Assessment of Campylobacter fetus subsp. venerealis molecular diagnosis using clinical samples of bulls

Background Campylobacter fetus subsp. venerealis (Cfv) is the pathogen responsible for Bovine Genital Campylobacteriosis (BGC), a venereal disease of cattle associated with impaired reproductive performance. Although several PCR assays were developed to identify this pathogen, most of them are still poorly evaluated in clinical samples. This study evaluated real-time PCR assays for Cfv detection in preputial samples of bulls (n = 308). Results The detection at the subspecies level (Cfv) compared four assays: two targeting ISCfe1 and two targeting parA gene. The detection at the species level (C. fetus) considered an assay targeting the nahE gene and a commercial kit for C. fetus identification. At the subspecies level, assays directed either to different targets (parA and ISCfe1), or to the same target (ISCfe1 or parA), showed a high percentage of disagreeing results. All samples positive at the subspecies level (n = 169) were negative in C. fetus detection assays, which strongly suggests the horizontal gene transfer of ISCfe1 and parA to other bacterial species. This was confirmed by microbiological isolation of three Campylobacter portucalensis strains responsible for false positive results. Sequences with a high level of identity with ISCfe1 and parA gene of Cfv were identified in C. portucalensis genome. Conclusions Overall, this study reveals that PCR assays solely directed to a subspecies target originate a high rate of false positive results, due to the presence of parA and ISCfe1 homologous sequences in other bacterial species, namely of the genus Campylobacter. Although the specificity of these methods may be higher if applied to bulls from herds with clinical features of BGC or in other geographical regions, current PCR diagnosis should couple subspecies and species targets, and further research must be envisaged to identify Cfv specific molecular targets. Supplementary Information The online version contains supplementary material available at 10.1186/s12917-020-02634-7.


Background
Bovine Genital Campylobacteriosis (BGC) is a venereal disease of cattle, caused by the bacterial pathogen Campylobacter fetus subsp. venerealis (Cfv), responsible for reproductive failure and significant economic losses, mainly in beef herds where natural breeding prevails [1,2]. Bulls are asymptomatic carriers, harbouring Cfv in the preputial crypts, and infect females during breeding [3]. By contrast, Cfv infection in females induces endometritis, early embryonic death and abortion [2]. An accurate diagnosis of BGC is essential for implementation of disease control programs and international trade of bulls and semen [4]. However, BGC diagnosis is hindered by the two cattle-associated Campylobacter fetus subspecies, C. fetus subsp. fetus (Cff) and Cfv, with distinct niche preferences but with similar genotypic and phenotypic characteristics [5,6]. Although Cff inhabits cattle intestinal tract, it can be occasionally recovered from bovine preputial samples and it is responsible for sporadic cases of abortion [1,7]. The OIE recommended method for BGC diagnosis is microbiological isolation of C. fetus, followed by the 1% glycine tolerance test, for which Cfv is intolerant and Cff is tolerant [2]. Due to the fastidious growth of Cfv and potential overgrowth of other microorganisms, microbiologic techniques are laborious, time-consuming and associated to poor sensitivity [1]. Additionally, Cfv can acquire glycine tolerance by mutation or transduction mechanisms being misidentified as Cff [8]. Therefore, molecular biology techniques have evolved as attractive tools for diagnosis of BGC. At the species level, real-time PCR assays to detect the nahE gene of C. fetus were developed and validated, showing high specificity and sensitivity [9,10]. However, the identification at the subspecies level is hampered by the high genetic similarity of Cfv and Cff [11]. Nevertheless, several PCR assays have been described to identify Cfv [12][13][14][15]. The most used molecular targets for Cfv identification are the parA gene and the ISCfe1 insertion element [9,10,13,14]. Assays based on ISCfe1 detection are described as more sensitive than parA-based assays to identify Cfv isolates [9,10]. Although real-time PCR assays targeting the parA gene were evaluated in clinical samples [13,[16][17][18], available data is insufficient to allow its routine use for diagnosis of BGC. Assays based on ISCfe1 detection were not yet assessed in clinical samples despite showing promising results on Cfv isolates [9,10]. Therefore, the diagnostic value of PCR assays for bull clinical samples has not been elucidated. This is a relevant clinical issue because bull testing is the most appropriate prophylactic measure for BGC control. This study evaluated different real-time PCR assays for Cfv identification on bull preputial samples in order to assess their suitability for BGC diagnosis.

