Detection of pseudorabies virus by duplex droplet digital PCR assay

Abstract

Aujeszky’s disease, caused by pseudorabies virus (PRV), has damaged the economy of the Chinese swine industry. A large number of PRV gene-deleted vaccines have been constructed based on deletion of the glycoprotein E (gE) gene combined with other virulence-related gene deletions, such as thymidine kinase (TK), whereas PRV wild-type strains contain an intact gE gene. We developed a sensitive duplex droplet digital PCR (ddPCR) assay to rapidly detect PRV wild-type isolates and gE gene–deleted viral vaccines. We compared this assay with a TaqMan real-time PCR (qPCR) using the same primers and probes. Both assays exhibited good linearity and repeatability; however, ddPCR maintained linearity at extremely low concentrations, whereas qPCR did not. Based on positive results for both gE and gB, the detection limit of ddPCR was found to be 4.75 copies/µL in contrast of 76 copies/µL for qPCR, showing that ddPCR provided a 16-fold improvement in sensitivity. In addition, no nonspecific amplification was shown in specificity testing, and the PRV wild-type was distinguished from a gE-deleted strain. The ddPCR was more sensitive when analyzing clinical serum samples. Thus, ddPCR may become an appropriate detection platform for PRV.

Introduction

Pseudorabies virus (PRV; species Suid alphaherpesvirus 1, genus Varicellovirus, subfamily Alphaherpesvirinae, family Herpesviridae) is the common herpesviral infectious agent of Aujeszky’s disease, and it can induce abortion, respiratory distress, neurologic disorders, and piglet death.¹¹ PRV is important economically to the Chinese swine industry. PRV consists of a double-stranded DNA genome of ~150 kbp, encompassed by a capsid, envelope, and tegument. Ten different glycoproteins (gB, gC, gD, gE, gH, gI, gK, gL, gM, gN) have been demonstrated to participate in replication, immunogenicity, and pathogenicity.^10,11 The viral replication–related glycoproteins, such as gB and gH, are essential for the replication of PRV. Virulence and immunogenicity-related glycoproteins, for instance gE, gI, and gC, are nonessential for replication in vitro; in turn, they contribute to the virulence of PRV.³ A large number of commercial live viral vaccines have been attenuated by the deletion of one or more genes that encode for nonessential proteins. Representative PRV commercial gE-deleted viral vaccines, such as the PRV strains Bartha and SA215,¹⁴ provide good protection against PRV infection. Despite these efforts, this pathogen is still challenging and damaging to the pig industry.

Conventional identification of PRV via viral isolation or detection of anti-PRV antibody by serologic tests is generally reliable, but the procedures can be time-consuming and demanding.^6,12 A multitude of molecular technologies and analysis methods, for instance TaqMan real-time PCR (qPCR), nanoPCR, recombinase polymerase amplification assays, and GenomeLab gene expression profiler,^2,4,9,17 are widely accepted for scientific research and pathogen detection.

In droplet digital PCR (ddPCR), the reaction mixture is separated into a multitude of separate reaction chambers called partitions.⁵ Results are obtained by counting the number of partitions in which the target gene is detected. By applying a Poisson correction analysis to the data on positive partitions, the quantitative mean number of target fragments can be obtained.¹⁹ The sensitivity of ddPCR is higher than that of qPCR, and this precision is also available in extremely low concentration samples without the requirement for standard curves.¹³ In addition, ddPCR can detect minority target sequences in a background of other normal sequences, and this ability has demonstrated advantages over other PCR detection methods.^1,15 In order to sensitively detect PRV and distinguish wild-type virus from gE-deleted strains, we developed a procedure for quantitative analysis performed by ddPCR.

