The Luciferase Immunoprecipitation System (LIPS) Targeting the Spike Protein of SARS-CoV-2 Is More Accurate than Nucleoprotein-Based LIPS and ELISAs for Mink Serology

Abstract

Since anthropo-zoonotic outbreaks of SARS-CoV-2 have been reported in mink farms, it is important to monitor the seroprevalence within this population. To investigate the accuracy of nucleo (N) or spike (S) protein-based assays to detect anti-SARS-CoV-2 antibodies in animal serum, we compared four assays, two commercial N-based enzyme-linked immunosorbent assays (ELISA) validated for animal sera and two luciferase immunoprecipitation systems (LIPS-N and LIPS-S), to the reference standard plaque reduction neutralisation test (PRNT). Samples included in this study were derived from a naturally infected mink population. For the first time in this study, serum samples of mink were collected over a 307-day period, at different time points, thus providing an overview of performances of four different rapid serological tests over time. The assays were compared by performing a correlation analysis using R2, Spearman's rank-order correlation coefficient, and Fleiss' and Cohen's kappa for analysis of agreement to PRNT, and an UpSet chart was created to visualize the number of shared positive samples between assays. Cohen's kappa test on categorical data showed an excellent agreement between PRNT and LIPS-S, while agreements between PRNT and N-based methods decreased from fair for LIPS-N to poor agreements for the ELISA kits. In addition, LIPS-S revealed the highest number of true-positive SARS-CoV-2 samples compared to N-based methods. Despite an excellent agreement between LIPS-S and PRNT, a weak correlation was detectable between PRNT titres and relative light units. This study shows that the LIPS-S assay can be used for serological surveillance within a naturally exposed mink population, while N-based serological assays are less accurate providing a higher number of false-negative results, especially at a later stage of infection, thus indicating that N antibodies are less persistent in naturally exposed mink. Our findings provide crucial information for veterinarians and competent authorities involved in surveillance and outbreak investigation in wild and farmed minks.

Figure 1

UpSet plot showing relationships of PRNT, LIPS-S, LIPS-N, Eradikit, and IDV. UpSet is sorted according to shared positive results between the assays (n = 77).

Figure 2

Cohen's kappa agreements for each time point (A–D) between IDV and Eradikit, LIPS-N and Eradikit, LIPS-N and IDV, LIPS-S and LIPS-N, LIPS-S and IDV, PRNT and LIPS-S, PRNT and Eradikit, PRNT and IDV, and PRNT and LIPS-N.

Figure 3

Distribution of RLU values generated using the sera from naturally SARS-CoV-2-infected mink samples across (a) the threshold line for the LIPS-S assay with the indication of PRNT results and sera sampling dates, (b) LIPS-N assay with threshold line, identification of PRNT results, and sera sampling dates, (c) LIPS-N assay with threshold line, identification of Eradikit results, and sera sampling dates, and (d) LIPS-N assay with threshold line, identification of IDV results, and sera sampling dates. Negative samples are located below the blue threshold line of mean plus five standard deviations.

Figure 4

Boxplot of PRNT titres at each sampling point using pairwise comparison with the Wilcoxon rank sum test and Bonferroni adjustment (alpha = 0.05).

Figure 5

Visualization of the correlation between LIPS-S and PRNT using Spearman's rank-order correlation coefficient.