Development and evaluation of a one-step multiplex real-time TaqMan® RT-qPCR assay for the detection and genotyping of equine G3 and G14 rotaviruses in fecal samples

Background Equine rotavirus A (ERVA) is the leading cause of diarrhea in neonatal foals and has a negative impact on equine breeding enterprises worldwide. Among ERVA strains infecting foals, the genotypes G3P[12] and G14P[12] are the most prevalent, while infections by strains with other genomic arrangements are infrequent. The identification of circulating strains of ERVA is critical for diagnostic and surveillance purposes, as well as to understand their molecular epidemiology. Current genotyping methods available for ERVA and rotaviruses affecting other animal species rely on Sanger sequencing and are significantly time-consuming, costly and labor intensive. Here, we developed the first one-step multiplex TaqMan® real-time reverse transcription polymerase chain reaction (RT-qPCR) assay targeting the NSP3 and VP7 genes of ERVA G3 and G14 genotypes for the rapid detection and G-typing directly from fecal specimens. Methods A one-step multiplex TaqMan® RT-qPCR assay targeting the NSP3 and VP7 genes of ERVA G3 and G14 genotypes was designed. The analytical sensitivity was assessed using serial dilutions of in vitro transcribed RNA containing the target sequences while the analytical specificity was determined using RNA and DNA derived from a panel of group A rotaviruses along with other equine viruses and bacteria. The clinical performance of this multiplex assay was evaluated using a panel of 177 fecal samples and compared to a VP7-specific standard RT-PCR assay and Sanger sequencing. Limits of detection (LOD), sensitivity, specificity, and agreement were determined. Results The multiplex G3 and G14 VP7 assays demonstrated high specificity and efficiency, with perfect linearity. A 100-fold difference in their analytical sensitivity was observed when compared to the singleplex assays; however, this difference did not have an impact on the clinical performance. Clinical performance of the multiplex RT-qPCR assay demonstrated that this assay had a high sensitivity/specificity for every target (100% for NSP3, > 90% for G3 VP7 and > 99% for G14 VP7, respectively) and high overall agreement (> 98%) compared to conventional RT-PCR and sequencing. Conclusions This new multiplex RT-qPCR assay constitutes a useful, very reliable tool that could significantly aid in the rapid detection and G-typing of ERVA strains circulating in the field.

Group A rotaviruses are transmitted through the fecal-oral route and infection in young foals is associated with life-threatening watery diarrhea induced by a combination of malabsorptive, osmotic and secretory mechanisms [1,20]. Control of ERVA infection in young foals is achieved by the routine vaccination of pregnant mares with an inactivated vaccine and strict husbandry/hygienic practices to reduce the viral burden in the environment [1,7,[21][22][23]. ERVA vaccines have been shown to aid in the reduction of the incidence and severity of diarrhea and also in the intensity and duration of viral shedding, however they do not guarantee full protection [1,21,22]. In addition, previous studies have shown that there is significant antigenic variation among ERVA genotypes, which leads to emergence of viruses that are not neutralized by antibodies elicited by the current vaccines [24][25][26][27][28][29]. Moreover, temporal and spatial variations in the prevalence and distribution of ERVA genotypes has been previously reported [2,29,30]. Therefore, it is important to perform genotypic characterization of ERVA strains in order to understand the molecular epidemiology of ERVA, identify novel viral reassortants and potential interspecies transmission, and assess vaccine performance in the field. Currently, sequencing of VP7, VP4 and other genome segments are required for genotyping circulating rotavirus strains. Conventional sequencing methodologies are generally labor intensive, low throughput and costly. Real-time reverse transcription quantitative polymerase chain reaction (RT-qPCR) assays, particularly TaqMan ® assays, offer a wide spectrum of advantages compared to conventional RT-PCR and sequencing. Some of these advantages include high throughput sample processing, increased sensitivity and specificity, faster turnaround time, and ability to multiplex. Even though several singleplex and multiplex RT-qPCR assays have been developed for the genotyping of human RVA genotypes [31][32][33][34][35], none have been developed for the genotyping of animal rotaviruses thus far, including ERVA. Here, we developed and evaluated the performance of a one-step multiplex RT-qPCR assay that allows the rapid detection of ERVA and the genotyping of the most frequent G-types affecting horses (G3 and G14) in fecal specimens. Overall, the one-step multiplex RT-qPCR assay developed in this study can simultaneously detect and genotype G3 and G14 ERVA strains with a performance equivalent to that of conventional VP7-specific RT-PCR and Sanger sequencing.

