IMMUNO-COV v2.0: Development and Validation of a High-Throughput Clinical Assay for Measuring SARS-CoV-2-Neutralizing Antibody Titers

Abstract

Neutralizing antibodies are key determinants of protection from future infection, yet well-validated high-throughput assays for measuring titers of SARS-CoV-2-neutralizing antibodies are not generally available. Here, we describe the development and validation of IMMUNO-COV v2.0, a scalable surrogate virus assay, which titrates antibodies that block infection of Vero-ACE2 cells by a luciferase-encoding vesicular stomatitis virus displaying SARS-CoV-2 spike glycoproteins (VSV-SARS2-Fluc). Antibody titers, calculated using a standard curve consisting of stepped concentrations of SARS-CoV-2 spike monoclonal antibody, correlated closely (P < 0.0001) with titers obtained from a gold standard 50% plaque-reduction neutralization test (PRNT50%) performed using a clinical isolate of SARS-CoV-2. IMMUNO-COV v2.0 was comprehensively validated using data acquired from 242 assay runs performed over 7 days by five analysts, utilizing two separate virus lots, and 176 blood samples. Assay performance was acceptable for clinical use in human serum and plasma based on parameters including linearity, dynamic range, limit of blank and limit of detection, dilutional linearity and parallelism, precision, clinical agreement, matrix equivalence, clinical specificity and sensitivity, and robustness. Sufficient VSV-SARS2-Fluc virus reagent has been banked to test 5 million clinical samples. Notably, a significant drop in IMMUNO-COV v2.0 neutralizing antibody titers was observed over a 6-month period in people recovered from SARS-CoV-2 infection. Together, our results demonstrate the feasibility and utility of IMMUNO-COV v2.0 for measuring SARS-CoV-2-neutralizing antibodies in vaccinated individuals and those recovering from natural infections. Such monitoring can be used to better understand what levels of neutralizing antibodies are required for protection from SARS-CoV-2 and what booster dosing schedules are needed to sustain vaccine-induced immunity. IMPORTANCE Since its emergence at the end of 2019, SARS-CoV-2, the causative agent of COVID-19, has caused over 100 million infections and 2.4 million deaths worldwide. Recently, countries have begun administering approved COVID-19 vaccines, which elicit strong immune responses and prevent disease in most vaccinated individuals. A key component of the protective immune response is the production of neutralizing antibodies capable of preventing future SARS-CoV-2 infection. Yet, fundamental questions remain regarding the longevity of neutralizing antibody responses following infection or vaccination and the level of neutralizing antibodies required to confer protection. Our work is significant as it describes the development and validation of a scalable clinical assay that measures SARS-CoV-2-neutraling antibody titers. We have critical virus reagent to test over 5 million samples, making our assay well suited for widespread monitoring of SARS-CoV-2-neutralizing antibodies, which can in turn be used to inform vaccine dosing schedules and answer fundamental questions regarding SARS-CoV-2 immunity.

Keywords: COVID-19; SARS-CoV-2; antibody titer; clinical validation; high-throughput assay; neutralizing antibodies; surrogate virus.

FIG 1

Overview of the IMMUNO-COV v2.0 assay. A VSV expressing SARS-CoV-2 spike and firefly luciferase (VSV-SARS2-Fluc) is incubated with test sera or plasma. In the absence of SARS-CoV-2-neutralizing antibodies (top), the virus retains infectivity and infects Vero-ACE2 monolayers. If the test sample contains SARS-CoV-2-neutralizing antibodies (bottom), the antibodies inhibit infection by blocking cell entry. As virus replication proceeds, infected cells express luciferase, which is used to quantitate virus infection. High luciferase signal means the test sample did not neutralize the virus, while decreased luciferase indicates the presence of SARS-CoV-2-neutralizing antibodies.

FIG 2

Generation and characterization of VSV-SARS2-Fluc. (A) Schematic representation of the VSV-SARS2-Fluc genome. The locations of the VSV N, P, M (M51R), and L genes are shown. In place of VSV-G, a codon-optimized SARS-CoV-2 spike gene with a 19-amino-acid C-terminal (CT) deletion (Δ19CT) is substituted. TM is the transmembrane domain. Firefly luciferase (Fluc) is inserted as an additional transcriptional element between S-Δ19CT and L. Not drawn to scale. (B) Immunoblot analysis. VSV-SARS2-Fluc or VSV-GFP control virus (5 × 10⁵ total PFU) or spike control from lysates of cells overexpressing SARS-CoV-2 spike was subjected to immunoblot analysis using anti-SARS-CoV-2 spike antibody (left) and anti-VSV-G antiserum (right). Arrows indicate the full-length S1/S2 variant and cleaved S2 variant of spike and the VSV-G proteins. (C) Infection of cell monolayers. Vero-ACE2 or BHK-21 cell monolayers were infected with VSV-SARS2-Fluc or control VSV-Fluc or mock infected. Images were taken 48 h postinfection at ×100 magnification. (D and E) Replication curves. Vero-ACE2 or BHK-21 cell monolayers were infected as in panel C, and the virus titers from culture supernatants collected at the indicated times postinoculation were determined. (F and G) Luciferase activity. Vero-ACE2 or BHK-21 cells were infected with VSV-SARS2-Fluc or control VSV-Fluc or mock infected in 96-well plates, and at the indicated times luciferase activity was measured. (H to K) Flow cytometry. Expression of ACE2 (H and I) and TMPRSS2 (J and K) was measured in Vero and Vero-ACE2 cells by flow cytometry using anti-ACE2 or anti-TMPRSS2, respectively. Controls were secondary antibody only.

