Comparison of culture versus quantitative real-time polymerase chain reaction for the detection of Taylorella equigenitalis in field samples from naturally infected horses in Canada and Germany

Abstract

A quantitative real-time polymerase chain reaction method (qPCR) was developed and tested for the detection of Taylorella equigenitalis. It was shown to have an analytical sensitivity of 5 colony-forming units (CFU) of T. equigenitalis when applied to the testing of culture swabs that mimicked field samples, and a high analytical specificity in not reacting to 8 other commensal bacterial species associated with horses. As designed, it could also differentiate specifically between T. equigenitalis and T. asinigenitalis. The qPCR was compared to standard culture in a study that included 45 swab samples from 6 horses (1 stallion, 5 mares) naturally infected with T. equigenitalis in Canada, 39 swab samples from 5 naturally infected stallions in Germany, and 311 swab samples from 87 culture negative horses in Canada. When the comparison was conducted on an individual sample swab basis, the qPCR had a statistical sensitivity and specificity of 100% and 96.4%, respectively, and 100% and 99.1% when the comparison was conducted on a sample set basis. A comparison was also made on 203 sample swabs from the 5 German stallions taken over a span of 4 to 9 mo following antibiotic treatment. The qPCR was found to be highly sensitive and at least as good as culture in detecting the presence of T. equigenitalis in post-treatment samples. The work demonstrates that the qPCR assay described here can potentially be used to detect the presence of T. equigenitalis directly from submitted sample swabs taken from infected horses and also for determining T. equigenitalis freedom following treatment.

Introduction

Contagious equine metritis (CEM), caused by Taylorella equigenitalis, is a highly contagious, non-systemic, venereal disease of horses (1–3). The disease can manifest itself in mares either i) acutely with a copious mucopurulent vaginal discharge accompanied by cervicitis/endometritis and vaginitis, or ii) sub-acutely with little or no apparent clinical presentation. The principal manifestation of infection in the mare is short-term infertility and, rarely, abortion. Clinically affected mares clear the infection in a minimum of 2 wk with a rapid return to estrus following a shortened diestrous period. However, they can remain carriers of the organism in the clitoral area and, to a lesser extent in the endometrium, for long periods of time. There is no clinical disease manifested in stallions, but stallions can remain carriers for life in the absence of antibiotic treatment. Spread of the organism is through coitus, contaminated fomites, artificial insemination with infective semen and even to foals born to undetected carrier dams. Due to the highly contagious nature of this pathogen and the negative potential economic impacts to equine breeding establishments, the World Organisation for Animal Health (OIE) has identified CEM as one of the listed equine diseases and an important consideration in the international trade of horses (3). The disease is endemic in Europe and other parts of the world but Canada is recognized as being CEM-free (2). Quarantine and sampling strategies that focus detection of T. equigenitalis in carrier animals before release are critical to maintaining disease freedom in resident herd populations.

Taylorella equigenitalis is a fastidious, capnophilic, small Gram-negative bacillus or coccobacillus. Two biotypes, streptomycin-sensitive and streptomycin-resistant, are recognized (1,3). Even under these conditions and with rigid quality control standards, a minimum of 3 to 4 d is needed before distinct colonies are observed on culture. The OIE recommends an incubation period of at least 7 d before negative culture results can be reported (3). In addition, the organism is relatively inert biochemically but there are a series of recognized laboratory tests that are used to confirm the identity of suspect isolates as T. equigenitalis (1–3). More recently a closely related bacterium, T. asinigenitalis, has been described that can also be found in the reproductive tract of horses and donkeys (4–6). The pathogenic potential of this latter bacterial species is questionable (3,5) and natural disease caused by this organism has yet to be reported. However, T. asinigenitalis can mimic T. equigenitalis in its growth and test identification patterns and so it is important to be able to differentiate between these 2 related species of bacteria using routine detection methods.

Both serological methods, primarily complement fixation (CF), and culture have been used to detect the presence of T. equigenitalis in infected animals. The CF test is reliable on naive populations of animals with no history of disease and positive titers can be detected from 15 to 80 d after infection (1,3). Nevertheless, its use is limited to mares since only mares, not stallions, will seroconvert and antibody titers will disappear and not be detected in the true carrier-only state. In addition, the presence of anti-complementary activity compounds problems with the use of the CF test. Consequently, culture is presently the only internationally recognized prescribed test for detection of this equine pathogen (3). However, for culture to work effectively, both mares and stallions must be sampled from at least 3 sites in the reproductive tract on at least 3 separate occasions with a minimum of 3 to 4 d between sampling. Even with this multiple sampling requirement, carrier stallions can be missed and must be mated to 2 test mares that are then sampled to determine infective status in the stallion. Both an enriched non-selective and selective blood-based agar media are recommended for culturing sample swabs (3,7). As such, the fastidious nature of T. equigenitalis, with the requirement for more expensive media and extended incubation periods, along with the large number of recommended samplings make culture labor-intensive and slow. In addition, the recent and extended CEM outbreak in the United States involving 48 states and over 1000 exposed and infected animals (2) has placed doubts on the ability of culture alone to guarantee full freedom of resident herds from T. equigenitalis in a timely manner, thereby compromising disease freedom in North America. Thus, alternative molecular based methods are needed in order to overcome the weaknesses of routine culture and enhance detection awareness.

