INTRODUCTION
The order Piroplasmida (Phylum Apicomplexa) is mainly composed of the genera Babesia and Theileria (Levine et al. Reference Levine, Corliss, Cox, Deroux, Grain, Honigberg, Leedale, Loeblich, Lom, Lynn, Merinfeld, Page, Poljansky, Sprague, Vavra and Wallace1980; Criado et al. Reference Criado, Martinez, Buling, Barba, Merino, Jefferies and Irwin2006). Classification of this order has been historically confined to parasite morphology, host cells in which schizogony occurs, levels of parasitaemia in the schizont and piroplasm stages, manifestation and pathology of clinical disease, as well as host and vector specificity (Barrnett, Reference Barnett and Kreier1977; Uilenberg, Reference Uilenberg2006). By these criteria, the genus Theileria is distinguished by its ability to infect host leukocytes and to multiply by schizogony as part of its life cycle (Neitz, Reference Neitz1957; Uilenberg, Reference Uilenberg2006). Theileria species were first discovered based on clinical disease manifestation, linked with the observation of piroplasms in red blood cells and xenodiagnosis by tick transmission, as was the case for Theileria parva that caused Corridor disease and East Coast fever and T. annulata the causative agent of Tropical Theileriosis (Koch, Reference Koch1903; Theiler, Reference Theiler1904; Dschunkowsky and Luhs, Reference Dschunkowsky and Luhs1904; Lawrence, Reference Lawrence1935). Observation of morphologically distinct piroplasm forms, with or without mild disease manifestation soon led to the discovery of more benign Theileria, such as T. mutans and T. velifera in cattle (Theiler, Reference Theiler1906; Uilenberg, Reference Uilenberg1964). New species were also discovered based on their presence in specific hosts, differential staining of blood smears and xeno-diagnosis, as in the case of T. taurotragi found in eland (Martin and Brocklesby, Reference Martin and Brocklesby1960; Brocklesby, Reference Brocklesby1962; Young et al. Reference Young, Grootenhuis, Kimber, Kanhai and Stagg1977).
Serology in the form of the indirect immuno-fluorescent antibody test (IFAT) was employed to distinguish species infecting the same host, but with different clinical manifestations, as is the case for T. parva and T. taurotragi (Young et al. Reference Young, Grootenhuis, Kimber, Kanhai and Stagg1977). The use of monoclonal antibodies and clonal cell culturing methods led to the discovery of antigenically distinct parasites from Cape buffalo (Syncerus caffer), thought to be T. parva (Conrad et al. Reference Conrad, Stagg, Grootenhuis, Irvin, Newson, Njamunggeh, Rossiter and Young1987). Sequencing of the 18S ribosomal RNA (rRNA) gene showed, however, that T. sp. (buffalo) was a distinct entity from T. parva (Allsopp et al. Reference Allsopp, Baylis, Allsopp, Cavalier-Smith, Bishop, Carrington, Sohanpal and Spooner1993). Subsequently, the systematics of the Piroplasmida has been extensively studied by analysis of the 18S rRNA gene (Allsopp et al. Reference Allsopp, Cavalier-Smith, De Waal and Allsopp1994; Chae et al. Reference Chae, Allsopp, Waghela, Park, Kakuda, Sugimoto, Allsopp, Wagner and Holman1999; Chansiri et al. Reference Chansiri, Kawazu, Kamio, Terada, Fujisaki, Philippe and Sarataphan1999; Criado-Fornelio et al. Reference Criado-Fornelio, Gónzalez-del-Río, Buling-Saraña and Barba-Carretero2004; Criado et al. Reference Criado, Martinez, Buling, Barba, Merino, Jefferies and Irwin2006; Reichard et al. Reference Reichard, Van Den Bussche, Meinkoth, Hoover and Kocan2005; Altay et al. Reference Altay, Dumanli and Aktas2007; Bhoora et al. Reference Bhoora, Franssen, Oosthuizen, Guthrie, Zweygarth, Penzhorn, Jongejan and Collins2009). Based on this, the reverse line blot is currently used for simultaneous detection of Babesia and Theileria species diversity within a vertebrate host (Gubbels et al. Reference Gubbels, de Vos, van der Weide, Viseras, Schouls, de Vries and Jongejan1999). This method depends on the amplification of a ∼400 bp fragment from the 18S rRNA hyper-variable region, using universal primers for Babesia and Theileria, followed by hybridization to a panel of probes specific for different Babesia and Theileria species (Gubbels et al. Reference Gubbels, de Vos, van der Weide, Viseras, Schouls, de Vries and Jongejan1999). A Babesia/Theileria genus catch-all is included to detect any species not recognized by the current probe set. This approach has been used successfully to identify new Babesia and Theileria species in antelope, black rhinoceros and dogs (Nijhof et al. Reference Nijhof, Penzhorn, Lynen, Mollel, Morkel, Bekker and Jongejan2003, Reference Nijhof, Pillay, Steyl, Prozesky, Stoltsz, Lawrence, Penzhorn and Jongejan2005; Matjila et al. Reference Matjila, Leisewitz, Jongejan, Bertschinger and Penzhorn2008; Oosthuizen et al. Reference Oosthuizen, Zweygarth, Collins, Troskie and Penzhorn2008).
The drawback of the reverse-line blot in the identification of novel species or genotypes is that it relies on animals infected with a single unique species. In the case of mixed infections, the presence of a positive signal for known species leads to the conclusion that no novel genotypes are present. This could be particularly relevant for buffalo and cattle, in which multiple Theileria spp. occurs (Young et al. Reference Young, Brown, Burridge, Grootenhuis, Kanhai, Purnell and Stagg1978a; Allsopp et al. Reference Allsopp, Theron, Coetzee, Dunsterville and Allsopp1999; Gubbels et al. Reference Gubbels, de Vos, van der Weide, Viseras, Schouls, de Vries and Jongejan1999). Random sequencing of the 18S SSU rRNA gene offers a way to obtain complementary results to the RLB for detecting Theileria species diversity. This approach does not depend on finding animals with single infections, while sequence coverage of an adequate sample set should give a reasonable estimate of Theileria sequence diversity.
