Development and Validation of a Novel ELISA for the Detection of Neospora Caninum Antibodies in Bovine Sera

Neospora caninum is a world-wide distributed apicomplexan parasite, causative agent of bovine neosporosis, which is one of the major causes of reproductive losses in cattle affecting both dairy and beef industries. Several techniques are regularly used for the diagnosis of bovine neosporosis. Amongst them, serological techniques are generally preferred, being IFAT the traditionally accepted reference method for diagnosis (gold standard). However, ELISA has become the most eligible method for large-scale screening of specific antibodies against N. caninum in diagnostic laboratories around the world, providing results that can be easily standardized between laboratories, within a few hours at a relative low-cost. There is no local production of commercial ELISA tests nor in Argentina neither in South America, therefore they must be purchased abroad, turning unfeasible the routinely use of ELISA tests. Here we have developed and validated accordingly to OIE specifications a novel ELISA test based on soluble antigens from N. caninum tachyzoites to accurately determine the presence of specific antibodies anti- N. caninum antigens in bovine sera samples, enabling the rapid, objective and low-cost diagnosis of bovine neosporosis in Argentina; replacing IFAT and avoiding delays and high importation taxes.

Keywords: Neospora Caninum; ELISA; Bovine Sera Samples; IgG

List of abbreviations: N. caninum: Neospora Caninum; T. gondii: Toxoplasma Gondii; IHQ: Immunohistochemical Staining; IB: Immunoblotting; DAT: Direct Agglutination Test; IFAT: Indirect Fluorescent Antibody Test; ELISA: Enzyme-Linked Immunoabsorbent Assays; OIE: World Organisation for Animal Health; sNcAg: Soluble Fraction of a Whole Tachyzoite Lysate; HRP: Horseradish Peroxidase; CV: Coefficient of Variation; OD: Optical Density; SnD: Diagnostic Sensitivity ; SpD: Diagnostic Specificity; S/P: Sample/Positive Ratio; ROC: Receiver Operating Characteristic Analysis; κ: Cohen's Kappa Coefficient; TP: True Positive; TN: True Negative; FP: False Positive; FN: False Negative; CI: Confidence Interval; PPV: Positive Predicted Value; NPP: Negative Predicted Value

Neospora caninum is a world-wide distributed apicomplexan parasite. It has a heteroxene life cycle, in which dogs and canids in general are the definitive hosts [1]; and a wide range of domestic and wild animals (cattle, buffalos, sheeps, horses, dogs, cats, etc) are the intermediate hosts [2]. N. caninum is the causative agent of bovine neosporosis, which is considered one of the major causes of reproductive losses in cattle as it can lead to abortion, stillbirth, reduced milk production and weight gain and increased birth-to-conception intervals and replacement rates [3-6] induced by mid-gestational exposure to the parasite or recrudescence of latent infections [2,7]. Globally, the estimated median losses due to N. caninum induced abortions were valued in US $1,298.3 million per annum (range US $633.4 million to US $2,380.1 million), affecting both dairy and beef industries [8]. It is closely related to other coccidian parasites, particularly to Toxoplasma gondii [9-12]. Both species share many common features, but Neospora cannot infect humans and does not share the same wide host range as Toxoplasma [1,2,12,13]. It also shows a higher efficiency of vertical transmission in cattle [14,15]. Considering this, several studies have assessed the cross-reactivity between both species in serological tests, finding no or faint cross-reactions for most of the immunodominant antigens present in the whole tachyzoite antigen [16-18].

Several experimental vaccines against neosporosis have been evaluated in different animal models. However, none of them is currently marketed, and only control actions regarding to the management of the animals herds are used [19-21]. The culling of infected cattle from the herd has been proposed to be a quite efficacious control method, though it is very expensive [22]. Other options include the selective breeding from only seronegative cows to both beef and milk industries or the breeding of seropositive cows only for the beef industry, together with the culling of those cows that have actually aborted with a confirmed diagnostic of neosporosis; besides strengthening the biosecurity measures to avoid horizontal transmission. In addition to this, it has been reported that seropositivity to N. caninum increases the risk of abortion, with estimates for the increased risk often quoted to range between 1.7 and 7.4-fold [5,23], but decreasing over time with increasing parity [5]. In this context, the development of reliable diagnostic tools to accurately detect the spread and circulation of the parasite is paramount.

Several techniques are regularly used for the diagnosis of bovine neosporosis, including histology and immunohistochemical staining (IHQ), nested-PCR [24], immunoblotting (IB), direct agglutination test (DAT), direct fluorescent antibody test (IFAT) and different enzyme-linked immunoabsorbent assays (ELISAs) [25-27]. Serological assays are preferred, being IFAT generally considered as the “gold standard” test [28,29]. However, due to the fact that it is easier and less time-consuming to obtain rapid results from large cattle herds with ELISAs than with IFAT, this assay has become the most eligible method for high-throughput screening of specific antibodies against N. caninum in diagnostic laboratories around the world [3]. ELISAs can provide results within a few hours at relatively low-cost do not rely on cell cultures or live parasites and do not need expensive specialized equipment, thus can be performed in almost any laboratory. ELISAs are also useful for determining the route of N. caninum transmission as avidity tests enable to differentiate between acute and chronic infections [30-34]. Furthermore, discrete results obtained by this assay allow the operator to minimize the bias introduced by the visual subjective analysis of the data achieved by IFAT. This is particularly relevant when comparing results obtained from different laboratories and countries, as the lack of harmonization between protocols involved in the diagnosis of neosporosis may interfere with the accurate estimation of its prevalence and with the management strategies described above [35,36].

