key: cord-0837067-1r6k7lvn
authors: Sholukh, A. M.; Fiore-Gartland, A.; Ford, E. S.; Hou, Y.; Tse, L. V.; Lempp, F. A.; Kaiser, H.; Saint Germain, R.; Bossard, E.; Kee, J. J.; Diem, K.; Stuart, A. B.; Rupert, P. B.; Brock, C.; Buerger, M.; Doll, M. K.; Randhawa, A. K.; Stamatatos, L.; Strong, R. K.; McLaughlin, C.; Jerome, K. R.; Baric, R. S.; Montefiori, D.; Corey, L.
title: Evaluation of SARS-CoV-2 neutralization assays for antibody monitoring in natural infection and vaccine trials
date: 2020-12-08
journal: medRxiv : the preprint server for health sciences
DOI: 10.1101/2020.12.07.20245431
sha: bfdb4b6fe02b98001804cd4d58157d869d17aa37
doc_id: 837067
cord_uid: 1r6k7lvn

Determinants of protective immunity against SARS-CoV-2 infection require the development of well-standardized, reproducible antibody assays to be utilized in concert with clinical trials to establish correlates of risk and protection. This need has led to the appearance of a variety of neutralization assays used by different laboratories and companies. Using plasma samples from COVID-19 convalescent individuals with mild-to-moderate disease from a localized outbreak in a single region of the western US, we compared three platforms for SARS-CoV-2 neutralization: assay with live SARS-CoV-2, pseudovirus assay utilizing lentiviral (LV) and vesicular stomatitis virus (VSV) packaging, and a surrogate ELISA test. Vero, Vero E6, HEK293T cells expressing human angiotensin converting enzyme 2 (hACE2), and TZM-bl cells expressing hACE2 and transmembrane serine protease 2 (TMPRSS2) were evaluated. Live-virus and LV-pseudovirus assay with HEK293T cells showed similar geometric mean titers (GMTs) ranging 141-178, but VSV-pseudovirus assay yielded significantly higher GMT (310 95%CI 211-454; p < 0.001). Fifty percent neutralizing dilution (ND50) titers from live-virus and all pseudovirus assay readouts were highly correlated (Pearson r = 0.81-0.89). ND50 titers positively correlated with plasma concentration of IgG against SARS-CoV-2 spike and receptor binding domain (RBD) (r = 0.63-0.89), but moderately correlated with nucleoprotein IgG (r = 0.46-0.73). There was a moderate positive correlation between age and spike (Spearman's rho=0.37, p=0.02), RBD (rho=0.39, p=0.013) and nucleoprotein IgG (rho=0.45, p=0.003). ND80 showed stronger correlation with age than ND50 (ND80 rho=0.51 (p=0.001), ND50 rho=0.28 (p=0.075)). Our data demonstrate high concordance between cell-based assays with live and pseudotyped virions.

4 unknown at this time is what immune responses are associated with protective immunity. 68

Recent studies suggest passive infusion of monoclonal antibodies can alter COVID-19 disease 69 progression. Infusion of convalescent sera is more controversial regarding its efficacy. In order 70 to determine what constitutes protective immunity in human populations, well-standardized, 71

reproducible antibody assays are required to establish correlates of risk and protection. The 72 current study evaluates several assays that are under validation for use to determine 73 correlates of protection in vaccine studies evaluating immune responses in persons with 74 symptomatic COVID-19. These assays include a live-virus neutralization assay, which is 75 notable because work with live SARS-CoV-2 requires Biosafety Level 3 (BSL-3) containment; 76

something not readily available at many institutions and not easily amenable to high-77 throughput experiments. 78

To avoid these barriers and improve assay feasibility, a variety of live-virus neutralization 79 assays use recombinant SARS-CoV-2 (rSARS-CoV-2) containing GFP or luciferase reporter 80 genes at the ORF7 locus of the viral genome (8, 9) . These recombinant viruses replicate 81 similarly to SARS-CoV-2 clinical isolates in vitro and successfully infect primary airway 82 epithelial cell cultures. A fluorescence-based rSARS-CoV-2 neutralization assay yielded 83 comparable results to plaque reduction neutralization test (PRNT) in nAb detection from 84 convalescent patient plasma (8). With a shorter turnaround time (24-48 hours for reporter virus 85 vs. 3 days for PRNT), rSARS-CoV-2 provides a useful high-throughput platform to study nAb 86 responses, but still requires a BSL-3 laboratory for assay set up and readout. 87

