Validation of a competitive enzyme-linked immunosorbent assay to improve the serological diagnosis of swine vesicular disease

Abstract

Swine vesicular disease (SVD) is an infectious viral disease of pigs. The clinical symptoms of SVD are indistinguishable from other vesicular diseases. In countries free of vesicular diseases, rapid SVD diagnosis and differentiation from other vesicular diseases are essential. In this report, a competitive enzyme-linked immunosorbent assay (cELISA) was developed and validated to improve the current SVD serological diagnosis. In this cELISA, an anti-SVD monoclonal antibody (mAb) captures the recombinant SVD virus-like particle (SVD-VLP) antigen, and 5B7 mAb is used as a competitor to compete with polyclonal antibodies in SVD-positive sera. The cut-off value of the SVD-VLP based cELISA (SVD-VLP cELISA) is ≥ 65% inhibition (%). The determined diagnostic specificity was 99.2%. SVD-VLP cELISA successfully detected SVD antibodies in the sera of SVD-infected animals and produced a diagnostic sensitivity of 100%. A panel of SVD positive sere including outbreak samples (n = 11) and samples (n = 5) from experimentally inoculated pigs, were correctly identified as positive by the SVD-VLP cELISA. In terms of reducing false positives detected by the currently used cELISA (5B7 cELISA), the performance of SVD-VLP cELISA is comparable to the gold standard virus neutralization test.

Swine vesicular disease (SVD) is a contagious viral disease of pigs caused by an enterovirus of the Picornaviridae family. The first known outbreak of swine vesicular disease (SVD) was in Italy in 1966 (1). Since then, many outbreaks have occurred throughout Europe, particularly during the seventies (2). The disease also occurred in parts of Asia, and the last outbreak was reported in China in 1999 (3). Swine vesicular disease eradication was formally recognized by the European Union in 2019 (4). Despite achievement of eradication, the SVD surveillance program is still in place in Italy (5).

Swine vesicular disease is characterized by the formation of vesicles and erosions around the calves and mouths of pigs. This disease is not usually fatal, but it is highly contagious and may cause economic losses due to export restrictions. Although SVD has not been listed in the World Organisation for Animal Health (OIE) since 2014, it is important to differentiate SVD from other very similar vesicular diseases, especially foot-and-mouth disease (FMD) and Senecavirus A (SVA) infection. Therefore, it is very important to diagnose SVD quickly and accurately.

For surveillance and export certification, the most used serological tests are competitive enzyme-linked immunosorbent assay (cELISA) and virus neutralization (VNT), in which cELISA is used as a screening test and VNT is used as a confirmation test. Virus neutralization is labor-intensive and takes several days to obtain results, whereas cELISAs are fast and easy to perform. The SVD serological test recommended by OIE for screening is the 5B7 cELISA (6,7), with inactivated SVD virus (SVDV) as the antigen. To prepare virus antigen, the infectious SVDV must be cultured in mammalian cells and inactivated by a hazardous chemical, binary ethyleneimine. Using non-infectious recombinant virus-like particle (VLP) as antigen will avoid handling live viruses. The preparation of recombinant antigens does not require a high-level biosafety laboratory and needs less regulatory control than the production of viral antigens. In our laboratory, recombinant SVD-VLPs have been previously generated (8). The results indicated that SVD-VLP could be used as an antigen in SVD-cELISA to replace SVD viral antigen. However, in the previous report, only a few SVD positive samples were tested. The cut-off value, diagnostic specificity and sensitivity were not yet determined and validated for use in diagnostic applications. Here we report on the development and validation of cELISA, using recombinant SVD-VLP as the antigen, SVD-specific mAb F44SVD as the capture antibody, and horseradish peroxidase (HRP) — conjugated mAb 5B7 as the competitive antibody (6,9). The comparison of SVD-VLP — based cELISA (SVD-VLP cELISA) with 5B7 cELISA and VNT confirmed that SVD-VLP could replace SVDV antigen for SVD serological diagnosis.

