Serological Assays Estimate Highly Variable SARS-CoV-2 Neutralizing Antibody Activity in Recovered COVID19 Patients

28 The development of neutralizing antibodies (nAb) against SARS-CoV-2, following infection or 29 vaccination, is likely to be critical for the development of sufficient population immunity to drive cessation 30 of the COVID19 pandemic. A large number of serologic tests, platforms and methodologies are being 31 employed to determine seroprevalence in populations to select convalescent plasmas for therapeutic trials, 32 and to guide policies about reopening. However, tests have substantial variability in sensitivity and 33 specificity, and their ability to quantitatively predict levels of nAb is unknown. We collected 370 unique 34 donors enrolled in the New York Blood Center Convalescent Plasma Program between April and May of 35 2020. We measured levels of antibodies in convalescent plasma using commercially available SARS-CoV-36 2 detection tests and in-house ELISA assays and correlated serological measurements with nAb activity 37 measured using pseudotyped virus particles, which offer the most informative assessment of antiviral 38 activity of patient sera against viral infection. Our data show that a large proportion of convalescent plasma 39 samples have modest antibody levels and that commercially available tests have varying degrees of 40 accuracy in predicting nAb activity. We found the Ortho Anti-SARS-CoV-2 Total Ig and IgG high 41 throughput serological assays (HTSAs), as well as the Abbott SARS-CoV-2 IgG assay, quantify levels of 42 antibodies that strongly correlate with nAb assays and are consistent with gold-standard ELISA assay 43 results. These findings provide immediate clinical relevance to serology results that can be equated to nAb 44 activity and could serve as a valuable ‘roadmap’ to guide the choice and interpretation of serological tests 45 for SARS-CoV-2. 46 47 . CC-BY-NC-ND 4.0 International licenseIt 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 (which was not certified by peer review) The copyright holder for this preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint

48 In late 2019, a cluster of patients in Wuhan, the capital city of China’s Hubei providence, were 49 reported to be afflicted with a severe respiratory illness of unknown origin.(1, 2) Patients presented with 50 symptoms that included high fever, pneumonia, dyspnea, and respiratory failure. The causative agent was 51 identified to be severe acute respiratory syndrome coronavirus variant 2 (SARS-CoV-2), the 7th coronavirus 52 strain to infect humans to date,(3) and the clinical syndrome was designated coronavirus disease of 2019 53 (COVID19). The pathogenesis of COVID19 is similar to previously documented respiratory distress 54 syndromes caused by related coronaviruses, including the 2005 SARS coronavirus (SARS-CoV) and the 55 middle east respiratory syndrome coronavirus (MERS).(4) However, the greater transmissibility of SARS-56 CoV-2 has enabled a swift global spread that has resulted in substantial mortality. Detection and tracking 57 SARS-CoV-2 spread has been difficult. Moreover, the spectrum of symptomatology observed in SARS-58 CoV-2 infection is wide, ranging from asymptomatic and mild, reminiscent of numerous seasonal 59 infections, including influenza and common cold viruses, all the way to life-threatening respiratory failure 60 that requires intensive care and invasive ventilation. Currently, increased age and comorbidities are the 61 factors most highly predictive of severe of COVID19 disease.(5) 62 The utility of serological tests to identify individuals who have acquired antibodies against SARS-63 CoV-2 is thus recognized as both an indication of the seroprevalence of SARS-CoV-2 infection and, 64 potentially, of immunity afforded to the seropositive individual.(3, 6-8) Seroconversion is determined by 65 detection of antibodies that recognize SARS-CoV-2 antigens. Coronaviruses have 4 major structural 66 proteins: spike (S) protein (including the S1 protein and receptor binding domain (RBD)), nucleocapsid (N) 67 protein, membrane (M) protein and envelope (E) protein.(9) Previous studies of SARS-CoV and MERS 68 found the most immunogenic antigens are the S- and N-proteins,(10) and development of serological tests 69 for SARS-CoV-2 antibodies has focused heavily on these viral proteins. 70 Three major platforms of serological testing have been adopted; 1) enzyme linked immunosorbent 71 assays (ELISA), 2) high-throughput serological assays (HTSA), and 3) lateral flow assays (LFA). ELISAs 72 offer wide flexibility for research laboratories to select virtually any antigen of interest and provide highly 73 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint sensitive, quantitative results. HTSAs are more suitable for clinical laboratories and offer limited antigen 74 diversity but allow high-throughput and sensitive, semi-quantitative results. LFAs also offer limited 75 antigen diversity, but function with small volumes (~20µL) of whole blood, plasma or sera and allow rapid 76 (£15 minutes) results at the point of care. The clinical community will undoubtedly employ multiple 77 SARS-CoV-2 serology platforms but a comparative analysis across platforms has not been undertaken. 78 Further, it is currently unknown whether the detection of antibodies that bind these proteins predicts 79 neutralizing activity or protection against infection.(11) 80 Convalescent plasma (CP) transfusion has been recognized as a potential treatment for critically ill 81 COVID19 patients and the New York Blood Center (NYBC) has led the first COVID19 CP donation 82 program in the United States. Using 370 unique CP donor samples deposited in our COVID19 Research 83 Repository (https://nybc.org/covid19repository), we conducted ELISA, HSTA and LFA assays as well as 84 SARS-CoV-2 pseudovirus neutralization assays. We find that CP donors have a wide range of antibody 85 titers measured across multiple COVID19 serological and neutralization assays. Notably, we show that 86 some HTSA and ELISA assays predict neutralizing activity in vitro and may thus serve to predict antiviral 87 activity against SARS-CoV-2 in vivo. 88