Despite different commercial ELISA tests are available in many countries [37], there is no local production of those kind of products in Argentina; therefore they must be purchased abroad [36]. This implies high costs and long waiting periods, turning unfeasible the routinely use of ELISA tests in the diagnosis of bovine neosporosis. The aim of this study was to develop and validate, a novel ELISA test based on soluble antigens from N. caninum tachyzoites to accurately determine the presence of specific antibodies in bovine serum samples, following the World Organization for Animal Health´s (OIE) guidelines [38]. This would enable the rapid, objective and low-cost diagnosis of bovine neosporosis in Argentina; replacing IFAT and avoiding delays and high importation taxes.

A total of 634 serum samples gently provided by Dr. Dadin P. Moore from the Animal Health Group (EEA INTA Balcarce) were involved in this study. The sample size, level of confidence and error range were determined following OIE recommendations. This theoretical number of samples from animals whose serostatus is known enables to determine the Diagnostic Sensitivity and Specificity (SnD and SpD, respectively) with an error range of 2% and a confidence level of 90% [38].

They correspond to two different panels. One of them consisted of 374 samples from 27 pregnant heifers from a herd with high prevalence of the disease at different time points; the other panel comprised 260 samples corresponding to 130 heifers at parturition and their calves before colostrum intake. Their serostatus was monitored by IFAT.

The presence of specific antibodies anti-N. caninum antigens was assessed by IFAT using a fluorescein isothiocyanate labeled affinity-purified rabbit anti-bovine IgG antibody (Sigma, St. Louis, USA), as it was previously described [39-41]. Briefly, N. caninum specific antibodies were measured in serum samples using dilutions starting at 1:25. Positive and negative control sera were used. Slides were examined with an epifluorescence microscope (Olympus Bx 51, Olympus Inc., Tokyo, Japan). Antibody titers were expressed as the reciprocal of the highest serum dilution that showed complete peripheral fluorescence of tachyzoites [42]. A titer greater than 1:25 was indicative for N. caninum infection [39].

Purification of the Antigen: The soluble fraction of a whole tachyzoite lysate (sNcAg), purified as previously described [9,43] was used as capture antigen. Briefly, tachyzoites (Nc1 strain) were partially purified using sephadex columns (Sephadex TM G-25 Medium, GE Healthcare, Sweden); disrupted by ultrasonic treatment and centrifuged. Soluble antigens (sNcAg) were recovered from the supernatant and quantified using a commercial protein assay method (Micro BCA Pierce, Rockford, USA).

Feasibility and Standardization: To standardize and optimize the ELISA protocol positive and negative control samples were generated by pooling sera samples from three positive or negative cattle, respectively, according to IFAT titers. Those controls were diluted 1:10 in stabilizing buffer and were used throughout the different steps of the validation process. Polystyrene microtiter plates (Grenier Bio One, Austria) were coated with different amounts of sNcAg per well (500, 750 and 1000 ng) diluted in coating buffer (0.05M carbonate-bicarbonate buffer, pH 9.6). Different dilutions of the sera (1:40; 1:60; 1:120) and the antibody anti-bovine IgG conjugated to horseradish peroxidase (HRP) (1:2500, 1:5000; Jackson ImmunoResearch, PA, USA) and two different presentations of the substrate (revealing solution) were evaluated. To reduce the background level, different blocking solutions based on casein and different washing conditions (in PBS-tween 0.05%) were also assessed. Plates were incubated with different blocking solutions for 60 and 90 minutes at 37 oC and for 16 hours at 4 oC. Several conditions related to the incubation of samples and conjugate were also assessed, ranging from 21 to 37 oC and from 10 to 60 minutes.

Accuracy Assessment: Repeatability, Reproducibility and Precision: The repeatability of the assay was evaluated by selecting a strong, mid and weak positive sample (IFAT titers 1:1600, 1:800 and 1:400, respectively) and a negative control (IFAT titer < 1:25). They were run by ELISA in triplicate in eight independent series within the same plate. The reproducibility was evaluated by the analysis of 20 randomly selected samples by two operators from different laboratories (Virology Institute and Experimental Agricultural Station Salta, Argentina). These samples were run 8 times by each operator in independent assays. A Man-Whitney test was developed for each sample to assess significant differences between operators. The precision of the positive and negative controls was determined running them 20 times in independent assays. For each determination Coefficient of Variation (CVs) were determined. Accuracy tests were approved when CVs between optical density (OD) values were 30% or lower (according to OIE specifications [38]).