The production of SVD-VLP was described previously (8). The recombinant SVD-VLPs were generated using the Bac-to-Bac baculovirus expression system (Invitrogen Corporation, Waltham, Massachusetts, USA). Analysis of SVD-VLP confirmed that the morphology and antigenicity of the recombinant SVD-VLP are similar to the natural SVDV. An in-house generated and wellcharacterized mAb F44SVD was selected as the capture mAb to replace the mAb 5B7. Although there is only one SVDV serotype, several strains with differing virulence have been identified previously. Genetic and antigenic analyses have classified SVDV isolates into at least 4 phylogenetically distinct groups (2). The mAb F44SVD binding site is located at site Ia which is conserved in all SVDV antigenic groups (9,10). For the SVD-VLP cELISA, 96-well Maxisorp plates (Nunc Company, Roskilde, Denmark) were coated with 100 μL (400 ng/mL) of the purified mAb F44SVD in carbonate buffer overnight at 4°C. After washing with a washing buffer [Phosphate-Buffered Saline with 0.05% Tween 20 (PBS-T)], 100 μL of SVD-VLP in 4% skimmed milk powder (SMP) in PBST was added and incubated at 37°C for 1 h. After washing, 50 μL of quality control (Q1: strong positive) and serum samples (1:10 in PBS-T) were added. Without washing, an equal volume of the HRP conjugated-5B7 mAb (1:300 in 4% SMP in PBS-T) was added at the same time and incubated simultaneously for 1 h. Following washing, the substrate, 3,3′,5′-tetramethyl benzidine dihydrochloride (TMB, Sigma-Aldrich, St. Lucia, Missouri, USA) was added and incubated for 15 min for color development. The reaction was stopped using 2N sulfuric acid and the optical density (OD) was determined at 450 nm using an Emax microplate reader (Molecular devices, San Jose, California, USA). Results were calculated based on Q1 and diluent control (DC) using the following formula:

$$\text{Percent\hspace{0.17em}inhibition\hspace{0.17em}}(\text{PI})=\hspace{0.17em}[(\text{DC}-\text{Sample\hspace{0.17em}OD})/(\text{DC}-\text{Q}1)\times 100\%].$$

Optimal antigen concentration, antibody dilutions, and buffer selection were pre-determined by checkerboard titrations and comparative trials. The cut-off value of SVD-VLP cELISA is calculated as the mean PI of negative pig serum plus 3 standard deviations (SD). According to the results of 1072 negative pig sera from different regions of Canada, the average PI of 1072 negative serum values is 24.2 plus 3 SD which equals 64.7. Therefore, the threshold of PI is set to ≥ 65%. Eight of the 1072 samples exceeded the 65% cut-off value, which produced a diagnostic specificity of 99.2%. Using SVD-VLP cELISA, the positive sera from 22 pigs [n = 120, 7 to 49 days post-infection (dpi)] (11) were all identified as positive results. Therefore, the diagnostic sensitivity of the SVD-VLP cELISA was 100%. SVD-VLP was previously reported used as an antigen in a SVD cELISA (12). However, only 6 EU SVD reference sera (RS1 to RS6) were tested in the report which was not enough to draw a definitive conclusion. To evaluate SVD-VLP cELISA, a broader panel of sera would be needed to obtain a conclusion. A panel of SVD positive sera was obtained from the Istituto Zooprofilattico Sperimentale della Lombardiae dell’Emilia-Romagna (IZSLER). The panel included 11 samples from outbreaks in Italy from 1994 to 2007 and 5 from experimentally inoculated pigs, correctly identified as positive by the SVD-VLP cELISA (Table 1).

Table I.

Results of SDV positive serum panel using 5B7 cELISA, SVD-VLP cELISA and virus neutralization test.