The first attempt to develop a conventional gel-based PCR method for the specific detection of T. equigenitalis targeted a region of the gene encoding the 16S rRNA with the option of confirming the nature of the 585 bp amplicon by dot blot and hybridization to an internal probe (8). Whereas evaluation of this PCR had suggested its specificity and potential value as an adjunct to culture (9), cross reactions with other bacterial species (10) and the isolation of the closely related T. asinigenitalis raised concerns of specificity and its ability to differentiate between the 2 Taylorella species without further time-consuming nucleotide sequence characterization of the amplicon (11). Accordingly, more specific PCR tests targeting greater sequence heterogeneity within the 16S rDNA gene of the 2 Taylorella species were developed (12,13) including a multiplex PCR that generated a distinct sized amplicon from each species (14). A cloned sequence described by Anzai et al (15) was also targeted for PCR detection of T. equigenitalis (15,16), but the nature of the gene fragment was not discussed in detail and the behavior of T. asinigenitalis in this assay was not reported. The subsequent development of quantitative real-time PCR (qPCR) platforms provided the opportunity to develop more sensitive and rapid methods that are less laborious and far less susceptible to carry-over contamination (17–19). While these qPCR assays, which also target the rDNA gene, have compared favorably with the results of culture, very low numbers of positive samples from naturally infected animals were included in those studies, placing doubts on the statistical evaluation and use of these qPCR assays under normal field conditions.

The purpose of the research work presented here was to address this latter weakness. It presents data describing the development and evaluation of a qPCR assay for the detection of T. equigenitalis and T. asinigenitalis and a comparison against culture for the detection of T. equigenitalis directly from sample swabs of naturally infected and uninfected horses in Canada and Germany.

Materials and methods

Bacterial strains

The Taylorella and other non-Taylorella bacterial strains used in this study are listed in Table I. All isolates were kept at −80°C in glycerol beads (CryoStor Micobiology Culture Preservation Vials; Innovatek Medical, Vancouver, British Columbia) according to the manufacturer’s instructions until required for testing. For testing, the Taylorella sp. were cultured on Eugon based agar medium supplemented with 5% chocolatized sterile defibrinated horse blood and 5 μg/mL of amphotericin B (MECA + A), as described by Atherton (20). All other bacterial strains were cultured to tryptic soy based agar supplemented with 5% sterile sheep blood.

Table I.

Bacterial strains used in this study and results of qPCR testing to the T. equigenitalis (EQUI-HEX) and T. asinigenitalis (ASINI-FAM) specific probes

Bacterial species	Strain (origin)	Detection by EQUI-HEX	Detection by ASINI-FAM
T. equigenitalis	ATCC 35865	Yes	—
	NCTC 11225	Yes	—
	96IMPN 671 (Canadaa, German stallion)	Yes	—
	99IMPN 099 (Canadaa, Dutch mare)	Yes	—
	01IM 263 (Canadaa, German stallion)	Yes	—
	01IM 463 (Canadaa, German mare)	Yes	—
	03IM 117 (Canadaa, Belgian mare)	Yes	—
	05IM 341 (Canadaa, Belgian stallion)	Yes	—
T. asinigenitalis	ATCC 700933 (California jack isolate)	—	Yes
	2329-98 (Kentucky grey jack isolate)	—	Yes
	BEL3751/05 (Swedish stallion)	—	Yes
Streptococcus equi ssp. zooepidemicus	AHML 2005B14-1	—	—
Bacillus cereus	ATCC 11778	—	—
Staphylococcus aureus	ATCC 25923	—	—
Klebsiella pneumoniae	ATCC 27736	—	—
Escherichia coli	ATCC 25922	—	—
Pseudomonas aeruginosa	ATCC 27853	—	—
Bordetella bronchiseptica	ATCC 10580	—	—
Oligella urethralis	ATCC 17960	—	—

^aT. equigenitalis strain isolated from imported horse in quarantine in Canada.

