Detection of pseudorabies virus antibody in swine oral fluid using a serum whole-virus indirect ELISA

Abstract

We evaluated the detection of pseudorabies virus (PRV) antibodies in swine oral fluid. Oral fluid and serum samples were obtained from 40 pigs allocated to 4 treatment groups (10 pigs/group): negative control (NC); wild-type PRV inoculation (PRV 3CR Ossabaw; hereafter PRV); PRV vaccination (Ingelvac Aujeszky MLV; Boehringer Ingelheim; hereafter MLV); and PRV vaccination followed by PRV inoculation at 21 d post-vaccination (MLV-PRV). Using a serum PRV whole-virus indirect IgG ELISA (Idexx Laboratories) adapted to the oral fluid matrix, PRV antibody was detected in oral fluid samples from treatment groups PRV, MLV, and MLV-PRV in a pattern similar to serum. Vaccination alone produced a low oral fluid antibody response (groups MLV and MLV-PRV), but a strong anamnestic response was observed following challenge with wild-type virus (group PRV). Analyses of the oral fluid PRV indirect IgG ELISA results showed good binary diagnostic performance (area under ROC curve = 93%) and excellent assay repeatability (intra-class correlation coefficient = 99.3%). The demonstrable presence of PRV antibodies in swine oral fluids suggests the possible use of oral fluids in pseudorabies surveillance.

Introduction

Pseudorabies virus (PRV; Suid alphaherpesvirus 1) is an enveloped DNA virus in family Herpesviridae, subfamily Alphaherpesvirinae. Swine are the natural host of PRV.³⁰ In domestic pigs (Sus scrofa domesticus), the clinical expression of PRV infection ranges from inapparent to nearly 100% mortality in susceptible piglets.^35,37,43 Typical of herpesviruses, PRV establishes latency in pigs that survive acute infection, with virus subsequently detectable in trigeminal ganglion, olfactory bulb, and tonsil tissues.^13,31,34,36

Beginning in the 1960s, clinical PRV became increasingly problematic in commercial swine herds in Europe, the Americas, and Southeast Asia, with annual losses estimated at US$21 to $25 million to U.S. producers.^21,25 The development of efficacious PRV marker vaccines and differential serum antibody ELISA made PRV eradication possible from domestic swine (e.g., Great Britain in 1991, Germany in 2003, New Zealand in 1997, the United States in 2004),^{3,14,23,27,42} but PRV continues to present a risk to commercial swine producers because it remains endemic in feral swine populations in Europe,^22,39,40 Asia,^16,19 and the Western Hemisphere.^24,26,29 In the United States, PRV has occasionally been introduced into “transitional” herds via contact with feral swine (e.g., Minnesota in 2002, Wisconsin in 2007). In France, PRV was detected in 2019 in 2 farms reportedly via contact with feral swine.¹¹ Under these circumstances, improvements in PRV testing, surveillance, control, and elimination remain relevant.

The purpose of disease surveillance is to monitor herd immunity and/or to determine population infection status. In the case of pathogens for which marker vaccines and corresponding assays have been developed, both functions can be achieved simultaneously. PRV control and eradication has been achieved in many parts of the world using glycoprotein E (gE)-deleted PRV MLVs and gB/gE serum ELISAs.^{2,10,12,41,44}

Current PRV differentiation of infected and vaccinated animals (DIVA) serum ELISAs are designed for testing individual pigs, whereas the industry is moving toward the use of population-based specimens (e.g., processing fluid samples and group-level oral fluid samples) as a means to achieve a higher probability of detection at a lower cost.⁴ Commercial serum ELISAs can often be adapted to the oral fluid matrix by accounting for the lower concentration of IgG antibody in oral fluid.^6,24 In general, this involves adjusting 4 variables: 1) sample volume and/or dilution ratio, 2) incubation time, 3) incubation temperature, and 4) conjugates and/or substrates. Including both nucleic acid and antibody testing, the veterinary diagnostic laboratories at Iowa State University, South Dakota State University, and the University of Minnesota performed > 350,000 tests on swine oral fluid specimens in 2016.⁴ These developments were driven by the ease of oral fluid sampling in the field and the emergence of sensitive and specific assays adapted to the oral fluid matrix. Continuing this line of research, we explored the detection of PRV antibody in oral fluid specimens using a PRV serum antibody indirect IgG ELISA.

