Quantifying absolute neutralization titers against SARS-CoV-2 by a standardized virus neutralization assay allows for cross-cohort comparisons of COVID-19 sera

Abstract

The global COVID-19 pandemic has mobilized efforts to develop vaccines and antibody-based therapeutics, including convalescent plasma therapy, that inhibit viral entry by inducing or transferring neutralizing antibodies (nAbs) against the SARS-CoV-2 spike glycoprotein (CoV2-S). However, rigorous efficacy testing requires extensive screening with live virus under onerous BSL3 conditions which limits high throughput screening of patient and vaccine sera. Myriad BSL-2 compatible surrogate virus neutralization assays (VNAs) have been developed to overcome this barrier. Yet, there is marked variability between VNAs and how their results are presented, making inter-group comparisons difficult. To address these limitations, we developed a standardized VNA using VSVΔG-based CoV-2-S pseudotyped particles (CoV2pp) that can be robustly produced at scale and generate accurate neutralizing titers within 18 hours post-infection. Our standardized CoV2pp VNA showed a strong positive correlation with CoV2-S ELISA and live virus neutralizations in confirmed convalescent patient sera. Three independent groups subsequently validated our standardized CoV2pp VNA (n>120). Our data show that absolute (abs) IC50, IC80, and IC90 values can be legitimately compared across diverse cohorts, highlight the substantial but consistent variability in neutralization potency across these cohorts, and support the use of absIC80 as a more meaningful metric for assessing the neutralization potency of vaccine or convalescent sera. Lastly, we used our CoV2pp in a screen to identify ultra-permissive 293T clones that stably express ACE2 or ACE2+TMPRSS2. When used in combination with our CoV2pp, we can now produce CoV2pp sufficient for 150,000 standardized VNA/week.

Figure 1.. Production of VSVΔG-rLuc bearing SARS-CoV-2 spike glycoprotein

(A) Overview of VSVΔG-rLuc pseudotyped particles bearing CoV-2 spike (top panel) with annotated spike glycoprotein domains and cleavage sites (bottom panel). As mentioned in the text, we refer to SARS-CoV as SARS-CoV-1 for greater clarity. (B) VSV-ΔG[Rluc] pseudotyped particles (VSVpp) bearing the Nipah virus receptor binding protein alone (NiV-RBPpp), SARS-CoV-2-S (CoV2pp), or VSV-G (VSV-Gpp) were titered on Vero-CCL81 cells using a 10-fold serial dilution. Symbols represent the mean +/− SEM (error bars) of each titration performed in technical triplicates. (C) Expression of the indicated viral glycoproteins on producer cells and their incorporation into VSVpp. Western blots performed as described in Methods using anti-S1 or anti-S2 specific antibodies. (D) CoV2pp entry is inhibited by soluble receptor binding domain (sRBD) derived from SARS-CoV-2-S. CoV2pp and VSV-Gpp infection of Vero-CCL81 cells was performed as in (B) in the presence of the indicated amounts of sRBD. Neutralization curves were generated by fitting data points using a variable slope, 4-parameter logistics regression curve (robust fitting method). The last point (no sRBD) was fixed to represent 100% maximal infection. Each replicate from an experiment performed in duplicate is shown. The calculated IC50 for sRBD neutralization of CoV2pp is 4.65μg/mL.

Figure 2.. CoV2pp entry is enhanced by trypsin treatment.

(A) Optimizing trypsin treatment conditions. Supernatant containing CoV2pp were trypsin-treated at the indicated concentrations for 15 min. at room temperature prior to the addition of 625 μg/mL of soybean trypsin inhibitor (SBTI). These particles were then titered on Vero-CCL81 cells in technical triplicates. Data shown as mean +/− SEM. (B) Dilution in serum free media (SFM, DMEM only) provides the highest signal:noise ratio for trypsin-treated CoV2pp entry. Particles were diluted 1:10 in Opti-MEM, SFM, or DMEM+10%FBS prior to infection of Vero-CCL81 cells and spinoculation as described in Fig. 1D. Cells infected without spinoculation show approximately 3x less signal:noise ratios (Supplemental Fig. 2). (C) Addition of soybean trypsin inhibitor at the time of infection reduces trypsin treated particle entry. This was performed in technical triplicates for two independent experiments. Shown are the combined results with error bars indicating SEM and **** indicating a p-value <0.0001. (D) Schematic showing overall view of how protease priming and SBTI treatment is working to enhance CoV2pp entry.

Figure 3.. Trypsin-treated CoV2pp depend on ACE2 and TMPRSS2 for entry.

(A) Parental and TMPRSS2 or ACE2 transduced VeroCCL81 cells were infected with the indicated pseudotyped viruses. All particles were diluted in serum free media in order to be within the linear range for the assay. Normalized infectivity data is presented as fold-over Vero-CCL81-WT for the various VSVpp shown. VSV-Gpp served as an internal control for the intrinsic permissiveness of various cell lines to VSV mediated gene expression. Data is presented as mean +/− SEM from two independent experiments done in technical triplicates. *, p <0.05, **, p- <0.01, and ****, p <0.0001. (B) Western blot of wild type and transduced Vero CCL81 cells. The numbers below each column show the relative ACE2 abundance was measured by densitometry and normalized as described in Methods.