89 Characteristics of the NYC CP Donor Population 90 Serological analysis of the CP donors was performed using 370 unique samples collected between 91 April and May of 2020 from the NYC area. CP donors enrolled in the program were required to have tested 92 positive for SARS-CoV-2 by PCR diagnostic tests and be symptom free for at least 2 weeks. To profile CP 93 donors, we cross-referenced donor demographic data to the 2010 U.S. Census database.(12) CP donors had 94 a median age of 41 years (95% CI: 39-44, range 17-75 years,) and showed a gaussian distribution (n=183, 95 r2=0.89) compared to the national median age of 38.2 years in 2018 (Figure 1A). The frequency of male 96 and female CP donors was 45.2% and 54.8%, respectively, and was not statistically different from the 97 national average of 49.2% and 50.8% (Figure 1B). The frequency of ABORh blood group antigens was 98 also largely consistent with the national frequency, with a slightly higher number of A- and O- donors and 99 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint slightly lower number of AB+ and B+ donors than expected (Figure 1C). Finally, CP donor ethnicity was 100 largely consistent with the national ethnic composition, with a slightly higher number of multiracial/other 101 donors and lower number of Black/African American donors than expected (Figure 1D). Overall, the 102 composition of NYC CP donors analyzed was reflective of the United States population demographic. 103 Neutralizing Activity of the CP Donor Population 104 Neutralization assays measure how effectively donor plasma or serum can inhibit virus infection of 105 target cells and are the gold standard for measuring the antiviral activity of antibodies. In the case of SARS-106 CoV-2, such assays require in biosafety level 3 (BSL-3) facilities and highly trained personnel. To 107 overcome this limitation and expedite testing, we employed pseudotyped virus assays based on either HIV-108 1 (human immunodeficiency virus type 1) or VSV (vesicular stomatitis virus). Both viruses were 109 engineered to lack their own envelope glycoproteins and to express a luciferase reporter gene. 110 Complementation in trans with the SARS-CoV-2 Spike (S) protein results in the generation of pseudotyped 111 virus particles that are dependent on the interaction between the S protein and its receptor ACE2 112 (angiotensin-converting enzyme 2) for entry into cells.(13) These reporter viruses were used to measure 113 infection of human cells engineered to express ACE2 (HIV-S assay) or expressed endogenous ACE2 114 (VSV-S assay) and to determine the ability of plasma dilutions to inhibit S-dependent virus entry. The NT50 115 values, reflecting the plasma dilution at which virus infection is reduced by 50%, were calculated for each 116 sample (Supplementary Figure 1A). 117 The neutralizing activity of CP donor samples was extremely variable and NT50 values obtained 118 ranged from <50 to over 20,000. The median NT50 values were 390.1 (95% CI: 278.3-499.7) and 450.6 119 (95% CI: 367.7-538.4) for the HIV-S or VSV-S assays, respectively (Figure 2A) and the two assays 120 showed a high degree of correlation (Supplementary Figure 1B-C). Fresh frozen plasma (FFP) samples 121 donated in 2019, before the SARS-CoV-2 outbreak, were used as negative controls (n=10). Importantly, the 122 NT50 values of all FFP samples were £50, which is the highest concentration of plasma used in the 123 neutralization assays and is hence designated as the signal cutoff (S/co) value. Overall, 83.1% and 92.7% of 124 the CP donor samples had detectable neutralization activity using HIV-S and VSV-S assays, respectively 125 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint (Figure 2B). Notably, 11.2% and 8.7% of CP donors had NT50 values at or greater than 2000 (40-fold over 126 S/co) using HIV-S and VSV-S, assays respectively while 55.8% and 52% of CP donors had NT50 values at 127 or less than 500 (10-fold over S/co) (Figure 2B). Thus, the majority of CP donors may have relatively 128 modest neutralizing activity and a small proportion of donors have high neutralization activity. 129 NT50 values were not statistically different between blood groups (Figure 2E, Supplementary 130 Figure 1G) or age groups (Figure 2C, Supplementary Figure 1E) and there was no linear correlation of 131 NT50 values with age (Supplementary Figure 1D) in contrast to previous reports.(14) However, in 132 agreement with recent studies,(15) NT50 values of male CP donor samples were ~1.7-fold higher than those 133 from female CP donors using HIV-S and VSV-S assays (Figure 2D and Supplementary Figure 1F, n = 134 195, p = 0.009 and <0.001, median difference 217 and 197, respectively). For CP donors where symptom 135 dates were reported, the time between last symptom and the date of donation was calculated. Interestingly, 136 CP donors 2-3 weeks post symptoms had a statistically significant increase in NT50 values compared to CP 137 donors >3 weeks post-symptom (Figure 2F and Supplementary Figure 1H, n=52, p = 0.03 and 0.04, 138 median difference 426 and 226, respectively). Overall, these data suggest CP donors possess a wide range 139 of neutralizing antibody levels that are proportionately distributed across demographic categories with the 140 exception of a small sex-dependent effect. 141 Serological Test Results of the CP Donor Population 142 Multiple platforms have been deployed to detect seroconversion against SARS-CoV-2. The 143 simplest tests are LFAs, which solubilize antibodies from whole blood, plasma or sera in an aqueous 144 mobile phase which moves across a nitrocellulose membrane coated with anti-human IgG and/or IgM to 145 distinguish between specific classes of immunoglobins while a control band ensures test function. Binding 146 of antibodies to antigen-conjugated enzyme, such as horseradish peroxidase, generates a colored band at 147 the test lines. Analysis of 144 CP donor samples showed that only 79.4% of CP donors tested positive for 148 SARS-CoV-2-specfic IgG antibodies and 24.8% for IgM antibodies (Figure 3A, top). While LFAs are not 149 designed to perform quantitatively, large discrepancies in band intensity between donors (Supplementary 150 Figure 2A) is often presumed to indicate semi-quantitative results. We performed densitometric analysis of 151 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint the test bands from LFA cassettes (Supplementary Figure 2B, 2C) and normalized each test to control 152 band intensity. LFAs showed an intensity range of 0% - 99.2% for IgG bands and 0% - 18.5% for IgM 153 bands, with a median intensity of 20% for IgG and <1% for IgM (Figure 3A, bottom). Thus, LFAs have a 154 high degree of variation in band intensity within the CP donor population. 