Materials and methods

Experimental design

We evaluated the detection of PRV antibody in swine serum (n = 350) and oral fluid (n = 1,540) samples of known infection and/or vaccination status using an indirect serum PRV IgG ELISA (Idexx Laboratories). Samples were obtained from 12- to 16-wk-old pigs in 4 treatment groups (10 pigs/group): negative control (NC), wild-type PRV inoculated (PRV), PRV vaccinated (MLV), and PRV vaccinated and challenged at 3 wk post-vaccination (MLV-PRV). Depending on the group, serum and oral fluid samples were collected from individual animals for up to 49 d (Table 1). Serum samples were tested as prescribed by the manufacturer, whereas oral fluid samples were tested using a modified protocol. Receiver operating characteristic (ROC) curve analysis was used to evaluate test performance. The repeatability of serum and oral fluid ELISAs was calculated for both quantitative and qualitative results. All procedures were approved by the U.S. Department of Agriculture–Agricultural Research Service and Iowa State University Institutional Animal Care and Use Committees.

Table 1.

Experimental design for the evaluation of pseudorabies virus antibody ontogeny in serum and oral fluid specimens.

Treatment group	Pigs (n)	Treatment		Sampling schedule
		Vaccination*	Inoculation†	Serum (d)	Oral fluid (d)
NC	10	NA	NA	0, 7, 11, 14, 17, 21, 24, 28, 35, 42, 49	0–49 daily
PRV	10	NA	Day 28	23, 28, 35, 42, 49	23–49 daily
MLV	10	Day 7	NA	2, 7, 11, 13, 16, 20, 23, 27	0–26 daily
MLV-PRV	10	Day 7	Day 28	2, 7, 11, 13, 16, 20, 23, 27, 34, 41, 48	0–49 daily

MLV = vaccination only; MLV-PRV = vaccination then pseudorabies virus inoculation; NA = not applicable; NC = negative control; PRV = PRV inoculation only.

^*Intramuscular injection of PRV modified live virus (MLV) vaccine (Ingelvac Aujeszky MLV; Boehringer Ingelheim).

^†Intranasal inoculation of PRV strain 3CR Ossabaw (1 × 10^3.5 TCID₅₀ per pig).

Virus inoculum

Animals were inoculated with wild-type PRV (PRV isolate 3CR Ossabaw) propagated on swine testicular (ST) cells.⁴⁷ In brief, a monolayer of ST cells in a 75-cm² flask was inoculated with virus (0.5 mL, 1 × 10^6.8 TCID₅₀/mL). When cytopathic effect (CPE) was observed in 80–90% of the monolayer, the flask was subjected to 2 freeze–thaw cycles (–80°C), the harvested contents were clarified by centrifugation (1,000 × g for 10 min), and the supernatant stored in 0.5-mL aliquots at –80°C.⁴⁷

Virus titrations were done on confluent monolayers of ST cells in 96-well plates. Virus solutions were serially 10-fold diluted (10^-1–10^-9) in minimum essential medium (MilliporeSigma) and then columns of wells preloaded with maintenance medium (100 µL) were inoculated with each dilution (100 µL). Plates were incubated at 37°C in a humidified 5% CO₂ incubator and then examined for CPE at 72 h. Virus titers and 95% CIs were calculated using the Spearman–Kärber method (http://www.klinikum.uni-heidelberg.de/Downloads.126386.0.html).^17,38

Animals and treatment groups

Forty 12- to 16-wk-old crossbred domestic pigs were obtained from a commercial swine herd in the north-central United States. All pigs were confirmed negative for porcine reproductive and respiratory syndrome virus antibodies before the start of the experiment. Pigs were individually housed in pens (1.5 × 1.5 m) constructed of gates allowing interaction between animals in neighboring pens. Each pen was equipped with a nipple drinker and a bracket from which to suspend a rope for oral fluid collection.