T. parva is the causative agent of buffalo-derived Corridor disease and cattle-derived East Coast fever and Zimbabwe theileriosis (Norval et al. Reference Norval, Perry and Young1992). Whereas cattle-derived T. parva is considered to be eradicated in South Africa, Corridor disease is a controlled disease that is endemic in the Kruger National Park (KNP), Hluhluwe-Imfolozi Park (Kwa-Zulu Natal) and regions between and surrounding these areas (Potgieter et al. Reference Potgieter, Stoltsz, Blouin and Roos1988). The movement of Cape buffalo (Syncerus caffer) in South Africa is thus regulated using results from a real-time hybridization PCR assay. This assay depends on amplification of a fragment of the V4 hyper-variable region of the 18S SSU rRNA gene using a T. parva-‘specific’ forward and a Theileria genus-specific reverse primer set (Sibeko et al. Reference Sibeko, Oosthuizen, Collins, Geysen, Rambritch, Latif, Groeneveld, Potgieter and Coetzer2008). This primer set has been shown to amplify at least 3 Theileria spp. to date, T. parva, T. sp. (buffalo) and T. sp. (bougasvlei) (Sibeko et al. Reference Sibeko, Oosthuizen, Collins, Geysen, Rambritch, Latif, Groeneveld, Potgieter and Coetzer2008; Zweygarth et al. Reference Zweygarth, Koekemoer, Josemans, Rambritch, Pienaar, Putterill, Latif and Potgieter2009). It is therefore important to determine whether other T. parva-related species exist in buffalo and cattle and whether this would affect the current test. We addressed this by random sequencing of the 18S SSU rRNA hyper-variable region.
MATERIALS AND METHODS
DNA extraction and real-time hybridization and conventional PCR
Genomic DNA was extracted from 200 μl of whole blood using the MagNa Pure Large Volume DNA Isolation Kit and MagNa Pure LC (Roche Diagnostics) according the manufacturer's instructions. DNA was eluted in 100 μl of elution buffer and 2·5 μl was used for each real-time hybridization assay as described (Sibeko et al. Reference Sibeko, Oosthuizen, Collins, Geysen, Rambritch, Latif, Groeneveld, Potgieter and Coetzer2008).
Selection of buffalo and cattle samples for analysis
Forty-nine cattle from private farms or rural communities in the vicinity of game farms and 62 buffalo samples from privately owned game farms or nature reserves were chosen from samples submitted to the Parasites, Vectors and Vector-Borne Diseases laboratory for routine T. parva diagnostics using the real-time hybridization PCR assay (Tables 1 and 2). The current real-time hybridization assay for T. parva uses 2 hybridization probe sets, the first specific for T. parva detectable at 640 nm and the second a Theileria genus-specific probe detectable at 705 nm (Sibeko et al. Reference Sibeko, Oosthuizen, Collins, Geysen, Rambritch, Latif, Groeneveld, Potgieter and Coetzer2008). Samples negative for the 640 nm, but positive for the 705 nm probe set are designated as being T. parva negative, but ‘T. sp. (buffalo)-like’ positive (Sibeko et al. Reference Sibeko, Oosthuizen, Collins, Geysen, Rambritch, Latif, Groeneveld, Potgieter and Coetzer2008). This latter designation will include all Theileria sequences that amplify with the primer set that is not detectable with the 640 nm probe. As the drive for this study was primarily to determine the extent of closely related species to T. parva in southern African buffalo and cattle, samples were chosen based on their T. parva positive or ‘T. sp. (buffalo)-like’ positive status (Tables 1 and 2). Samples were also chosen if their melting profiles differed from that of the known T. parva-positive controls.
Table 1. Cattle samples amplified, cloned and sequenced
Table 2. Buffalo samples amplified, cloned and sequenced
Reverse-line blot
Reverse-line blot analysis was performed as described and modified (Gubbels et al. Reference Gubbels, de Vos, van der Weide, Viseras, Schouls, de Vries and Jongejan1999; Matjila et al. Reference Matjila, Penzhorn, Bekker, Nijhof and Jongejan2004), with the exception that the initial 10 cycles used for touchdown PCR decreased with 1°C steps instead of 2°C as reported (Matjila et al. Reference Matjila, Penzhorn, Bekker, Nijhof and Jongejan2004). Probes used included those used for the detection of Theileria species in southern Africa (Matjila et al. Reference Matjila, Leisewitz, Jongejan, Bertschinger and Penzhorn2008). These included probes for T. annulata, T. bicornis, T. buffeli, T. equi, T. mutans, T. parva, T. sp. (buffalo), T. sp. (duiker), T. sp. (kudu), T. sp. (sable), T. taurotragi and T. velifera, as well as a Theileria/Babesia catch-all.
Cloning and sequencing of 18S hyper-variable region
A 1101 bp fragment of the 18S rRNA gene was amplified using Theileria genus-specific primers designated as 989 and 990 by Allsopp et al. (Reference Allsopp, Baylis, Allsopp, Cavalier-Smith, Bishop, Carrington, Sohanpal and Spooner1993). Reaction conditions included the use of 25 μl GoTaq® Green Ready reaction mix (Promega), 2·5 μl of DNA template and 10pmol of each primer up to a total volume of 50 μl. Conditions for PCR included a pre-melt of 2 min (94°C), 40 cycles of melting at 94°C (30s), annealing at 52°C (30s) and extension at 72°C (120s), with a final extension for 7 min at 72°C. Amplified products were analysed by 1% agarose gel electrophoresis and 1 μl used for cloning into the pGEM®-T Easy cloning kit according to the manufacturer's instructions (Promega). Blue-white selection was used to identify clones with inserts, and clones were screened using the 989/990 primers. Amplified products (1101 bp) were purified for 10 positive clones from each sample using the DNA extraction kit (Fermentas). Purified fragments were sequenced using the BigDye® Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems) and the Theileria genus-specific primer 989. To confirm the novel genotypes observed, selected samples were re-amplified, cloned using high fidelity Pfu polymerase (Fermentas) and sequenced from both ends.