Reporter assays using pseudotyped viruses, which are restricted to a single round of 88 intracellular replication, allow experiments to be carried out in BSL-2 environments. 89

Pseudotyped viral particles can be created with two packaging platforms: lentiviral/retroviral 90 8 VSV-pseudovirus. The codon-optimized sequence of the SARS-CoV-2 spike protein 157 (YP_009724390.1) with a truncation of the 19 C-terminal amino acids (D19) was cloned into a 158 pcDNA3.1(+) vector (ThermoFisher) under control of the human CMV promoter to generate 159 pcDNA3.1(+)-SARS-CoV-2-D19. The C-terminal truncation leads to a deletion of the ER-160 retention signal, localizing the spike protein to the cell surface, which enhances pseudovirus 161 packaging (30). VSV(G*ΔG-luciferase) system was purchased from Kerafast (13, 31) . four hours prior infection with VSV(G*ΔG-luciferase), 293T cells were transfected with pcDNA-163

WuhanCoV-S-D19. Next day, supernatant was harvest, centrifuged for 5 min at 1,000xg, 164 aliquoted and stored at -80 ºC. TCID50 was measured by infecting Vero cells (catalog number 165 CCL-81; ATCC) with serial 2-fold dilutions of the prepared pseudovirus. Luminex SARS-CoV-2 IgG binding antibody assay. Protein antigens were coupled to the 188 Bio-Plex Pro Magnetic COOH beads in a ratio of 10 μg of antigen per 2.5 x 10 6 beads in a two-189 step carbodiimide reaction. First, beads were washed and resuspended in Activation Buffer 190 (100 mM MES, pH 6) and then incubated with N-hydroxysulfosuccinimide (Sulfo-NHS, catalog 191 number 24520; ThermoFisher) and 1-ethyl-3-[3-dimethlyaminopropyl]carbodiimide-HCl (EDC, 192 catalog number 77149; ThermoFisher) also dissolved in Activation Buffer for 20 minutes on an 193 end-over-end rotational mixer at room temperature protected from light. Activated beads were 194 washed three times in Activation buffer. For coupling, antigen was mixed with activated beads 195 and reaction was carried out for 2 h on a rotational mixer at room temperature protected from 196 light. Conjugated beads were washed three times with Wash buffer (PBS, 197 1% BSA, 0.1% NaN3) and finally resuspended in Wash buffer at 10 7 beads/ml. Beads were 198 stored at 4 ºC for no longer than 30 days. 199

Antigen-specific IgG was measured using two replicate dilutions. Beads were blocked with 200 phosphate buffered saline (PBS; Gibco) containing 5% Blotto (Bio-Rad) and 0.05% Tween-20 201 (Sigma) and incubated for 1 hour with serially diluted plasma samples. Next, beads were 202 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 December 8, 2020. ; https://doi.org/10.1101 https://doi.org/10. /2020 washed 3 times with 0.05% Tween-20 in PBS and incubated with anti-human IgG Fc-PE 203 (catalog number 2048-09; Sothern Biotech). After incubation with secondary antibody, beads 204 were washed and resuspended in PBS with 1% BSA and 0.05% Tween-20 and binding data 205 were collected on Bio-Plex 200 instrument (Bio-Rad). Median Fluorescence Intensity (MFI) 206 was measured for a minimum of 50 beads per region. Background was established by 207 measuring the MFI of beads conjugated to antigens but incubated in Assay buffer. Background 208 MFI values were subtracted from all readings. We also trialed unconjugated beads and beads 209 conjugated to a decoy antigen with the same plasma samples used in testing and did not 210 detect non-specific binding above the assay background described above. Pooled sera from 211 normal human donors collected in 2015 -2016 was included as the negative control for 212 SARS-CoV-2 antigens. For the positive control we used convalescent plasma from a subject 213 with PCR-confirmed severe An IgG standard curve run in duplicate was used to estimate IgG concentration. Anti-human 215