Limitations in facility space dictated that pigs were received in 2 lots (10 and 30 pigs) and likewise determined the length of observation for each group. Groups MLV, PRV, and MLV-PRV were housed for 27, 38, and 62 d, respectively, in a biosafety level 3 (BSL3-Ag) high-containment animal facility at the USDA–National Animal Disease Center (Ames, IA). Group NC was held for 54 d in a BSL-2 livestock infectious disease isolation facility at Iowa State University (Ames, IA; Table 1). On day 0, pigs were randomly assigned to treatment groups by blindly selecting ear tags from a bag and then housed individually in pens. On day 7, pigs in the MLV and MLV-PRV groups were vaccinated intramuscularly with a 2-mL dose of a commercial vaccine (Ingelvac Aujeszky MLV; Boehringer Ingelheim). On day 28, pigs in the PRV and MLV-PRV groups were intranasally exposed to 2 mL of an inoculum containing 1 × 10^2.9 TCID₅₀/mL (95% CI: 1 × 10^2.5 and 1 × 10^3.3) PRV isolate 3CR Ossabaw into each naris (i.e., 1 × 10^3.5 TCID₅₀ per pig; Table 1).

Throughout the experiment, pigs were observed at least twice daily for general health and signs of infectious disease. All animals were managed identically and, at the termination of the study, all animals were euthanized by captive bolt or chemical injection followed by exsanguination, as specified in the American Veterinary Medical Association Guidelines for the Euthanasia of Animals.¹

Biological specimens

Blood samples (Table 1) were collected from the external jugular vein using 20-ga needles (Exelint International), reusable hubs (Becton Dickinson), and 12.5-mL serum separator tubes (Covidien). Serum was harvested by centrifugation (1,000 × g for 15 min), aliquoted into 2-mL cryogenic vials (Cryovial; Greiner Bio-One), and stored at –20°C.

Oral fluid samples (Table 1) were collected from individual pigs daily by hanging 1.27 cm (0.5 in) diameter, 3-stranded, 100% cotton rope (Web Rigging Supply) from metal brackets installed in each pen.^15,46 Rope length and bracket height were adjusted to the size of pigs and placed to avoid contact with the floor. After allowing pigs to interact with the ropes for 30–45 min, the wet portion of the rope was placed in a plastic bag (Elkway Plastics) and passed through a chamois wringer (Dyna-Jet Products). Liquid that pooled in the bag was decanted into 50-mL conical centrifuge tubes (Falcon; Fisher Scientific) and stored at –20°C.

PRV indirect IgG ELISA

Serum and oral fluid samples were tested for PRV IgG using a PRV indirect serum ELISA based on the Shope strain of PRV (Idexx ADV(S) indirect ELISA; Idexx Laboratories). Serum samples were tested and the results interpreted according to the manufacturer’s instructions. Negative (3 wells), weak-positive (3 wells), and strong-positive (2 wells) controls for plate validation and calculating results were provided by the manufacturer. Results were standardized as sample-to-positive (S/P) ratios (equation 1), and samples with S/P ratios ≥0.4 were classified as PRV antibody positive (OD = optical density).

$$S/P=ratio=\frac{\left(\mathrm{sample}\mathrm{OD}-\mathrm{negative}\mathrm{control}\mathrm{mean}\mathrm{OD}\right)}{\left(\mathrm{weak-positive}\mathrm{control}\mathrm{mean}\mathrm{OD}-\mathrm{negative}\mathrm{control}\mathrm{mean}\mathrm{OD}\right)}$$ (1)

Oral fluid samples were tested by the PRV indirect IgG serum ELISA using a modified protocol. Samples (100 µL) were first diluted (1:1) with kit diluent, then 100 µL was added to each well and the plates incubated for 16 ± 2 h at 4°C. Wells were washed 5 times with kit wash solution (300 µL), then goat anti-pig IgG (Bethyl Laboratories) diluted 1:2,000 was added to each well (100 µL), and the plates incubated at 37°C for 2 h. The remainder of the procedure was performed as per the manufacturer’s instructions. Negative and positive controls for plate validation and S/P calculation consisted of oral fluid samples derived from the animal inoculation study with oral fluid OD values of 0.09 ± 0.01 and 1.02 ± 0.06, respectively. Oral fluid S/P ratios were calculated as shown in equation 2.