Fecal samples
A total of 177 fecal samples from diarrheic foals were used in this study. Among these, 112 fecal samples were collected from farms in central Kentucky [29] while 65 were from outbreaks of diarrhea that occurred in Argentina between 2009 and 2014 [29,30]. Ten percent fecal suspensions in serum-free EMEM were prepared, centrifuged at 2500 X g for 15 min at 4°C, then filtered through a 0.45 μm syringe filter. Aliquots of fecal suspensions were stored at − 80°C.

Nucleic acid isolation
Nucleic acid isolation was performed using the taco™ mini nucleic acid extraction system (GeneReach USA, Lexington, MA, USA) as previously described [37]. Two hundred microliters of 10% fecal suspension or tissue culture supernatant was used as sample input and elution was performed with 200 μl of elution buffer and stored at − 80°C for future use.

RT-PCR amplification of ERVA VP7 gene (segment 9)
We established a VP7-specific (gene segment 9) standard RT-PCR assay using the Qiagen One-Step RT-PCR kit (Qiagen, Valencia, CA, USA) as previously described [38]. This assay was used as the gold-standard method for ERVA detection in fecal specimens [2,39]. Briefly, a 25 μl reaction mixture was composed of 5 μl 5X One-Step RT-PCR Buffer, 1 μl dNTP Mix, 1 μl of VP7-specific forward and reverse primers (Table 2, 20 μM, final concentration 0.8 μM), 1 μl of One-Step RT-PCR Enzyme Mix, 11 μl of RNase-free water and 5 μl of template previously subjected to a denaturing step at 95°C for 5 min. The cycling conditions included a reverse transcription step (50°C for 30 min) followed by a PCR activation step at 95°C for 15 min; 35 cycles of denaturation (94 Table 1 A panel of viruses and bacteria associated with diarrhea in horses, cattle and simians was used to assess the specificity of the singleplex and multiplex RT-qPCR assays for detection and genotyping of ERVA °C for 1 min), annealing (47°C for 1 min) and extension (72°C for 2 min); and a final extension at 72°C for 2 min. PCR amplification products yielded a 1062 bp band following electrophoretic separation in a 1% agarose gel.

Sequencing of ERVA VP7 gene for G-typing
Sequencing of the full-length VP7 gene (genome segment 9) was performed using a high fidelity One-Step RT-PCR kit (Qiagen One-Step Ahead RT-PCR kit) and the forward and reverse primers RVAVP7-Gra-5 and RVAVP7-Gra-3 (Table 2) as previously described [29]. Briefly, a 25 μl reaction mixture was composed of 10 μl 2.5X One-Step Ahead RT-PCR Master Mix, 1 μl of VP7-specific forward and reverse primers (20 μM, final concentration 0.8 μM), 1 μl of 25X One-Step Ahead RT-Mix, 7 μl of RNase-free water and 5 μl of template previously subjected to a denaturing step at 95°C for 5 min. The cycling conditions included a reverse transcription step (45°C for 15 min) followed by a PCR activation step at 95°C for 5 min; 40 cycles of denaturation (95°C for 15 s), annealing (47°C for 15 s) and extension (68°C for 2 min); and a final extension at 68°C for 5 min. PCR products (1062 bp) were gel-purified using the QIAquick ® Gel Extraction kit (Qiagen) according to the manufacturer's recommendations. DNA was submitted for Sanger sequencing to a commercial company (Eurofins Genomics LLC, Louisville, KY, USA). Both DNA strands of VP7 amplicons were sequenced using a panel of primers specified in Table 2. Sequence analysis was performed using Geneious R7 (Biomatters Inc., Newark, NJ, USA). G-types were identified using an automated genotyping tool for RVA (RotaC 2.0, http://rotac.regatools.be/) [40].
ERVA-specific singleplex TaqMan ® real-time RT-PCR assays targeting G3 VP7, G14 VP7 and NSP3 genes Primers and probes specific for ERVA G3 VP7 and G14 VP7 were designed as described above (Table 3). The reaction was set up using the QuantiTect™ Probe RT-PCR kit (Qiagen) following the manufacturer's recommendations. Briefly, the 25 μl reaction contained 12.5 μl of 2X Quanti-Tect™ Probe RT-PCR Master Mix with ROX, 0.25 μl QuantiTect™ RT Mix, 200 nM TaqMan ® fluorogenic probe, 500 nM of each primer, and 5 μl of template RNA (previously subjected to a denaturing step at 95°C for 5 min). Reverse transcription and amplification were carried out in an ABI 7500 Fast Real-time PCR System (Applied Biosystems ® , Life Technologies, Grand Island, NY). The program included 30 min at 50°C (reverse transcription step), 15 min at 95°C (PCR initial activation step), followed by 45 cycles at 94°C for 15 s (denaturation) and 60°C for 1 min (combined annealing/extension). The NSP3-specific (gene segment 7; pan-rotavirus A) assay was established in the laboratory as previously described (