Pure culture preparation of bacterial isolates for DNA extraction and qPCR testing

For the Taylorella strains, a 0.5 McFarland nephelometer standard, diluted 1:1000 in sterile Milli-Q (MQ) water, was made and 1.0 mL transferred into a sterile microcentrifuge tube. For the non-Taylorella bacterial species, a sterile bacteriological loop was used to transfer 1 to 2 colonies of each pure culture bacterial isolate from culture plates into 1.0 mL of sterile phosphate buffered saline (PBS, pH 7.2) in a sterile 1.5 mL microcentrifuge tube. Each bacterial suspension was washed once by centrifugation (13 000 rpm, 10 min), the pellets resuspended in 1.0 mL of sterile MQ water and the DNA extracted and tested as described below. For determination of the assay’s analytical sensitivity for Taylorella sp. detection, a tenfold serial dilution of the isolate, starting with a 0.5 McFarland nephelometer standard (10° dilution), was made in sterile MQ water down to a 10⁻⁷ dilution. Total bacterial counts were determined by spread plating 100 μL of the 3 highest dilutions on triplicate plates of MECA + A and incubating plates for 4 d at 36°C in 7.5% CO₂. Colony counts were used to determine original bacterial cell concentrations [colony forming units (CFU)/mL] in each dilution tube of the original series. A 100 μL aliquot of each serial dilution tube was inoculated into the absorbent end of a sterile swab and the swab inserted into vials of Amies transport medium with charcoal (Amies plus charcoal). The vials were incubated at 4°C for 24 h and DNA extracted and tested as described below.

Clinical sampling of horses

A total of 45 swab samples from 6 T. equigenitalis naturally infected positive animals (1 imported stallion, 2 imported mares, and 3 test mares bred to 2 imported CEM positive stallions) and 311 swab samples from 87 CEM culture negative animals (82 mares and 5 stallions) were tested from Canada. Thirty-nine swabs from 5 stallions in a single stud with confirmed T. equigenitalis infection, along with 203 swabs from the same stallions post antibiotic treatment were tested from Germany. In Canada, swabbing of animals consisted of 3 sets of 3 or 4 sample swabs collected with a minimum of 3 d between sampling sets. Each set from mares included a sample swab from: i) the clitoral fossa, ii) the clitoral medial sinus, iii) the clitoral lateral sinuses, and occasionally, iv) the endometrium or cervix. Each set from stallions included a sample swab from i) the prepuce and sheath, ii) urethral sinus, and iii) the fossa glandis including the diverticulum. In Germany, 2 sets of 4 swabs were taken from i) the fossa glandis, ii) urethra, iii) surface of prepuce, and iv) pre-ejaculatory fluid. Post-antibiotic treatment samples from Germany could also include the urethral sinus. All sample swabs tested were placed into Amies plus charcoal, shipped on ice and received and cultured in the laboratory within 48 h.

Culture, isolation, and identification of T. equigenitalis from sample swabs

The culture procedure was the same in both countries. Briefly, swabs were cultured to 3 different agar media: i) MECA + A, ii) MECA + AS (containing 5 μg/mL amphotericin B and 400 μg/mL streptomycin), and iii) MECA + CTA (containing 5 μg/mL clindamycin, 1 μg/mL trimethoprim, and 5 μg/mL amphotericin B). The basal media formulation and antibiotics for the 3 media, with the exception of the streptomycin concentration in MECA + AS, were as described by Atherton (20) including the use of 5% sterile lysed horse blood to the MECA + CTA to prevent inactivation of the trimethoprim. All agar plates were incubated at 36°C in 7.5% CO₂ and observed for suspect T. equigenitalis colonies at 4, 7, 10, and 14 d incubation in Canada and 4, 7, and 10 d in Germany. Any suspect colonies were subcultured to MECA + A and all Gram-negative rods were identified as T. equigenitalis according to negative reactions to: i) aerobic growth on 5% sheep blood agar, ii) motility in BHI broth, iii) glucose utilization in Hugh-Leifson glucose medium, and positive reactions to: i) catalase, ii) oxidase, iii) phosphatase, and iv) rapid strong agglutination in the MonoTayl test (Bionar Laboratories AS, Skien, Norway). Once cultured, all swabs were placed at −20°C until required for DNA extraction and qPCR testing as described below.

Real-time PCR design

An alignment of 16S rDNA gene sequences of T. equigenitalis, strain NCTC 11184 (GenBank accession X68645), and T. asinigenitalis, strain UCD-1 (GenBank accession AF067729), was generated using CLUSTALX (21). A region at position 434–454 of the T. equigenitalis sequence (corresponding to positions 454–474 of the T. asinigenitalis sequence) that included 8 base pair differences was suitable for the generation of species-specific probes. The 2 PCR primers, TAYQF and TAYQR, targeted conserved sequences flanking this region as follows: TAYQF: 5′-CGCGTGTGCGATTGAA-3′ was located at positions 378–393 and 398–413, respectively, in T. equigenitalis and T. asinigenitalis; TAYQR: 5′-GCCGGTGCTTATTCTTCA-3′ was located at positions 485–468 and 505–488, respectively, in T. equigenitalis and T. asinigenitalis. The T. equigenitalis specific probe (EQUI-HEX) was: HEX-5′-AGGTTTGTGTTAATACCATGGACTGCTGAC-3′-BHQ1 (positions 432–461) and the T. asinigenitalis specific probe (ASINI-FAM) was: FAM -5′-AGTTTTAGGATAATACCCTAGGATG CTGACGGT-3′-BHQ1 (positions 450–482). Reagent design was optimized using the on-line Oligo Analyzer 3.1 tool (http://www.idtdna.com/analyzer/Applications/OligoAnalyzer) and BLAST analysis of the GenBank database (http://blast.ncbi.nlm.nih.gov/Blast.cgi).