$$S/P=ratio=\frac{\left(\mathrm{sample}\mathrm{OD}-\mathrm{negative}\mathrm{control}\mathrm{mean}\mathrm{OD}\right)}{\left(\mathrm{positive}\mathrm{control}\mathrm{mean}\mathrm{OD}-\mathrm{negative}\mathrm{control}\mathrm{mean}\mathrm{OD}\right)}$$ (2)

Statistical analysis

ELISA repeatability

The repeatability of the PRV serum and oral fluid ELISAs was quantified by calculating intra-class correlation coefficients (ICC) using the R package irr (v.0.84.1) and the results interpreted based on established guidelines (https://cran.r-project.org/web/packages/irr/index.html).¹⁸ Among the methods described for quantifying repeatability, ICC was chosen because it is better suited to manage > 2 test repeats.⁵ The repeatability analysis was based on ELISA testing of a subset of 40 oral fluid and 40 serum samples randomly selected from each of the 10 animals in each of the 4 treatment groups. A sample size of 40 was calculated using equation 3, in which n is the number of samples, m is the number of repetitions for each sample, and Z_1-(α/2) indicates the 100(1-α/2)th percentile of the standard normal distribution.²⁰ Solving equation 3 for n produced equation 4. The 2-tailed α was set at 0.05, therefore, the half-width of the 95% CI of the within-subject standard deviation (S_w) estimation was set at 10%. Variation between plates and kit lots was assumed to be equal across all tests (i.e., measurement variation was the variable of interest).

$$Z_{1-\left(\alpha /2\right)}\times \frac{S_w}{\surd 2n(m-1)}=10\%\times S_w$$ (3)

$$n=\frac{1}{2(m-1)}\times {\left(\frac{Z_{1-\left(\alpha /2\right)}}{10\%}\right)}^2$$ (4)

This subset of samples was tested 6 times using the previously described specimen-specific procedures, thereby producing 240 S/P ratios for each of the 2 sample types. Thereafter, the ICC was estimated using a 3-step approach: 1) a 2-way mixed-effects model was used to estimate intra-replicate reliability and avoid the influence of heterogeneity, 2) S/P ratios from the first replicate were used as a baseline, and 3) “absolute agreement” was used to determine whether different replicates assigned the same measurement to the same samples.¹⁸ Higher correlation suggested fewer measurement errors and better repeatability.⁷ Previously established ICC classification guidelines¹⁸ were used to interpret the repeatability of the indirect PRV IgG ELISA testing serum and oral fluid samples: 1) < 50% (poor); 2) 50–75% (moderate); 3) 75–90% (good); 4) > 90% (excellent).

Serum versus oral fluid ELISA responses

The relationship between serum and oral fluid testing was evaluated by fitting the data to a cubic linear regression model. Before fitting the model, S/P ratios from paired serum (n = 350) and oral fluid samples (n = 350; pig by day of study) were transformed (log₁₀) to normalize the model residuals. In addition, the average ranked ELISA S/P ratios from serum and oral fluid samples were compared using a nonparametric test (Kruskal–Wallis) because the assumption of normality was rejected (Shapiro–Wilk test, p < 0.001) for both the raw and transformed (square root, cube root, and log) ELISA results from serum (n = 350) and oral fluid (n = 350) samples paired by day of study.