IgG Fab-specific (Southern Biotech) was conjugated to the same bead regions used to 216 conjugate to antigen proteins. IgG-coupled beads were blocked, washed and incubated with 217 serially diluted human standard IgG (catalog number I4506; Sigma) for 1 h. Standard beads 218

were washed and incubated with anti-human IgG Fc-PE and MFI was measured as described 219

above. MFI readings and associated IgG concentrations were fitted to a four-parameter logistic 220 curve (4PL) using the R packages nCal and drc. A standard curve for each experiment was 221 used to obtain the effective concentrations of IgG in serum using the MFI measured with 222 antigen-coated beads. Since serum samples were also run as a dilution series we used the 223 median of the estimated concentrations from the dilutions that yielded MFIs between 100 and 224 10,000. Serum with all values above (below) this range were right (left) censored at the 225 concentration of the minimum (maximum) MFI. 226

Live SARS-CoV-2 neutralization assay. Assay was carried out in BSL-3 suite. Vero E6 cells 227

were seeded at 2x10 4 cells/well in a 96-well plate 24 h before the assay. Seventy five pfu of 228 the recombinant SARS-CoV-2-nanoLuc virus (rSARS-CoV-2-nLuc) (9) were mixed with Ab at 229 1:1 ratio and incubated at 37ºC for 1h. A 8-points, 3-fold dilution curve was generated for each 230 sample with starting concentration at 1:50. Virus and Ab mix was added to each well and 231 incubated at 37ºC + 5% CO2 for 48h. Luciferase activities were measured by Nano-Glo 232

Luciferase Assay System (Promega) following manufacturer protocol using SpectraMax M3 233 luminometer (Molecular Devices). Percent neutralization was calculated by the following Tyvek suits wearing personal powered-air purifying respirators. 

to manufacturer (GenScript) protocol and recommendations as follows. Capture plate was 279 incubated with plasma samples diluted 1:10, washed and probed with secondary antibody. 280

Assay was developed via TMB (ThermoFisher) and OD at 450 nm was measured using 281 SpectraMax M2 reader (Molecular Devices). Positive and negative controls were provided in 282 the kit. Binding inhibition was determined via the following formula: Inhibition = (1 -(OD of 283 sample / OD of Negative control)) × 100%. Percent binding inhibition was interpreted as a 284 percent neutralization. In order to determine ND50, plasma samples were serially diluted 285 starting from 1:10 and assay was performed as described above. 286

Statistical Analysis and Visualization. Neutralization titers were defined as the plasma 287 dilution that reduced relative luminescence units (RLU) by 50% or 80% relative to virus control 288

and 80 percent neutralization titers (ND50 and ND80) were estimated using the nCal and drc 290 packages in R. RLU was first transformed to neutralization using the formula neut = 1 -291 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 December 8, 2020. ; https://doi.org/10.1101/2020.12.07.20245431 doi: medRxiv preprint

4PL model that was used to estimate the dilution at which there would be 50% or 80% 293 neutralization. For samples with all dilutions having <50% neutralization the result was right 294 censored at the highest concentration. Patient demographic information (sex and age) was 295 extracted from a RedCap survey database. Abbott assay results (including index value) were 296 extracted from the laboratory information system (Sunquest Laboratory). 297

Correlations were estimated between pairs of neutralization or binding antibody readouts using 298

were logged prior to estimating correlation. Log-transformed ND50 values and IgG 300

concentrations were approximately normally distributed with few outliers and a low level of 301 censoring, justifying use of Pearson's correlation and linear regression. Left censored values 302 were given a value of half the level of detection, which corresponded to the first dilution for 303 each neutralization assay. Student's t test was also on log-transformed values. Association of 304 neutralization and IgG concentration with age and BMI were conducted using Spearman's 305 rank-based correlation and Wilcoxon's rank-sum tests were used to compare neutralization 306 and IgG in two groups of individuals (e.g. gender, presence of symptom score >2). Statistical 307 significance was determined based on a p-value < 0.05. 308