An internal positive control (IPC), designed to detect inhibitors in the qPCR assay, was developed using a human adenovirus 5 (HA5) probe together with the TAYQF/TAYQR primers. The DNA construct, used as the control, was generated using a set of primers containing the TAYQ primer sequences linked downstream to HA5 sequences (primers Ad5For and Ad5Rev) as described elsewhere (22). These hybrid primers were used with a clone of HA5 DNA to generate an amplicon of 148 bp corresponding to HA5 co-ordinates 28297–28407 (GenBank accession no. M73260) and flanking TAYQF/R primer sequences. This product was cloned into vector PCRII-TOPO using TOPO TA cloning kit as described (Life Technologies, Burlington, Ontario) and a clone with the expected sequence, as confirmed by Sanger sequencing, was recovered. Plasmid DNA of this clone was included in each PCR assay in an amount (0.125 fg/μL), at which point it was detected specifically by the IPC-specific probe (HA5-CY5, CY5-5′-TCTACGAGAGAACCTCTCCGAGCTCAGCT-3′-BHQ2) with an expected Cycle threshold (Ct) value of 35. Increase of the Ct value by more than 3 above the running mean value was used to indicate possible interference with the amplification by inhibitors.

All primers and probes were synthesized (Integrated DNA Technologies, Inc., Coralville, Iowa, USA) and dissolved in 100 μM stocks in 10 mM Tris-HCl, pH 8.0.

Preparation of DNA and qPCR testing

Five separate DNA template control vials were tested each time qPCR was conducted as follows: i) T. equigenitalis and IPC DNA, ii) T. asinigenitalis and IPC DNA, iii) Oligella urethralis and IPC DNA, iv) IPC DNA only, and v) no IPC nor Taylorella sp. DNA.

For initial trials using cultured bacterial strains, 0.1 mL of each washed culture suspension was heated at 100°C for 15 min. The suspension was centrifuged for 30 s at 13 000 rpm to pellet any cellular debris, and 2 μL of supernatant tested by qPCR as described below.

For work with swab samples from horses, known positive control samples were produced by immersing sterile swabs into a 0.5 McFarland nephelometer standard of T. equigenitalis ATCC 35865, diluted 1:1000 in sterile MQ water, which were subsequently placed into Amies plus charcoal for 24 h at 4°C. The negative control samples were naturally contaminated field sample swabs, in Amies plus charcoal, from the reproductive tract of 33 horses in Canada that were culture negative for T. equigenitalis. A total of 15 positive and 75 negative control swabs in Amies plus charcoal were inserted randomly in the Canada and Germany field samples and all test field samples and controls were renumbered and tested blind by qPCR as follows.

Frozen field, or refrigerated control or dilution series, swabs in Amies plus charcoal were brought to room temperature. The absorbent end of each swab was aseptically cut from the main shaft and placed into 1.0 mL of sterile MQ water in a 1.5 mL microcentrifuge tube, gently vortexed for 5 s, and allowed to stand for 5 min. Following centrifugation for 30 s at 13 000 rpm, the swab end was gently removed using sterile forceps, being careful not to disturb the pellet, the supernatant discarded, and the pellet re-suspended in 20 μL of nuclease-free sterile water. The suspension was heated at 100°C for 15 min in a heating block. Lysates were cooled and either tested immediately or stored at −20°C before testing by qPCR. Lysates were centrifuged for 30 s to pellet any remaining media debris and 2 μL of supernatant was added to 18 μL of ABI TaqMan master mix (Life Technologies) containing 400 nM TAYQF and TAYQR primers, 200 nM of EQUI-HEX probe, 100 nM of each of the other 2 probes (ASINI-FAM and HA5-CY5) and 2.5 fg IPC DNA. The PCR cycling was conducted on an ABI 7500 (Life Technologies) starting with a 2-minute incubation at 50°C (to allow uracil-N-glycosylase digestion of any pre-formed amplicon), an initial 10-minute denaturation at 95°C for Taq polymerase activation, followed by 40 cycles of denaturation at 95°C for 10 s, and primer annealing and extension at 55°C for 1 min. A Ct value ≤ 40 for any test lysate was considered a positive qPCR result.

Analysis of data

Statistics included use of mean and standard deviation in the determination of analytical sensitivity and standard 2 × 2 tables for comparison of culture and qPCR test results of sample swabs with the sensitivity, specificity, Kappa coefficient and confidence intervals (CI) defined and calculated using the on-line Clinical Calculator 1 tool (http://vassarstats.net/clin1.html).