155 HTSA systems offer the advantage of performing semi-quantitative seroconversion assays using 156 clinical laboratory testing infrastructure at large scale. We performed the Ortho-Clinical Diagnostics 157 VITROS SARS-CoV-2 total Ig assay, the VITROS SARS-CoV-2 IgG assay and the Abbott Labs Architect 158 SARS-CoV-2 IgG assay using between 100 and 330 CP donor plasma samples. We found 96.4% and 159 91.0% of CP donor samples were positive using the Ortho total Ig and IgG assays, respectively, and 91.4% 160 were positive using the Abbott IgG assay (Figure 3B). The median value of CP samples using the Ortho 161 total Ig assay was 101 arbitrary units (AU) (n=333, 95% CI: 78.5 – 123, S/co = 1, range 0 to 1000 AU) 162 while that of FFP healthy controls was 0.01 AU (n=8, 95% CI: 0.01 – 0.02). Similarly, the median value of 163 CP samples using the Ortho IgG assay was 11.7 AU (n=100, 95% CI: 8.3 – 16.07, S/co = 1, range 0 to ~30 164 AU). For the Abbott assay, the median value of CP samples was 6.04 AU (n=315, 95% CI: 5.48 – 6.44, 165 S/co = 1.4, range 0 to ~10 AU) while that of FFP healthy controls was 0.02 AU (95% CI: 0.01 – 0.15). 166 These results clearly show HTSA platforms detect a wide variation in antibody levels in the CP donor 167 population and offer greater dynamic range than LFA assays. 168 The gold standard for quantification of antigen-specific antibodies is ELISA assays. Studies of 169 antibody responses during SARS-CoV and MERS outbreaks identified the S- and N-proteins as the 170 dominant antigens. Therefore, we designed three indirect ELISA assays using SARS-CoV-2 recombinant, 171 His-tagged, spike protein S1 domain (S1), spike protein RBD domain (RBD) and nucleocapsid protein (N). 172 We utilized monoclonal antibodies demonstrated to bind antigen in a dose-dependent manner to generate 173 standard curves from which antibody concentrations were calculated and FFP from healthy controls to 174 determine signal cutoffs. Thus, we report our ELISA results as monoclonal antibody (mAb) titers. These 175 ELISA assays showed that 85.2%, 89.1%, and 96.3% of CP donor samples were positive for antibodies 176 against S1, RBD and N antigens, respectively (Figure 3C). Using the S1 ELISA, the median value for CP 177 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint donor samples was 445µg/mL (n=285, 95% CI: 342 – 536µg/mL, S/co = 120µg/mL) and for FFP controls 178 100.9µg/mL (n=10, 95% CI: 78 – 120µg/mL). In the NP ELISA the median value for CP donor samples 179 was 6432µg/mL (n=271, 95% CI: 2811 – 13792µg/mL, S/co = 700µg/mL) while in the RBD ELISA the 180 median value of CP donor samples was 15.6µg/mL (n=43, 95% CI: 12.55 – 25.6µg/mL, S/co = 4µg/mL). 181 Notably, the range of S1 and NP-binding antibody concentrations observed in the ELISAs was extreme, 182 constituting a 1,000-fold difference in titers within the CP donor population. Taken together, these data 183 demonstrate that CP donors have a wide range of concentrations of antibodies specific to immunogenic 184 SARS-CoV-2 antigens, as measured across multiple serological platforms. 185 Correlation of Serology Tests with Neutralizing Activity 186 It is not logistically feasible to implement neutralization assays as a measurement of antiviral 187 antibodies at a scale of the general population. While quantification of seroconversion is practiced, 188 controlled studies that determine the relationship between quantitative SARS-CoV-2 serology test results 189 and neutralizing activity is sparse. We examined the correlation between serology and neutralization assays 190 in the CP donor samples (Figure 4A, Supplementary Figure 3A, Supplementary Figure 4C). As 191 expected, S1 ELISA titers showed a positive linear regression with NT50 values (r2 = 0.35) while the RBD 192 ELISA titers showed slightly higher linearity (r2 = 0.38), commensurate with the fact that the RBD is a key 193 target for neutralizing antibodies. Conversely, NP ELISA titers showed a comparatively lower degree of 194 linear regression with neutralization activity (r2 = 0.09). By comparison, both the Ortho HTSA total Ig 195 assay and the IgG assay showed higher (r2 = 0.45 for both) while the Abbott HTSA IgG assay showed 196 lower linear regression with neutralization activity (r2 = 0.24). Although Ortho HTSAs and the Abbott 197 HTSA IgG platforms quantify antibodies against S1 and NP antigens, respectively, a linear regression of 198 r2=0.33 was calculated between these two HTSAs (Supplementary Figure 3B). As expected, linear 199 regression between the Ortho total Ig and IgG assay was strong (r2 = 0.72) since the two assays measure the 200 same epitope. LFA IgG densitometry measurements showed the poorest correlation with neutralization 201 activity (r2 = 0.22). 202 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint Correlation between serological results and neutralization activity was also examined using the non-203 parametric Spearman test that does not assume linear dependence (Figure 4B). As expected, a high 204 correlation between the HIV-S and VSV-S neutralization assays was obtained (r=0.89). The Ortho and 205 Abbott HTSA platforms exhibited the highest degree of correlation with neutralization among the serology 206 assays tested (r = 0.75 and 0.72, respectively for the HIV-S assay; 0.70 and 0.69 for the VSV-S assay). The 207 S1, RBD, and NP ELISAs also showed a high degree of correlation, particularly with the HIV-S 208 neutralization assay (r = 0.69, 0.65, and 0.65) while the LFA IgG and IgM assay showed the poorest 209 correlation (r = 0.56, 0.41). Taken together, the data demonstrate that all quantitative serological assays 210 correlate to some degree with neutralization activity. However, HTSA and S1 ELISA assays that measure 211 anti-spike protein antigens have the highest predictive value as a surrogate for pseudovirus neutralization 212 assays. Importantly, correlation between HTSA scores and NT50 values suggest presumptive ranges of 213 neutralizing activity based on ranges of HTSA values (Figure 4C, Supplementary Figure 4A). 214 While ELISA assays revealed S1 and N antibody titers correlated with each other, these titers were 215 not always proportional among CP donor samples. To examine the coincidence of S1 and NP antibody 216 titers and using FFP plasma samples as negative controls, we categorized S1 and N antibody titers that fell 217 below S/co values as ‘negative’ and titers greater 10-fold over S/co as ‘high’ (Supplementary Figure 4B). 218 Using 241 CP donor samples that were assayed with both the S1 and N ELISA assays, we found that 81% 219 of donors were double positive (DP), while 16% of samples were single positive (14% N and 2% S1, 220 respectively) (Figure 4D). Only 2.5% of CP donors were double negative for S1 and NP antibodies. Within 221 the double positive population, we found that 23% of samples were DPhigh while 5% and 30% of samples 222 were only S1high or Nhigh and the remaining 42% were DPlow. We then examined the distribution of NT50 223 values from the HIV-S neutralization assay within these populations (Figure 4E). Notably, DN samples 224 showed NT50 values at the S/co observed for FFP healthy control samples while DPlow samples had 225 relatively low NT50 values (median value 327, 95% CI: 186 – 444). Importantly, the DPhigh donors had 226 NT50 values that were 7-fold higher than DPlow donors (median value 2130). Additionally, NT50 values in 227 the Nhigh and S1high groups were 2.5- and 4-fold higher than those of the DPlow group. 