Sample number	Serum type	Virus Origin	5B7		SVD-VLP
			cELISA	VNT****	cELISA
			PI (%)***	Titer	PI (%)***
1	Field*	Torino 265914/2002	96	1440	107
2	Field*	Torino 265915/2002	97	1024	107
3	Field*	Torino 265916/2002	96	1440	105
4	Field*	Torino 265917/2002	98	> 2048	111
5	Field*	Pig 1. Rubiera, BS	92	2048	94
6	Field*	Pig 12. Rubiera, BS	98	> 2048	106
7	Field*	Pig 13. R1569, CR/2007	96	1024	105
8	Exp. Infec.**	Pig 4. R1076/1992	99	> 2048	113
9	Exp. Infec.**	Pig 761. R 1394, Firenze/2002	95	708	104
10	Exp. Infec.**	Pig 1. R1076/1992	99	> 2048	107
11	Field	Pig 1. R1569, CR/2007	83	355	86
12	Field	Pig 2. R1569, CR/2007	89	1024	96
13	Exp. Infec.**	Pig 836. R 1394, Firenze/2002	95	> 2048	110
14	Field	Pig 64. Coprolama	99	> 2048	113
15	Field	Pig 7. 5/2003	91	> 2048	95
16	Exp. Infec.**	Pig 0934. Iperimmune, Italy/72	99	> 2048	110

^*Field;

^**Exp. Infec.;

^***PI — Percent inhibition;

^****VNT — Virus neutralization titer.

SVD reference sera was obtained from the Istituto Zooprofilattico Sperimentale della Lombardiae dell’Emilia-Romagna (IZSLER), including 11 outbreak samples in Italy from 1994 to 2007 and five sample from experimentally inoculated pigs. The sera were tested using 5B7 cELISA, VNT, and SVD-VLP cELISA. Sera giving a percent inhibition ≥ 55% for 5B7 cELISAs, titer ≥ 1:64 for VNT, and a percent inhibition ≥ 65% for SVD-VLP cELISAs were considered positive results. Through three tests, all sera were correctly identified as positive.

The kinetics of antibody responses to SVDV in the sera was examined using the SVD-VLP cELISA. Serial serum samples were obtained from 16 pigs (identified as pig #109–124) experimentally infected with SVDV UKG 27/72 strain (16 pigs, 0 to 14 dpi, 6 of 16 pigs up to 21 dpi) (13). Thirteen of the 16 pigs showed positive seroconversion at 4 dpi and all animals were positive at 5 dpi (Figure 1).The serum samples remained positive until the end of the experiment (21 dpi). Results were compared with those obtained with the 5B7 cELISA. The original protocol of 5B7 cELISA has been modified in our laboratory with the following modifications: i) using a final serum dilution of 1/15 as opposed to 1/7.5; and ii) according to current validation results using the sera from Canadian swine, the cut-off value was set at ≥ 55% PI, whereas the OIE manual indicates that inhibition of ≥ 80% is positive. The 5B7 cELISA used in this report refers to the modified 5B7 cELISA. The SVD-VLP cELISA detected seroconversion 1 day earlier than the 5B7 cELISA in 7 pigs. The correlation coefficient (R²) between the 2 cELISAs, evaluated with results of the 142 serial bleed samples from 16 pigs was 0.96 (P < 0.0001) (Figure 2).

Figure 1.

The kinetics of antibody responses to SVDV using SVD-VLP cELISA. Pigs #109–124 were experimentally infected with SVDV and serum samples were collected from 0 to 21 days post-inoculation (dpi). Sera were collected daily up to 10 dpi, and then at 7-day intervals. Serum samples were tested using SVD-VLP cELISA.

Figure 2.

Correlation between the 5B7 cELISA and the SVD-VLP cELISA. Scatter plot representing the percent inhibition of 5B7 cELISA versus SVD-VLP cELISA for 142 samples (0 to 21 dpi) collected from 16 pigs (#109–124) experimentally infected with SVDV. The line is fitted using linear regression and the correlation coefficient is 0.96 (P < 0.0001).

The SVD-VLP cELISA was also compared with the virus neutralization test (VNT), which is recommended as confirmatory test (7). Following internal standardization, sera with titers ≥ 1:64 (1.8 log10) were considered positive. The comparison was conducted using serial serum samples from 6 experimentally inoculated pigs (pig #119–124). The (R²) determined between VNT and SVD-VLP cELISA was 0.88 (P < 0.0001) (Figure 3a), and was 0.82 (P < 0.0001) between VNT and 5B7 cELISA (Figure 3b).