Cohort characteristics, demographics, survey participation, and serology clinical 311 testing. A total of 1,359 email invitations were sent to 2,655 phase 1 study volunteers and 63 312 phase 2 volunteers. Among phase 1 and 2 volunteers invited to participate, 973 (72%) and 53 313 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 December 8, 2020. ; https://doi.org/10.1101/2020.12.07.20245431 doi: medRxiv preprint (84%) people consented and completed the enrollment survey. Of these, 967 participants 314 presented for specimen collection between May 4-19, 2020, and 222 (22.8 %) with blood 315 drawn had IgG antibodies to SARS-CoV-2 nucleoprotein according to the Abbott Architect test 316 (index value ≥1.40).Out of these, we randomly selected forty positive samples to evaluate 317 different platforms of SARS-CoV-2 neutralizing antibody assays. Participants had a median 318 age of 51.5 years and a range between 23 and 81 years (Table 1 ). According to the survey, 319 only one participant reported being hospitalized and four participants (10%) were self-320 described as asymptomatic. Among participants reporting different symptoms (Table S1) , 321 57.5% had fever while fatigue (87.5%), cough (72.5%), headache (67.5%) and chills (65%) 322

were more prevalent. Based on this, our cohort can be described as representing mild-to-323 moderate symptomatic infections. Fifty percent neutralizing dilution (ND50) is a standard numerical parameter to compare virus-334 neutralizing potency between different samples and studies. To reflect the ultimate capacity of 335 serum antibodies to neutralize virus both ND50 and ND80, the dilutions at which 50% and 80% 336 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 December 8, 2020. ; https://doi.org/10.1101/2020.12.07.20245431 doi: medRxiv preprint neutralization is observed, are used together. We serially diluted plasma samples to generate 337 titration curves and estimated ND50 and ND80 relative to positive and negative controls (Fig.  338 S1-S3). All four cell-based neutralization assays performed comparably and generated 339 titration curves necessary for ND50 and ND80 estimation using a four-parameter logistic model 340 ( Fig. S1-S3 ). On average, the slope parameter for neutralization curves with rSARS-CoV-2-341 nLuc was higher compared to other assays (slope B=3.3 vs. 0.6, 1.4, 1.5 for LV-pseudo/293T, 342 LV-pseudo/TZM-bl and VSV-pseudo/Vero, respectively; all p < 0.001). Geometric mean ND50 343 from the assay with rSARS-CoV-2-nLuc (141, 95% CI 94-213) did not differ (p=0.2) from the 344 LV-pseudo/293T assay (178, 95%CI 112-283; Fig. 1A, Fig. S4A ). However, the LV-345 pseudo/TZM-bl assay showed significantly lower geometric mean ND50s compared to LV-346 pseudo/293T (Fig. 1A, Fig. S4A ). The VSV-pseudo/Vero assay produced significantly higher 347 ND50 values (geometric mean ND50 of 310, 95%CI 211-454) compared to both the SARS-348

CoV-2/VeroE6 assay and the two LV-pseudovirus assays (Fig. S4A ) suggesting that it is 349 easier to neutralize VSV-based pseudovirus in Vero cells compared to other approaches. 350

The assay platforms also differed in their capacity to detect neutralization. In the live-virus 351 assay, LV-pseudo/293T and LV-pseudo/TZM-bl neutralization was detectable at the lowest 352 dilution for 31, 37, and 34 samples, respectively, and therefore permitted estimation of the 353 ND50; ND50 of the remaining samples was censored at the lowest dilution (Fig. 1A) . In 354 contrast, VSV-pseudo/Vero permitted estimation of ND50 for all 40 samples (Fig. 1A) . With the 355 sVNT, only 31 of 40 samples showed neutralization above 20%, a negative cutoff value 356 according to the manufacturer's protocol, at the lowest dilution 1:10 (Fig. S3A ). To estimate 357 ND50, we selected 13 of these 31 samples and tested them in serial dilutions (Fig. 1A, Fig.  358 S3B). Samples were selected to represent different percent neutralization observed at 1:10 359 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 December 8, 2020. ; https://doi.org/10.1101/2020.12.07.20245431 doi: medRxiv preprint dilution. The resulting ND50 was significantly lower (29.5 95%CI 18. 2-47.9 ) than in the cell-360 based assays, further supporting the conclusion that the surrogate assay had lower sensitivity 361 compared to the cell-based assays. 362