Results

As summarized in Table I, using DNA prepared from bacterial cultures, it was established that the qPCR assay specifically detected and differentiated established strains of T. equigenitalis and T. asinigenitalis with no cross-reactivity observed between the 2 probes. No signal was generated by either probe for the 8 other non-Taylorella commensal bacterial species (Table I). The analytical sensitivity of the qPCR test (Figure 1) for detection of T. equigenitalis ATCC 35865 directly from sample swabs placed into Amies plus charcoal was 5 CFU at a Ct value of 40.

Figure 1.

Standard line graph showing qPCR testing of T. equigenitalis (ATCC 35865) from a dilution series inoculated to swabs placed into Amies + charcoal and incubated for 24 h at 4°C.

^a Each point along the line represents the mean ± standard deviation of 3 replicates.

^b Extrapolation of the line graph to a Ct of 40 which is equivalent to 5 colony-forming units (CFU).

The comparison of culture and qPCR testing on field sample swabs in Amies plus charcoal from Canada and Germany included 15 randomly inserted positive and 75 randomly inserted negative control sample swabs. All 15 positive control swabs were qPCR positive for T. equigenitalis only and the 75 negative control swabs were all negative by qPCR for both Taylorella species. Results of both culture and qPCR testing of sample swabs and sets from all 11 infected horses in both countries are shown in Table II. A total of 84 swabs were collected from these 11 animals of which 61 swabs (72.6%) were culture positive/qPCR positive, 12 swabs (14.3%) were culture negative/qPCR positive, 11 swabs (13.1%) were culture negative/qPCR negative, and no swabs (0%) were culture positive/qPCR negative. Table III shows the 2 × 2 table of culture versus qPCR for these 84 swabs from 11 infected horses, as described above, together with 311 swabs from 87 T. equigenitalis culture negative animals in Canada. When compared to culture, the diagnostic sensitivity and specificity of the qPCR was 100% and 96.4%, respectively (Kappa coefficient = 0.893), when based upon individual swab sample results. If the comparison is performed based on swab sets (Table IV), where a single positive swab result in a set would make the complete swab set positive and where all negative swab results in a set were required to produce a negative set result, the sensitivity of the qPCR test was 100% with a specificity of 99.1% (Kappa coefficient = 0.974).

Table II.

Results of field sample swabs (Amies plus charcoal), tested by culture and qPCR, taken from T. equigenitalis naturally infected horses in Canada and Germany

Animal number	Number of submitted samplesa	Culture +, qPCR +	Culture +, qPCR −	Culture −, qPCR +	Culture −, qPCR −
Canada
1	6 (2)	5	0	1	0
2	9 (3)	9	0	0	0
3	6 (2)	4	0	1	1
4	9 (3)	9	0	0	0
5	9 (3)	9	0	0	0
6	6 (2)	2	0	1	3
Germany
1	8 (2)	8	0	0	0
2	7b (2)	6	0	1	0
3	8 (2)	3	0	5	0
4	8 (2)	6	0	2	0
5	8 (2)	0	0	1	7
Totals
11	84 (25)	61	0	12	11

^aNumbers in brackets represent the number of swab sample sets saved and tested. A single swab sample set consisted of 3 separate sample swabs taken at the same time (from different sampling sites in the reproductive tract) from horses in Canada versus 4 separate sample swabs from horses in Germany.

^bOne of the swab samples in the first sample set did not have a culture result and thus the sample set consisted of results from 3 swab samples instead of 4.

Table III.

Individual field swab sample results tested by culture and qPCR from known T. equigenitalis naturally infected and uninfected horses from Canada and Germany

		Culture
		Positive	Negative
qPCR	Positive	61	12
	Negative	0	322
$$\begin{array}{ll}\text{Sensitivity}\hfill & =\text{number\hspace{0.17em}of\hspace{0.17em}qPCR\hspace{0.17em}positive\hspace{0.17em}swabs}/\text{number\hspace{0.17em}of\hspace{0.17em}culture\hspace{0.17em}positive\hspace{0.17em}swabs}\times 100\%\hfill \\ \hspace{0.17em}\hfill & =100\%\hspace{0.17em}(95\%\hspace{0.17em}\text{Cl}=92.6-100\%)\hfill \\ \text{Specificity}\hfill & =\text{number\hspace{0.17em}of\hspace{0.17em}qPCR\hspace{0.17em}negative\hspace{0.17em}swabs}/\text{number\hspace{0.17em}of\hspace{0.17em}culture\hspace{0.17em}negative\hspace{0.17em}swabs}\times 100\%\hfill \\ \hspace{0.17em}\hfill & =96.4\%\hspace{0.17em}(95\%\hspace{0.17em}\text{Cl}=93.6-98.0\%)\hfill \\ \text{Kappa}\hfill & =0.892\hspace{0.17em}(95\%\hspace{0.17em}\text{Cl}=0.833-0.952)\hfill \end{array}$$

Table IV.