Bioinformatic analysis of the 18S gene
Sequences for each sample and a master sequence set that included known Theileria species sequences were aligned using a consideration of the secondary structure of RNA (Q-INS-i) as implemented in MAFFT (Katoh and Toh, 2007). Alignments were manually inspected and adjusted and edges were trimmed to give a ∼320 bp region that spanned the V4 hyper-variable region that was used for phylogenetic analysis. Neighbor-Joining analysis was performed using Mega4 software (Tamura et al. Reference Tamura, Dudley, Nei and Kumar2007). Gaps were treated as pairwise deletion and both transitions and transversions were included in the analysis. Due to the relatively limited number of phylogenetically informative positions available in the fragment analysed, pairwise-distances were calculated using the p-distance parameter, which was also used as nucleotide substitution model (Nei and Kumar, Reference Nei and Kumar2000). Patterns among lineages were assumed to be homogeneous and rates among sites to be uniform. Branch support was estimated using bootstrap analysis (100 000 replicates). Sequences assigned to putative species were aligned as a group and manually checked for single nucleotide polymorphisms.
Cloning and sequencing of the S5 gene
Primers for the S5 gene (S5F: ATG ACA AAC ACA GAA GTC GCC CT; S5R: ATT TCA TCC TTC TTC TTG ATT GCG T) were designed based on conserved regions between the T. annulata and T. parva S5 ribosomal genes that were published as part of their genome sequencing projects (Gardner et al. Reference Gardner, Bishop, Shah, de Villiers, Carlton, Hall, Ren, Paulsen, Pain, Berriman, Wilson, Sato, Ralph, Mann, Xiong, Shallom, Weidman, Jiang, Lynn, Weaver, Shoaibi, Domingo, Wasawo, Crabtree, Wortman, Haas, Angiuoli, Creasy, Lu, Suh, Silva, Utterback, Feldblyum, Pertea, Allen, Nierman, Taracha, Salzberg, White, Fitzhugh, Morzaria, Venter, Fraser and Nene2005; Pain et al. Reference Pain, Renauld, Berriman, Murphy, Yeats, Weir, Kerhornou, Aslett, Bishop, Bouchier, Cochet, Coulson, Cronin, de Villiers, Fraser, Fosker, Gardner, Goble, Griffiths-Jones, Harris, Katzer, Larke, Lord, Maser, McKellar, Mooney, Morton, Nene, O'Neil, Price, Quail, Rabbinowitsch, Rawlings, Rutter, Saunders, Seeger, Shah, Squares, Squares, Tivey, Walker, Woodward, Dobbelaere, Langsley, Rajandream, McKeever, Shiels, Tait, Barrell and Hall2005). Conventional PCR was performed using an initial denaturation step at 94°C (2 min), followed by 40 cycles of denaturation at 94 °C (30s), annealing at 53°C (30s) and extension at 72°C (60s). A final extension step at 72°C (7 min) was performed to ensure that all PCR products were extended. Reaction conditions included the use of 25 μl GoTaq® Green Ready reaction mix, 2·5 μl of DNA template and 10pmol of each primer up to a total volume of 50 μl. Amplified products (1 μl) were used for cloning into the pGEM®-T Easy vector as described above for the 18S gene. DNA products were purified for 10 positive clones and sequenced from both sides using the BigDye® Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems) and the M13 forward and reverse primers. Control samples chosen were previously confirmed by sequencing of the 18S gene and included T. taurotragi (18565), T. sp. (buffalo) (25317, 26705), T. sp. (bougasvlei) (26531, 27071), T. parva (KNP102, 9288, 9446, 9468), T. parva variant 1 (32806), T. parva variant 3 (34431) and T. lestoquardi (Morocco). Numbers in parentheses refer to laboratory numbers of samples submitted for diagnostic analysis.
Bioinformatic analysis of the S5 gene
Sequences were aligned using ClustalX (Jeanmougin et al. 1996) and manually checked. Particular attention was paid to the intron-exon boundaries and open reading-frames. Neighbor-joining analysis was performed with pairwise-deletion of gapped positions. The Tamura 3-parameter was used as substitution model which included transversions, transitions, first, second and third codon positions as well as the non-coding intronic regions. Patterns among lineages were treated as heterogenous and rates among sites (787 sites) as uniform. Branch confidence was estimated using bootstrap analysis for 100 000 replicates.
RESULTS
Detection of species diversity
Cloning and sequencing of 10 clones each of the 18S gene from 49 cattle and 62 buffalo samples yielded 463 and 598 high-quality sequences, respectively. Sequences were analysed by multiple sequence alignment and phylogenetic analysis using a master sequence set of known Theileria genotypes (Fig. 1). Twenty-six unique genotypes were found in buffalo and cattle and grouped into 5 major evolutionary groups or clades (Fig. 1). The ‘T. taurotragi’ group included T. parva (8 genotypes), T. sp. (bougasvlei), T. sp. (buffalo) (3 genotypes) and T. taurotragi. The Theileria antelope group included a single genotype, T. sp. (sable)-like found in a bovine. The ‘T. mutans’ group had 5 genotypes, while the ‘T. buffeli’ and ‘T. velifera’ groups, possessed 4 and 3 genotypes, respectively. Of these, T. taurotragi, T. sp. (sable)-like, T. buffeli-Warwick and T. buffeli-like B was found exclusively in cattle. T. buffeli-like C, T. buffeli D-like, T. mutans-like 1, T. mutans-like 2, T. mutans-like 3, T. sp. (bougasvlei), T. sp. (buffalo) and its variants, as well as T. velifera B were found exclusively in buffalo.