We used the same 4PL models to estimate ND80 titers. Though ND80 was consistently lower, 363 the correlation with ND50 was high ranging from Pearson's r=0.87 for the live-virus assay to 364 r=0.97 for the VSV-pseudovirus assay. Similar to ND50, the ND80 titers also differed among 365 the assays (Fig. 1B, Fig. S4B ) with the SARS-CoV-2/VeroE6 assay reporting the lowest 366 number of samples with ND80 titers above the limit of detection (24/40) and VSV-pseudovirus 367 assay showing the highest (39/40). The live-virus assay also showed the smallest difference 368 between ND50 and ND80 titers (Fig. S5, Table S2 ), a direct consequence of the steeper 369 titration curves observed for this assay (Fig. S1A) . For pseudovirus-utilizing assays the 370 difference between ND50 and ND80 was greater and ranged between 2.7 and 4.2-fold. Due to 371 inability to reach 80% neutralization for the many samples in sVNT, we could not calculate 372 ND80 (Fig. S3) . 373

Strong correlation among neutralization assays. Next, we conducted a correlation analysis 374 of the ND50 and ND80 values derived from each of the five neutralization assays (Fig. 2, Fig.  375 between the percent neutralization measured at 1:10 dilution and the outcomes of the cell-381 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 December 8, 2020. ; https://doi.org/10.1101/2020.12.07.20245431 doi: medRxiv preprint based assays (Pearson r = 0.73 -0.80); correlation between ND50 and percent neutralization 382 at 1:10 dilution was modest (r = 0.59). 383

Similar correlation was observed for ND80 outcomes for cell-based assays (Fig. 2B) with 384

Person's r ranging between 0.69 -0.88. Only a few samples showed neutralization greater 385 than 80% in sVNT (Fig. S3) and thus correlation between sVNT ND80 and other assays was 386 not estimated. 387

Prior studies of SARS-CoV-2 individuals showed that the serum titer of spike and RBD-binding 389

IgG antibodies was a correlate of neutralizing potency (14, 34, 35) . Using quantitative, 390

Luminex-based immunoassay, we measured concentration of IgG to SARS-CoV-2 spike, RBD 391 and nucleoprotein in each of the serum samples; IgG to tetanus toxoid was also measured as 392 a proxy for overall IgG level and state of humoral immunity (Fig. 1C) Abbott SARS-CoV-2 IgG assay is designed and used for qualitative detection of IgG against 398 the SARS-CoV-2 nucleoprotein, the instrument reports index values that can be used in 399 quantitative analyses (Fig. 1D) . 400

Pearson correlation analysis revealed that levels of RBD and spike IgG correlated strongly 401 (Pearson's r = 0.89, 95% CI [0.81, 0.94]) ( Fig. 2A) . Nucleoprotein-specific IgG measured in our 402

Luminex immunoassay was highly correlated with the quantitative index of the Abbott Architect 403 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. with SARS-CoV-2 spike, spike RBD or nucleoprotein IgG (all p > 0.05). 408

We then examined the relationship between concentration of SARS-CoV-2 IgG and virus 409 neutralization (Fig. 2, Fig. S7 ). We found that IgG concentrations to each of the SARS-CoV Tetanus-specific IgG did not correlate with any of SARS-CoV-2-associated IgG concentrations 420 or neutralization titers. 421

parameters were associated with SARS-CoV-2 specific IgG or neutralization. Previously, body 423 mass index (BMI), female sex and age were reported to positively correlate with antibody titers 424 against SARS-CoV-2 (36). We asked whether ND50 titers obtained from each neutralization 425 assay correlated with age, gender, BMI or self-reported disease symptoms (Table 3) . We 426 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 December 8, 2020. ; 20 found a moderate positive correlation between age and concentrations of spike-specific 427 (Spearman's rho=0.37, p=0.02), spike RBD-specific (rho=0.39, p=0.013) and nucleoprotein-428 specific (rho=0.45, p=0.003) IgG. Similarly, there were positive correlations between age and 429 neutralization titer, though the correlations tended to be higher with ND80 compared to ND50 430 titer (Fig. 3) . For example, the correlation coefficient of age with live-virus neutralization ND80 431 was rho=0.51 (p=0.001), compared to rho=0.28 (p=0.075) for ND50. No consistent significant 432 correlations with BMI, gender or symptoms were observed (Table 3) . with mild-to-moderate disease involved in a county-wide outbreak of COVID-19. These data 439