Materials and methods

Strains, plasmids, and specimens

Based on previous research,^9,14 the strain Bartha K61 was chosen for the specificity experiment along with the following vaccine viruses: classical swine fever virus (CSFV), porcine reproductive and respiratory syndrome virus (PRRSV), Japanese encephalitis virus (JEV), porcine parvovirus (PPV), and porcine circovirus 2 (PCV-2; Table 1). To construct the recombined plasmid, the target sequences of gB and gE were inserted into the pMD19-T plasmid to generate pMD-gB and pMD-gE plasmids by Taq-amplified cloning. Subsequently, concentrations of 2 purified recombinant plasmids were obtained (NanoDrop 2000, Thermo Fisher Scientific, Waltham, MA) and stored at −70°C. Twenty-three clinical porcine specimens and 21 serum samples were collected from pig farms in the Sichuan Province of China and preserved by the Key Laboratory of Animal Disease and Human Health of Sichuan Province (Table 2). All procedures performed in our study involving animals had been approved by and carried out in accordance with guidelines of the Institutional Animal Care and Use Committee of the College of Veterinary Medicine, Sichuan Agricultural University. Oral consent was obtained from the pig farmers when collecting clinical specimens and serum samples. All of the clinical specimens were tested by qPCR⁷ as recommended in the OIE Standard (Manual of Diagnostic Tests and Vaccines for Terrestrial Animals 2016, chapter 2.1.2. Available from: https://goo.gl/keV5pG).

Table 1.

Strains of viruses and bacteria assayed in the current study.

Object	Strain	Source
Pseudorabies virus	Bartha-K61	Commercial vaccine
	Clinical isolate	Laboratory*
Classical swine fever virus	Cell line origin	Commercial vaccine
Porcine reproductive and respiratory syndrome virus	NVDC-JXAI	Commercial vaccine
Japanese encephalitis virus	SA14-14-2	Commercial vaccine
Porcine parvovirus	WH-1	Commercial vaccine
Porcine circovirus 2	GL	Commercial vaccine
pMD19-T (2,692-bp clone vector, Amp, Lac Z)		Takara (Dalian, China)
pMD-gB (containing a 96-bp fragment, Amp, Lac Z, 208.7 ng/µL)		Laboratory*
pMD-gE (containing a 90-bp fragment, Amp, Lac Z, 157.85 ng/µL)		Laboratory*

^*Key Laboratory of Animal Disease and Human Health of Sichuan Province, Sichuan Agricultural University, Chengdu, China.

Table 2.

Clinical tissue samples and their pseudorabies viral copy content as determined by real-time PCR.

Case	Copies/mg*	Specimen type	Collection site
1	3,724	Lung	Suining, Sichuan
2	937	Lymph node	Pujiang, Sichuan
3	928	Tonsil	Yaan, Sichuan
4	4,846	Lung	Suining, Sichuan
5	2,902	Lung	Mianyang, Sichuan
6	9,709	Lung	Suining, Sichuan
7	3,716	Spleen	Pujiang, Sichuan
8	380	Brain	Suining, Sichuan
9	5,063	Spleen	Chengdu, Sichuan
10	491	Lung	Yaan, Sichuan
11	3,691	Lung	Suining, Sichuan
12	6,388	Lung	Pujiang, Sichuan
13	260	Spleen	Suining, Sichuan
14	5,655	Lung	Chengdu, Sichuan
15	245	Tonsil	Suining, Sichuan
16	181	Brain	Mianyang, Sichuan
17	128	Tonsil	Pujiang, Sichuan
18	4,207	Lung	Suining, Sichuan
19	Neg	Lung	Suining, Sichuan
20	Neg	Lung	Suining, Sichuan
21	Neg	Brain	Suining, Sichuan
22	Neg	Tonsil	Suining, Sichuan
23	Neg	Lung	Mianyang, Sichuan

^*The copies/mg data were obtained by using quantification cycle values, formula of standard curve, and the volume ratio of extracted DNA and template DNA (30:1) to calculate the gene copies in every 100 mg of clinical samples, and the copies/mg data could be calculated as well. Neg = negative.

Primers and probes

The gB primers were designed to detect sequences of glycoprotein B (nucleotide positions 18385–18480, a 96-bp fragment in the gB gene, GenBank accession KU315430.1; Table 3). The gE primers were selected according to the sequence of glycoprotein E (nucleotide positions 123684–123773, a 90-bp fragment in the gE gene, GenBank accession KU315430.1).

