Semiquantitative Multiplexed Tandem PCR for Detection and Differentiation of Four Theileria orientalis Genotypes in Cattle

Abstract

Oriental theileriosis is an emerging, tick-borne disease of bovines in the Asia-Pacific region and is caused by one or more genotypes of the Theileria orientalis complex. This study aimed to establish and validate a multiplexed tandem PCR (MT-PCR) assay using three distinct markers (major piroplasm surface protein, 23-kDa piroplasm membrane protein, and the first internal transcribed spacer of nuclear DNA), for the simultaneous detection and semiquantification of four genotypes (Buffeli, Chitose, Ikeda, and type 5) of the T. orientalis complex. Analytical specificity, analytical sensitivity, and repeatability of the established MT-PCR assay were assessed in a series of experiments. Subsequently, the assay was evaluated using 200 genomic DNA samples collected from cattle from farms on which oriental theileriosis outbreaks had occurred, and 110 samples from a region where no outbreaks had been reported. The results showed the MT-PCR assay specifically and reproducibly detected the expected genotypes (i.e., genotypes Buffeli, Chitose, Ikeda, and type 5) of the T. orientalis complex, reliably differentiated them, and was able to detect as little as 1 fg of genomic DNA from each genotype. The diagnostic specificity and sensitivity of the MT-PCR were estimated at 94.0% and 98.8%, respectively. The MT-PCR assay established here is a practical and effective diagnostic tool for the four main genotypes of T. orientalis complex in Australia and should assist studies of the epidemiology and pathophysiology of oriental theileriosis in the Asia-Pacific region.

INTRODUCTION

Tick-borne diseases (TBDs) pose a major threat to livestock production worldwide and can have a significant impact on farming communities due to economic losses (1). Theileriosis is one of the important TBDs of cattle, sheep, and/or other ruminants, mainly in tropical and subtropical regions of the world (2). In cattle, East Coast fever (ECF) and Mediterranean/tropical theileriosis are due to Theileria parva and Theileria annulata, respectively, whereas oriental theileriosis is caused by Theileria orientalis. The prevalence of various forms of theileriosis in different parts of the world is dependent on the occurrence of suitable tick vectors for their transmission (3).

Oriental theileriosis is caused by one or more genotypes of the T. orientalis complex and is transmitted by ixodid ticks, primarily Haemaphysalis spp. (4–6). Presently, 11 genotypes of T. orientalis complex (designated Chitose or type 1, Ikeda or type 2, Buffeli or type 3, types 4 to 8, and N-1 to N-3) have been identified using a number of molecular markers, including major piroplasm surface protein (MPSP) (7, 8), 23-kDa piroplasm membrane protein (p23) (9–11, 60), small-subunit (SSU) rRNA gene (8, 12, 13), and/or the first and second internal transcribed spacers of nuclear ribosomal DNA (ITS-1 and ITS-2, respectively) (12, 14). Of these genotypes, Ikeda and Chitose are recognized to be associated with clinical outbreaks of oriental theileriosis, mainly in the Asia-Pacific region (15–21). The major clinical signs of this disease include fever, anemia, jaundice, lethargy, weakness, abortion, and/or mortality (16–18), with significant production losses in dairy cattle (22). Thus far, four genotypes (Buffeli, Chitose, Ikeda, and type 5) of T. orientalis have been reported in Australia (13, 18, 20–23).

Currently, the diagnosis of oriental theileriosis is usually based on the observation of clinical signs, the detection of piroplasms of T. orientalis in blood smears (19, 24, 25), and/or the use of serological (26) or conventional molecular techniques (7, 27, 28). Each of these approaches has limitations. For example, clinical diagnosis is subjective and usually requires further laboratory investigations to confirm the presence of infection/disease. Microscopy is commonly used and involves the detection of T. orientalis piroplasms in blood smears. Although microscopy might be used to quantify the level of parasitemia (28), it is relatively time-consuming and inaccurate and does not provide any genetic information on the parasite. Serological tests can detect anti-T. orientalis antibodies early in an infection (29), but there are issues with immunological cross-reactivity among genotypes of T. orientalis (G. J. Eamens, personal communication), and it is not possible to unequivocally differentiate among exposure, current infection, and past infection by Theileria spp. (30). Conventional PCR techniques can be more sensitive than the aforementioned methods; however, their diagnostic performance can be affected by blood constituents (e.g., hemoglobin and lactoferrin) that are inhibitory to PCR, and they do not allow the quantitation of parasites (21, 31–33). Some of these issues can be overcome using real-time PCR assays, which allow the relative or absolute quantification of the parasites present in blood (34). Such assays have been developed for Theileria sergenti (35), T. parva (36, 37), and T. equi (38, 39) but have not yet been established for members of the T. orientalis complex.