show a high level of congruency among cell-based SARS-CoV-2 neutralization assays. The 440 50% and 80% neutralization titer readouts of cell-based assays were highly correlated with 441 each other and with the concentration of RBD and spike-specific IgG. The results of the 442 ELISA-based sVNT were also positive correlated with the other neutralization assays, however 443 the correlation was modest in comparison. Though levels of spike-specific IgG were highly 444 correlated with neutralization, this does not indicate that all spike-specific binding IgG have 445 neutralization activity, rather it implies that individuals who produce spike-specific binding 446 antibodies are also likely to make neutralizing IgG. The correlation between nucleoprotein-447 specific IgG and neutralization was consistently lower than the correlations of spike and spike 448 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 December 8, 2020. ;

RBD-specific IgG with neutralization. This is consistent with known mechanisms of 449 neutralization, which involve binding and/or blocking the spike:ACE2 receptor binding domain; 450 the moderate correlations of nucleoprotein-specific IgG with neutralization may indicate that 451 presence of any SARS-CoV-2 specific IgG is a biomarker of the presence of neutralizing 452 antibodies as well. The association of age with both the plasma concentration of spike-specific 453

IgG and neutralization titer suggests that the previously reported association of high 454 neutralization titer among older individuals may be mediated by higher concentrations of spike 455 and spike RBD-specific IgG. Whether this is a result of prior infections with seasonal 456 coronaviruses or an effect of age on the developing immune response to SARS-CoV-2 is not 457

Previously, in a cohort of severely ill COVID-19 patients, deceased individuals were reported to 459 have higher concentrations of nucleoprotein-than spike-and RBD-specific IgG, and the 460 opposite scenario was associated with survival (37). In our study, individuals with mild-to-461 moderate disease also demonstrated higher concentrations of nucleoprotein IgG compared to 462 spike and RBD IgG suggesting that the immune response to spike and nucleoprotein differ in 463 milder forms of disease. Of interest, we did not see any correlation between the neutralizing 464 potency of plasma and BMI. As our cohort was largely uniform with regard to disease severity, 465 our study cannot comment on the association between spike and nucleoprotein antibodies and 466 disease severity. 467

With the set of cell lines used for the assays in this study, we were able to address questions 468 regarding the influence of proteolytic cleavage of SARS-CoV-2 spike on virus neutralization by 469 serum antibodies which is important for choosing an assay that would provide more 470 physiologically relevant outcomes. Although there was no significant difference observed in 471 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 December 8, 2020. ; https://doi.org/10.1101 https://doi.org/10. /2020 virus titer at 48 h post infection on the wildtype Vero cells and Vero cells expressing furin, at 472 the early time point cells expressing protease showed a higher virus titer (9). Proteolytic 473 cleavage was also shown to be essential for SARS-CoV-2 infectivity on other cell types (9, 17) . 474

Therefore, cell lines expressing TMPRSS2 or a related protease would allow testing for the 475 possible role of proteolytic cleavage of the spike glycoprotein in virus infectivity. In contrast to 476 this reasoning, comparison of different cell types used for pseudovirus assays revealed that 477 the presence of TMPRSS2 is not critical for assay performance. As such, TZM-bl cells 478 designed to express both ACE2 and TMPRSS2 showed no significant difference compared to 479 293T cells that do not express TMPRSS2 endogenously and were only expressing ACE2. 480

