Automated and virus variant-programmable surrogate test qualitatively compares to the gold standard SARS-CoV-2 neutralization assay

Abstract

The ongoing emergence of new severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants underscores the need for rapid, adaptable, high-throughput testing. However, assays for neutralizing antibodies, which are a good measure of viral protection, usually require cell culture and either infectious SARS-CoV-2 or pseudotyped viral particles. To circumvent the challenges of cell-based assays, SARS-CoV-2 surrogate virus neutralization tests (sVNTs) measure inhibition of the binding of the spike (S) protein receptor binding domain (RBD) to its receptor, human angiotensin-converting enzyme 2 (hACE2) by neutralizing antibodies. Here we tested a prototype automated microfluidic cartridge-based sVNT platform using SARS-CoV-2 wild-type (WT) and B.1.617.2 (Delta) variant RBDs. This sVNT showed a high correlation with cell-based neutralization assays for biospecimens collected post-COVID-19 vaccination and post-SARS-CoV-2 infection as well as for pre-pandemic SARS-CoV-2 negative sera. Thus, this assay, which takes less than 80 min, is a relatively simple, safe, and accurate alternative to traditional VNTs.

Fig. 1. Schematic diagram of the sVNT for SARS-CoV-2.

The capture reagent is a DIG-labeled RBD (red) that binds to the anti-DIG antibody attached to the GNR wall (blue). Capture-protein-specific antibodies (gold) in the serum compete with His-tagged ACE-2 (brown and orange) to decrease the signal. The ACE-2-His bound to the DIG-labeled RBD is measured using a biotinylated anti-His antibody (pink) that produces a fluorescent signal when the biotin binds to a fluorescent streptavidin conjugate (green).

Fig. 2. Determination of inhibition cutoff percent.

ROC plot showing specificity and sensitivity for (a) WT and (b) Delta RBD. c The percent inhibition for post-vaccination/infection samples vs. WT-RBD (blue circles) or Delta-RBD (gray circles) and a panel of negative controls (white circles) comprised of pre-vaccination sera collected prior to 2019 and from other respiratory infections. The dotted line represents the 27% cutoff. Vaccination samples (n = 103) from the PASS cohort received the two-dose BNT162b2 (Pfizer) series, whereas the infection samples (n = 51 serum samples from 17 individuals at 3 time points) represent individuals positive for SARS-CoV-2 with varied vaccination status from the EPICC cohort. The negative control sample panel (n = 203) comprised pre-vaccination confirmed negative samples (n = 103), sera collected prior to the outbreak of SARS-CoV-2 (n = 50), and samples from other respiratory viruses and human coronaviruses (n = 50): 229E (n = 9), HKU1 (n = 9), NL63 (n = 10), OC43 (n = 8), influenza A (n = 6) and influenza B (n = 8).

Fig. 3. Correlations for the sVNT, mVNT, and pVNT antibody titers. (log EC₅₀).

Correlations for sVNT, mVNT, and pVNT antibody titers (log EC₅₀) against WT RBD. a–c Post-vaccination (n = 36), (d–f) early post-infection (n = 14), (g–i) 6 months post-infection (n = 14), and (j–l) 12 months post-infection (n = 14). The correlation was determined by Spearman’s rho. Linear regression line (solid line), Spearman’s 95% confidence interval (CI) (dashed line), Spearman’s P-value, and n are shown.

Fig. 4. Comparison of titers of antibodies against WT RBD for the mVNT, pVNT, and sVNT.

Differences between mean log EC₅₀ titers post-vaccination (n = 36), early post-infection (n = 14), 6 months post-infection (n = 14), and 12 months post-infection (n = 14). The upper and lower bars represent the minimum/maximum, ns = not significant (p > 0.05). Dunn’s statistical significance following normality testing was calculated using GraphPad

Fig. 5. Comparison of antibody titers using the WT-sVNT and the Delta-sVNT.

Log EC₅₀ antibody titers for WT-sVNT and Delta-sVNT for samples (A) post-vaccination (n = 102), (B) early post-infection, (n = 13), (C) 6 months post-infection (n = 12), and (D) 12 months post-infection (n = 12). The upper and lower bars represent the minimum/maximum, ns = not significant (p > 0.05). Wilcoxon statistical significance was calculated using GraphPad.