A real-time PCR method that shows major promise is multiplexed tandem PCR (MT-PCR) (40). This technique can use multiple primer pairs for the detection of multiple pathogens. It consists of two amplifications: (i) multiplexed amplification (primary target enrichment), which involves a small number of PCR cycles and multiplexed or outer primer sets, and (ii) a subsequent quantification amplification which utilizes a diluted product from the primary amplification as a template and specific, nested, or inner primers (40). Although MT-PCR was originally developed to quantify gene transcription (40), MT-PCR has been applied to the sensitive and simultaneous detection of some fungi, such as Candida spp. (41, 42), enteric pathogens of humans (43, 44), gastrointestinal nematodes of sheep (45), and toxigenic cyanobacteria (46). As two genotypes of the T. orientalis complex (i.e., Chitose and Ikeda) are presently recognized to relate to clinical disease, there is a need to identify and differentiate each of them from nonpathogenic genotypes (i.e., Buffeli and type 5) of T. orientalis known to occur in southeast Australia (21). MT-PCR could offer a useful means of achieving such differential diagnosis as well as estimating the infection intensities of individual T. orientalis genotypes in bovines.

The aim of the present study was to establish and evaluate an MT-PCR assay for the simultaneous detection and differentiation of the four distinct genotypes, Buffeli, Chitose, Ikeda, and type 5, representing the T. orientalis complex known to occur in Australasia as well as for the semiquantitation of DNA of each of these genotypes in blood samples from cattle.

MATERIALS AND METHODS

Blood and genomic DNA samples.

Blood samples were available from 200 cattle (group 1; symptomatic or asymptomatic animals) from a previous study from 19 farms on which clinical outbreaks of oriental theileriosis were recorded (Table 1) (21). These blood samples had already been characterized using a conventional PCR-based approach (i.e., 170 test-positive samples, 20 samples showing PCR inhibition [using 2 μl of template], and 10 previously test-negative samples). In addition, blood samples were collected from 110 cattle (group 2) from the coccygeal vein (using an 18-gauge needle) into EDTA tubes by registered, practicing veterinarians (see Acknowledgments) from the Western District in Victoria, a region in which no outbreaks of oriental theileriosis and/or T. orientalis infections have been reported to date (Table 1). Genomic DNAs were extracted from individual blood samples (200 μl) using a DNeasy blood and tissue kit (catalog no. 69506; Qiagen, USA), according to the manufacturer's protocol, and eluted in 100 μl. In addition, genomic DNAs of other common blood parasites of cattle, including T. parva, T. annulata, Babesia bovis, and Anaplasma centrale, were available from colleagues (see Acknowledgments).

TABLE 1.

Demographic and characteristics of cattle farms selected for this study from various locations in Victoria