Fig. 1. Phylogenetic analysis of Theileria species and genotypic variants found in buffalo and cattle. Major phylogenetic groups are indicated by grey circles. The number of each species found by reverse-line blot analysis and the number of sequenced clones found in cattle (Ca) and buffalo (Bu) are indicated, as well as the number of animals (#Animals) from which these clones were derived. Unique genotypes identified in this study are underlined. Species are indicated by a 3–5 letter abbreviation, followed by their GenBank Accession code. The tree was rooted using Babesia microti as outgroup. Bootstrap support above 50% is indicated.
Species diversity in buffalo and cattle
The Theileria genotypic diversity found in buffalo, both in regard to number of genotypes harboured and number of co-infecting genotypes, are more extensive than in cattle when the random sequencing data are compared (Figs 1 and 2). For the random sequencing approach, a total of 16 genotypes (9 unique) were found in cattle, while 17 genotypes (10 unique) were found in buffalo (Fig. 1). In contrast, RLB analysis of the same sample set detected only 6 genotypes in cattle (T. buffeli, T. mutans, T. parva, T. sp. (sable), T. taurotragi, T. velifera) and 5 in buffalo (T. buffeli, T. mutans, T. parva, T. sp. (buffalo), T. velifera). This can be expected as the unique genotypes observed in this study are not included in the current RLB probe set.
Fig. 2. Number of Theileria genotypes found in buffalo and cattle by random sequencing and RLB. Indicated is the number of species found per animal against the total percentage.
For the sequenced samples up to 4 different Theileria genotypes were found per cattle sample, with the majority (94%) harbouring a maximum of 1–3 genotypes (Fig. 2). In contrast, buffalo carried up to 7 different genotypes per sample, with the majority (93%) harbouring between 2 and 6 genotypes. Up to 4 genotypes per sample were found in cattle and buffalo with the RLB. Reverse-line blot analysis of the cattle samples indicate that, in general, more genotypes are detected per sample than observed for the random sequencing approach with the majority (85%) harbouring 2–4 genotypes (Fig. 2). In the case of the buffalo, 45% of all samples harboured 4 genotypes.
T. sp. (sable)-like sequences in cattle
All clones sequenced for 1 bovine were identical to each other and similar (2 nucleotide differences) to sequences recently obtained from R. evertsi evertsi ticks collected from a gemsbok (Oryx gazella gazella) (Tonetti et al. Reference Tonetti, Berggoetz, Rühle, Pretorius and Gern2009). These sequences are related to T. sp. (sable) and T. sp. (roan) derived sequences and cross-hybridize with the RLB probe for T. sp. (sable) (Nijhof et al. Reference Nijhof, Pillay, Steyl, Prozesky, Stoltsz, Lawrence, Penzhorn and Jongejan2005). Differences across the fragment analysed (5 nucleotide differences) are large enough to be considered unique.
The T. buffeli clade
Four different genotypes were found within the T. buffeli group of which the complementary region to the RLB probe is more than 95% conserved (Fig. 3). However, significant differences exist upstream from this region to designate all as unique genotypes. T. buffeli D-like were found in 3 buffalo samples and is similar to T. buffeli type D and T. sp. (Thung Song) found in cattle in Thailand, the USA and Korea (Chae et al. Reference Chae, Allsopp, Waghela, Park, Kakuda, Sugimoto, Allsopp, Wagner and Holman1999; Chansiri et al. Reference Chansiri, Kawazu, Kamio, Terada, Fujisaki, Philippe and Sarataphan1999) and a sequence designated as T. sp. MK from sheep and goats in Turkey (Altay et al. Reference Altay, Dumanli and Aktas2007), but differed by at least 8 nucleotide substitutions. Twelve buffalo samples (T. buffeli-like C) presented sequences similar to unpublished T. buffeli sequences from China and India. Four cattle possessed a genotype (T. buffeli-like B) similar (1 nucleotide difference) to T. buffeli (Marula), previously found in a bovine in Kenya (Allsopp et al. Reference Allsopp, Cavalier-Smith, De Waal and Allsopp1994). Two cattle samples presented sequences identical to T. buffeli (Warwick) originally isolated from cattle in Australia (Stewart et al. Reference Stewart, De Vos, Shiels and McGregor1987; Chansiri et al. Reference Chansiri, Kawazu, Kamio, Terada, Fujisaki, Philippe and Sarataphan1999).
Fig. 3. Alignment of the 18S SSU RNA V4 hyper-variable region of the genotypes found in buffalo and cattle. Indicated are the regions for the Theileria parva ‘specific’ forward primer and the 640 anchor-probe pair for the real-time hybridization test for T. parva (Sibeko et al. Reference Sibeko, Oosthuizen, Collins, Geysen, Rambritch, Latif, Groeneveld, Potgieter and Coetzer2008). Reverse-line blot probe regions are underlined and grey shaded areas indicate positions that differ from the T. parva sequence. Accession codes for sequences can be found in Fig. 1.
The T. velifera clade
The sequence for T. velifera, as described from the original isolate obtained from cattle (Uilenberg and Schreuder, Reference Uilenberg and Schreuder1976; Gubbels et al. Reference Gubbels, de Vos, van der Weide, Viseras, Schouls, de Vries and Jongejan1999), was found to be prevalent in both cattle and buffalo (64 animals). Two additional genotypes differed from the published T. velifera sequence with 2 (T. velifera B) and 7 (T. velifera A) nucleotides, respectively. T. velifera B was limited to buffalo. In all genotypes the complementary region to the RLB probe for T. velifera is more than 90% conserved (Fig. 3).