Performance of the real-time PCR assays
The performance of the real-time PCR assays was evaluated using DNA from preputial samples spiked with Cfv genome copies. All assays were able to detect 10 to 10 5 Cfv genome copies within 40 cycles of amplification and considering a positivity threshold set to Ct ≤ 35, were able to detect 100 Cfv genome copies. The efficiency (Table 1) was within the acceptable range of 90-110% [19] and a linear relationship between the cycle threshold (Ct) value and the log of copy number was evidenced by an r 2 ≥ 0.98. The intra-and inter-assay coefficients of variation were less than 5%, as shown in Table 1.
Detection of Cfv and C. fetus molecular targets in preputial samples Four different assays for Cfv identification were tested in clinical samples with unknown BGC sanitary status, targeting the ISCfe1 sequence (ISC-A and ISC-B assays) and the parA gene (parA-A and parA-B assays). The agreement between parA-A and ISC-A assays was evaluated in all the 308 bovine preputial samples. The parA-A assay gave a lower number of positive results (n = 78, 25.3%) than the ISC-A assay (n = 155, 50.3%) (P < 0.001). This originated a high percentage of disagreeing results (34.1%, Kappa = 0.32), mainly represented by ISCfe1 positive and parA negative samples ( Table 2).
The conservation of parA and ISCfe1 sequences was evaluated on a subset of 141 samples, comparing results obtained with assays targeting two different regions of each nucleotide sequence (ISC-A/ISC-B and parA-A/ parA-B assays). The parA-A and parA-B assays provided 28.4 and 7.8% positive results, respectively, resulting in a high percentage of disagreeing results (20.6%; Kappa = 0.35) (Additional file 1). Likewise, the ISC-A and ISC-B assays originated 58.9 and 17.0% positive results, respectively, also resulting in a high percentage of disagreeing results (41.8%; Kappa = 0.25) (Additional file 1). These differences observed between assays directed towards the same target are statistically significant (P < 0.01). The number and percentage of positive samples detected by each assay, or double combination is shown in Fig. 1, for all samples tested by the four assays (n = 141). Results obtained by different assays were highly inconsistent. ParA-B assay provided the lowest number of positive samples (n = 11), while ISC-A assay detected the highest number of positive samples (n = 83). In contrast, there were no samples tested simultaneously positive by ISC-B and parA detection assays.
The presence of genes from the Cfv-associated genomic island (GI), which may contain the parA-A gene, was evaluated by PCR. Results showed that genes of this GI (fic1, fic2, virB9, virB11) were detected in 95% (n = 18) of the parA-A positive samples (Table 3), and a significant association was found between the presence of the parA gene and C. fetus GI genes (φ = 0.873, P < 0.001).
To evaluate the specificity of the parA gene and ISCfe1 as Cfv diagnostic targets, samples tested positive for at least one of these targets were also tested using a real-time PCR assay to detect the C. fetus specific nahE gene and a commercial diagnostic kit (VetMAX C. fetus kit) approved for C. fetus detection in DNA isolated from clinical samples. These assays detect targets common to both C. fetus subspecies (Cff and Cfv). All samples (n = 169) that tested positive to Cfv specific targets (parA gene and/or ISCfe1) were negative in C. fetus detection assays.