Farm condition and no.	Location	Geographical coordinates	Farm enterprise			Cattle breed(s)	Sample collection date (day/mo/yr)	No. of individuals tested
			Beef	Dairy	Mixed
With oriental theileriosis outbreaks
1	Bairnsdale	37°82′ S, 147°62′ E	−	−	+	Mixed beef and dairy breeds	3/14/2012	11
2	Balmattum	36°65′ S, 145°64′ E	+	−	−	Mixed beef breeds, including Brangus	3/26/2012	8
3	Bena farm 1	38° 41′ S, 145° 76′ E	−	+	−	Friesian	3/14/2012	9
4	Bena farm 2	38 41′ S, 145°76′ E	−	+	−	Friesian	3/14/2012	18
5	Bena farm 3	38° 41′ S, 145° 76′ E	−	+	−	Friesian	3/4/2012	4
6	Benallab	36°55′ S, 145°98′E				Angus	7/1/2012	3
7	Bete Bolong	37°69′ S, 148°39′ E	−	+	−	Friesian	3/6/2012	21
8	Bethanga	36°12′ S, 147°09′ E	+	−	−	Angus	3/8/2012	10
9	Bunyip	38°09′ S, 145°72′ E	+	−	−	Angus × Belgian blue	3/20/2012	16
10	Corryong	36°19′ S, 147°91′ E	+	−	−	Angus	4/3/2012	10
11	East Gippsland	37° 45′ S, 148° 18′E	+	−	−	Angus	3/8/2012	9
12	Freeburgh	36°76′ S, 147°03′ E	+	−	−	Angus	3/20/2012	3
13	Girgarre	36°40′ S, 144°98′E	−	+	−	Illawarra Shothorn	3/28/2012	3
14	Katandra	36°24′ S, 145°63′ E	−	+	−	Holstein	4/10/2012	12
15	Orbost	37° 71′ S, 148° 45′ E	−	−	+	Angus	3/6/2012	9
16	Pranjip	36°76′ S, 145°39′ E	+	−	−	Hereford	3/28/2012	20
17	Staghorn	36°24′ S, 146°93′ E	+	−	−	Hereford	3/5/2012	7
18	Tallangatta	36°28′ S, 147°43′ E	+	−	−	Simmental	5/3/2012	18
19	Warragul	38°16′ S, 145°93′ E	+	−	−	Angus	6/3/2012	9
With no history of oriental theileriosisa
20	Curdievale	38°51′ S, 142°88′E	−	+	−	Friesian and Holstein	11/21/2013	21
21	Jancourt East	38°41′ S, 143°13′ E	−	+	−	Friesian and Holstein	2/4/2014	27
22	Princetown	38° 64′ S, 143° 21′ E	−	+	−	Holstein	2/4/2014	29
23	Timboon West	38°56′ S, 142°92′ E	−	+	−	Holstein and Jersey	11/21/2013	33

^aWestern District of Victoria.

^bEpidemiological data for this farm could not be collected.

MT-PCR.

The Easy-Plex platform (AusDiagnostics Pty., Ltd., Australia) was used, which includes a Rotor-Gene 6000 real-time PCR thermocycler (Qiagen, Germany) and a Gene-Plex CAS1212 liquid handling robot (AusDiagnostics). The primary amplification (target enrichment) was conducted using primer pairs designed to the sequences of the ITS-1 and the p23 gene of T. orientalis for genotypes Ikeda and Buffeli, respectively, and to the MPSP gene for both Chitose and type 5. Current information on the T. orientalis genome (Japanese Ikeda strain) indicates that it has two copies of ITS-1 and one copy each of the p23 and MPSP genes (47). The secondary amplification for semiquantification used nested primer pairs to internal regions of these loci (catalog no. 4023; AusDiagnostics); these internal primer pairs amplify a region of 107 bp from ITS-1 (genotype Ikeda), a region of 115 bp from the p23 gene, and regions of 70 to 112 bp from the MPSP gene (Chitose and type 5). In addition, an independent primer pair is included in each reaction mixture as a reference for quantitation and to assess the efficiency of amplification from 10,000 copies of a synthetic oligonucleotide template (internal spike control).

The final protocol was as follows: for primary amplification (15 cycles of 10 s at 95°C, 20 s at 60°C, and 20 s at 72°C), 5 μl of genomic DNA representing each test sample or 5 μl of water (negative control) was dispensed into 0.2-ml PCR strips and placed into a 24-well thermocycling block within the Gene-Plex robotic platform. Following the dispensing of each sample and the initiation of the assay, the following setup process and analysis were executed by the program Easy-Plex Assay Setup (AusDiagnostics), with the secondary amplification in MT-PCR and the melting curve analysis being semiautomated (44, 48). A sample was recorded as test positive using the auto-call function of the Easy-Plex software (AusDiagnostics) if the amplicon produced a single melting curve which was within 1.5°C of the expected melting temperature, the height of the peak was higher than 0.2 dF/dT (where dF/dT is the derivative of fluorescence over temperature), and the peak width was ≤3.5°C. Cycle threshold (C_T) values were recorded for each test-positive sample, and the DNA copy number for each genotype in each sample was determined by comparison with C_T data determined for an internal spike control (40) for each sample tested. In instances (n = 9) where the internal spike control did not reach the expected DNA copy number of 10,000, the genomic DNA sample was diluted to 1:10 or 1:100 and retested, and the DNA copy number was calculated for the undiluted sample. Using this protocol, a minimum of 2.5 DNA copies (1 fg) could be detected. Finally, genotypes Buffeli, Chitose, Ikeda, and type 5 were assigned according to their mean (± standard deviation) expected peak melting temperatures of 83.6 ± 1.5°C, 82.1 ± 1.5°C, 87.4 ± 1.5°C, and 81.6 ± 1.5°C, respectively. The DNA copy number determined can be used as a measurement of the intensity of infection for each genotype. The relative intensities of infection by genotypes Buffeli, Chitose, and type 5 were estimated as the DNA copy number recorded for individual genotypes, while relative intensity of infection by genotype Ikeda was estimated by dividing the DNA copy number recorded by 2. In order to verify the specificity of MT-PCR as well as to assess nucleotide variation among amplicons in relation to peak melting temperature, selected samples (n = 100) were subjected to single-strand conformation polymorphism analysis (SSCP) and sequenced (n = 10) using an established cloning-based protocol (49).