Table 3.

Primers and probes used to detect pseudorabies virus.

Object	Sequence (5’-3’)	Length (bp) of amplicon
gB
F	TGAAGCGGTTCGTGATGG	96
R	CCCCGCACAAGTTCAAGG
Probe	HEX-CGCGTACGTGCTCCCGGACC-BHQ
gE
F	CTTCCACTCGCAGCTCTTCTC	90
R	GTRAAGTTCTCGCGCGAGT
Probe	FAM-TTCGACCTGATGCCGC-BHQ

Extraction of DNA and RNA

Samples (100 mg) of porcine lung, brain, liver, and spleen were placed in lysing matrix tubes containing 500 µL of PBS, and then homogenized (MP Fastprep-24, MP Biomedicals, Santa Ana, CA). After centrifugation at 8,000 × g for 5 min, the supernatant was collected and used immediately for DNA/RNA extraction with a magnetic viral DNA/RNA kit (Tiangen Biotech, Beijing, China). Serum samples and vaccines were extracted in the same way without the grinding step. Bacterial DNA was isolated using a magnetic genome kit (Tiangen Biotech), and all of the extracts were stored at −70°C for further analysis.

PCR assays

We performed 2 types of quantitative detection: a standard qPCR amplification assay (CFX 96, Bio-Rad Laboratories, Hercules, CA) and a ddPCR assay (QX 100 droplet digital PCR system, Bio-Rad). The same primers and probes were used in both assays (Table 3). As for qPCR, the procedure was optimized with regard to the annealing temperature and reagent concentrations. The reaction mixture was prepared with Bestar qPCR master mix (TaqMan Probe, DBI Bioscience, Ludwigshafen, Germany), the primer-probe set for gB (750/250 nM) and gE (900/250 nM), and 1 µL of DNA template in a final volume of 25 µL. For optimization of the annealing temperature, the initial denaturation step was 95°C for 5 min, followed by 40 cycles of 95°C for 15 s, and a gradient temperature of 50–60°C for 34 s. For ddPCR, the reaction mixture, prepared according to the manufacturer’s instructions, consisted of 10 µL of 2× ddPCR supermix (Bio-Rad), primer-probe sets at final concentrations of 900 nM and 250 nM for both gB and gE, and 1 µL of DNA template in a final volume of 20 µL. Subsequently, the reaction mixture was transferred into the sample wells of the droplet generation cartridge (Bio-Rad), and then 70 µL of droplet generation oil (Bio-Rad) was loaded into the bottom wells of the cartridge, which was then placed in a droplet generator. The water-in-oil droplets were transferred to a 96-well PCR plate (Bio-Rad), and the plate was heat-sealed with foil (PX1 PCR plate sealer, Bio-Rad). In order to optimize the annealing temperature, the initial denaturation step (95°C for 10 min) was followed by 40 cycles of 95°C for 30 s and a gradient temperature of 50–60°C for 1 min, then 1 cycle of 98°C for 10 min. When amplification was complete, the products were analyzed (QX100 droplet reader, Bio-Rad).

Analysis of TaqMan qPCR and ddPCR

Data analysis of the correlations and regressions of the standard curves from qPCR was performed by the CFX manager. To analyze the ddPCR data, the copy numbers of the initial templates were analyzed by Poisson statistics using QuantaSoft analysis software (Bio-Rad). Kappa statistics were applied to determine the agreement of clinical detection results between qPCR and ddPCR.