The T. mutans clade
Sequences that were identical to the Intona strain of T. mutans from a bovine in Kenya and T. mutans MSD from a bovine in South Africa (De Vos and Roos, Reference De Vos and Roos1981; Chae et al. Reference Chae, Allsopp, Waghela, Park, Kakuda, Sugimoto, Allsopp, Wagner and Holman1999), were found in both buffalo and cattle. In addition, sequences limited to buffalo that are identical to 2 unpublished sequences (T. mutans-like1 and T. mutans-like 2) obtained from buffalo in the Kruger National Park, were found in 47 buffalo samples. A fourth unique genotype (T. mutans-like 3) was found in 22 buffalo samples.
The T. taurotragi clade
A well-supported clade is formed by the established species T. taurotragi, T. annulata, T. lestoquardi and T. parva, which have known vectors, host and disease aetiologies (Neitz, Reference Neitz1957; Young et al. Reference Young, Grootenhuis, Kimber, Kanhai and Stagg1977). Also found in this clade are the less well-characterized parasites, T. sp. (buffalo) and T. sp. (bougasvlei) known to occur in buffalo and thought to be apathogenic (Allsopp et al. Reference Allsopp, Baylis, Allsopp, Cavalier-Smith, Bishop, Carrington, Sohanpal and Spooner1993; Zweygarth et al. 2008). Of these, T. parva was found in buffalo and cattle, while T. taurotragi was limited to cattle. Conversely, T. sp. (buffalo) and T. sp. (bougasvlei) were only found in buffalo.
T. parva variants
Pairwise genetic distances based on the p–distance model of nucleotide substitution show a linear correlation with the number of pairwise nucleotide differences (Fig. 4). Genetic distances between well-described, established and closely related species show that the smallest genetic distance (0·01) found was between T. annulata/T. lestoquardi and T. parva/T. sp. (buffalo), respectively. This translates to 3 nucleotide differences between each pair, respectively for the ∼300 bp fragment analysed of the SSU 18S RNA gene. In the other evolutionary related groups much higher variation is observed. Genotypes from the T. mutans group had pairwise distances above 0·043, which corresponds to more than 13 nucleotide differences, while the T. buffeli group had distances above 0·017, which corresponds to 5 or more nucleotide differences.
Fig. 4. Genetic and nucleotide differences of the 18S SSU V4 hypervariable region. Indicated are the ranges within and between the major groups and genera, as well as what is considered to be variation within species. Genetic distances are based on the p-distance measure and pairwise nucleotide differences were calculated with all pairwise gaps deleted.
On the other hand, genetic distances smaller (less or equal than 3 nucleotide differences) would fall in a grey area where species discrimination would be more difficult. Pairwise distances of 0·003 (1 nucleotide difference) were obtained and corresponded with sequences that were designated as T. parva or T. sp. (buffalo) variants (Figs 1 and 3). These differences all occur within the 640 nm probe region of the hybridization probe for T. parva (Fig. 3).
In order to find further confirmation that the T. parva variants do exist, field samples for which variant sequences were obtained were analysed by real-time PCR (Fig. 5). The positive T. parva control sample (KNP102) gave 2 peaks as previously described (Sibeko et al. Reference Sibeko, Oosthuizen, Collins, Geysen, Rambritch, Latif, Groeneveld, Potgieter and Coetzer2008). This includes a main peak at ∼64°C as well as a shoulder peak at ∼52°C. The samples that possess variant sequences have profiles that differ from that of the control, in that the Tm of the shoulder peak shifts to ∼57°C. Such profiles were observed in the case of variant 1 for samples 32797 and 32806. In both of these samples, no canonical T. parva sequences were obtained and in the case of 32806, 5 clones from the 10 sequenced, represented the variant 1 sequence (Table 1). For variant 3, profiles that differed from the norm were obtained for samples 30069, 30423 and 34431 and in the case of 34431 six out of 10 clones gave the sequence for variant 3 and no canonical T. parva clones were found. In the case of T. parva variant 5, all animals that possessed sequences for this variant also possessed canonical T. parva sequences. One sample, 35524, gave a melting profile that differed from that of the standard T. parva (Fig. 5), suggestive of the presence of variants in this animal.
Fig. 5. Real-time PCR melting profiles of Theileria parva variant samples. Indicated are melting profiles obtained with the 640 nm anchor-probe pair. Vertical grey lines indicate the shoulder (51°C) and main (64°C) melting peaks for positive control KNP102.
Species differentiation using the S5 ribosomal gene
To confirm that the variants observed is indeed T. parva and not another closely-related species, a gene (S5 ribosomal protein), with more phylogenetic informative characters than the 18S gene was identified. The primers designed exclusively amplified a 779 bp fragment from members of the ‘T. taurotragi’ clade that is composed of a 554 bp stretch of coding region, with 3 interspersed introns (86 bp, 113 bp and 30 bp, respectively). As such, the S5 fragment should be useful to investigate the genetic relationship that exists between T. parva, T. sp. (buffalo) and T. sp. (bougasvlei) and their variants in more detail. The primer set should also allow for the species-specific amplification of this gene from animal samples that harbour parasites from the other major groups. Phylogenetic analysis of the S5 gene shows that T. parva, T. sp. (buffalo) and T. sp. (bougasvlei) group into distinct and well-supported clades (Fig. 6). Furthermore, the T. parva variant 1 and variant 3 sequences group within the T. parva clade suggesting that these 18S SSU rRNA variants are indeed T. parva. A linear correlation can be observed when nucleotide and genetic distances are compared and correlate with the same plot for the 18S gene (Figs 4 and 6). Demarcation between genotypes are, however, much more distinct than observed for the 18S gene, with within-species variation grouping in the lower left quadrant of the plot with pairwise genetic distances less than 0·02 (<15 nucleotide differences) observed. Pairwise genetic-distances for between-species variation are 0·039 for T. annulata and T. lestoquardi (30 nucleotide differences) and above 0·067 (>50 nucleotide differences) for the rest.