Statistical analyses.

To assess repeatability of the MT-PCR assay, the coefficient of variation (CV) was estimated using the program Microsoft Excel (2010). Owing to a positive skew, copy number data were log transformed and are presented as medians and geometric means (the back-transformed mean of the log-transformed copy number estimates) according to the following equation:

$$\text{geometric mean}=\text{exp}\left[\frac{1}{n}{\displaystyle \sum _j\text{ln}\left(x_j\right)}\right]$$

where n is the sample size, j is 1, …, n, and x_j is the jth data value.

Pairwise comparisons of the relative intensity of each genotype in mixed genotypic infections were conducted (using Ikeda as the reference). For samples in which two genotypes were present (e.g., Buffeli and Ikeda), pairwise comparisons were conducted with paired-sample t tests of the geometric mean copy numbers, whereas for infections of more than two genotypes (e.g., Buffeli, Chitose, and Ikeda), linear mixed models were used to estimate the difference in geometric means using the program Stata (release 13; StataCorp LP, College Station, TX) by incorporating a random effect term to account for nonindependent observations (i.e., multiple genotypes in each individual). Models were of the following form: log₁₀(gene copy number) = β₀ + β₁ · Chitose + β₂ · Buffeli + β₃ · type 5 + Individual_j + ε, where β₀ is an intercept which can be interpreted as the expected geometric mean copy number for the reference category (Ikeda), and β₁, β₂, and β₃ are regression coefficients for categorical variables and may be interpreted as the difference in geometric mean copy number between genotype Ikeda and the genotypes Buffeli, Chitose, and type 5, respectively. We assumed that Individual_j, which is the random effect term for the jth of N individual cows (where N is the total sample size of the study), was normally distributed, along with the residual error (ε), with a standard deviation of S_individual according to the following: Individual_j ∼ Normal(0, S_individual), for j = 1, …, N.

The diagnostic specificity and sensitivity of the MT-PCR were estimated following the recommended Bayesian latent class modeling approach (50, 51) for two conditionally dependent tests on two populations (groups 1 and 2) in the absence of a gold standard (i.e., reference samples of known disease status). Conventional PCR cannot be considered a gold standard because it has been shown that the analytical sensitivity of the MT-PCR assay was 1,000 times higher than that of conventional PCR. Conventional PCR is a suitable diagnostic technique to detect T. orientalis. However, in MT-PCR, depending on the selected cutoff DNA copy number, a higher diagnostic sensitivity or higher diagnostic specificity than that of conventional PCR can be achieved. The Bayesian latent class modeling approach makes no assumptions about the status of animals from the two populations (groups 1 and 2). Prevalence was assumed to be distinct in each population, and diagnostic specificity and sensitivity were assumed to be constant across the two populations. The tests were assumed to be dependent (conditional on infection status) because they had the same biological basis, that is, the detection of nucleic acids of genotypes of T. orientalis. Prior information about the diagnostic specificity and sensitivity of the MT-PCR assay was modeled using independent and informative beta distributions elicited from a technical expert (R. B. Gasser) with knowledge of the populations and test performance yet not involved in the sample collection or testing (52). The most likely (modal) value and the (α − 100)th percentile of the corresponding beta distribution were elicited by asking the expert to specify that he was (100 − α)% sure that the diagnostic sensitivity of the MT-PCR was >X, and the most likely value for this parameter was Y (51). Prior information was similarly elicited for the prevalence in each population, while diagnostic specificity and sensitivity of the conventional PCR assay were specified as diffuse priors based on elicited modal values only, following Branscum et al. (51). Dependence parameters were specified as uninformed independent uniform distributions, and Bayesian inferences were based on the joint posterior distribution, numerically approximated using the program WinBUGS (53), running 110,000 model iterations, discarding the first 10,000 iterations as burn-in, and thinning by 10 to minimize auto-correlation. Agreement statistics (prevalence-adjusted bias-adjusted kappa [PABAK]) (54) were directly calculated as model outputs. Final inferences were presented as the 50%, 2.5%, and 97.5% quantiles of the marginal posterior distributions for each of the parameters, corresponding to a posterior median point estimate and 95% probability interval (PI), respectively.