ROC analysis

ROC curve analyses were done to evaluate the performance of the serum and oral fluid ELISAs using R software package pROC (v.1.16.2; https://cran.r-project.org/web/packages/pROC/index.html).³² Diagnostic sensitivity and specificity were derived from the ROC analyses for specific assay cutoffs. Associated 95% CIs were estimated using a procedure for normally distributed correlated data.⁹ Before performing the analyses, the S/P data of serum and oral fluid were normalized by 4/7 and 1/3 power transformations, respectively. ROC analyses require designation of the status (positive or negative) of the sample. Based on prior reports,^28,45 sample status was defined as: 1) serum and oral fluid samples collected ≤ 7 d post-vaccination (dpv) or day post-inoculation (dpi) were classified as PRV antibody negative; 2) serum samples collected ≥ 10 dpv/dpi were considered antibody positive; 3) oral fluid samples collected ≥ 14 dpv/dpi were considered antibody positive. The nonparametric DeLong method⁸ was used to estimate the 95% CIs for the area under the curve (AUC) and to compare serum and oral fluid AUCs. Typical methods for estimating CIs for proportions are based on binomial distribution and do not account for the correlated structure of longitudinal data (i.e., repeated observations from the same animals over time). Therefore, a method for deriving CIs for correlated, normally distributed data was utilized.⁹ In brief, the correlation of the data was considered by fitting normalized data into a linear mixed model.

$$Y_{ij}=\mu +{\gamma }_i+\tau s_{ij}+{\epsilon }_{ij}$$ (5)

In equation 5, Y_ij is the jth observation for the ith subject; µ is the overall mean for samples classified as PRV antibody negative; γ_i is the random effect of the ith subject; τ is the fixed effect indicating the mean difference between PRV antibody-negative and -positive groups; s_ij is the disease status of the jth observation for the ith subject; and ε_ij is the random error of the jth observation for the ith subject. The variances from model 5 were used to calculate the 95% CIs for ROC-derived diagnostic sensitivity and diagnostic specificity estimates. Logit transformation was used to prevent the estimated intervals from exceeding the range of probability (i.e., [0, 1]).

Results

Clinical observations

Animals in groups MLV-PRV, MLV, and NC remained clinically healthy throughout the observation period. Clinical signs were observed in 3 of 10 animals in group PRV (i.e., lethargy, ataxia, and tremors beginning 7 dpi), but the clinical signalment did not fulfill the IACUC-approved criteria for euthanasia and all animals were clinically normal by 14 dpi. Individual oral fluid samples were successfully collected (≥ 2 mL) from pigs in groups MLV-PRV, MLV, and NC in 1,182 of 1,230 attempts (96.1%). In group PRV, 246 of 270 oral fluid collections (91.1%) were successful throughout the study. Coincident with the appearance of clinical signs, oral fluid samples were collected from 7 of 10 pigs at 7 dpi, 8 of 10 pigs at 8–11 dpi, and either 9 or 10 pigs, thereafter.

PRV ELISA testing and performance

No PRV antibodies were detected in serum or oral fluid samples from negative control animals. In contrast, vaccination (group MLV) or inoculation (group PRV) produced detectable serum antibody responses by 10 dpv or 14 dpi, and vaccination followed by inoculation (group MLV-PRV) produced a clear anamnestic response at 7 dpi (Fig. 1). Similar, but lower, antibody responses were observed in oral fluid specimens across all treatment groups (Kruskal–Wallis test, p < 0.001). The serum ELISA ICC was estimated at 99.0% (95% CI: 98.3, 99.5) and the oral fluid ELISA ICC was 99.3% (95% CI: 98.9, 99.6; i.e., both assays had excellent repeatability). The relationship between paired serum and oral fluid antibody S/P ratios was shown to fit the cubic curvilinear regression model shown in equation 6, in which oral fluid and serum stand for the transformed (log₁₀) S/P ratios (adjusted R² = 0.6503, standard error = 0.4407).

Figure 1.

