N protein also had a region recognized by IgG and IgA identified by two 132 100 aa fragments. Together these antibody-reactive regions encompass about two thirds of the 133 N protein that likely contains at least two epitopes. The S1 protein also showed greatest IgG 134 binding near its carboxy terminus, in the penultimate 100 aa fragment. This antigenic region of 135 S1 was defined further by IgG and IgA reactivity with 50 aa fragments from aa 550 to 600. The 136 region containing the RBD was not strongly reactive when produced by IVTT. In contrast, the S2 137 protein of SARS-CoV-2 showed three regions of strong IgG, IgA and IgM binding and differential 138 reactivity with full-length, 100 aa and 50 aa fragments. Only the region near the carboxy terminus, 2). Less reactivity was seen in non-structural proteins, but significant reactivity of COVID-19 155 patient sera compared to control sera could be identified in fragments of the 3a and 7a accessory 156 proteins (Supplemental Figure 1 ). . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 15, 2021. 19 patient sera, but others were reactive in a subset of individuals. Heterogeneity was higher and 187 overall signal intensities were lower for IgA and IgM than for IgG. There were no significant 188 associations between age and sex with antibody levels in the positive group after adjustment for 189 the false discovery rate for any of the three isotypes (Supplemental Table 1 ). . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.14.21249690 doi: medRxiv preprint The S2 protein was reactive with patient IgG at a much higher frequency than in the controls for 1.1 x 10 -3 (Fig. 3) . The COVID-19 patient sera used in this study had less coronavirus reactive 236 IgM than IgG or IgA, perhaps because the samples were obtained during the convalescent phase 237 of disease. Nevertheless, significantly greater IgM reactivity was seen in patient sera compared 238 to control donor sera for four proteins and two protein fragments produced in vitro (Fig. 3) . These To visualize the relative importance of antibody isotype binding in differentiating COVID-19 253 positive sera from negative sera, the samples were projected in two dimensions for each isotype 254 using t-distributed stochastic neighbor embedding (tSNE; Fig. 4A ), a nonlinear machine learning 255 dimensionality reduction method which clusters together similar sets of multidimensional data. The thirty most reactive proteins for all isotypes were selected for this analysis to reduce the effect 257 of differing isotype background levels that would be notable in low-reactivity proteins (Fig. 4B ). is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.14.21249690 doi: medRxiv preprint The full-length SARS-CoV-2 N and S2 proteins as well as several fragments of both proteins had 277 the top nine largest mean differences in IgG reactivity between COVID-19 patients and healthy 278 controls (Fig. 4B) . These results were also statistically significant with t-test p values ranging from 279 2.1 x 10 -6 to 4.3 x 10 -2 (Supplemental Table 1 ). Antibody responses to HCoV-NL63, HCoV-OC43 280 and MERS-CoV proteins were also among the thirty most discriminatory antigens for 281 differentiating COVID-19 patients from control donors due to high reactivity with COVID-19 282 positive sera, while also demonstrating a considerable reactivity with negative. Nearly all the same 283 epitopes and regions of reactivity found for IgG were recapitulated by IgA reactivity as well, when 284 reactivity to the overlapping 100 aa, 50 aa and 30 aa protein fragments was analyzed (Fig. 3) . This includes the epitopes mapped in the SARS-CoV-2 N, S1, S2 and M proteins (Fig. 2) . . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. In this study of twenty COVID-19 patients, the strongest antibody responses to the SARS-CoV-2 356 proteins used on this array, for all antibody isotypes, were directed to the N and S2 proteins as 357 has been previously seen in other studies (5, 8, 14, 15) . We also detected antibody responses to 358 S1, M and accessory proteins 3a and 7a. Moreover, we were able to localize regions of each of 359 these SARS-CoV-2 proteins to which antibodies bound, by antibody reactivity with overlapping but not all of these predicted epitopes. In particular, of the top six predicted B cell epitopes in the 371 S protein we found significantly stronger reactivity with COVID-19 patient sera compared to 372 healthy donor sera for regions containing three epitopes: DIADTT, residues 568-573 near the 373 . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.14.21249690 doi: medRxiv preprint carboxy terminus of S1, PPIKD, residues 792-796 near the amino terminus of S2 and 374 VYDPLQPELDSF, residues 1137-1148 near the carboxy terminus of S2. The other three top 375 predicted B cell epitopes of the S protein, residues 405-428, 440-450 and 496-507, were not in 376 highly reactive regions of the S protein in our experiments, perhaps due to the overall low reactivity 377 of the S1 protein except for its carboxy terminal region or a need for native structure not found in 378 protein fragments produced in vitro. Similarly, we found COVID-19 specific reactivity for regions 379 including nine of the fourteen B cell epitopes in the N protein and one of three B cell epitopes in 380 the M protein predicted by Crooke et al. A few other groups have used protein or peptide arrays to map antibody reactivity to SARS-CoV-382 2 protein (15, (20) (21) (22) (23) . Two studies included full-length purified structural proteins from SARS-CoV-383 2, other human coronaviruses and diverse human retroviruses (15, 20) . Their results are Recently two groups published epitope maps of SARS-CoV-2 using phage display (33, 34) . One Limitations of our study are the small sample size and the inclusion of only convalescent samples. Despite these