228 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint Finally, we sought to determine if the frequency of peripheral blood immune cells varied as a 229 function of antibody titer. We stained peripheral blood mononuclear cells (PBMCs) isolated from CP donor 230 buffy coats for classical surface markers associated with B-cell or T-cell populations (Supplementary 231 Figure 5A, 5B). We examined T cell subsets including T central memory (CD45RO+CD62L+) and T 232 effector memory (TEM; CD45RO+CD62Lneg) while B cell (CD20+) subsets analyzed included memory B 233 cells (CD27+CD24+), plasmablasts (CD24negCD38hiCD138neg) and the more mature plasma cells 234 (CD24negCD38hiCD138+) (Supplementary Figure 5C). We found statistically significant differences in 235 naïve CD4 and CD8 T-cell populations in donors with high S1 ELISA titers compared to those with low 236 titer. Decrease in CD24+CD27+ memory B cells was detected in individuals with higher anti-S titers. 237 Although the cause of this lower frequency is not known, it could raise the possibility that individuals with 238 reduced memory B cells may develop a less robust antibody response with future infections. Although our 239 phenotypic analysis of B and T cell compartments was limited, these data suggest phenotypic differences in 240 canonical B and T cell populations are insufficient to explain the large differences in antibody titers or 241 neutralization activity observed in CP donors and warrants future studies designed to study B and T cell 242 function from individual donors. 243

244 Demographic limitations of the CP donor population 245 Recent studies have noted a disproportion in COVID19 morbidity and mortality among minority 246 communities.(16) In this study, of the 370 CP samples analyzed, only 204 donors (55%) elected to identify 247 ethnicity, representing the least reported demographic category we collected. Nevertheless, we did not 248 observe a significant difference in nAb or serology results as a function of any demographic metric, 249 including ethnicity. Although we showed that the CP donor samples analyzed in this study comprised a 250 relatively normal distribution of demographic indicators, based on the U.S. census data, we acknowledge 251 that some factors, including ethnicity, are underrepresented in this cohort and limit the interpretation of the 252 study beyond the population aggregate. The potential explanations of this phenomenon are complex and 253 extend beyond the scope of this study.(17) The blood banking community is continuously working to 254 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint recruit minority donors, who are consistently underrepresented amongst regular blood donors.(18) Efforts 255 to increase public participation in local blood and CP donor programs would both improve blood product 256 diversity of transfusion products and strengthen the rigor of epidemiological disease. Thus, studies 257 designed to characterize serological responses to COVID19 specifically in minority groups are warranted 258 and necessary to augment our current understanding of the pandemic. 259 Seroconversion assays of the population 260 Quantification of antiviral antibodies in recovered individuals is an important metric for 261 determining population immunity conferred by exposure to SARS-CoV-2. Our study suggests that most 262 New York City convalescent plasma donors have antibodies against SARS-CoV-2. Indeed, our data 263 demonstrate that the HTSA, including Ortho and Abbot assays, which have received EUA from the FDA, 264 are well suited to quantify a wide range of antibody titers and reported that 91 – 96.4% of the CP 265 population possesses detectable SARS-CoV-2 antibodies. LFAs performed less well, and individuals with 266 low antibody titers scored weakly positive or negative in LFAs. Such outcomes could be interpreted 267 incorrectly, thus increasing the rate of false negative results. Ultimately, studies that accurately document 268 SARS-CoV-2 seroprevalence in diverse populations will require highly sensitive, high quality assays such 269 as HSTA or ELISA to be reliable. 270 Correlation between serological assay measurements and neutralizing activity 271 Since patient recovery often precedes the development of efficacious and safe therapeutics, a 272 longstanding treatment strategy for infectious diseases is passive antibody transfer. Therefore, refining 273 strategies to improve CP infusion efficacy benefits both the current treatment options of COVID-19 and 274 will inform the medical community for future pandemics. Our serological analyses are consistent with 275 previous publications that show a considerable range in antibody titers in recovered COVID19 patients.(19) 276 However, this study provides a comprehensive analysis of the correlation of quantitative serological test 277 values with neutralization activity. Importantly, high dynamic range serological assays, such as the HTSA 278 and S1 ELISA, had a significant linear correlation with neutralization activity. We show, for the first time, 279 the extent to which three widely available SARS-CoV-2 HTSAs correlated to nAb activity as well as to 280 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint each other, providing the clinical and scientific community with a comprehensive overview of clinical 281 serology test performance. To this end, investigators from the Mayo Clinic’s COVID-19 Expanded Access 282 Program (EAP) performed an exploratory analysis on the efficacy of CP as a therapeutic agent using data 283 from over 35,000 transfusions.(20) Although the study showed uncertainty as to the statistical significance 284 of effect, the authors noted that patients transfused with high antibody titer CP units, quantified by the 285 Ortho IgG assay, showed a notable reduction in the odds ratio of mortality at both 7 and 30 days after 286 transfusion. These data support the assertion that antibody quantification of CP units using high dynamic 287 range HSTA assays may further improve therapeutic options for COVID-19 and, perhaps, future pandemic 288 responses. This knowledge will also be necessary for deriving potential serologic correlates of 289 protection,(21) and may aid in predicting immunity at the individual and population levels.