Analyses were repeated by applying different DNA copy number cutoff values for dichotomizing the MT-PCR results as test positive, which enabled estimations of the two-way receiver-operator-characteristic (ROC) curve and optimal cutoff. Sensitivity analyses were performed as recommended previously (50, 52) to test for the influence of elicited priors on the final results, inputting vague (flat) priors, and comparing all model outputs.

RESULTS

Establishment of the MT-PCR assay.

In setting up the MT-PCR assay, a series of experiments was conducted to establish the optimum cycling protocol, the specificity and sensitivity of the MT-PCR, and the repeatability of results. The analytical specificities of individual primer sets (genotypes Buffeli, Chitose, Ikeda, and type 5 of the T. orientalis complex) were assessed using well-defined genomic DNA samples representing each of the four genotypes (positive controls; n = 4) (from Perera et al. [21]) as well as from T. annulata, T. parva, A. centrale, and B. bovis (negative controls; n = 4). Each of the four primer sets designed and tested amplified products exclusively from the expected genotypes (Fig. 1A and B). The identity of individual products was confirmed by SSCP analysis and sequencing, and no products were amplified from T. annulata, T. parva, A. centrale, or B. bovis DNA. Using the same, well-defined samples, repeatability of the copy number was greater within a run (CV of 12%) than among runs (CV of 26%), and genotypes were always correctly assigned (CV of 0%) for samples with ≥30 DNA copies.

FIG 1.

Detection of various genotypes of the Theileria orientalis complex using the MT-PCR assay. Cycling (A) and melting (B) curves for the genotypes Buffeli, Chitose, Ikeda, and type 5 of T. orientalis are shown. (C) Cycling curves of a blood DNA sample showing partial inhibition/delayed amplification of the spike control. (D and E) Cycling curves using undiluted template and showing no inhibition when the template was diluted at 1:10 (D) and 1:100 (E).

Validation of the MT-PCR assay.

Two hundred DNA samples representing cattle from 19 farms on which oriental theileriosis outbreaks had occurred (group 1) and 110 samples representing cattle farms where no outbreaks had occurred in Victoria (group 2) were tested in MT-PCR. Of the 200 samples from cattle in group 1, all genomic DNA samples that tested positive in a previous conventional PCR study (21) also tested positive (>0 DNA copies) by MT-PCR (n = 170). In addition, 17 of the samples that showed PCR inhibition (using 2 μl of template) in conventional PCR (n = 20) (21) did not inhibit MT-PCR. Of 10 samples that were previously negative by conventional PCR (21), eight samples tested positive by MT-PCR, with all eight positive samples containing 4 to 15 DNA copies. In addition, of 110 samples from group 2, 2 tested positive for T. orientalis by both MT-PCR and conventional PCR, and a further 6 samples tested positive by MT-PCR only. Of all 200 samples from group 1, 9 samples showed inhibition using 5 μl of template but did not when the original template was diluted to 1:10 or 1:100 and retested (Fig. 1C to E). SSCP analysis of 100 amplicons representing all four genotypes (Buffeli, n = 30; Chitose, n = 25; Ikeda, n = 30; type 5, n = 15) of T. orientalis revealed four main profiles (see Fig. S1 in the supplemental material); minor SSCP profile variation was repeatedly observed within genotypes Buffeli and Chitose, which was reflected in differences in the peak melting temperatures (0.9 to 1.0°C). DNA sequencing of amplicons revealed that nucleotide variation of 1.4 to 1.7% was associated with these differences (data not shown).

Of 200 blood samples collected from group 1, 198 tested positive in MT-PCR (applying a cutoff >0 DNA copy number detected). In this group, the prevalences of individual genotypes (i.e., Buffeli, Chitose, Ikeda, and type 5) of the T. orientalis complex in cattle included in these outbreaks were 92.9% (184/198), 57.1% (113/198), 95.5% (189/198), and 32.3% (64/198), respectively. The prevalence of T. orientalis infections with single or mixed genotypes detected is shown in Fig. 2. The number of infections with mixed genotypes was higher (93.43%; 185/198) than those with single genotypes (6.57%; 13/198). There was a high prevalence (38.9%) of mixed infections with genotypes Buffeli and Ikeda, followed by infection with all four genotypes (31.3%) and with genotypes Buffeli, Chitose, and Ikeda (21.2%) (Fig. 2). Most of the oriental theileriosis outbreaks (31.6%; 6/19) had a high prevalence of genotype Ikeda, followed by infection with genotype Buffeli (Table 2). Ten of 19 farms had a prevalence of 100% for genotype Ikeda. Type 5 showed the lowest prevalence among the four genotypes (Table 2). Compared with other regions, a comparatively high average relative intensity of infection by genotype Ikeda was recorded in Bairnsdale, Balmattum, Benalla, Bena farm 1, Bethanga, Bunyip, Corryong, Katandra, and Tallangatta, where deaths and/or abortions were reported (Table 2).