limitations, we were able to identify clear differences in the antibody response from 424 COVID-19 patients and healthy, non-exposed controls. The ideal dataset to further investigate 425 associations between preexisting antibody to specific epitopes and protection from severe is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.14.21249690 doi: medRxiv preprint Student's t-tests were used for comparison of the individual antibody response means between 516 the positive and negative groups. Comparison of the medians was done using Wilcoxon's rank 517 sum test. The area under the receiver operating characteristics curve (AUC) was calculated to 518 estimate delineation of groups for each antigen. The t-SNE analyses were calculated after 25,000 519 iterations with a perplexity parameter of 30 using the R package Rtsne (35) . Comparisons of the 520 proportions of responders to each protein between groups was done using two-proportions z-521 tests implemented by the 'prop.test' function in R. Correlation between antibody features and 522 between protein microarray and ELISA measurements used Pearson's correlation coefficient (ρ), 523 and association between antibody measurements and sample information such as sex, age and 524 cohort were modeled using linear regression. The association of specific antibody responses with 525 virus neutralization titers was estimated using linear regression with the values below detection 526 levels (<20) coded as zero, or by converting neutralization titers to ordinal values and estimating 527 the proportional odds ratio by ordinal logistic regession, whereby p values were estimated by 528 comparing the t-value against the standard normal distribution. Adjustment for the false discovery 529 rate was performed using the "p.adjust" function in R (36) . Data visualization was performed using 530 the circlize (37), ComplexHeatmap (38), ggplot2, heatmap2 and corrplot (39) packages in R. Unadjusted p values were shown in graphics. We thank the Laboratory Task Force of the CDC COVID_19 response for their project review and 535 resource support. This research was made possible using samples obtained from the CDC 536 Biorepository. The findings and conclusions in this report are those of the author(s) and do not necessarily CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.14.21249690 doi: medRxiv preprint Projecting transmission 562 dynamics of SARS-CoV-2 through the postpandemic period Estimates of the severity of coronavirus disease 2019: 566 a model-based analysis Antibody responses to SARS-CoV-2 in patients with COVID-19 Antibody responses to SARS-CoV-2 in patients of novel coronavirus disease Distinct features of SARS-CoV-2-specific IgA response in COVID-19 patients Profiling Early Humoral Response to Diagnose Novel Coronavirus Disease (COVID-19) Distinct Early Serological Signatures Track with SARS-CoV-2 Survival Heterogeneous antibodies against SARS-CoV-2 spike receptor binding domain 581 and nucleocapsid with implications on COVID-19 immunity Evasion of antibody neutralization in emerging severe acute respiratory syndrome 584 coronaviruses Severe Acute Respiratory Syndrome Coronavirus Spike Antibodies Trigger 587 Infection of Human Immune Cells via a pH-and Cysteine Protease-Independent FcgammaR 588 Antibody-dependent infection of human macrophages by severe acute respiratory 597 the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted Multi-epitope vaccine design using an immunoinformatics approach for 603 2019 novel coronavirus in China Potential T-cell and B-cell Epitopes of 2019-nCoV Analysis of Serologic Cross-Reactivity Between Common Human 609 Coronaviruses and SARS-CoV-2 Using Coronavirus Antigen Microarray Linear epitopes of SARS-CoV-2 spike protein elicit neutralizing antibodies in COVID-19 613 patients Potent neutralizing antibodies in the sera of 1 convalescent COVID-19 patients 616 2 are directed against conserved linear epitopes on the SARS-CoV-2 spike 3 protein Proteome-wide analysis of differentially-expressed SARS-CoV-2 antibodies in 620 early COVID-19 infection Immunodominant SARS Coronavirus Epitopes in Humans Elicited both 623 Enhancing and Neutralizing Effects on Infection in Non-human Primates SARS-CoV-2 infection protects against rechallenge in rhesus 627 macaques Primary exposure to SARS-CoV-2 protects against reinfection in rhesus 630 macaques 2020) the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted Evaluation of current medical approaches for COVID-19: a systematic review 636 and meta-analysis Convalescent Plasma Therapy and Its Effects On COVID-19 Patient 639 Outcomes: A Systematic Review of Current Literature Treatment of COVID-19 with convalescent plasma: lessons from 643 pastcoronavirus outbreaks Cryo-EM structure 647 of the 2019-nCoV spike in the prefusion conformation Validation of a SARS-CoV-2 spike protein ELISA for use 650 in contact investigations and serosurveillance. bioRxiv Viral epitope profiling of COVID-19 patients reveals cross-reactivity and 655 correlates of severity Multiplex Diagnostic Pipeline, Pans Human Sera for SARS-658 Controlling the false discovery rate: a practical and 661 powerful approach to multiple testing 2014) circlize implements and enhances circular visualization in R Complex heatmaps reveal patterns and correlations in multidimensional provided with only clinical and demographic information retained. The majority of samples (7/10) 468 were from patients that were not hospitalized with blood collected between 26 and 60 days post 469 symptom onset. Negative control sera were collected pre-COVID-19, in the fall of 2019. This is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 15, 2021. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)The copyright holder for this preprint this version posted January 15, 2021. ; https://doi.org/10.1101/2021.01.14.21249690 doi: medRxiv preprint