(15) 290 Yet, the levels of plasma neutralizing activity required to prevent SARS-CoV-2 re-infection are 291 currently unknown. Anecdotal results have been reported for seasonal coronavirus experimental infection 292 studies. For example, one study of 229E HCoV found a positive correlation between pre-infection antibody 293 titer and neutralization activity with symptom clinical severity.(22) In another study, 7 of 8 individuals with 294 low neutralizing titers excreted virus upon re-exposure, compared to only 1 of 4 subjects with higher 295 titers.(23) However, the conclusions of these studies are not directly comparable to the current SARS-CoV-296 2 pandemic. As such, the necessity of human epidemiological or vaccination studies are necessary to 297 determine the minimum threshold of neutralizing activity necessary to prevent SARS-CoV-2 re-infection. 298 Conversely, sub-neutralizing antibody levels have been reported to facilitate, rather than inhibit, viral entry 299 of the some coronaviruses in vitro, through antibody dependent enhancement (ADE).(24-26) While ADE 300 dependent replication has not been demonstrated to occur in SARS-CoV, viral uptake into macrophages via 301 antibody association with Fc receptors does induce IL-6 and TNFa cytokines which may promote 302 inflammation and tissue damage.(27) Insights gained from an accurate analysis of antibody levels and 303 neutralization activity in SARS-CoV-2-infected individuals will help address these important questions and 304 the corresponding health consequences. 305 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint A key biological question is: what underlies the large variation in antibody titers (neutralizing or 306 otherwise) observed in CP donors? Numerous variables, including the effectiveness of innate immune 307 responses, SARS-CoV-2 exposure dose, anatomical site of initial infection, and partial cross-reactive 308 immunity conferred by prior seasonal coronavirus infection, could all impart variation on the amount and 309 dissemination of SARS-CoV-2 antigen. Variation in the exposure of the adaptive immune system to SARS-310 CoV-2 antigen would, in turn, likely impact the magnitude of immune responses. Our observation that the 311 levels of antibody to N, as well as S, correlates with S-specific neutralizing titer suggests that quantitative 312 differences in the overall adaptive immune response to SARS-CoV-2, rather than intrinsic differences in 313 the ability of individuals to mount neutralizing responses, at least partly explains the large variation in 314 neutralizing capacity of CP. This notion is consistent with recent findings that all individuals examined, 315 generated very similar, potent monoclonal SARS-CoV-2 neutralizing antibodies, but at very different 316 levels.(15) 317 Future utility for vaccine and CP donor strategies 318 The development of efficacious vaccines against SARS-CoV-2 may be necessary for ending the 319 COVID19 pandemic. Clinical trials will undoubtedly include a battery of serological and neutralization 320 assays in test subjects to assess candidate vaccine efficacy. Surrogate serology tests to neutralizing activity 321 could help to rapidly inform as to the likely effectiveness, as well as immunogenicity, of vaccines against 322 SARS-CoV-2. To this end, real-time analyses using scalable HTSA testing platforms is effectuate while 323 future studies are conducted to more precisely measure in vivo neutralization activity. 324 Finally, the utility of convalescent plasma in the treatment of infection has been recognized since 325 the turn of the 20th century.(28) CP transfusion is thought to be effective through passive immunization, 326 specifically the transfer of neutralizing antibodies from a recovered individual to another individual 327 manifesting life-threatening symptoms.(29, 30) Previously CP therapy has been used to treat both SARS 328 and MERS,(31) and currently can be rapidly deployed against SARS-CoV-2 while other therapies are 329 under development.(32) Nevertheless, many questions remain regarding the optimal antibody levels 330 necessary to treat patients at varying stages of COVID19 disease. Accurate quantification using serological 331 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint assays that predict neutralization activity may improve clinical outcomes through refinement of CP unit 332 selection for patients of varying symptomatology. In summary, we demonstrate that HTSA and S1 ELISA 333 assays show the strongest correlation with neutralization activity and may serve to predict the degree of 334 antiviral antibody activity present in recovered patients or vaccine recipients. 335 336 Authors’ Contributions 337 LLL conceptualized the study, designed and performed serology experiments and managed the collection 338 of data, figures and statistical analyses. LLL, PB, TH and CDH co-wrote the manuscript. TH, PB, FM, YW 339 and FS designed and performed the neutralization assays. BR, DJ, WB, SJ, JP, MR and NT performed most 340 of the serology experiments. CG, MP, ES and HZ processed and preserved donor plasma and PBMCs. DS 341 coordinated donor demographic information. KY contributed to PBMC flow cytometry and interpretation. 342 DJ and BR contributed equal authorship to the manuscript. LLL, PB and TH contributed equal 343 corresponding authorship. 344 345

346 We thank Jill Alberigo, Amanda Brites and Kelly Brightman from Rhode Island Blood Center for their help 347 with performing the Ortho Anti-SARS-CoV-2 Total Test and the Abbott SARS-CoV-2 IgG test. We thank 348 Chockalingam Palaniappan and Paul Contestable for their assistance with performing the Ortho Anti-349 SARS-CoV-2 IgG Test. We thank Haidee Chen for assistance with editing the manuscript. 350 351 Conflicts of Interest 352 The authors declare no conflicts of interest. 353 354 Role of the Funding Source 355 Funding source for TH, PBD, FM, YW and FS were NIH R01AI78788 and R37AI064003. Funding 356 sources did not have a role in the writing of the manuscript or the decision to submit for publication. 357 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint 358 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint

359 Cell lines 360 Huh7.5 cells were a gift from Charles Rice (33). The 293T/ACE2cl.13 cell clone was generated by 361 transducing 293T cells (ATCC ® CRL-3216™) with a CSIB -based ACE2 lentivirus expression vector 362 containing a cDNA encoding a catalytically inactive ACE2 mutant. Single cell clones were isolated by 363 limiting dilution and one clone (293T/ACE2cl.13) was used in these studies. 364 Collection of CP donor information, isolation of convalescent plasma and PBMCs 365 Disclosure of demographic information was elective at the time of donation and showed that of the 370 CP 366 donors analyzed, 71.1% indicated age, 95.4% indicated blood type, 95.6% indicated sex and 55.1% indicated 367 ethnicity. To examine the demographic characteristics within the convalescent plasma (CP) donor population, 368 we used the 2010 U.S. Census demographic data as expected frequencies. Plasma was isolated from EDTA- 369 anticoagulated human whole blood sa mples. Samples were shipped from the NYBC Sample Management 370 Facility overnight at 4C and centrifuged for 5 min at 500 xg to facilitate plasma/cell phase separation. The 371 resulting upper plasma layer was extracted, aliquoted to minimize future freeze -thaw cycles, and stored at -372 80 C. Samples were cryopreserved and stored in the NYBC COVID19 Research Repository 373 (https://nybc.org/covid19repository). 374 Plasmid constructs 375 The env-inactivated HIV-1 reporter construct (pHIV-1NL4-3 ΔEnv-NanoLuc) was generated from a pNL4-3 376 infectious molecular clone (obtained through NIH AIDS Reagent Program, Division of AIDS, NIAID, NIH 377 from Dr Malcolm Martin). It contains a NanoLuc Luciferase reporter gene in place of nucleotides 1-100 of 378 the nef-gene and a 940 bp deletion 3’ to the vpu stop-codon. The rVSVΔG/NG/NanoLuc plasmid was 379 generated by insertion of a cassette containing an mNeonGreen/FMDV2A/NanoLuc luciferase cDNA into 380 rVSVΔG (Kerafast) (PMID: 20709108) between the M and L genes. The pSARS-CoV-2 S-protein 381 expression plasmid containing a C-terminally truncated SARS-CoV-2 S protein (pSARS-CoV2Δ19) was 382 generated by insertion of a synthetic human-codon optimized cDNA encoding SARS-CoV-2 S1 spike protein 383 lacking the C-terminal 19 codons into pCR3.1. An ACE2 lentiviral expression vector was constructed by 384 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint inserting a cDNA encoding a catalytically inactive ACE2 mutant into the lentivirus expression vector CSIB 385 (PMID: 30084827). 386 SARS-CoV-2 pseudotype particles 387 To generate (HIV/NanoLuc) -SARS-CoV-2 pseudotype particles, 293T cells were transfected with pHIV -388 1NL4-3 ΔEnv-NanoLuc reporter virus plasmid and pSARS-CoV-2-SΔ19 at a molar plasmid ratio of 1:0.55. The 389 transfected cells were washed twice with PBS the following day, and at 48h after transfection, supernatant 390 was harvested, clarified by centrifugation, passed through a 0.22 µm filter, aliquoted and frozen at -80oC. 391 To generate (VSV/NG/NanoLuc)-SARS-CoV-2 pseudotype particles, 293T cells were infected with 392 recombinant T7-expressing vaccinia virus (vTF7 -3) and transfected with rVSV ΔG/NG/NanoLuc, pBS-N, 393 pBS-P, pBS -L, and pBS -G (PMID: 20709108). At ~24h post transfection the superna tant was collected, 394 filtered and used to infect 293T cells transfected with a VSV -G expression plasmid, for amplification. To 395 prepare stocks of (VSV/NG/NanoLuc)-SARS-CoV-2 pseudotype particles, 293T cells were transfected with 396 pSARS-CoV2Δ19 and infected with the VSV-G complemented rVSVΔG/NG/NanoLuc virus. At 16h later the 397 supernatant was collected, clarified by centrifugation, filtered, pelleted through a 20% sucrose cushion and 398 stored at -80oC. The viral stock was incubated with 20% I1 hybridoma supernatan t (ATCC CRL-2700) for 399 1h at 37oC before use. 400 Neutralization assays 401 To measure neutralizing antibody activity in convalescent plasma, five -fold serial dilutions of plasma were 402 incubated for 1 hour at 37 oC in 96 -well plates with an aliquot of HIV -1 or VSV -based SARS -CoV-2 403 pseudotyped virus containing approximately 1x10 3 infectious units. Thereafter, 100 µl of the plasma/virus 404 mixture was added to target cells (293TAce2 cl.13, or Huh7.5) cells in 96-well plates. Cells were cultured for 405 48h (HIV-1 pseudotype viruses) or 16h (VSV pseudotype viruses). Then, cells were washed twice, lysed and 406 NanoLuc Luciferase activity in lysates was measured using either the Nano -Glo Luciferase Assay System 407 (Promega) and a Modulus II Microplate Multimode reader (Turner BioSystem) or a Glowmax Navigator 408 luminometer (Promega). The half maximal neutralizing titer (NT 50) for plasma, was determined using a 4 -409 parameter nonlinear regression in Prism 8.4 (GraphPad). 410 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint Lateral Flow ImmunoAssay (LFA) 411 Lateral flow immunoassays (LFAs) were pro vided by external companies. Assay cartridges contained 412 detection bands for IgG and IgM against SARS-CoV2 specific epitopes as well as an internal positive control. 413 For each assay, 20 µL convalescent plasma or serum was applied to the sample pad, followed by two drops 414 of proprietary running buffer. After 30 minutes, high resolution pictures of the detection zone were taken and 415 saved as .JPEG files. All tests were performed at room temperature. 416 LFA Densitometry Analysis 417 Relative quantification of anti -SARS-CoV-2 IgG and IgM in convalescent plasma samples was performed 418 using built-in gel analysis macros in FIJI (https://fiji.sc/). A rectangular selection covering the detection zone 419 was analyzed using Analyze>Gels>Plot Lanes. Integrated density values were outlined manually and 420 extracted from the resulting plot. Using MS Excel, IgG and IgM values were normalized against the density 421 of the control band. 422 The remaining whole blood cellular phase was supplemented with 2 mL of 35 g/L HSA/DPBS and diluted 423 1:1 with DPBS. Diluted whole blood was layered over 7 mL Ficoll-Paque Premium 1.078 g/mL (GE 424 Healthcare) and centrifuged for 20 minutes at 20C and 400xg without braking. Buffer coats were extracted, 425 counted with AOPI viability stain using the Cellometer Auto2000 (Nexelom Bioscience LLC), and frozen in 426 PBMC freezing media (10% DMSO in Knockout SR). 427 SARS-CoV-2 Binding-Antibody ELISA 428 Flat-well, nickel-coated 96 well ELISA plates (Thermo Scientific) were coated with 2µg/mL of recombinant 429 S1 spike protein, nucleocapsid protein, or Receptor Binding Domain (RBD) spike protein specific to SARS-430 CoV-2 in resuspension buffer (1% Human Serum Albumin in 0.01% PBST) and incubated in a stationary 431 humidified chamber overnight at 4 C. On the day of the assay, plates were blocked for 30 min with ELISA 432 blocking buffer (3% W/V non-fat milk in PBST). Standard curves for both S1 and RBD assays were 433 generated by using mouse anti-SARS-CoV spike protein monoclonal antibody (clone [3A2], ABIN2452119, 434 Antibodies-Online) as the standard. Anti-SARS-CoV-2 Nucleocapsid mouse monoclonal antibody (clone 435 [7E1B], bsm-41414M, Bioss Antibodies) was used as a standard for nucleocapsid binding assays. 