and pseudoviruses due to the assay robustness and reproducibility (38, 39) . 482

sensitivity among assays tested. This could be explained by lower affinity of interaction 484 between SARS-CoV-2 and simian ACE compared to the human ACE2 and by different density 485 of spike glycoprotein on the surface of VSV pseudovirus. While increased sensitivity may lead 486 to overestimation of neutralization potency, it can also be useful for specimens with low 487 neutralizing activity or when sample volume is limited such as for mucosal secretions and 488 washes. However, the high correlation between the live-virus assay and Vero/VSV-489 pseudovirus assay suggests that data obtained in the latter can accurately reflect the sample 490 potency to neutralize wildtype SARS-CoV-2. 491 ELISA-based assays have two major limitations: i) inability to account for synergistic action of 492 antibodies targeting different epitopes; and ii) detection only of antibodies that block interaction 493 between RBD and ACE2, thus omitting antibodies that neutralize virus via non-RBD sites on 494 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 December 8, 2020. ; the virus glycoprotein (24, 40). For example, synergistic action of antibodies against RBD and 495 the S2 domain has been reported (41). There are two ways of performing such a surrogate 496 assay: soluble biotinylated ACE2 competing with serum antibodies for binding to immobilized 497 RBD or spike, or an opposite version with soluble RBD and immobilized ACE2 (21). Abe et al. 498

found that an assay with soluble ACE2 and immobilized RBD was more sensitive and yielded 499 ND50 values that correlated with ND50 titers obtained in the classical cell-based PRNT with a 500 coefficient of determination of 0.6. The GenScript assay that we have tested in our study 501 utilizes immobilized ACE2, which likely explains why we were not able to measure ND50 titers 502 for the majority of samples. Of note, samples used by Abe et al. were also collected from mild-503 to-moderate COVID-19 patients. In conclusion, a surrogate assay can be used cautiously as 504 an alternative to cell-based assays to obtain preliminary qualitative results, to rapidly 505 distinguish between samples with high and low neutralizing potency, and when a cell-based 506 assay is not available or reasonably feasible. 507 508 Acknowledgements 509

We thank Dr. Mindy Minor for critical reading of the manuscript and editorial help, X for 510 technical assistance, and Sara Thiebaud for assistance with data management and analysis. 511

This work was funded by: NIAID Service Agreement 225472-99 to RKS, R01AI134878 and 512 UM1AI068614 to LC, Fred Hutch Evergreen grant to AMS. 513 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 December 8, 2020. ; https://doi.org/10.1101/2020.12.07.20245431 doi: medRxiv preprint systematic review of antibody mediated immunity to coronaviruses: kinetics, correlates of 529 protection, and association with severity. Nat Commun 2020 111 11:1-16. 530

Long QX, Liu BZ, Deng HJ, Wu GC, Deng K, Chen YK, Liao P, Qiu JF, Lin Y, Cai XF, 531

Wang DQ, Hu Y, Ren JH, Tang N, Xu YY, Yu LH, Mo Z, Gong F, Zhang XL, Tian WG, 532 Hu L, Zhang XX, Xiang JL, Du HX, Liu HW, Lang CH, Luo XH, Wu SB, Cui XP, Zhou Z, 533 Zhu MM, Wang J, Xue CJ, Li XF, Wang L, Li ZJ, Wang K, Niu CC, Yang QJ, Tang XJ, 534

Zhang Y, Liu XM, Li JJ, Zhang DC, Zhang F, Liu P, Yuan J, Li Q, Hu JL, Chen J, Huang 535 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 December 8, 2020. (which was not certified by peer review) 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 December 8, 2020. ; https://doi.org/10.1101/2020.12.07.20245431 doi: medRxiv preprint All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 December 8, 2020. ; https://doi.org/10.1101/2020.12.07.20245431 doi: medRxiv preprint Figure 1 . Boxplots of SARS-CoV-2 neutralization and binding antibody concentration for 40 plasma samples from 40 COVID-19 convalescent patients. (A) ND50 and (B) ND80 neutralization titer measured using five SARS-CoV-2 neutralization assays. Each assay defined its own lower limit of detect (LOD) based on the initial dilution: 50-fold for SARS-CoV-2/VeroE6, 20 for the LV and VSV pseudovirus assays and 10 for the sVNT. Data below the LOD (open triangle) is plotted at LOD/2. Number and percent of samples above the LOD is indicated above each plot. (C) Antigen-specific IgG concentration measured using a Luminex bead-based assay. (D) Index values for each sample from the Abbott Architect nucleoprotein IgG assay. For each assay the box represents the extend of the inter-quartile range (IQR) with a line indicating the median; whiskers extend to 1.5 times the IQR. All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 December 8, 2020. ; https://doi.org/10.1101/2020.12.07.20245431 doi: medRxiv preprint Figure 2 . Correlation among assay readouts measuring neutralization or antigen-specific IgG concentration in plasma. Heatmap color is determined by the Pearson's correlation coefficient (r, annotations). Each panel includes either ND50 titers (A) or ND80 titers (B) and their correlation with sVNT % neutralization, SARS-CoV-2 specific IgG concentration (Luminex bead-based assay), the quantitative index of the Abbott nucleoprotein assay and tetanus toxoid-specific IgG concentration. (which was not certified by peer review) 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 December 8, 2020. ; https://doi.org/10.1101/2020.12.07.20245431 doi: medRxiv preprint Figure 3 . Correlation analysis of plasma neutralizing potency and age of participants. A, ND50 versus age; B, ND80 versus age.