Impairment of Cfv detection in clinical samples by Campylobacter portucalensis
To investigate the cause of false positive results, microbiological culture and isolation of Campylobacter spp. was performed in three samples positive to parA and/or ISCfe1 and negative in C. fetus detection assays. Three strains identified biochemically and by 16S rRNA gene sequencing as Campylobacter portucalensis [20] were isolated from these samples. These strains were confirmed to be responsible for false positive results in the above Cfv detection assays. Whole genome sequencing (WGS) data of the type strain C. portucalensis FMV-PI01 T (NCBI accession no: VWSJ00000000) allowed the identification of the parA gene (98.1% identity with sequence from Cfv strain WBT011/99) and of an insertion sequence highly similar to ISCfe1 (93.5% identity with sequence from Cfv strain zaf3). The WGS of FMV-PI01 also showed that parA gene is located in a GI, which is highly homologous to the Cfv-associated GI with 97.42% of identity in 80% of the sequence. The arrangement of parA, fic and T4SS genes in the GI is schematically represented in Fig. 2.
Two of the C. portucalensis strains harbouring Cfv molecular markers, tested positive for both ISCfe1 and parA gene by ISC-A and parA-A assays, and GIassociated genes, whereas one strain tested positive for ISC-A and negative for parA-A and GI-associated genes.

Discussion
The analytical performance of the assays used in this study was evaluated using preputial samples spiked with Cfv DNA. Assays directed to species and subspecies targets showed similar analytical sensitivities, detecting 10 1 to 10 2 genome copies of Cfv NCTC 10354. All assays provided reproducible results in the same experiment and between experiments, as evidenced by the coefficient of variation (CV) below 5%. Overall, the results showed a good analytical performance of real-time PCR assays and indicate their suitability for detection of Cfv in the bovine preputial sample matrix. A considerable disagreement was observed between results obtained with parA-A and ISC-A assays. This was also observed for assays targeting the same molecular target (ISC-A/ISC-B and parA-A/parA-B), which suggests divergences in the nucleotide sequences of parA and ISCfe1, specifically in primer and probe binding sites. Moreover, in a considerable number of samples only one of the molecular targets was detected. This finding evidences a sensitivity and/or specificity failure of one or both detection assays. Previous studies showed that the parA-based PCR assays have a lower sensitivity when compared with assays targeting ISCfe1 for the identification of Cfv isolates [9,10]. In fact, parA sequence variation may potentially hinder parA-based diagnostic methods [16,21]. On the other hand, at least two different ISCfe1 sequences were found in Cfv strains, sharing 98.7% sequence homology [10]. Therefore, inconsistent results obtained by assays directed towards the same target may result from divergences in the nucleotide sequences.
Samples positive for subspecies specific targets revealed negative in the C. fetus detection assays. These samples were expected to be positive for C. fetus detection, since the species-specific assays are based on targets common to both C. fetus subspecies (Cff and Cfv). The nahE detection assay showed an analytical sensitivity comparable to that obtained with the subspecies detection assays, indicating that the absence of amplification is not related with differences in the analytical sensitivity. Moreover, although all results obtained with the VetMAX C. fetus kit were negative, the amplification of an internal positive control allowed to validate results. These findings strongly suggest that subspecies targets were horizontally transferred to other bacterial species. Although the parA gene is still used for Cfv identification, the horizontal gene transfer of parA to C. hyointestinalis isolates and Cff strains was reported [22,23]. Also, parA detection assays identified putative positive Cfv bulls in herds without impairment of the reproductive performance [24] and in virgin bulls [17,25]. Indeed, the parA gene can be found in mobile genetic elements, namely in a GI almost exclusive of Cfv [26,27], in the chromosome and in extra-chromosomal plasmids [21,27]. On the other hand, genome sequencing data recently deposited in the NCBI database revealed that ISCfe1 can be found in plasmids (e.g. acc. no. CP043436.1), which may facilitate the horizontal transfer of this sequence to non-Cfv microorganisms. As a transposable element, ISCfe1 may spread in the genome, thus justifying the variable number of ISCfe1 copies in different Cfv strains [14].
Although the possibility of the above targets being transferred horizontally has already been described or suggested, there is no indication in the literature that assays based on these targets can lead to such a high rate of false positive results as those found in this study. In fact, these targets are still used as sole molecular targets for Cfv detection, as evidenced in recent studies [28][29][30]. The rate of false positive results may depend on the geographical region, since previous studies in Canada [18] or Brazil [28,29] reported lower rates of positive results with parA-based assays.
C. portucalensis, which is an inhabitant of the bull's prepuce [20], was identified as a cause of false positive results. The high identity between the sequences found in C. portucalensis and ISCfe1 and parA sequences of Cfv suggests the horizontal transfer of these molecular targets. Although C. portucalensis and Cfv are not phylogenetically close species [20], the sharing of the same niche by Cfv and C. portucalensis may have facilitated the horizontal gene transfer of parA and ISCfe1. The genomic island associated to Cfv, in which the parA gene is included, was found in the genome of C. portucalensis, supporting the results of parA positive samples, with amplification of GI genes (fic1, fic2, virB11, virB9) and without amplification of C. fetus specific molecular targets. One of the C. portucalensis strains is positive for ISCfe1, without amplification of GI genes or parA. This finding suggests an independent horizontal gene transfer of parA and ISCfe1 and may justify the high percentage of samples positive in ISC-A assay and negative in parA-A assay (n = 91, 29.5%). The high similarity between the sequences found in C. portucalensis and ISCfe1 and parA gene sequences of Cfv is enough to impair the accuracy of molecular diagnostic methods based on these targets' detection.
Overall, this study evidenced that ISCfe1 and parA gene-based assays are associated with considerable specificity failures and, consequently, these targets are unsuitable for Cfv identification when used solely. However, the detection of ISCfe1 or parA may be used as part of a diagnostic strategy, with validation of positive results to parA or ISCfe1 with previous or subsequent C. fetus detection assays. This strategy overcome the specificity failures associated to ISCfe1 and parA detection assays found in this study. Still, as described in previous studies [22,23,31], these molecular targets were sporadically found in Cff isolates, leading to false positive results even with the above diagnostic strategy. However, the specificity of these methods is expected to be higher if applied to bulls from farms with reproductive failure, where other causes were ruled out, as described previously [18].