Swab sample set results tested by culture and qPCR from known T. equigenitalis naturally infected and uninfected horses in Canada and Germany

		Culture
		Positive	Negative
qPCR	Positive	23a	1b
	Negative	0	112c
$$\begin{array}{ll}\text{Sensitivity}\hfill & =\text{number\hspace{0.17em}of\hspace{0.17em}qPCR\hspace{0.17em}positive\hspace{0.17em}swab\hspace{0.17em}sets}/\text{number\hspace{0.17em}of\hspace{0.17em}culture\hspace{0.17em}positive\hspace{0.17em}swab\hspace{0.17em}sets}\times 100\%\hfill \\ \hspace{0.17em}\hfill & =100\%\hspace{0.17em}(95\%\hspace{0.17em}\text{Cl}=82.1-100\%)\hfill \\ \text{Specificity}\hfill & =\text{number\hspace{0.17em}of\hspace{0.17em}qPCR\hspace{0.17em}negative\hspace{0.17em}swab\hspace{0.17em}sets}/\text{number\hspace{0.17em}of\hspace{0.17em}culture\hspace{0.17em}negative\hspace{0.17em}swab\hspace{0.17em}sets}\times 100\%\hfill \\ \hspace{0.17em}\hfill & =99.1\%\hspace{0.17em}(95\%\hspace{0.17em}\text{Cl}=94.5-100\%)\hfill \\ \text{Kappa}\hfill & =0.974\hspace{0.17em}(95\%\hspace{0.17em}\text{Cl}=0.924-1.000)\hfill \end{array}$$

^aIf a single swab sample result in a swab sample set was positive by both culture and qPCR then this made the entire swab sample set positive by both tests.

^bIf a single swab sample result in a swab sample set was culture negative and qPCR positive then this made the entire swab sample set culture negative/qPCR positive.

^cIf all swab sample results in a sample set were negative by both culture and qPCR then this made the entire swab sample set negative by both tests.

The results generated by qPCR for swabs obtained from the 5 German stallions following antibiotic treatment are shown in Table V. A total of 203 sample swabs (51 swab sample sets) in Amies plus charcoal were taken from the 5 stallions over a 4- to 9-month period following treatment. In one of the stallions (animal number 1) T. equigenitalis was detected by both culture and qPCR in 2 of 4 swab samples 28 d following re-commencement of sampling to prove freedom from infection. This positive finding occurred in the 7th set of 4 culture swabs taken following treatment. In addition, there was a 3rd swab, in the same swab set from this same animal, which was culture positive/qPCR negative and a 4th swab, again in the same swab set from this same animal, which was culture negative/qPCR positive. This stallion was treated a second time with antibiotics and re-sampled and tested for freedom 6 additional times over an 8 mo period with no additional culture or qPCR positive results detected. There were 2 other stallions (animal numbers 2 and 4) that both had a single swab in a single set that was culture negative/qPCR positive. In both animals the detection occurred in the 3rd set of swabs taken 4 d after the re-commencement of sampling for pathogen freedom. The 2 remaining stallions (animal numbers 3 and 5) had no culture positive and/or qPCR positive swab results following treatment.

Table V.

Results of field sample swabs (Amies plus charcoal), tested by culture and qPCR, taken from 5 T. equigenitalis naturally infected stallions in Germany following antibiotic treatment

Animal number	Number of submitted samplesa	Culture +, qPCR +	Culture +, qPCR −	Culture −, qPCR +	Culture −, qPCR −
1	51 (13)	2 (1)b	1 (1)b	1 (1)b	47 (12)
2	40 (10)	0	0	1 (1)c	39 (9)
3	36 (9)	0	0	0	36 (9)
4	40 (10)	0	0	1 (1)c	39 (9)
5	36 (9)	0	0	0	36 (9)
Totals
5	203 (51)	2	1	3	197

^aNumbers in brackets represent the number of swab sample sets represented. A single sample set consisted of 4 separate sample swabs taken at the same time from different sampling sites in the reproductive tract.

^bThese swab sample results occurred in the same sample set (the 7th set) of 4 sample swabs taken 28 d post-antibiotic treatment to prove pathogen freedom. This animal was treated a second time with antibiotics and tested 6 additional times over an 8-month period with negative culture and qPCR results.

^cThis single qPCR positive swab result occurred in the third sample set taken 4 d post-antibiotic treatment to prove pathogen freedom. These 2 animals were not treated further with antibiotics.

Discussion

The qPCR procedure described in this work targeted the same region of the 16S rDNA gene as has been used by others for the detection and differentiation of T. equigenitalis and T. asinigenitalis (17–19) and although the primers and probes used here were designed independently, they were close in nature to those reagents previously reported. In agreement with those previous publications, high analytical sensitivity and specificity is demonstrated in this work. However, there are novel aspects of the work presented here that include: i) determination of analytical sensitivity of the qPCR procedure for detection of T. equigenitalis from swabs in Amies plus charcoal, ii) the nature of the internal positive control to identify PCR inhibitors, iii) the testing of greater numbers of naturally contaminated samples, especially positives, and iv) an evaluation of the use of qPCR versus culture on infected animals following antibiotic treatment. The first 2 novel aspects are related to the optimization of the assay during the evaluation stage and connected to the determination of analytical and diagnostic sensitivity, and the 3rd and 4th aspects are important for obtaining more accurate statistical data for detection of T. equigenitalis by this procedure under field conditions.