FIG 3

Inhibition of VSV-SARS2-Fluc by monoclonal antibodies and convalescent sera. (A) Infectivity of different Vero cell lines. VSV-SARS2-Fluc was incubated with 2 or 0.2 μg/ml of monoclonal anti-SARS-CoV-2 spike antibody mAb10914 in pooled seronegative sera or pooled seronegative sera alone (negative matrix). After 30 min, virus mixes were overlaid onto Vero, Vero-ACE2, or Vero-ACE2/TMPRSS2 cells. Luciferase activity was measured after an additional 24 h. Values represent the average (mean) relative light units (RLU) ± standard deviation. (B) Optimization of cell density. The indicated numbers of Vero-ACE2 cells were seeded in 96-well plates. The following day, virus mixes as described for panel A were overlaid onto the cell monolayers. Luciferase activity was measured after an additional 24 h. Values represent the average (mean) RLU ± standard deviation. (C) Neutralization by convalescent sera. VSV-SARS2-Fluc was incubated with pooled seronegative sera at a 1:80 dilution or serum samples from 11 donors (6 seronegative and 5 seropositive for anti-SARS-CoV-2 antibodies by ELISA) at a 1:80 dilution. After 30 min, virus-serum mixes were overlaid onto Vero-ACE2 cells. Luciferase activity was measured after an additional 24 h. Values represent average (mean) luciferase activity relative to the pooled seronegative serum sample control ± standard deviation.

FIG 4

Assay performance of VSV-SARS2-Fluc. (A) Susceptibility of virus to antibody neutralization. The indicated amounts (PFU) of VSV-SARS2-Fluc were incubated with 2 or 0.2 μg/ml anti-SARS-CoV-2 spike monoclonal antibody mAb10914, a SARS-CoV-2-seropositive plasma sample at a 1:80 dilution, or pooled seronegative serum (negative matrix, 1:80). After 30 min, virus mixes were overlaid onto Vero-ACE2 cells, and luciferase activity was measured after an additional 24 h. Values represent the average (mean) luciferase activity relative to the negative matrix control ± standard deviation. (B) Consistency of virus lots. Various amounts (PFU) of two different lots (A and B) of VSV-SARS2-Fluc were incubated with the indicated concentrations of mAb10914. Luciferase activity was measured after an additional 24 h. Values represent the average (mean) RLU ± standard deviation. (C and D) Virus stability. Aliquots of VSV-SARS2-Fluc were removed from the freezer, thawed, and either used immediately for assay (Immediate use) or stored either at room temperature or on ice for the indicated time (h). For assay, 300 PFU of VSV-SARS2-Fluc was incubated with 0.154 (QC-High) or 0.031 (QC-Low) μg/ml of anti-SARS-CoV-2 spike monoclonal antibody mAb10922 in pooled seronegative sera or in pooled seronegative sera alone (Neg. Control). After 30 min, virus mixes were overlaid onto Vero-ACE2 cells, and luciferase activity was measured after an additional 24 h. Values represent the average (mean) RLU ± standard deviation.

FIG 5

Effect of sample matrix on assay performance. (A and B) Effect of heat inactivation of sera or plasma. Matched serum (A) and sodium-heparin plasma (B) samples from 20 donors (11 seronegative and 9 seropositive for anti-SARS-CoV-2 antibodies by ELISA) were split and incubated either on ice or at 56°C for 30 min. Following incubation, plasma samples were clarified by centrifugation. Samples were then incubated at a 1:80 dilution with VSV-SARS2-Fluc. Pooled seronegative sera or plasma were used as assay controls. After 30 min, virus mixes were overlaid onto Vero-ACE2 cells, and after an additional 24 h, luciferase activity was measured. Values represent the average (mean) luciferase activity relative to the pooled seronegative matrix control for each sample. Statistical analysis was performed using a two-way repeated-measures (RM) analysis of variance (ANOVA) followed by a Bonferroni multiple-comparison test. n.s., not significant; * = <0.0001. (C to F) Characterization of matrix interference. Seronegative serum (C, n = 40), sodium-heparin plasma (D, n = 40), ACD plasma (E, n = 26), or K₂-EDTA plasma (F, n = 49) samples were serially diluted as indicated and incubated with VSV-SARS2-Fluc. Virus mixed with medium only was used as a control. After 30 min, virus mixes were overlaid onto Vero-ACE2 cells, and after an additional 24 h, luciferase activity was measured. Values represent the average (mean) luciferase activity relative to the medium control ± standard deviation.