FIG 2.

Prevalence of genotypes of the Theileria orientalis complex detected by the MT-PCR assay. Letters C, B, I, and T denote single infections by genotypes Chitose, Buffeli, Ikeda, and type 5, respectively. Various combinations of letters with the plus sign denote mixed infections with two or more genotypes.

TABLE 2.

Numbers of cows that died or aborted and average relative intensity of infection by genotypes of Theileria orientalis in each outbreak at each location (farm)

Farm no.	Location (n)a	No. of cows that died due to theileriosis	No. of cows that aborted due to theileriosis	Prevalence of genotype (%)				Avg intensity of infection by genotype (DNA copies)
				Ikeda	Chitose	Buffeli	Type 5	Ikeda	Chitose	Buffeli	Type 5
1	Bairnsdale (11)	1	0	100	36.4	100	9.1	165,636	14,797	110,165	9
2	Balmattum (8)	1	2	87.5	37.5	100	37.5	120,177	107,017	102,661	4,355
3	Bena farm 1 (9)	16	6	77.8	0	88.9	11.1	65,023	0	21,800	1,000
4	Bena farm 2 (18)	1	0	94.4	88.9	94.4	55.6	238,319	270,6797	106,535	16
5	Bena farm 3 (4)	0	1	100	0	75	0	38,514	0	15,145	0
6	Benalla (3)b			33.3	100	33.3	0	20,689	9	10,750	0
7	Bete Bolong (21)	7	0	100	90.5	100	42.9	59,959	22,843	137,234	33,42
8	Bethanga (10)	4	4	100	10	100	0	195,658	15	72,115	0
9	Bunyip (16)	5	0	100	6.3	100	12.5	150,092	6,033	106,745	1,602
10	Corryong (10)	4	0	100	20	90	10	50,331	6,305	16,673	9
11	East Gippsland (9)	22	12	88.9	88.9	88.9	55.6	166,249	221,099	505,150	2,163
12	Freeburgh (3)	0	1	66.7	33.3	0	0	8	16	0	0
13	Girgarre (3)	0	3	66.7	66.7	66.7	0	84,918	108,368	302,627	0
14	Katandra (12)	6	0	100	58.3	83.3	16.7	110,907	97,482	47,436	20
15	Orbost (9)	1	0	100	100	100	100	24,857	156,553	338,766	44,072
16	Pranjip (20)	0	2	100	100	100	95	111,315	186,715	115,938	8,403
17	Staghorn (7)	2	1	100	14.3	85.7	14.3	3,268	108,368	40,302	16,847
18	Tallangatta (18)	1	1	94.4	50	94.4	5.6	191,611	147,184	120,811	201
19	Warragul (9)	0	0	88.9	77.8	88.9	0	24,587	770	27,561	0

^an, number of cows.

^bEpidemiological data for this farm could not be collected.

Although all four genotypes were detected in cattle experiencing clinical oriental theileriosis, the relative intensity of infection by each of these genotypes showed that genotypes Ikeda and Buffeli dominated the other two genotypes (Chitose and type 5) (Table 3). For the most prevalent, mixed infections (i.e., with genotypes Buffeli and Ikeda), the genotype Ikeda showed a significantly higher relative intensity of infection than the genotype Buffeli (P < 0.001) (Table 3; Fig. 3). Genotype Ikeda was significantly more dominant (P < 0.001) than genotype Chitose in mixed infections with genotypes Buffeli, Chitose, and Ikeda. Of 110 DNA samples from group 2, eight samples tested positive in MT-PCR; four had single infections with genotype Buffeli (copy number range, 11 to 26), and four had mixed infections with genotypes Buffeli (copy number range, 5 to 30,019) and Ikeda (copy number range, 3 to 90,547).

TABLE 3.