Fig. 6. Analysis of the S5 ribosomal gene. (A) Neighbor-joining analysis of the S5 ribosomal gene. Bootstrap support values (100 000 replicates) are indicated. Numbers indicate animal numbers (5 digits) or GenBank Accession codes. (B) Genetic and nucleotide differences are indicated within and between species. Genetic distances are based on the p-distance measure and pairwise nucleotide differences were calculated with all pairwise gaps deleted.
DISCUSSION
The discovery of species diversity in the Theileria originated from direct observation of clinical disease development and parasite morphology linked to tick transmission as encapsulated in the classical paradigm of Koch's postulates. This progressed to observation of benign Theileria forms and the confusion engendered by morphological similarities in the absence of clinical disease progression, to the current state of species definition based on the uniqueness of their 18S SSU rRNA molecular sequences.
The current study showed how a random sequencing approach can be an effective discovery vehicle for detection of new genotypes and possibly new species. The question is raised whether the genotypic diversity observed in the current study is exhaustive, given the sample set and sequence depth coverage obtained. In this regard, no new genotypes were observed after ∼200 clones were sequenced each for buffalo and cattle samples, respectively. This suggests that genotypic diversity was saturated with the current sample set. However, the possibility that new genotypes will be found if different geographical areas are screened cannot be excluded. This might be true for the T. buffeli group, of which a number of diverse genotypes have been observed (Gubbels et al. Reference Gubbels, Hong, van der Weide, Qi, Nijman, Guangyuan and Jongejan2000).
The RLB analysis was more sensitive than the random sequencing approach in detecting genotypes in the case of cattle samples. This could be expected, given the fact that only 10 clones were sampled per specimen for sequencing. The larger number of genotypes found in the buffalo samples using RLB could, however, also indicate that parasitaemia in cattle compared to buffalo for any given genotype is much higher. The sampling size for cattle-derived clones may therefore affect the number of genotypes detected. Alternatively, based on the similarity of the RLB probes and the genotypes that was found in buffalo, cross-hybridization is expected to occur between members of the T. mutans and T. buffeli clades. This could explain the fact that buffalo seem to have higher numbers of genotypes per animal compared to cattle.
The discovery of numerous genotypes that may cross-hybridize with current RLB probes highlights problems that can occur with probe-based technologies. In this regard, T. sp. (sable) was detected previously in buffalo and cattle (Nijhof et al. Reference Nijhof, Pillay, Steyl, Prozesky, Stoltsz, Lawrence, Penzhorn and Jongejan2005). It should be noted that the current T. sp. (sable) probe will also cross-hybridize with the T. velifera group, making previous identification of this genotype in buffalo and cattle, dubious (unpublished experimental observation).
The questions are raised whether the 26 different genotypes found in buffalo and cattle in the current study constitute different species, or mere variants or subspecies of the major groups described, or if multiple variable 18S SSU rRNA copies reside inside a single species. No consensus has yet been reached on the degree of variation within the 18S SSU rRNA or other genes that would define a species (Chae et al. Reference Chae, Allsopp, Waghela, Park, Kakuda, Sugimoto, Allsopp, Wagner and Holman1999; Gubbels et al. Reference Gubbels, Hong, van der Weide, Qi, Nijman, Guangyuan and Jongejan2000; Bhoora et al. Reference Bhoora, Franssen, Oosthuizen, Guthrie, Zweygarth, Penzhorn, Jongejan and Collins2009). It has been suggested that multiple copies of the 18S SSU RNA could account for genetic variation within a species, such as the case for Plasmodium and Babesia, where differential expression of 18S rRNA occurs as a response to various host environments and life-cycle stages (Gunderson et al. Reference Gunderson, Sogin, Wollett, Hollingdale, de la Cruz, Waters and McCutchan1987; Chae et al. Reference Chae, Allsopp, Waghela, Park, Kakuda, Sugimoto, Allsopp, Wagner and Holman1999; Lau, Reference Lau2009; Laughery et al. Reference Laughery, Lau, White, Howell and Suarez2009). It has been shown that 2 copies of the 18S gene exist in the genomes of T. annulata and T. parva and is as such probably conserved in all Theileria spp. (Gardner et al. Reference Gardner, Bishop, Shah, de Villiers, Carlton, Hall, Ren, Paulsen, Pain, Berriman, Wilson, Sato, Ralph, Mann, Xiong, Shallom, Weidman, Jiang, Lynn, Weaver, Shoaibi, Domingo, Wasawo, Crabtree, Wortman, Haas, Angiuoli, Creasy, Lu, Suh, Silva, Utterback, Feldblyum, Pertea, Allen, Nierman, Taracha, Salzberg, White, Fitzhugh, Morzaria, Venter, Fraser and Nene2005; Pain et al. Reference Pain, Renauld, Berriman, Murphy, Yeats, Weir, Kerhornou, Aslett, Bishop, Bouchier, Cochet, Coulson, Cronin, de Villiers, Fraser, Fosker, Gardner, Goble, Griffiths-Jones, Harris, Katzer, Larke, Lord, Maser, McKellar, Mooney, Morton, Nene, O'Neil, Price, Quail, Rabbinowitsch, Rawlings, Rutter, Saunders, Seeger, Shah, Squares, Squares, Tivey, Walker, Woodward, Dobbelaere, Langsley, Rajandream, McKeever, Shiels, Tait, Barrell and Hall2005). The copies are identical in the 18S hyper-variable region of the genomes sequenced, probably due to concerted evolution (Eickbush and Eickbush, Reference Eickbush and Eickbush2007). Co-amplification of different copies would indicate their presence in the same genome, but in the current study, no consistent pattern of co-amplification was observed for the different genotypes. It is therefore unlikely that the high degree of variation observed within the different groups could be due to copy variation or for that matter variation within the species population.