Standard curves and detection limits of qPCR and ddPCR

For the convenience of data collection and analysis, the template mixture consisted of the plasmids pMD-gB and pMD-gE, which were mixed and adjusted to equal concentrations, used as initial standard, and serially diluted for both of the assays. The experiments on standard curves and detection limits must be performed based on simultaneously detectable concentrations of gE and gB. To determine the standard curves and detection limits of PRV-gB and PRV-gE, serially diluted plasmids at concentrations of 10⁸–1 copy/µL were used in the qPCR assay. Given the limited generated droplet numbers, too many templates (>10⁵ copies/µL) could cause inaccurate quantitative results.⁵ Thus, 10-fold serially diluted plasmids (copy numbers 10⁴–1 copy/µL) and 2-fold serially diluted plasmids (copy numbers 76–2.3 copies/µL) were used for the construction of ddPCR standard curves and the determination of detection limits. In addition, 5 replicates of each dilution were performed in these 2 assays. With regard to the possibility of false-positive droplets, we monitored potential contamination carefully.

Specificity of primers and probes

The primers and probes were designed to detect glycoprotein B and glycoprotein E proteins of PRV by qPCR and ddPCR. In both assays, positive and negative DNA/complementary DNA extracted from viral and bacterial strains (Table 1) were amplified in an optimized PCR reaction condition to determine primer-probe specificity.

Robustness and reproducibility test

Serially diluted standards were used to determine the robustness and reproducibility of the qPCR and ddPCR assays. Thus, each template was tested in triplicate at each concentration to evaluate intra- and interassay reproducibility by statistical analysis of the standard deviation (SD) and the coefficient of variation (CV).

Clinical sample detection by qPCR and ddPCR assays

Clinical tissue samples of lung, brain, liver, and spleen collected from 23 pigs with suspected PRV infection were tested to evaluate detection precision and agreement of the 2 assays. The gE-deleted Bratha-K61 strain was previously identified as a gE-negative sample. Twenty-one PRV gE and gB antibody-positive sera that had been identified by using a commercial PRV gB ELISA kit (IDEXX, Westbrook, ME) and gE ELISA kit (Hipra, Barcelona, Spain) were also used to determine and compare the detection sensitivity of qPCR and ddPCR in low-abundance samples.

Results

Primer annealing temperature optimizing

For qPCR, the optimized annealing temperature was 58°C. For ddPCR, annealing temperature gradients from 50–60°C were performed to optimize the separation between positive and negative partitions. The distinction in signals between the fluorescent channels peaked when the annealing temperature was 58.3°C (lane E06, Fig. 1A, 1B), which was chosen as the optimized temperature for ddPCR.

Figure 1.

Optimization of annealing temperature for detection of pseudorabies virus. A. Lane A06: 55°C; lane B06: 55.5°C; lane C06: 56.6°C; lane D06: 58.3°C; lane E06: 60.1°C; lane F06: 61.7°C; lane G06: 62.6°C; lane H06: 63°C. Blue plots indicates target gene gE (FAM signal). B. Green plots indicate target gene gB (HEX signal).

Analysis of standard curves and detection limits

For the TaqMan qPCR assay, fluorescent signals of the FAM-labeled gE probe and HEX-labeled gB probe were analyzed by a CFX 96 system. The standard curves exhibited good linearity (Fig. 2A). The R² value of gE was 0.999 and gB was 1. In addition, the values of slope and PCR efficiency were in the appropriate range (Fig. 2A). For the standard curves of ddPCR, log-linear diagrams with the x-axis showing the log assumed quantity and the y-axis showing the log starting quantity were obtained from the results. In testing the 10-fold diluted standards of ddPCR, the standard curves exhibited good linearity with R² values of 0.998 (gB) and 0.999 (gE; Fig. 2B). Also, the 2-fold diluted standards were used to construct standard curves for which the R² values were 0.997 (gB) and 0.997 (gE; Fig. 2C). As shown above, for qPCR, template concentrations ranging from 7.6 × 10⁷ to 76 copies/µL fit within the linear region of the standard curve; a copy number between 7.6 × 10⁴ and 4.75 copies/µL conformed to this region for ddPCR. In the sensitivity tests, the lowest detectable concentrations were 76 copies/µL for qPCR and 4.75 copies/µL for ddPCR. Consequently, the detection limit of ddPCR was ~16-fold lower than that of qPCR, which indicated that ddPCR was considerably more sensitive for PRV detection with the primers and probes used in our study (Table 4).