Figure 3.

Correlation between cELISAs and VNT. Scatter plots representing the percent inhibition of cELISAs versus VNT titers for 53 serum samples from 6 pigs (#119–124) infected with SVDV. Sera were collected daily up to 10 dpi, and then at 7-day intervals. The lines are fitted by linear regression (a) percent inhibition of SVD-VLP cELISA versus VNT log10 titer, R² = 0.88, P < 0.0001, and (b) percent inhibition of 5B7 cELISA versus VNT log10 titer, R² = 0.82, P < 0.0001.

During surveillance or export certification, a small percentage (0.2 to 0.4%) of pigs without a history of SVDV exposure give false positive results using 5B7 cELISA, and up to half of these pigs will also have false positive results for VNT. These sera are called “singleton reactors” (SR) (6,7,4–17). The reason for the false positives is not yet clear. These SRs cause delays in the release of results for export testing. According to our protocol, any 5B7 cELISA positive samples will be retested. If it remains positive, the isotyping ELISA and VNT are executed. Any determined positive samples will be rebled and retested. If it is still positive and the SVDV genome is detected in saliva and rectal swabs, SVD will be suspected and investigated further. The exporting test results of the National Center for Foreign Animal Diseases of Canada from 2013 to 2017 showed that 295 sera were tested positive (PI ≥ 55%) using 5B7 cELISA and were considered 5B7 cELISA positive sera. To evaluate the ability of SVD-VLP cELISA to reduce the percentage of the false-positives, these archived 5B7 cELISA positive samples were tested using SVD-VLP cELISA and compared with VNT results. SVD-VLP cELISA detected 29 positives out of 295 samples, thus reducing false-positives by 90.1%; VNT was performed on 219 sera with sufficient volume available for testing, reduced false-positives by 91.7% (Figure 4). Only one sample is positively detected by all three methods (60%PI for 5B7 cELISA, 96.4%PI for VLP-cELISA, and VNT titer 128). Others detected positive samples by VNT are not the same samples detected by SVD-VLP cELISA. This indicates that the false-positive samples are not true SVD-positive but caused by nonspecific reactions. The results show that the SVD-VLP cELISA results are equivalent to the VNT gold standard test in reducing the false positive percentage of 5B7 cELISA. The following various reasons may explain why 5B7 cELISA shows higher false positive results: i) archived results of 5B7 cELISA were used instead of parallel testing with VNT and SVD-VLP cELISA, and the generally weak/borderline SVDV-reactivity of most false-positives may have declined due to the age of samples; ii) in this study, the 5B7 cELISA was used for screening and the SVD-VLP cELISA or VNT as confirmatory tests. It is possible that, in the opposite order, the 5B7 cELISA would not confirm the false-positives revealed by the SVD-VLP ELISA or VNT; iii) the cut-off value of the 5B7 cELISA currently used in our laboratory may need to be further evaluated and should be re-determined based on the latest samples from Canada, rather than following the cut-off value determined in 1998 (11).

Figure 4.

Scatter plot of 5B7 cELISA, VNT and SVD-VLP cELISA results of false positive sera identified by 5B7 cELISA from Canadian pig population (2013–2017). Sera giving a percent inhibition ≥ 55% for 5B7 cELISAs, titer ≥ 1:64 (1.8 log10) for VNT, and a percent inhibition ≥ 65% for SVD-VLP cELISA were considered positive results (a) 5B7 cELISA, (b) VNT, and (c) SVD-VLP cELISA. One sample that is positive by the 3 methods is marked with triangles.

In summary, a SVD-VLP cELISA using recombinant SVD-VLP, a new SVD-specific mAb and HRP-conjugated 5B7 mAb was developed and validated. The diagnostic specificity and sensitivity of SVD-VLP cELISA are 99.2% and 100% respectively. Comparable to VNT results, SVD-VLP cELISA reduced the false-positive samples identified by 5B7 cELISA. SVD-VLP cELISA can be used as a firstline test for routine export testing, which has the advantage of using recombinant SVD-VLP instead of SVD virus.