Conclusion
The results of this study have major implications in the molecular diagnosis of BGC, invalidating the use of ISCfe1 and parA as sole targets for Cfv identification. Diagnosis of BGC based solely on these genomic targets originates false positive results, which may lead to the unnecessary treatment or culling of animals. The combined use of Cfv and C. fetus detection assays should be encouraged as part of the BGC diagnostic strategy, in order to improve the specificity of molecular diagnostic methods. This study also described, for the first time, the presence of the C. fetus GI in another Campylobacter species inhabitant of the bull's preputial mucosa, evidencing that this genomic element can be transferred to other Campylobacter species. A highly homologous sequence of ISCfe1, previously considered a promising Cfv molecular marker, and the parA gene, were also found in another Campylobacter species. This evidences the horizontal gene transfer of current diagnostic Cfv molecular targets to other Campylobacter species and prompts for the development of reliable molecular diagnostic tools for BGC diagnosis.

Samples and DNA extraction
This observational study considered samples collected by certified veterinarians, within a bull breeding soundness examination, for diagnostic purposes.The number of tested bulls (n = 308) and herds (n = 61) is a representative sampling subset of the province of Alentejo, which represents the main beef cattle production area in Portugal, accounting over 10.000 beef cows of several breeds, in natural mating, distributed in the different regions of the province. Herds had unknown sanitary status for BGC, although some showed clinical features compatible with the disease (low breeding season fertility, extended time from bull introduction to conception depicted from calving records, embryo-fetal mortality observed at pregnancy diagnosis). Preputial samples were collected using a scraping/washing technique [32] and were sent to the laboratory under refrigeration conditions. Total DNA extraction was performed using the DNeasy Blood and Tissue Kit (Qiagen), according to the manufacturer's instructions. Extracted DNA was quantified in a Nanodrop 2000C spectrophotometer (Thermo Scientific) and stored at − 20°C until use.