Figure 2.

Standard curves for real-time (q)PCR and droplet digital (dd)PCR detection of pseudorabies virus. A. Standard curves of qPCR and ddPCR. In standard curves of TaqMan qPCR, the green line was generated from the signal of HEX-labeled probe for detection of the PRV gB gene. The blue line was obtained from the signal of FAM-labeled probe targeting the gE region of PRV. B. Standard curves of ddPCR with 10-fold serially diluted plasmids. C. Standard curves of ddPCR with 2-fold serially diluted plasmids.

Table 4.

Detection of pseudorabies virus by real-time (q)PCR and droplet digital (dd)PCR using serially diluted plasmids.

Assumed plasmid concentrations (copies/µL)	Mean Cq value of qPCR		Mean copies/µL and number of positive replicates (%)
	gE	gB	gE	gB
7.6 × 107	16.09	13.56	NT	NT
7.6 × 106	19.22	16.81	NT	NT
7.6 × 105	22.63	20.10	NT	NT
7.6 × 104	26.17	23.53	7,724/100	7,668/100
7.6 × 103	29.65	26.84	739/100	762/100
7.6 × 102	33.25	30.21	65/100	69/100
76	36.56	33.51	7.3/100	7.3/100
38	Neg	Neg	3.2/100	3.5/100
19	NT	NT	1.1/100	1.7/100
9.5	NT	NT	0.89/100	1.1/100
4.75	NT	NT	0.38/80	0.49/100

Cq = quantification cycle; Neg = negative; NT = not tested.

Analysis of specificity, robustness, and reproducibility

The PRV wild-type strain and gE-deleted strain were tested, along with other major swine pathogens. In the one-dimensional scatter plot (Fig. 3), the distinction between the wild-type and gE-deleted strains was visible (lanes D11 and E11), given that there was no FAM signal observed in lane D11 (Bratha-K61) in contrast to the positive FAM signal in lane E11 (wild-type). Other templates, consisting of CSFV, PRRSV, PPV, PCV-2, JEV, and the negative control, tested negative in both the FAM and HEX channels. This demonstrated that there was no nonspecific amplification in the genomes of the pathogens mentioned above when using the primers and probes designed for our study. CV values (Table 5) of intra-assay variation of 0.3–1.84% and interassay variation of 0.39–1.86% were used for evaluation of the robustness and reproducibility of ddPCR. For robustness, mean CV values were 0.66% for gE and 1.46% for gB. For reproducibility, mean CV values were 0.74% for gE and 1.48% for gB.

Figure 3.

Specificity analysis of droplet digital PCR detection of pseudorabies virus (PRV). A. Lane A11: negative control; lane B11: classical swine fever virus; lane C11: porcine reproductive and respiratory syndrome virus; lane D11: PRV Bartha K61; lane E11: PRV wild-type strain; lanes G11–H12: Japanese encephalitis virus, porcine parvovirus, porcine circovirus 2, and porcine genome DNA, respectively. Green plots indicates target gene gB (HEX signal). B. Blue plots indicate target gene gE (FAM signal).

Table 5.

Robustness and reproducibility analysis of droplet digital (dd)PCR.

Sample	Intra-assay variation (robustness)						Interassay variation (reproducibility)
	Mean copies/µL		SD		CV (%)		Mean copies/µL		SD		CV (%)
	gB	gE	gB	gE	gB	gE	gB	gE	gB	gE	gB	gE
1	6,002	4,680	110.3	14.1	1.84	0.3	5,892	4,654	77.8	18.38	1.86	0.39
2	752	526.5	11.3	3.5	1.50	0.67	732	522	14.1	3.18	1.52	0.6
3	67.5	48.65	0.71	0.49	1.05	1.02	67	47	0.35	0.6	1.06	1.24

CV = coefficient of variation; SD = standard deviation.