In order to maximize analytical sensitivity of most qPCR techniques, centrifugation has been used to concentrate bacteria. This practice has also been used here and we were able to obtain an analytical sensitivity of one CFU of T. equigenitalis (one CFU also for T. asinigenitalis, data not shown) with the qPCR procedure described here and use of testing from a pure culture dilution series. This is in agreement with other described qPCR assays developed for T. equigenitalis (17–19), which also used centrifugation, and reported analytical sensitivities of 1–20 CFU (40 CFU for T. asinigenitalis) from pure culture titrations. However, routine testing for T. equigenitalis is not conducted from pure culture titrations but from sample swabs in Amies plus charcoal and the effects of this medium have to be considered for the proper evaluation of the test procedures. Previously published work on use of PCR for T. equigenitalis detection (15) speculated that charcoal was responsible for observed inhibition of PCR product amplification in that study. This speculation was supported by qualitative experiments conducted here (data not shown) which demonstrated that, although variations were observed between trials, the activated charcoal component of the medium could repeatedly inhibit the qPCR results by as much as 30% as could the complete Amies plus charcoal transport medium but without agar. In addition to charcoal, agar, an ingredient also present in the Amies plus charcoal medium, has been shown to be inhibitory in gel based PCR amplification procedures (23) and could be an inhibitory ingredient here as well. While centrifugation allowed for concentration of bacteria, washing of sediments and the elimination of soluble ingredients such as salts, it did not necessarily eliminate all larger insoluble transport media components, such as charcoal and agar, from pellets and test vial solutions that could affect testing and sensitivity. Wakeley et al (18) reported that 5.8% of the swab samples tested during their evaluation of a qPCR for T. equigenitalis produced inconclusive results and were thus eliminated in the final assessment. Several possible reasons were given including that the extraction failed for those particular samples. This allows speculation that potential inhibitory substances, such as transport media components, could have been carried over during sample preparation and prevented adequate amplification of target sequences. The problem of inconclusive results was not encountered in this study and media carry over was overcome by careful manipulations and volume extractions of test lysates that prevented disturbing and transferring any extraneous pellet material to final test reaction vials. Also related to sample preparation, it was found that a 5 min initial settling period of the sample swab in solution, before initial centrifugation, was advantageous in optimizing sensitivity of the assay for field trials. It can only be speculated that this subtle difference allowed for greater separation of bacteria from the swab to solution. Using these techniques, a more realistic analytical sensitivity of approximately 5 CFU of T. equigenitalis from swabs in Amies plus charcoal was achieved with the qPCR procedure described here. It is possible that use of a sterile pre-moistened swab placed into a transport vial without any transport medium may be a much better sampling procedure for qPCR testing of T. equigenitalis than the use of Amies plus charcoal. However, because of long transportation distances requiring as much as 48 h for laboratory arrival, use of an additional sampling procedure or changes away from use of Amies plus charcoal would be impractical for routine testing in Canada at this time.

Internal positive controls (IPC) are used routinely in qPCR test procedures and serve an important quality control function in each test. Internal positive controls used in previous qPCR assays for Taylorella have used 16S rDNA primers designed to amplify a region conserved in many other eubacteria, such as Klebsiella pneumoniae, Salmonella Typhimurium, T. asinigenitalis, O. urethralis, Streptococcus zooepidemicus, and Staphylococcus aureus (19). With the exception of S. Typhimurium and T. asinigenitalis, the other bacterial species are considered common commensals of horses. We also tested and were able to confirm that O. urethralis, K. pneumoniae, Bordetella bronchiseptica, Escherichia coli, Pseudomonas aeruginosa, and S. aureus could produce amplicons from the IPC primers described previously in qPCR assays for T. equigenitalis (data not shown). The probes to these IPC amplicons do not interfere with the specific recognition of the individual Taylorella species and vice versa, and thus, do not interfere with analytical specificity. However, the amplicons produced from these common bacterial species are competitive with the target sequences for reagents in the master mix and can cause non-specific depletion of reagents and the potential of a false negative result. This would decrease analytical sensitivity of the procedure for the T. equigenitalis target sequence. However, if little or no bacterial DNA is present or the commensal bacterial DNA present in a sample are not recognized by the control primers, it is possible that amplification using this type of competitive IPC will not occur, again compromising quality control and the test result. Wakeley et al (18) used this type of positive control but it differed in that it was set up in a separate reaction vessel and was not internal. Regardless, an alternative reason given for the 5.8% rate of inconclusive sample results in that study was failure of the positive control and this could have been due to inadequate numbers of bacteria on the sample swab or lack of bacterial cross-hybridization with the 16S rDNA gene probe. In this study, an alternative type of competitive IPC was employed using an approach similar to that described by Das et al (24) for detection of avian influenza using real-time reverse transcriptase (RT)-PCR. The IPC used the same primers as the Taylorella PCR but incorporated a separate probe (HA5-Cy5) that recognized an internal sequence specific only to the IPC. This IPC design has the advantage that it can be multiplexed with the PCR and thus acts as a true internal control for each individual assay. The amount of the IPC template included in each assay was adjusted so as to yield Ct values around 35 and thereby ensure minimal depletion of reagents required for amplification of the Taylorella target. This approach avoids any influences from the presence or absence of other bacterial DNA in the test lysate and has the advantage of allowing an evaluation of the sample aliquot within each reaction tube, an improvement in quality control.