Relative intensity of infection (DNA copies) by genotypes of Theileria orientalis in regions where outbreaks occurred

Type of mixed infection (n)a	Genotype	Category	Median DNA copy no. (min, max)b	Geometric mean DNA copy no.	Difference of geometric mean DNA copy no. (95% CI)c	P valued
Ikeda + Chitose (3)	Ikeda	0	8 (8, 2,500)
	Chitose	1	19 (8, 9,000)
Ikeda + Buffeli (77)	Ikeda	0	22,061 (3, 706,132)	11,599	−6,744 (−8,063, −4,933)	<0.001
	Buffeli	1	10,000 (4, 792,570)	4,855
Ikeda + Chitose + Buffeli (42)	Ikeda	0	28,577 (6, 621,062)	16,604	0
	Chitose	1	207 (10, 266,168)	460	−16,144 (−16,450, −15,233)	<0.001
	Buffeli	2	24,812 (11, 1,324,465)	21,510	4,906 (−9,389, 47,519)	0.642
Ikeda + Buffeli + type 5 (1)	Ikeda	0	8,424
	Buffeli	1	21,778
	Type 5	2	3,175
Ikeda + Chitose + Buffeli + type 5 (62)	Ikeda	0	40,958 (6, 7,06,132)	38,004	0
	Chitose	1	76,866 (8, 1,412,263)	38,815	811 (−19,553, 43,651)	0.97
	Buffeli	2	89,387 (9, 1,412,263)	67,702	29,698 (−5,821, 104,419)	0.13
	Type 5	3	35 (5, 249,621)	217	−37,787 (−37,900, −37,548)	<0.001

^an, number of animals.

^bmin, minimum; max, maximum.

^cCI, confidence interval.

^dP values were obtained by comparing each category with the reference group (i.e., category 0). For three mixed infections, data were analyzed by pair-wise comparisons using genotype Ikeda as the reference category. Statistical analysis for two mixed infections was not performed as the sample size was low. Results were estimated using linear mixed models, adjusting for variability among samples. Significant values are in boldface.

FIG 3.

Box plot diagrams showing the number of DNA copies of genotypes in mixed infections as indicated on each panel. The DNA copy number recorded for genotype Ikeda was divided by 2 to determine the DNA copy numbers shown in the figure.

The diagnostic specificity of the MT-PCR (94.0%; 95% PI, 90.1 to 96.8%) was lower than that of the conventional PCR (96.8%; 95% PI, 93.0 to 98.8%); the diagnostic sensitivity of the MT-PCR was 98.8% (95% PI, 96.7 to 99.7%) if test positivity was defined based on a cutoff of >0 DNA copies, compared with 95.1% (95% PI, 91.6 to 97.5%) for the conventional PCR. When the MT-PCR was interpreted using the test-positive cutoff of >20 DNA copies (Fig. 4), diagnostic performance was equivalent to that of the conventional PCR (see Fig. S2 and Table S1 in the supplemental material). There was excellent agreement between the two diagnostic tests in both groups of samples (posterior median PABAK of >0.864 in all iterations), and the prevalence estimates in the two populations were relatively stable (group 1, >95.1%; group 2, 1.3%) even as the MT-PCR cutoff was altered. Changes in inference were negligible when the Bayesian latent class model was populated with flat (uninformative) priors.

FIG 4.

Diagnostic sensitivity and specificity of the MT-PCR at different cutoff points.

DISCUSSION

The present study established and validated an MT-PCR assay for the detection, differentiation, and semiquantitation of four genotypes (i.e., Buffeli, Chitose, Ikeda, and type 5) of the T. orientalis complex in Australia in blood samples from cattle. Bayesian latent class analysis estimated that 95% of 200 cattle specifically selected from infected farms and only 1.3% of 110 cattle from farms from an area in Victoria (Western District) where theileriosis cases have not been reported tested positive for one or more of the four genotypes. Moreover, the levels of parasite DNA in blood were substantially higher (30 times) in most cattle in the region of endemicity (group 1) than in the eight cattle from the Western District of Victoria (group 2) that tested positive in the MT-PCR (see Table S2 in the supplemental material). It is possible that these test-positive cattle were recently introduced into this district as cattle transport from regions of endemicity to regions where the parasite is not endemic within Victoria as well as from New South Wales is common (Department of Primary Industries, New South Wales Government [www.dpi.nsw.gov.au]). Conventional PCR (21) detected DNA of T. orientalis in only two of the eight cattle with the highest intensity of infection inferred from the MT-PCR.