Detection of PRV in clinical samples by qPCR and ddPCR

Twenty-three clinical tissue samples from suspected PRV-infected pigs were tested using the 2 assays, and both assays correctly identified PRV DNA in 18 of 23 samples (Table 6); the remaining 5 samples were gE and gB negative by qPCR in contrast to 1 gB-positive sample found by ddPCR. Kappa statistics were used to measure agreement; the kappa value was 0.876 after recalculation based on 1 additional positive, which implies good agreement between the 2 assays. In addition, 21 low-abundance PRV serum samples were used to test the difference in the sensitivity of the 2 assays. In qPCR, there were 10 positive samples consisting of 8 gE and gB double-positive samples and 2 gB-positive samples; the total positive rate was 48% (Table 7). In contrast, 14 double-positive samples and 3 gB-positive samples were detected by ddPCR (i.e., 81% of the 21 samples; Table 7). Thus, ddPCR was found to be more sensitive than qPCR for the detection of low concentrations of PRV in serum samples.

Table 6.

Detection of clinical tissue samples of pseudorabies virus by real-time (q)PCR and droplet digital (dd)PCR.

Assay	Total	gE (+), gB(+)	gE (−), gB(−)	gE (−), gB (+)
qPCR	23	18	5	0
ddPCR	23	18	4	1

+ = positive; − = negative.

Table 7.

Detection of serum samples of pseudorabies virus by real-time (q)PCR and droplet digital (dd)PCR.

Assay	Total	gE (+), gB(+)	gE (−), gB(−)	gE (−), gB (+)
qPCR	21	8	11	2
ddPCR	21	14	4	3

+ = positive; − = negative.

Discussion

We used serially diluted plasmids to determine the sensitivities of qPCR and ddPCR. The ddPCR was more sensitive. In plasmid tests, a concentration of 4.75 copies/µL was detectable by ddPCR, whereas the detection limit was 76 copies/µL for qPCR. These values indicate a 16-fold sensitivity difference. Furthermore, the ddPCR standard curves obtained from low-concentration plasmids that were 2-fold serially diluted (4.75–76 copies/µL) also exhibited good linear relation. The experiments using clinical serum samples that were identified by ELISA generally verify the superior sensitivity of ddPCR, with an increase in the positive rate from 48% to 81%. This demonstrated that ddPCR quantitative detection conformed to the linear region, even when target sequence concentrations varied (in certain ranges) by 1–6 orders of magnitude. However, in spite of the advantages indicated above, ddPCR is sometimes less versatile than qPCR and may provide nonlinear results when the initial concentration is too high because of multiple partitions generated in the experiment. According to the manufacturer’s instruction, the template concentration for ddPCR is limited in 1:1 × 10⁵ copies in a 20-µL reaction mixture. Therefore, the viral loads of clinical samples used in our study were limited to this range. The qPCR method, as a companion detection method, could be used when the specimens have very high viral loads (>10⁵ copies in the added template), whereas ddPCR is suitable for the detection of low viral load PRV samples (<76 copies/µL in the added template according to the sensitivity test). However, ddPCR is relatively expensive for routine diagnostic work, which may limit its usefulness.

In related molecular amplification studies, limits of detection were reported (qPCR—gE: 1.63 × 10⁰ copies, gB: 5.75 × 10¹ copies,¹⁸ both gE and gB: 1 × 10¹ copies¹⁶; nanoPCR—both gE and gB: 1 × 10¹ copies⁸). In our study, the detection limit (4.75 × 10⁰ copies) by ddPCR for both genes was lower overall than the published results. When PCR is used to distinguish PRV wild-type and vaccine strains, the presence of both gE and gB must be tested. With low viral load samples, ddPCR with equally low sensitivities to both genes might be more suitable for the differentiation of PRV wild-type and vaccine strains.

There was a degenerate base at the third position of the gE reverse primer. According to NCBI BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi), A-G nucleotide variations are observed at this position in different reference strains, especially in emerging wild-type strains. Therefore, this strategy could ensure the detection of positive samples and prevent missed detection, in turn making the results less clustered in a one-dimensional scatter plot.