Performance characteristics: To assess the SnD and SpD and to determine the optimal threshold value a positive and negative serum samples (according to IFAT titers, considering this technique as the gold standard) were run in different dilutions (1:40, 1:60 and 1:120). Results were normalized as sample/positive control ratios (S/P%). Frequency distribution curves and a receiver operating characteristic analysis (ROC) of positive and negative samples were constructed to determine the area under the curve and positive and negative predicted values. Test agreement, SnD and SpD were calculated using GraphPad Prism 7 software. The level of agreement between the different tests was expressed as Kappa values (κ). The agreement was considered low, moderate and high, i.e., 0.4 ≤ κ < 0.6; 0.6 ≤ κ ≤0.8 and 0.8 ≤ κ < 1, respectively. Mann-Whitney test was used to evaluate whether differences between tests were significant or not.

In order to assess whether this protocol could be applied to the quick high-yielded screening of bovine neosporosis, different pooled samples were generated. For this purpose, a strong, mid and weak positive sample (IFAT titers 1600, 800 and 400, respectively) was mixed with either 25, 50, 75 or 100 negative samples. Samples were run at different dilutions (1:40 and 1:120) and both the limit of detection and the performance characteristics were determined as described above.

A total of 634 bovine sera samples were analyzed by IFAT. This technique was considered as the gold standard. From those, 327 samples were classified as negative (titers lower than 1:25) and 307 as positive; with titers ranging from 1:25 to 1:12800.

In a first attempt to determine the feasibility of the technique, different concentrations of capture antigen (500, 750 and 1000 ng/well), different dilutions of the sera (1:40, 1:60, 1:120) and two different presentations of the revealing solution were evaluated following a standard protocol. To standardize the final conditions of use sera samples at the optimized dilution were incubated 30 or 40 minutes either at 37 oC or at room temperature. Different dilutions of the secondary antibody (peroxidase conjugated) were assessed (1:2500 or 1:5000), whether incubated 20 or 30 minutes at 37 oC or at room temperature. Optimal conditions finally stablished are summarized on Table 1. Briefly, plates are coated with 750 ng of sNcAg/well and incubated for 6h at 4 oC. After washing 3 times, 125 μl/well of a casein-based blocking solution is added and plates are incubated ON at 4 oC. Then samples are added (50μl/well) diluted 1:120 in blocking solution and incubated for 30 min at room temperature. Plates are washed 3 times and an anti-bovine IgG diluted 1:5000 in blocking solution is added and incubated for 20 min at room temperature. Plates are washed again and the substrate (50μl/well) is incorporated. Plates are incubated at room temperature in a dark place for 10 min. Then stop solution (H2SO4, 50μl/well) is added and plates are read immediately at 450nm. The obtained results are normalized calculating the S/P(%). Samples with S/P(%) greater than 40% are considered positive, and those with S/P(%) values smaller than 30% are negative. S/P(%) values between 30 and 40 correspond to doubtful samples that should be confirmed by running them again diluted 1:40 in blocking solution.

A strong (Pos+++), mid (Pos++) and weak (Pos+) positive sample and a negative control (Neg) were run by triplicate in eight independent series in order to determine the intra-plate variability (repeatability). Within each series (triplicate) of the same operator the coefficient of variation for each sample was less than 10% for the 3 out of 4 samples analyzed. Higher variation (10-14%) was observed in 4 series of the weak positive control (Table 2 and Figure 1). Variation lower than 10% was observed between the 8 series for each sample (Table 3), within the same plate.

To assess the reproducibility of the technique, 20 samples were run in 8 independent assays by two operators from different laboratories. The coefficient of variation intra-operator was determined for each sample using raw data (OD values). The mean variation for the operator one was 22.52% and for the operator two was 8.59%, both acceptable levels according to OIE [38] (<30%, Tables 4a and b).

A Mann-Whitney test was carried out to compare the OD values obtained for each sample from operator one (laboratory A) and operator two (laboratory B). No significant differences were observed for 10 out of 20 of the analyzed samples (p>0.05). All of them, however, exhibited acceptable CVs (<30%) (Figure 1).

To assess the precision of the technique the dispersion observed between different series of the negative and positive control samples in 20 independent assays was determined. Results obtained for each sample were very precise, as low level of dispersion was observed, particularly from series 4 to 20 (Figure 2).

Cut off values: The identity of the samples was unknown to the laboratory operators to avoid any testing bias. Relative frequency distribution curves (percentages) of positive (n=307) and negative (n=327) samples (final dilution 1:120) were constructed. Two histograms were built, and their intercept point was considered as the cut-off value (Figure 3). A S/P ratio of 40% was defined as the appropriate cut-off threshold, but as certain overlapping of both curves was observed, samples with S/P values between 30 and 40% were considered doubtful and were run again in a final dilution of 1:40. The cut off value of this confirmatory test was defined as 30% (Figure 4).