Differences observed within the T. buffeli, T. mutans and T. velifera groups probably indicate unique species. Previously, a higher degree of genetic diversity has been observed within the T. buffeli-group for various genetic markers compared to other Theileria species (Gubbels et al. Reference Gubbels, Hong, van der Weide, Qi, Nijman, Guangyuan and Jongejan2000). While it was acknowledged that this could be an indication of multiple species within this group, the limited knowledge at the time precluded a definitive conclusion regarding the designation of new T. buffeli-related species. The current study is similarly limited and can, as yet, not conclude that the novel genotypes observed are new species. However, host distributions of the various genotypes do support the idea that distinctive biological differences exist for the various genotypes. As such, T. velifera B, T. buffeli-like C and T. buffeli-like D were restricted to buffalo compared to T. velifera that was found in both buffalo and cattle and T. buffeli-Warwick and T. buffeli-like B that were limited to cattle. Similarly, T. mutans-like 1, T. mutans-like 2 and T. mutans-like 3 were found solely in buffalo, while T. mutans and T. mutans MSD were found in cattle and buffalo. The within-group genetic differences in the T. buffeli, T. mutans and T. velifera clades are also much larger than that of members of the T. taurotragi clade. The basal positions of the T. buffeli, T. mutans and T. velifera clades relative to that of the T. taurotragi clade suggest that they are more ancient, which could explain the larger genetic differences. This implies that much larger genetic variation can be expected across their genomes, that will likely result in genic incompatibility and hybrid sterility during recombination (Coyne and Orr, Reference Coyne and Orr2004).
As the tick midgut is the site of sexual recombination for Theileria parasites (Norval et al. Reference Norval, Perry and Young1992), vector specificity as a means of geographical isolation could be a viable mechanism for speciation. Even so, host specificity as some of the current data would suggest, could be a viable alternative mechanism for speciation, given the fact that transmission of Theileria occurs transtadially. Closely related species, such as T. sp. (buffalo), T. sp. (bougasvlei) and T. parva which seem to occur sympatric in their non-recombining stages would most possibly have different tick vectors. The same would hold for the ‘T. buffeli’ and ‘T. mutans’ groups. Several strains of T. mutans and T. buffeli were previously characterized that differed in regard to tick vector specificity (de Vos et al. 1981; Stewart et al. Reference Stewart, Uilenberg and De Vos1996; Young et al. Reference Young, Purnell, Payne, Brown and Kanhai1978b). Whether this would imply that these different strains were different species remains to be determined.
The Theileria species diversity found in Cape buffalo relative to cattle parasites suggest that buffalo could be the ancestral reservoir host for some Theileria parasites found in bovids in Africa as suggested for T. parva, T. annulata and T. mutans (Uilenberg et al. 1981; Young et al. 1981; Conrad et al. Reference Conrad, ole-MoiYoi, Baldwin, Dolan, O'Callaghan, Njamunggeh, Grootenhuis, Stagg, Leitch and Young1989; Norval et al. Reference Norval, Lawrence, Young, Perry, Dolan and Scott1991, Reference Norval, Perry and Young1992). In the case of genotypes limited to cattle (T. taurotragi, T. sable-like, T. buffeli Warwick and T. buffeli-like B), it is possible that antelopes could have been the ancestral hosts (Young et al. Reference Young, Grootenhuis, Kimber, Kanhai and Stagg1977). Alternatively, parasites such as T. buffeli could have been introduced from other geographical areas, where these parasites were adapted to wild ungulates other than the Cape buffalo, such as Asiatic buffalo (Bubalus bubalis) (Stewart et al. Reference Stewart, Uilenberg and De Vos1996; Gubbels et al. Reference Gubbels, Hong, van der Weide, Qi, Nijman, Guangyuan and Jongejan2000). In contrast, some genotypes have only been found in buffalo, which suggests that these parasites have not yet adapted to cattle. Sampling of larger numbers of buffalo and cattle could show that there is no host preference. Evidence from literature indicates, however, that T. taurotragi was unable to transform buffalo-derived lymphocytes (Stagg et al. Reference Stagg, Young, Leitch, Grootenhuis and Dolan1983), while certain T. buffeli strains derived from Asiatic buffalo were unable to infect cattle (Stewart et al. Reference Stewart, Uilenberg and De Vos1996). Similarly, no ‘T. sp. (buffalo)-like’ profiles have been observed in ∼2500 cattle samples tested with the real-time hybridization PCR assay during routine diagnostics (personal observation), suggesting that cattle are not infected by T. sp. (buffalo) or T. sp. (bougasvlei).
While some of the aims of the current study were to investigate the existence of variants, the possibility that single nucleotide polymorphism might be PCR errors should be considered. This could certainly be the case for T. parva variants no. 2, 4, 6 and 7, where only 1 variant sequence from a single animal was obtained, while the same samples also gave canonical or variant T. parva sequences. However, in the case of variants 1, 3 and 5, as well as the T. sp. (buffalo) variants, multiple sequences from a number of different animals were obtained, suggesting that these genotypes do occur within the local cattle and buffalo populations. In addition, all novel genotypes from the major groups, as well as variants 1, 3 and 5 were confirmed by re-cloning using a high fidelity polymerase. As such, the possibility that these genotypes observed is due to PCR error can be excluded. We could not confirm the sequences from variants 2, 4, 6 and 7 as we only obtained T. parva or variant sequences after re-cloning of the original samples and they could therefore be PCR artifacts. It is, however, likely that these sequences exist at low copy number and were outcompeted during cloning. Given the fact that variants 1, 3 and 5 have been confirmed to exist, there is no theoretical reason why other polymorphisms within the 18S hyper-variable region for T. parva would not exist as well, although they might not be prevalent within the population.