Electrophoretic mutation scanning analysis and targeted sequencing demonstrated specificity for all four sets of primers, all of the amplicons produced, and the conditions of MT-PCR. In addition, the DNA samples from four heterologous blood pathogens (T. annulata, T. parva, A. centrale, and B. bovis) tested were, as expected, all negative. Nonetheless, future studies should reevaluate the specificity of the MT-PCR assay in regions where other blood-borne bovine pathogens are endemic (including viruses and bacteria). Although intended for genotypic detection/differentiation and semiquantitation, the present MT-PCR assay might also be useful as a mutation scanning tool to detect genetic variability within individual genotypes of T. orientalis because SSCP-coupled sequencing was able to show that subtle variation (∼0.9°C) in peak melting temperature linked to sequence difference of 1.4 to 1.7% (one or two nucleotide alterations) in loci for the Buffeli and Chitose types was readily detectable.

The minimum amount of DNA detectable (i.e., 1 fg or 2.5 DNA copies) by MT-PCR was comparable to that of previous studies using the same platform (44, 46), and the test was approximately 1,000 times more sensitive than conventional PCR (21). Given the ability of the MT-PCR to detect ≥1 fg of T. orientalis DNA, the present study has shown that most infections are multigenotypic, in contrast to previous results achieved by conventional (one-step) PCRs. The performance of MT-PCR is comparable to or better than that reported for some real-time TaqMan PCRs established for T. equi and T. sergenti (35–39). In our MT-PCR, the DNA copy number estimate for each genotype and sample relative to the internal spike control (i.e., 10,000 DNA copies of a synthetic oligonucleotide template amplified by specific primers) is likely to be more accurate and repeatable than for other assays used previously. The DNA copy number determined can be used as a measurement of the intensity of infection for each genotype. Given that high DNA copy numbers of pathogenic genotypes (Chitose and Ikeda) of T. orientalis in cattle might relate to disease (our unpublished data), the ability to estimate intensity could be useful to predict the risk of an outbreak, but this proposal warrants testing.

Coinfections with multiple T. orientalis genotypes were commonly detected by MT-PCR, consistent with previous studies in Australia (13, 20, 21, 23). However, here, Ikeda was the most common genotype, followed by the Buffeli, Chitose, and type 5 genotypes, in contrast to previous evidence showing that Chitose was the second most prevalent genotype (21–23). This difference in prevalence is likely due to the ability of the present MT-PCR to detect tiny amounts (≥1 fg) of parasite DNA compared with conventional PCRs (21–23). The sensitivity of the MT-PCR assay also explains why the prevalence of the Buffeli type was higher than recorded in previous studies (20, 21) and also provides additional support for the proposal that the Buffeli genotype is endemic in Australia (55, 56); however, a large-scale nation-wide survey would be needed to establish the geographical distribution of different genotypes of the T. orientalis complex. Currently, the MT-PCR assay has been designed for the four genotypes of T. orientalis known to occur in Australia (20, 21, 23). The assay could be readily modified to include loci or gene regions for genotypes not included in the present assay, provided that the markers to be used have been prevalidated for specificity to detect additional genotypes prior to their inclusion in the assay.

In conclusion, the semiautomated MT-PCR assay established here is a cost-effective, time-efficient, and practical diagnostic tool. It provides a major advance because it allows a qualitative and quantitative evaluation of four distinct genotypes of T. orientalis at once. Currently, the estimated cost per sample is Aus$19, which is approximately half that of our conventional PCR-based testing (21, 23), and the time required from sample preparation to test result is about one-fifth (about 1 day) of that using the conventional approach. In our opinion, the MT-PCR assay has broad applicability and can now be utilized to support investigations into the epidemiology, pathophysiology, and transmission of oriental theileriosis. For example, the assay could be readily used to explore the temporal changes in genotypes that occur within individual cattle (proposed by Perera et al. [22]), population dynamics suggested to occur during transmission from cattle to ticks and vice versa (57), and/or to test the hypothesis that definitive and intermediate hosts other than cattle and Haemaphysalis, respectively, are involved in disease spread (58, 59). For instance, it would be interesting to explore whether water buffaloes or deer might act as reservoir hosts. Importantly, the present MT-PCR assay will be useful for the surveillance and monitoring of oriental theileriosis in Australasia and should be readily applicable in other countries in the Asia-Pacific region where this disease significantly impacts livestock health, welfare, and production.