Statistical analysis
Standard curves were performed using IVT RNA (10 7 to 0.1 IVT RNA copies/μl). Coefficients of determination (R 2 ) were used to assess curve fitness. PCR amplification efficiencies (%) were calculated after regression analysis using the following formula: E ¼ ½10 − 1 slope −1 Â 100. Limit of detection with 95% confidence (LOD 95% ) was determined by statistical probit analysis (a non-linear regression model) using the commercial software SPSS 14.0 (SPSS Inc., Chicago, IL, USA) for all assays with 9 replicates per dilution ranging from 10 5 to 1 IVT RNA copies/μl. Cycle threshold (Ct) cut-off values were determined as the average Ct + 3 standard deviations of nine replicates of the endpoint dilution [42]. Precision (within-run and between-run imprecision) of the ERVA multiplex RT-qPCR assay was determined by performing 9 replicates of IVT RNA containing 10 5 , 10 4 and 10 3 RNA copies/μl on the same run (within-run imprecision) or three replicates of IVT RNA containing 10 5 , 10 4 and 10 3 RNA copies/μl tested on two different operational days. The coefficient of variation (%) was determined for each target concentration (G3 VP7, G14 VP7, and NSP3). The performance of the ERVA multiplex RT-qPCR assay was evaluated in fecal specimens and compared to the VP7-specific RT-PCR and G-typing by Sanger sequencing. Contingency tables (2 × 2) were generated to determine the sensitivity, specificity and agreement (kappa statistic) of each target within the multiplex RT-qPCR assay.

Analysis of fecal samples by VP7-specific RT-PCR and sequencing for determination of G-types
A total of 177 fecal samples were included in the study, from which 92 samples were confirmed negative for ERVA, while 85 were positive as determined by VP7-specific standard RT-PCR [29,30]. From the 85 ERVA-positive samples, 58 were collected in Argentina and 27 were collected from the USA (Kentucky). Among these, 41 were confirmed as G3 genotype while 44 were confirmed as G14 genotype by sequencing of the VP7 gene. Extensive genetic and phylogenetic analysis of these samples was recently published in a separate article [29].