Real time PCR assays
Samples were tested by real-time PCR assays directed to subspecies (Cfv) and species (C. fetus) molecular targets. Detection at the subspecies level (Cfv) was performed using four assays, of which two targeting the parA gene (parA-A and parA-B assays) and the other two targeting the insertion element ISCfe1 (ISC-A and ISC-B assays). These pairs of assays, although directed to the same molecular target, detect different nucleotide regions as shown in Table 4. Additionally, two assays were performed for C. fetus detection, including an assay targeting the nahE gene and a commercial real-time PCR Kit (VetMAX C. fetus kit, Life Technologies). These two assays are designed to amplify sequences common to both C. fetus subspecies (Cfv and Cff), and, consequently, do not allow the identification at the subspecies level.
The flow chart of the sampling procedure for Cfv and C. fetus detection assays is schematically represented in Fig. 3. Briefly, all samples (n = 308) were tested with ISC-A and parA-A assays. In addition, a subset of these samples (n = 141) was blindly selected to be tested with ISC-B and parA-B assays. Detection of C. fetus specific targets was performed in all ISCfe1 and/or parA gene positive samples (n = 169).
For all assays, a PCR positive control was used in every experiment, which was prepared by adding 100 copies of Cfv genomic DNA from strain NCTC 10354 to 25 ng of DNA extracted from a Cfv-negative preputial sample to simulate a Cfv-positive preputial sample.

Detection of ISCfe1 insertion sequence: ISC-A and ISC-B assays
For the detection of the insertion element ISCfe1 two different Taqman MGB probe-based assays were used (ISC-A and ISC-B assays; Table 4). The ISC-A assay was developed to detect the tnpB gene of ISCfe1 (Table 4). The primers and probe (Additional file 2) were designed to target a 109 bp sequence using Primer Express software v2.0 and specificity was assessed using BLAST search in the NCBI database. PCR reactions were performed in duplicate, in 20 μL mixtures containing 1x SensiFAST probe Hi-ROX mastermix (Bioline Reagents Ltd), 400 nM of each primer, 100 nM of Taqman MGB probe and 25 ng of total DNA. Amplifications were performed on a StepOnePlus system (Applied Biosystems), using the following thermal cycle conditions: 2 min incubation step at 50°C, denaturation for 4 min at 95°C, followed by 35 cycles of 10 s at 95°C and 30 s at 60°C. The ISC-B assay targets the tnpA gene of ISCfe1, using previously published primers and probe [10]. PCR reaction mixtures and thermal cycle conditions were as previously described for ISC-A. Samples with a Ct < 35 were considered positive.