The qPCR method, reported by Premanandh et al (19), analyzed 172 field samples of which none were culture positive. However, they reported 10 samples as culture negative but qPCR positive. They concluded that their qPCR method had a greater sensitivity than the culture method for T. equigenitalis detection even though they were unable to provide diagnostic sensitivity or specificity of their procedure. Ousey et al (17) were also unable to report diagnostic sensitivity and specificity of their commercial qPCR test for T. equigenitalis because there were only 9 positive samples (out of 2072 total) and 6 of these samples were prepared quality assurance T. equigenitalis spiked sample swabs, not natural field samples. Using a different qPCR assay for T. equigenitalis, Wakeley et al (18) reported a sensitivity of only 83.3% and a specificity of 99.7% and this was determined with just 12 culture positive sample swabs (out of 733 total), of which 2 swabs were qPCR negative. Hence, in addition to the contradictions to claims of higher sensitivity, the inability to obtain appropriate numbers of positive field samples from animals naturally infected with T. equigenitalis has, unfortunately, prevented a proper statistical analysis and evaluation of previously described qPCR methods for detection of this equine pathogen. The work described in this paper overcomes this problem by providing comparative results from 84 field samples, in Amies plus charcoal, from 11 CEM naturally infected horses and a further 311 field samples from 87 uninfected horses from both Canada and Germany. Compared to culture, the qPCR had a diagnostic sensitivity and specificity of 100% and 96.4%, respectively, from individual sample swabs and 100% and 99.1% from swab sample sets, respectively. The specificity was somewhat affected by the presence of a single stallion from Germany (animal number 5) that was culture negative but had a single positive qPCR swab result on initial testing. Without test mating to uninfected mares, the true infection status of this animal cannot be stated with certainty but its presence in a known CEM infected stud with 4 remaining proven CEM positive stallions would be highly suggestive of a very low level carriage rate of T. equigenitalis. Under these conditions, given the highly contagious nature of this equine pathogen and the serious nature of this sexually transmitted disease, it would be prudent and reasonable to treat this animal as truly infected without further confirmatory testing. Regardless of any questions about the results from this single animal, the kappa values of 0.892 for individual samples and 0.974 for swab sample sets indicate an almost perfect level of agreement between the culture and qPCR procedures in this study.

The qPCR was also used to assess the 5 T. equigenitalis infected stallions from Germany following antibiotic treatment. Two of the stallions (animal numbers 3 and 5, Table V) were both culture and qPCR negative following treatment. Two additional stallions (animal numbers 2 and 4) were culture negative following treatment but a single swab sample taken 4 d after recommencement of sampling from both animals was qPCR positive. Reasons for this observation include the generation of a false positive result or the presence of a few dead cells of T. equigenitalis remaining in the reproductive tract of both animals after the antibiotic treatment that could have only been detected using the qPCR test. A fifth stallion (animal number 1) was clearly positive by culture and qPCR following antibiotic treatment but this finding was not detected from Amies plus charcoal swabs until the seventh swab set sampling that occurred 28 d following recommencement of sampling after treatment. It accounts for the greater number and sets of swabs taken from this stallion compared to the others in order to ensure freedom following a second antibiotic treatment regimen. Whether this second observation of T. equigenitalis infection in this stallion represented a new infection or a failure of initial treatment is not known with certainty. However, regardless of the speculation on these observations, the results presented here indicate that the more sensitive nature of the qPCR procedure strongly suggests its usefulness in detecting T. equigenitalis presence in post antibiotic treatment of animals and assessing freedom from this pathogen.

In summary, a modified qPCR procedure has been developed and shown to be highly sensitive and specific for the detection of T. equigenitalis directly from field samples as compared to culture. The robustness of the procedure was also observed from the fact that positive qPCR results were obtained from sample swabs that were frozen for several years following culture. It also has the potential of being very valuable to use, in conjunction with culture, in enhancing the assessment of antibiotic treatment of previously T. equigenitalis infected animals to prove freedom.