Correlates of protection induced by vaccination

Complete mapping of 519 mutations to the SARS-CoV-2 spike receptor-binding domain that escape antibody

Neutralizing Antibody Responses to SARS-CoV-2 in a 523 COVID-19 Recovered Patient Cohort and Their Implications

surrogate virus neutralization test based on antibody-mediated blockage of ACE2-spike 603 protein-protein interaction

Analysis of a SARS-607

CoV-2-Infected Individual Reveals Development of Potent Neutralizing Antibodies with 608

Isolation of potent SARS-CoV-2 neutralizing antibodies 614 and protection from disease in a small animal model

Human neutralizing antibodies elicited by SARS-CoV-2 618 infection

Cryo-EM Structure of the 2019-nCoV Spike in the Prefusion 621

Daedalus: a robust, turnkey platform for rapid production of decigram quantities of active 624 recombinant proteins in human cell lines using novel lentiviral vectors

A novel multi-affinity tag 627 system to produce high levels of soluble and biotinylated proteins in Escherichia coli

Engineered tobacco etch 630 virus (TEV) protease active in the secretory pathway of mammalian cells

Characterization of spike glycoprotein of 634

SARS-CoV-2 on virus entry and its immune cross-reactivity with SARS-CoV

Ebolavirus Fusion Loop as a Site of Vulnerability

Efficient transfer, integration, 642 and sustained long-term expression of the transgene in adult rat brains injected with a 643 lentiviral vector

SARS-CoV-2 IgG Assay and Seroprevalence in

A serological assay to detect 652

SARS-CoV-2 seroconversion in humans

Clinical performances of an ELISA for SARS-CoV-2 antibody assay and 655 correlation with neutralization activity

Humoral Immune Response to SARS-CoV-2 668 in Iceland

Distinct Early Serological Signatures Track with SARS-CoV-2 Survival

Measuring HIV Neutralization in a Luciferase Reporter Gene Assay 674

Tiered Categorization of a 679

Diverse Panel of HIV-1 Env Pseudoviruses for Assessment of Neutralizing Antibodies

Broad neutralization of SARS-related viruses by human monoclonal antibodies

Antibody potency, effector function and 691 combinations in protection from SARS-CoV-2 infection in vivo

Figure 1. Boxplots of SARS-CoV-2 neutralization and binding antibody concentration for 40 707 plasma samples from 40 COVID-19 convalescent patients

neutralization titer measured using five SARS-CoV-2 neutralization assays. Each assay 709 defined its own lower limit of detect (LOD) based on the initial dilution

20 for the LV and VSV pseudovirus assays and 10 for the sVNT

LOD (open triangle) is plotted at LOD/2. Number and percent of samples above the LOD is 712 indicated above each plot. (C) Antigen-specific IgG concentration measured using a Luminex 713 bead-based assay. (D) Index values for each sample from the Abbott Architect nucleoprotein 714

For each assay the box represents the extend of the inter-quartile range (IQR) with 715 a line indicating the median

SARS-CoV-2 virus/Vero LV-pseudovirus/293T

SARS-CoV-2 virus/Vero LV-pseudovirus/293T

Vero sVNT % Neutralization SARS-CoV-2 spike SARS-CoV-2 RBD 1