436 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint Monoclonal antibody standard curves and serial dil utions of convalescent donor plasma were prepared in 437 assay buffer (1% non -fat milk in PBST) and added to blocked plates in technical duplicate for 1 hr with 438 orbital shaking at room temperature. Plates were then washed three times with PBST and incubated fo r 1 hr 439 with ELISA assay buffer containing Goat anti-Human IgA, IgG, IgM (Heavy & Light Chain) Antibody-HRP 440 (Cat. No. ABIN100792, Antibodies-Online) and Goat anti-Mouse IgG2b (Heavy Chain) Antibody-HRP (Cat. 441 No. ABIN376251, Antibodies -Online) at 1:30000 and 1:3000 dilutions, respectively. Plates were then 442 washed three times, developed with Pierce TMB substrate for 5 min, and quenched with 3 M HCl. 443 Absorbance readings were collected at 450 nm. Standard curves were constructed in Prism 8.4 (Graphpad 444 Software Inc.) using a Sigmoidal 4PL Non-Linear Regression (curve fit) model. 445 High-throughput Serology Assays 446 Convalescent donor plasma samples were barcoded and dispatched to Rhode Island Blood Center (RIBC). 447 Samples were analyzed using the Abbott SARS -CoV-2 IgG chemiluminescent microparticle immunoassay 448 with the Abbott Architect i2000SR (Abbott Core Laboratories), as well as the VITROS Immunodiagnostic 449 Products Anti-SARS-CoV-2 Total Test and the Anti-SARS-CoV-2 IgG Test with the VITROS 5600 (Ortho 450 Clinical Diagnostics). All assays were performed by trained RIBC employees according to the respective 451 manufacturer standard procedures. 452 Flow cytometric analysis of PBMCs 453 Cryopreserved PBMCs were thawed, filtered and stained with a B -cell or T -cell antibody cocktail for 3 0 454 minutes in PBS. Cells were washed with PBS and analyzed with a BD LSR Fortessa 4 laser cytometer. 455 Cytometric analysis was performed using RUO FCS Express 7 (DeNovo Software). 456 457 458 459 460 461 462 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint 463 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint

464 1. F. Wu et al., A new coronavirus associated with human respiratory disease in China. Nature 579, 265-269 (2020). 465 2. P. Zhou et al., A pneumonia outbreak associated with a new coronavirus of probable bat origin. Nature 579, 270-273 466 (2020). 467 3. K. G. Andersen, A. Rambaut, W. I. Lipkin, E. C. Holmes, R. F. Garry, The proximal origin of SARS -CoV-2. Nat Med 468 26, 450-452 (2020). 469 4. M. Xie, Q. Chen, Insight into 2019 novel coronavirus - An updated interim review and lessons from SARS -CoV and 470 MERS-CoV. Int J Infect Dis 94, 119-124 (2020). 471 5. Q. Yao et al., Retrospective study of risk factors for severe SARS -Cov-2 infections in hospitalized adult patients. Pol 472 Arch Intern Med 10.20452/pamw.15312 (2020). 473 6. A. J. Kucharski et al., Early dynamics of transmission and control of COVID-19: a mathematical modelling study. Lancet 474 Infect Dis 20, 553-558 (2020). 475 7. C. M. Chan et al. , Examination of seroprevalence of coronavirus HKU1 infection with S protein -based ELISA and 476 neutralization assay against viral spike pseudotyped virus. J Clin Virol 45, 54-60 (2009). 477 8. C. Y. Lee, R. T. P. Lin, L. Renia, L. F. P. Ng, Serological Approaches for COVID -19: Epidemiologic Perspective on 478 Surveillance and Control. Front Immunol 11, 879 (2020). 479 9. X. Zhong et al., B-cell responses in patients who have recovered from severe acute respiratory syndrome target a dominant 480 site in the S2 domain of the surface spike glycoprotein. J Virol 79, 3401-3408 (2005). 481 10. W. Tai et al. , Characterization of the receptor -binding do main (RBD) of 2019 novel coronavirus: implication for 482 development of RBD protein as a viral attachment inhibitor and vaccine. Cell Mol Immunol 10.1038/s41423-020-0400-4 483 (2020). 484 11. L. A. VanBlargan, L. Goo, T. C. Pierson, Deconstructing the Antiviral Neutralizing-Antibody Response: Implications for 485 Vaccine Development and Immunity. Microbiol Mol Biol Rev 80, 989-1010 (2016). 486 12. DHHS, U.S. Census Bureau. 487 13. M. Hoffmann et al., SARS-CoV-2 Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically Proven 488 Protease Inhibitor. Cell 181, 271-280 e278 (2020). 489 14. A. W. Fan Wu, Mei Liu, Qimin Wang, Jun Chen, Shuai Xia, Yun Ling, Yuling Zhang, Jingna Xun, Lu Lu, Shibo Jiang, 490 Hongzhou Lu, Yumei Wen, Jinghe Huang, Neutralizing antibody responses to SARS-CoV-2 in a COVID-19 recovered 491 patient cohort and their implications. medRxiv 10.1101/2020.03.30.20047365 (2020). 492 15. C. G. Davide F. Robbiani, Frauke Muecksch, Julio C. C. Lorenzi1 , Zijun Wang, Alice Cho, Marianna Agudelo, 493 Christopher O. Barnes, Anna Gazumyan, Shlomo Finkin, Thomas Hagglof, Thiago Y. Oliveira, Charlotte Viant, Arlene 494 Hurley, HansHeinrich Hoffmann, Katrina G. Millard, Rhonda G. Kost, Melissa Cipolla, Kristie Gordon,, S. T. C. Filippo 495 Bianchini1, Victor Ramos, Roshni Patel, Juan Dizon, IrinaShi meliovich1, Pilar Mendoza, Harald Hartweger, Lilian 496 Nogueira, Maggi Pack, Jill Horowitz, Fabian Schmidt, Yiska Weisblum, Eleftherios Michailidis, Alison W. Ashbrook, 497 Eric Waltari,John E. Pak, Kathryn E. Huey -Tubman, Nicholas Koranda, Pauline R. Hoffman, An thony P.West, Jr., 498 Charles M. Rice, Theodora Hatziioannou, Pamela J. Bjorkman, Paul D. Bieniasz, Marina Caskey, Michel C. 499 Nussenzweig, Convergent Antibody Responses to SARS -CoV-2 Infection in Convalescent Individuals. 500 10.1101/2020.05.13.092619. (2020). 501 16. C. P. Gross et al. , Racial and Ethnic Disparities in Population -Level Covid -19 Mortality. J Gen Intern Med 502 10.1007/s11606-020-06081-w (2020). 503 17. B. H. Shaz, A. B. James, K. L. Hillyer, G. B. Schreiber, C. D. Hillyer, Demographic patterns of blood donors and donations 504 in a large metropolitan area. J Natl Med Assoc 103, 351-357 (2011). 505 18. B. H. Shaz, C. D. Hillyer, Minority donation in the United States: challenges and needs. Curr Opin Hematol 17, 544-549 506 (2010). 507 19. A. W. Fan Wu, Mei Liu, Qimin Wang, Jun Chen, Shuai Xia, Yun Ling, Yuling Zhang, Jingna Xun, Lu Lu, Shibo Jiang, 508 Hongzhou Lu, Yumei Wen, Jinghe Huang, Neutralizing antibody responses to SARS-CoV-2 in a COVID-19 recovered 509 patient cohort and their implications. 10.1101/2020.03.30.20047365. 510 20. M. J. Joyner et al., Effect of Convalescent Plasma on Mortality among Hospitalized Patients with COVID -19: Initial 511 Three-Month Experience. medRxiv 10.1101/2020.08.12.20169359 (2020). 512 21. P. Kellam, W. Barclay, The dynamics of humoral immune responses following SARS-CoV-2 infection and the potential 513 for reinfection. J Gen Virol 10.1099/jgv.0.001439 (2020). 