Figure 4.. CoV2pp viral neutralization assay and absIC50/80 versus spike binding of patient sera.

(A) 36 patient sera screened for CoV2pp neutralization. CoV2pp were used to infect Vero-CCL81 cells in the presence of a 4-fold serial dilution of patient sera as described in the Methods. Samples in light purple do not neutralize CoV2pp. Neutralization curves were fit using a variable slope, 4-parameter logistics regression curve with a robust fitting method. (B) The same 36 samples are shown as a neutralization heat map, which were generated in R as described in the Methods. Here, red represents complete neutralization and blue represents no neutralization. Samples are sorted by the average from the first four dilutions with the most neutralizing samples on the left. (C) Correlation of CoV2pp neutralization titers to spike binding (ELISA AUC) and live virus microneutralization (MN) activity. Absolute IC50 (absIC50, top) and IC80 (absIC80, bottom) for CoV2pp neutralizations and live virus MNs were calculated in R using a 4-parameter logistic regression model as described in the Methods. Presented are the added IgG and IgM ELISA AUC. AUC and live virus neutralizations were performed as described in the Methods. Presented are the r and p value from a simple linear regression. (D) Positive serum samples and their CoV2pp reciprocal absIC50 (top) and absIC80 (bottom). The IC50 graph is colored and ordered to display samples with low, average, or high IC50 as blue, grey or red circles, respectively. The IC80 graph below retains the coloring from the IC50 graph, but the samples are now ordered from left to right to show samples with the lowest to highest IC80 values. Tukey box and whisker plots show median with interquartile range (IQR) and whiskers extending to 1.5x the IQR. All points outside that range are depicted.

Figure 5.. CoV2pp viral neutralization assay validated against patient sera by external groups.

(A) Patient sera neutralization of CoV2pp for 88 samples run by three different independent groups. This is visualized as in Fig. 4B where red represents complete neutralization and blue represents no neutralization. (B) Correlations of CoV2pp reciprocal AbsIC80 to spike ELISAs. AbsIC80 was calculated as previously described in Fig. 4C. For LSUHS ELISAs, spike ectodomain was used and sera was diluted to a 1:100 dilution. For the COVIDAR ELISAs, a mixture of sRBD and spike was utilized as previously described and AUC was calculated as described in the Methods. (C) Summary AbsIC80 of 89 positive sera CoV2pp neutralizations. Samples from all 4 groups are depicted on the X-axis. AbsIC80 was calculated as described in Fig. 4C and Tukey box and whisker plots are shown as described in Fig. 4D.

Figure 6.. 293T stably transduced with ACE2 and TMPRSS2 (293T-ACE2+TMPRSS2) are ultra-permissive for SARS-CoV-2pp infection.

(A) Infection of 293T cells lines transduced to stably express, TMPRSS2, ACE2, or both. Cell lines were generated as described in the Materials and Methods. A single dilution of particles was used to infect cells prior to spinoculation as described in the Methods. Infections were done in technical triplicates. Presented are the aggregated results from two independent replicates and error bars show SEM. For statistics, ns = not significant, ** is a p-value <0.01, and **** is a p-value <0.0001. (B) Western blot of ACE2 expression in 293T cell lines. This was performed as described in the methods and the values below each column represents the relative abundance of ACE2. (C) Normalized CoV2pp entry into single cell clones. Entry was normalized to the wild type parental cell line and further normalized to VSV-G entry. Presented are the average of one experiment in technical triplicates. Error bars show the median and interquartile range. Raw entry data for each cell clone is shown in Supplementary Fig. 9A (D) Entry inhibition of CoV2pp in by Nafamostat mesylate, a serine protease inhibitor. Nafamostat was mixed with CoV2pp (left panel) or VSV-Gpp (right panel) prior addition to cells. Shown are the results from one experiment in technical triplicates. Data are presented as described in Fig. 4A and error bars show SEM.

Figure 7.. Ultra-permissive 293T-ACE2+TMPRSS2 cell clones retains the same phenotypic sensitivity to convalescent COVID-19 sera.

(A) Selection of pooled sera samples. Results from Fig. 4A are reproduced here for the reader’s convenience. Presented are the subset of samples that were pooled for use in viral neutralization assays (VNAs) in the adjacent panel. (B) Vero CCL81 and transduced 293T cells were used for VNAs. Sera previously shown to be negative, weakly positive, or strongly positive for CoV2pp neutralizations were selected to be pooled in equal volumes. These were subsequently used for VNAs, which were performed and presented as described in Fig. 4A. Notably, these VNAs were performed in the absence of exogenous trypsin or spinoculation.