Serum and oral fluid pseudorabies virus (PRV) antibody responses (x̅ ± standard error) based on ELISA testing (ADV (S) iELISA kit; Idexx Laboratories). Day 7: intramuscular administration of modified live virus vaccine (Ingelvac Aujeszky MLV; Boehringer Ingelheim). Day 28: intranasal administration of PRV isolate 3CR Ossabaw (1 × 10^3.5 TCID₅₀ per pig). S/P = sample-to-positive ratio.

$$\mathrm{oral}\mathrm{fluid}=-\mathrm{1}.5658+1.\mathrm{5336}\left(\mathrm{serum}\right)+1.1924{\left(\mathrm{serum}\right)}^{\mathrm{2}}+0.2772{\left(\mathrm{serum}\right)}^3$$ (6)

The AUCs, and hence the performance, for serum (n = 350) and oral fluid (n = 1,630) PRV ELISAs was different for the 2 curves (DeLong method, p < 0.001; Fig. 2). Likewise, estimated diagnostic sensitivities and specificities differed between the 2 assays (Table 2). For cases in which all test results were classified either as positive or negative at a specific cutoff (i.e., 100%), 95% CIs were estimated using sensitivity or specificity as 99.99% because logit(100%) is undefinable.

Figure 2.

Receiver operating characteristic (ROC) curve analysis of the pseudorabies virus (PRV) indirect ELISA (ADV (S); Idexx Laboratories) results in swine serum and oral fluid samples. Serum samples collected ≤7 dpv/dpi were classified as PRV antibody negative; samples collected ≥10 dpv/dpi were considered positive. Oral samples collected ≤7 dpv/dpi were classified as PRV antibody negative; samples collected ≥14 dpv/dpi were considered positive.

Table 2.

Pseudorabies virus indirect ELISA^* aggregate diagnostic sensitivity (DSe; %) and specificity (DSp; %) by specimen and by cutoff.

Cutoff (S/P)	Serum		Oral fluid
	DSe	DSp	DSe	DSp
0.1	100.0 (0, 100)†	66.5 (57.3, 74.7)	38.6 (29.6, 48.6)	99.2 (98.6, 99.5)
0.2	99.2 (0, 100)	90.5 (85.7, 93.8)	33.1 (24.4, 43.1)	100.0 (100.0, 100.0)†
0.3	98.4 (6.2, 100)	96.4 (93.8, 97.9)	26.2 (17.7, 36.9)	100.0 (100.0, 100.0)†
0.4	98.4 (52.4, 100)	99.1 (98.2, 99.6)	21.4 (12.9, 33.3)	100.0 (99.8, 100.0)†
0.5	98.4 (81.8, 99.9)	99.1 (97.8, 99.6)	20.7 (11.0, 35.6)	100.0 (99.1, 100.0)†
0.6	97.7 (86.6, 99.6)	99.1 (97.0, 99.7)	17.9 (7.9, 35.9)	100.0 (88.9, 100.0)†

S/P = sample-to-positive ratio. Numbers in parentheses are 95% CIs. Point estimates of diagnostic sensitivities and specificities derived from ROC analyses.

^*ADV (S) iELISA kit, Idexx Laboratories.

^†95% CIs were estimated at sensitivity or specificity = 99.99% rather than 100.0% to avoid calculation error (logit(100%) = undefined).

Discussion

The PRV indirect oral fluid IgG ELISA that we used provided “excellent” repeatability (ICC = 99.3%).¹⁸ Our adapted serum assay detected PRV antibody at ≥14 dpi in oral fluid specimens from virus-inoculated pigs and revealed a strong anamnestic response in vaccinated or challenged animals. In contrast, vaccine alone induced a low concentration of antibody in oral fluid (7–21 dpv).

PRV antibody has been detected in oropharyngeal swabs using a 4-layer enzyme immunoassay.³³ We detected PRV IgG (10 dpv/dpi) in oropharyngeal swabs from vaccinated (n = 5) and inoculated (n = 7) pigs, whereas IgM (7 dpi) and IgA (10 dpi) were found only in inoculated pigs. Using procedures similar to those described herein, PRV antibodies were detected in oral fluids between 10 and 17 dpi following exposure to a wild-type PRV Thai isolate.⁴⁷ Thus, our data agree with these earlier reports.^28,33

Our results demonstrate the feasibility of detecting PRV antibodies in oral fluid specimens. This represents the first step toward the development of a DIVA-compatible oral fluid–based ELISA for use in PRV control and elimination programs.