514 22. K. A. Callow, H. F. Parry, M. Sergeant, D. A. Tyrrell, The time course of the immune response to experimental 515 coronavirus infection of man. Epidemiol Infect 105, 435-446 (1990). 516 23. A. F. Bradburne, M. L. Bynoe, D. A. Tyrrell, Effects of a "new" human respiratory virus in volunteers. Br Med J 3, 767-517 769 (1967). 518 24. M. Jaume et al., Anti-severe acute respiratory syndrome coronavirus spike antibodies trigger infection of human immune 519 cells via a pH- and cysteine protease-independent FcgammaR pathway. J Virol 85, 10582-10597 (2011). 520 25. S. F. Wang et al., Antibody-dependent SARS coronavirus infection is mediated by antibodies against spike proteins. 521 Biochem Biophys Res Commun 451, 208-214 (2014). 522 26. Y. Wan et al., Molecular Mechanism for Antibody-Dependent Enhancement of Coronavirus Entry. J Virol 94 (2020). 523 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint 27. A. Iwasaki, Y. Yang, The potential danger of suboptimal antibody responses in COVID -19. Nat Rev Immunol 20, 339-524 341 (2020). 525 28. G. Marano et al., Convalescent plasma: new evidence for an old therapeutic tool? Blood Transfus 14, 152-157 (2016). 526 29. E. M. Bloch et al., Deployment of convalescent plasma for the prevention and treatment of COVID -19. J Clin Invest 527 10.1172/JCI138745 (2020). 528 30. K. Duan et al., Effectiveness of convalescent plasma therapy in severe COVID-19 patients. Proceedings of the National 529 Academy of Sciences of the United States of America 117, 9490-9496 (2020). 530 31. B. L. Brown, J. McCullough, Treatment for emerging viruses: Convalescent plasma and COVID-19. Transfus Apher Sci 531 10.1016/j.transci.2020.102790, 102790 (2020). 532 32. A. Budhai et al., How did we rapidly implement a convalescent plasma program? Transfusion 10.1111/trf.15910 (2020). 533 33. K. J. Blight, J. A. McKeating, C. M. Rice, Highly permissive cell lines for subgenomic and genomic hepatitis C virus 534 RNA replication. J Virol 76, 13001-13014 (2002). 535 536 537 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint Figure 1 538 539 540 541 542 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint Figure 2 543 544 545 546 547 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint Figure 3 548 549 550 551 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint Figure 4 552 553 554 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint Figure 1: Demographics of convalescent plasma donors. 555 556 A; Distribution of convalescent plasma donor age (left, blue bars) compared to U.S. population (right, red 557 bars). Dotted line represents Gaussian distribution curve fit. N=263; Pearson’s correlation coefficient. 558 559 B; Distribution of convalescent plasma donor blood group antigen (left, blue bars) compared to U.S. 560 population (right, red bars). N=370, binomial test for discrepancy versus U.S. population; * p < 0.05. 561 562 C; Distribution of convalescent plasma donor sex (blue bars) compared to U.S. population (red bars). N=354, 563 binomial test for discrepancy versus U.S. population. 564 565 D; Distribution of convalescent plasma donor ethnicity (blue bars) compared to U.S. population (right, red 566 bars). N=204, binomial test for discrepancy versus U.S. population; * p<0.05. 567 568 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint Figure 2: Neutralizing activity analysis of convalescent plasma donors. 569 570 A; Distribution of neutralization IC 50 values (NT50, reciprocal plasma dilution) of convalescent donor 571 plasma using HIV (red) or VSV pseudovirus (blue) overexpressing the SARS -CoV-2 spike protein (S). 572 573 B; Frequency of convalescent plasma donor NT50 values within indicated groups using HIV-S (top) or VSV-574 S pseudovirus constructs. 575 576 C; Frequency distribution of convalescent plasma HIV -S NT50 values versus age groups. Signal to cutoff 577 (S/co, dotted grey line) and 10x S/co (solid grey line) thresholds are indicated. n=5 -38, Kruskal-Wallis test; 578 * p < 0.05. 579 580 D; Frequency of convalescent plasma donor NT50 values versus sex. Signal to cutoff (S/co, dotted grey line) 581 and 10x S/co (solid grey line) thresholds are indicated. n=190, Mann-Whitney test, ** p < 0.01. 582 583 E; Frequency of convalescent plasma donor NT50 values versus blood group antigen. Signal to cutoff (S/co, 584 dotted grey line) and 10x S/co (solid grey line) thresholds are indicated. n=15-82, Kruskal-Wallis test, * p < 585 0.05. 586 587 F; Frequency of convalescent plasma donor NT50 values versus time (days) since last reported symptom. 588 Signal to cutoff (S/co, dotted grey line) and 10x S/co (solid grey line) thresholds are indicated. n=19 -33, 589 Mann-Whitney t-test, *p < 0.05. 590 591 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint Figure 3: Serological analysis of convalescent plasma donors. 592 593 A; Frequency of densitometric IgG (left) or IgM (right) results from LFA bands relative to control band. 594 Median values (red band) with 1st and 3rd quartiles (thin red lines) are shown. 595 596 B; Frequency of HTSA results using the total Ig or IgG assays derived from the Ortho Diagnostics platform 597 (left) or Abbott IgG assay platform (right). Results from fresh frozen plasma (FFP) units collected before 598 COVID19 are shown as healthy controls. 599 600 C; Frequency of S1 spike protein (left), Nucleocapsid (NP) protein (center) and RBD spike protein (right) 601 ELISA titer results. Titers reflect concentrations calculated using a mAb standard curve and not absolute 602 plasma concentrations. Median values (red band) with 1st and 3rd quartiles (thin red lines) are shown. 603 604 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint Figure 4: Correlation of serology assays versus neutralization activity of convalescent plasma donors. 605 606 A; Linear regression of HIV-S NT50 values (abscissa) versus serological assay values (ordinate). N 607 indicated in each graph, r2 = goodness of fit. 608 609 B; Spearman correlation coefficients, r, of neutralization and serological assays. N=137 samples. 610 611 C; Distribution of CP donor sample HTSA scores within indicated HIV-S NT50 groups using Ortho total Ig 612 (left), Ortho IgG (center) or Abbott IgG (right) assays. 613 614 D; Frequency of convalescent donor S1 and NP ELISA values defined in C. n=241 samples. 615 616 E; Distribution of NT50 values corresponding to populations defined in C. n=4-51, Kruskall-Wallis test, * 617 p < 0.05, ** p < 0.01. 618 619 . CC-BY-NC-ND 4.0 International licenseIt 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 preprintthis version posted September 9, 2020. ; https://doi.org/10.1101/2020.06.08.20124792doi: medRxiv preprint