FIG 6

Establishment of a standard curve for titer calculations. (A) Antibody-specific neutralization of VSV-SARS2-Fluc. The indicated concentrations of anti-SARS-CoV-2 spike monoclonal antibody mAb10914 or mAb10922 or isotype control antibody were incubated with VSV-SARS2-Fluc. After 30 min, virus mixes were overlaid onto Vero-ACE2 cells, and luciferase activity was measured after an additional 24 h. Values represent the average (mean) luciferase activity relative to the medium control ± standard deviation. (B) Standard curve performance. VSV-SARS2-Fluc was incubated with 0.8, 0.4, 0.2, 0.1, 0.05, or 0.025 μg/ml (corresponding to the indicated equivalent VNTs) of mAb10914 or negative pooled sera alone. After 30 min, virus mixes were overlaid onto Vero-ACE2 cells, and luciferase activity was measured after an additional 24 h. Values represent average (mean) luciferase activity relative to the pooled negative serum control ± standard deviation from 242 unique assay runs. (C) Limit of detection. Five different seronegative serum samples (at a 1:80 dilution) were spiked with anti-SARS-CoV-2 spike monoclonal antibody mAb10914 at 0.01, 0.02, 0.04, 0.06, 0.08, and 0.1 μg/ml (corresponding to the indicated equivalent VNTs) and incubated with VSV-SARS2-Fluc. VSV-SARS2-Fluc incubated with unspiked serum samples (Neg) or medium alone were included as controls. After the 30-min incubation, virus mixes were overlaid onto Vero-ACE2 cells, and luciferase activity was measured after an additional 24 h. Box and whisker diagrams display the interquartile range in the box, with the center line representing the median for the data set and whiskers representing the lower 5% and upper 95% value. Values are based on a total of 12 different assay runs performed on three separate days by six analysts using two different virus lots.

FIG 7

Interassay variability of standards and controls. Standards consisting of monoclonal anti-SARS-CoV-2 spike antibody mAb10914 at 0.8, 0.4, 0.2, 0.1, 0.05, and 0.025 μg/ml in media and QC High and QC Low controls consisting of 0.154 and 0.031 μg/ml antibody mAb10922, respectively, in pooled seronegative sera were incubated with VSV-SARS2-Fluc. Pooled seronegative serum alone was used as a negative control. After 30 min, virus mixes were overlaid onto Vero-ACE2 cells, and luciferase activity was read after an additional 24 h. A total of 207 assay runs were performed over 5 days, by five analysts, using two different virus lots. Box plot represents the 25th to 75th percentile of the data with the line representing the medium titer equivalent (VNT) value. Whiskers display the minimum and maximum values.

FIG 8

Correlation of virus-neutralizing units with PRNT50%. Fifty-eight SARS-CoV-2-seropositive serum samples were assayed using IMMUNO-COV v2.0 starting at a 1:80 dilution. Established controls, including a standard curve (0.8, 0.4, 0.2, 0.1, 0.05, and 0.025 μg/ml mAb10914 in medium), were included on each assay plate. The IMMUNO-COV v2.0 titer (VNT) was determined using the standard curve, where 1 VNT equals the concentration of mAb10914 multiplied by 400. All samples were subjected to PRNT using a clinical isolate of SARS-CoV-2. Statistical comparison of VNT to PRNT50% was performed using Spearman’s rank order correlation analysis as both data sets had a non-Gaussian distribution (P < 0.0001).

FIG 9

The strength of neutralizing antibody responses correlates with disease severity. As part of assay validation (Table 2), neutralizing antibody titers were determined for 46 donors who self-reported COVID-19 disease symptoms at least 2 weeks prior to sample donation. Disease symptoms were classified as severe (acute respiratory distress or pneumonia), moderate (shortness of breath), mild (fever, feverishness, cough, chills, myalgia, rhinorrhea, sore throat, nausea/vomiting, headache, abdominal pain, or diarrhea), or none (asymptomatic). The graph indicates the titer value (VNT) for each donor grouped based on disease symptoms. Bars represent the average (mean) titer for each group. Differences in antibody titers based on disease severity were not statistically significant (n.s.) by one-way ANOVA (P = 0.1904).

FIG 10

Durability of neutralizing antibody responses. (A to C) Samples were collected from donors in April and October 2020 (n = 13). Neutralizing antibody levels were measured using IMMUNO-COV v2.0 (A), the PRNT assay (B), or the c-PASS SARS-CoV-2 neutralization antibody detection kit (C), which is a binding assay that utilizes the SARS-CoV-2 spike RBD. The reductions in antibody titers were statistically significant (*) for IMMUNO-COV v2.0 and the PRNT assay but not significant (n.s.) for the sVNT binding assay (P = 0.0007, 0.0004, and 0.4669, respectively, from paired t test).