Analytical sensitivity of ERVA-specific multiplex RT-qPCR assay
Standard curves generated for the three targets (G3 VP7, G14 VP7 and NSP3) under multiplex conditions also demonstrated perfect linearity (R 2 > 0.99, Table 4 and Fig.  1). However, while the amplification efficiencies for the G3 VP7 and NSP3 targets were ± 10% of that determined under singleplex conditions (96 and 100%, respectively), a lower amplification efficiency was determined for the G14 VP7 target when multiplexing (88%). Detection rates (100%) for the multiplex RT-qPCR assay are shown in Table 4. While the 100% detection rate limit for the NSP3 assay was equal between the singleplex and multiplex formats, a 100-fold difference was observed for the G3 VP7 and G14 VP7 assays when these were multiplexed (Table  4). In comparison to the singleplex format, the LOD 95% were higher (716, 215 and 42 copies/μl of IVT RNA for the G3 VP7, G14 VP7 and NSP3 targets, respectively). Ct cut-off points were determined at 32, 34 and 34, respectively.
Analytical specificity of ERVA-specific singleplex and multiplex RT-qPCR assays To evaluate the analytical specificity of the singleplex and multiplex RT-qPCR assays, a panel of rotavirus strains along with other viruses and bacteria associated with diarrhea in horses was used ( Table 1). The ERVA-specific G3 and G14 VP7 primer-probe combinations were exclusively specific for the respective ERVA genotype, did not cross-react between each other, did not amplify other rotavirus genotypes from other species and, interestingly, did not amplify the simian SA11 strain (G3P2). The NSP3-specific primer-probe combination in both singleplex and multiplex format was specific for RVA and amplified the reference G3 and G14 strains of ERVA as well as bovine and simian rotavirus strains as previously reported [34]. None of the assays (G3 VP7, G14  Precision assessment of the ERVA-specific multiplex RT-qPCR assay To evaluate the precision of the multiplex RT-qPCR assay, within-run and between-run imprecision was determined as recommended [42]. In all cases, the coefficient of variation was less than 3%, indicating that the multiplex assay has a high repeatability (within-run) and reproducibility (between-run) within the range of detection (Table 5).
Clinical performance of the ERVA-specific multiplex RT-qPCR assay targeting G3 VP7, G14 VP7 and NSP3 The clinical performance of the ERVA-specific multiplex RT-qPCR assay was evaluated in a total of 177 fecal samples. The NSP3 (pan-RVA) assay was able to successfully detect ERVA in all positive samples (85/85) while no non-specific amplifications were observed in negative samples (n = 92; Table 6a). Therefore, the assay presented 100% sensitivity and specificity when compared to the VP7-specific standard RT-PCR assay, along with perfect agreement (kappa = 1). In the case of the G3 VP7 assay, the assay was able to correctly genotype 38/41 ERVA G3 samples while non-specific amplifications were not observed in G3-negative samples (n = 136, Table 6b). Only three ERVA G3 positive samples were unable to be genotyped by the multiplex assay, however these were correctly genotyped by the G3-specific singleplex RT-qPCR assay. Overall, the G3 VP7 assay presented a 92.7% sensitivity and 100% specificity when compared to the VP7-specific standard RT-PCR assay, and a high agreement (98.31% [kappa = 0.951]). Finally, the G14 VP7 assay was able to correctly identify 44/44 ERVA G14-positive samples and did not amplify 132/133 ERVA G14 negative samples (Table 6c). Consequently, the G14 VP7 assay presented a 100% sensitivity and 99.2% specificity when compared to the VP7-specific standard RT-PCR assay. The agreement between assays was high (99.44% [kappa = 0.985]). Regarding the presumed false positive sample, although this sample was determined to be an ERVA G3P [12] by Sanger sequencing, it yielded a concurrent positive amplification by the G3 and G14-specific RT-qPCR assays in both their singleplex and multiplex formats, suggesting a possible co-infection with both genotypes of ERVA.