Detection of parA gene: parA-A and parA-B assays
The detection of parA gene also used two different Taqman MGB probe assays (parA-A and parA-B assays; Table 4). The parA-A assay is a real-time PCR assay previously described [13], with minor modifications (Additional file 2). The reverse primer was modified according to sequencing data of PCR products amplified with primers VENSF/VENSR [12] in the bovine preputial samples. All reactions were carried out in 20 μL reaction volume, with 900 nM of each primer, 250 nM of Taqman MGB probe, 1x SensiFAST probe Hi-ROX mastermix and 25 ng of DNA. Amplifications were performed on a Step One Plus System, using the thermal cycle conditions previously described [13]. The parA-B assay was designed to amplify a different sequence of the parA gene with 78 bp ( Table 4). The primers and probe (Additional file 2) were designed using Primer Express software v2.0 and specificity was assessed using BLAST  search in the NCBI database. PCR reaction mixtures and thermal cycle conditions were the same as described for ISC-A assay. Samples with a Ct < 35 in parA-A or parA-B assays were considered positive.
Detection of C. fetus species-specific targets: nahE and VetMAX C. fetus kit C. fetus detection was performed using a real-time PCR assay targeting the nahE gene and a commercial diagnostic kit (VetMAX C. fetus kit, Life Technologies) in all samples positive to ISCfe1 (ISC-A and/or ISC-B assays) and/or parA gene (parA-A and/or parA-B assay). The nahE gene was detected using previously described primers and probe [10], with the conditions described above for ISCfe1 amplification. Samples with a Ct < 35 were considered positive. Samples were also tested with VetMAX C. fetus kit following the instructions and validation criteria recommended by the manufacturer.
Real-time PCR assays' performance: linearity, amplification efficiency and reproducibility The analytical performance of the real-time PCR assays was assessed in DNA extracted from preputial samples, spiked with genomic DNA from Cfv strain NCTC 10354. Standards were prepared using DNA from Cfv NCTC 10354, previously extracted with Qiagen DNeasy Blood and Tissue kit (Qiagen). The genomic DNA was quantified using a Nanodrop 2000C spectrophotometer and the number of Cfv genome copies was determined based on the strain's genome size (1,874,244 bp), according to the formula: Number of copies = (DNA concentration (ng/μL) x [6.022 × 10 23 ]) / (length of template (bp) x [1 × 10 9 ] × 650). Ten-fold dilutions were performed from 1.25 × 10 6 to 1.25 × 10 2 genome copies/μL. Then, DNA preputial samples of three bulls, previously tested as negative in the PCR assays under study, were spiked with ten-fold dilutions of Cfv NCTC 10354, representing 10 1 to 10 5 genome copies per 25 ng of preputial sample DNA in 2 μL of template (PCR reaction). These three sets of standards were aliquoted and stored at − 20°C until use. Three independent experiments were performed for each real-time PCR assay, one for each set of standards, all tested with two replicate wells per dilution. The performance of the real-time PCR assays was evaluated based on the linearity (r 2 ), amplification efficiency (E) and reproducibility. The amplification efficiency was estimated based on the slope of the standard curve, using the formula: E = 10 (− 1/slope) -1. The reproducibility was evaluated using the inter-assay and intra-assay coefficients of variation (CV) for each dilution.

Detection of genomic island genes in bovine preputial samples
The presence of GI genes formerly considered Cfv-specific, which include type 4 secretion system-coding genes (virB2-virB11/virD4 genes), was assessed by conventional PCR. This Cfv-associated genomic island may also include the parA gene and fic genes. Amplification of virB9 and virB11 was performed with previously described primers [11]. Primers for amplification of fic1 and fic2 (Additional file 2) were designed with Primer-BLAST [33] using the sequence of Cfv strain NCTC 10354 as a reference. These assays for Amplifications were performed in a Doppio thermal cycler (VWR) using the following conditions: 3 min at 94°C, followed by 35 cycles of 94°C for 30 s, annealing temperature for 30 s, and 72°C for 1 min, with a final extension step of 5 min at 72°C. The annealing temperatures selected for amplification of fic1/fic2, virB9 and virB11 were 57°C, 53°C and 56°C, respectively. Amplification products were detected by electrophoresis on a 1.5% agarose gel stained with ethidium bromide and bands were visualized in a ChemiDoc XRS+ System (Biorad).

Statistical analysis
Statistical data analysis was performed using IBM SPSS Statistics for Windows, version 26.0 (IBM Corporation).
To evaluate inter and intra-assay reproducibility of realtime PCR assays, the coefficient of variation of the Ct was calculated as follows: % CV = (standard deviation Ct / mean Ct) × 100. Agreement between results of different real-time PCR assays was evaluated using the Cohen's Kappa coefficient. Additionally, results were analysed using McNemar's test for testing the null hypothesis that methods were equally likely to identify samples as positive or negative. The presence of association between the detection of GI-associated genes and the detection of the parA gene was assessed using the phi (φ) coefficient. Values of P < 0.05 were considered statistically significant.