Discussion
Group A rotaviruses are a primary cause of diarrhea in children and animal species, including horses [1-6, 43, 44]. Even though seven G-types and six P-types of ERVA have been identified in horses, the G3P [12] and G14P [12] constitute the most epidemiologically relevant genotypes [1,2,[17][18][19]. Spatial as well as temporal fluctuations between these predominant G-types (G3 and G14) of ERVA circulating in equine populations have been reported around the world [2,30]. Interestingly, the emergent pattern of G14 ERVA and the temporal shift in the prevalent genotype has been observed in association with the implementation of widespread vaccination programs in Argentina, Japan and Ireland [2,30,45,46], which rely on the use of inactivated vaccines containing only the H2 or HO-5 (G3P [12]) strains of ERVA. The difficulties faced to date in establishing cell-culture adapted G14P [12] or other strains of ERVA has precluded their inclusion into vaccine formulations. However, we have recently isolated and cell-culture adapted, three G14P [12] ERVA strains with the potential  to be used as reference G14P [12] strains to study the molecular biology of this genotype and perform vaccine efficacy studies following heterologous challenge in the future [29].
In light of the antigenic differences between ERVA genotypes, their spatial and temporal distribution and their impact on vaccine efficacy, molecular surveillance and genotypification of circulating strains is critical. Since genomic arrangements of ERVA other than G3P [12] and G14P [12] are rare and the outer capsid protein VP7 contains the major neutralizing epitopes, we developed a one-step multiplex TaqMan ® real-time RT-PCR for the rapid detection and G-typing of the most prevalent genotypes of ERVA (G3 and G14) in fecal specimens. Compared to the conventional methods for ERVA genotyping (RT-PCR and Sanger sequencing), the multiplex RT-qPCR assay has a significantly faster turnaround time, is high-throughput, less labor-intensive and exhibits a high sensitivity, specificity and agreement as demonstrated in this study. While multiplexing did not have an impact on the detection limit of the NSP3-target, the G3 and G14 targets demonstrated a 100-fold difference in their analytical sensitivity under multiplex conditions. However, this difference in analytical sensitivity did not have a significant impact on their clinical performance on fecal specimens and only three G3 ERVA-positive samples were unable to be typed by the multiplex RT-qPCR assay (false negatives). Interestingly, these samples were correctly G-typed when the G3 VP7-specific assay was performed under singleplex conditions. Such differences are probably due to a combination of low target nucleic acid in these fecal specimens along with the 100-fold higher analytical sensitivity of the singleplex compared to the multiplex assay. Despite the low number of false negative samples (n = 3), all three targets (G3 VP7, G14 VP7 and NSP3) showed a high sensitivity and specificity (> 90%) along with a high level of agreement (> 98%) in the clinical specimens tested under multiplex conditions.
Noteworthy, a single sample, G-typed as G3 by means of conventional methods (RT-PCR and Sanger sequencing), exhibited specific amplification of both the G3 VP7 and G14 VP7 targets simultaneously under singleplex and multiplex conditions. Although confirmation would require RT-PCR using genotype-specific primers or next generation sequencing, due to the fact that both G3 and G14 ERVA strains were identified to be co-circulating in the same farm during the same time period, these results suggest that this dual-positive fecal specimen most likely derived from a foal that was co-infected with both G3 and G14 ERVA strains. Consequently, this may indicate that the multiplex RT-qPCR assay developed can be advantageous for the diagnosis of co-infections with G3 and G14 strains of ERVA that are currently challenging to identify. Further assessment using spiked specimens is required in order to analyze this multiplex RT-qPCR assay's capability to identify co-infected animals. Due to the lack of reference strains and uncommon occurrence of other ERVA G-types, these were not included in this study. Therefore, it is imperative to perform Sanger sequencing on those samples that test positive for ERVA by amplification of NSP3 but are not genotyped as G3 or G14 by the current assay. In this regard, the genotyping assay developed here will facilitate rapid genotyping of circulating strains and identify rare G-types that can then be incorporated into this assay depending on their epidemiological relevance.

Conclusions
In conclusion, the study presented herein describes the development and evaluation of a one-step multiplex TaqMan ® RT-qPCR assay for the detection and genotyping of the most frequent G-types of ERVA infecting horses. This assay demonstrated to have a high sensitivity, specificity and agreement compared to conventional RT-PCR and sequencing, providing rapid and reliable G-typing of ERVA strains. Therefore, this assay is highly suitable for routine diagnostics as well as to aid current surveillance programs of ERVA by rapidly characterizing circulating strains. Finally, the number of specific targets included in this assay can be updated and expanded as other genomic arrangements of ERVA emerge and become prevalent in equine populations.