Samples with variants 1, 3 and 5 sequences all gave aberrant melting profiles that differ from that of the canonical T. parva, thus providing phenotypic evidence that these variants do exist. A precedent for the presence of the T. parva variant 1 genotype exists in the literature, as the sequence for the 18S SSU rRNA gene of T. parva in GenBank (Accession number: 2335189) possess an R at this position (Collins and Allsopp, Reference Collins and Allsopp1999), indicating that it could either be a G as found in the canonical T. parva sequence, or an A as found for variant 1. One of the gold standard positive control animals (9288) used in the development of the real-time hybridization assay (Sibeko et al. Reference Sibeko, Oosthuizen, Collins, Geysen, Rambritch, Latif, Groeneveld, Potgieter and Coetzer2008) gave a profile similar to that of variant 3. Direct sequencing of its 18S PCR product indicated a mixed infection with similar sequence trace peak intensities for the canonical as well as variant 3 genotypes (data not shown).
Previously, both cattle-derived and buffalo-derived parasites were shown to possess identical 18S SSU rRNA sequences (Collins and Allsopp, Reference Collins and Allsopp1999). The discovery of variant sequences in the probe region for T. parva is therefore of considerable interest in regard to accurate diagnosis and the question is raised whether the variants observed are variant genotypes of T. parva, or closely related species, or subspecies. In this regard, originally 3 separate species were distinguished for the related disease syndromes known as East Coast fever (T. parva), Corridor disease (T. lawrencei) and January disease (T. bovis) (Theiler, Reference Theiler1904; Neitz et al. Reference Neitz, Canham and Kluge1955; Neitz, Reference Neitz1957). The latter 2 were later synonymized with T. parva, although a trinomial system (T. parva bovis, T. parva lawrencei and T. parva parva) was proposed to differentiate between the different forms, but was subsequently dropped (Uilenberg, Reference Uilenberg1976; Lawrence, Reference Lawrence1979; Norval et al. Reference Norval, Perry and Young1992). However, studies based on monoclonal antibodies and nucleic acid sequences continue to show that antigenic and sequence variation exist, keeping the notion of subspecies within the T. parva group viable (Conrad et al. Reference Conrad, Stagg, Grootenhuis, Irvin, Newson, Njamunggeh, Rossiter and Young1987; Allsopp et al. Reference Allsopp, Carrington, Baylis, Sohal, Dolan and Iams1989; Collins and Allsopp, Reference Collins and Allsopp1999).
Analysis of the S5 gene suggests that the variants observed do not comprise subspecies. All T. parva positive control samples (KNP102, 9288, 9446 and 9468) originally came from South African buffalo, while variant 1 (32806) derived from a buffalo from the Niassa district in Mozambique. Variant 3 originated from a bovine (34431) that presumably contracted the parasites from buffalo-cattle contact. As such, the T. parva clade does not show any demarcation between buffalo or cattle-derived parasites for this gene and no support for subspecies classification of the variants observed in this study. This supports the notion that at genetic level no evidence for speciation can be detected between buffalo- and cattle-derived T. parva (Sibeko et al. Reference Sibeko, Geysen, Oosthuizen, Matthee, Troskie, Potgieter, Coetzer and Collins2010).
The S5 ribosomal gene exists as a single copy gene in the genomes of T. parva and T. annulata, so phylogenetic analysis will not be biased by pseudogenes or multi-gene copies (Gardner et al. Reference Gardner, Bishop, Shah, de Villiers, Carlton, Hall, Ren, Paulsen, Pain, Berriman, Wilson, Sato, Ralph, Mann, Xiong, Shallom, Weidman, Jiang, Lynn, Weaver, Shoaibi, Domingo, Wasawo, Crabtree, Wortman, Haas, Angiuoli, Creasy, Lu, Suh, Silva, Utterback, Feldblyum, Pertea, Allen, Nierman, Taracha, Salzberg, White, Fitzhugh, Morzaria, Venter, Fraser and Nene2005; Pain et al. Reference Pain, Renauld, Berriman, Murphy, Yeats, Weir, Kerhornou, Aslett, Bishop, Bouchier, Cochet, Coulson, Cronin, de Villiers, Fraser, Fosker, Gardner, Goble, Griffiths-Jones, Harris, Katzer, Larke, Lord, Maser, McKellar, Mooney, Morton, Nene, O'Neil, Price, Quail, Rabbinowitsch, Rawlings, Rutter, Saunders, Seeger, Shah, Squares, Squares, Tivey, Walker, Woodward, Dobbelaere, Langsley, Rajandream, McKeever, Shiels, Tait, Barrell and Hall2005). The translated amino acid sequence for the fragment analysed was 100% identical for all Theileria species, with the majority of coding differences due to silent substitutions at the third codon position. This codon position is known to have greater numbers of parsimony-informative characters and higher overall rates of evolution compared to the first or second codon positions making it advantageous for phylogenetic inference (Simmons et al. Reference Simmons, Zhang, Webb and Reeves2006). Combined with the non-coding substitutions, the S5 gene seems to be a much better marker for species differentiation than the 18S hyper-variable region and it provided supporting evidence that T. parva, T. sp. (buffalo) and T. sp. (bougasvlei) are distinct species.
Cloned PCR amplicons for the 26 genotypes found in the current study were analysed with the real-time hybridization assay. No amplification at 705 nm was observed, except for T. parva, T. sp. (buffalo), T. sp. (bougasvlei) and their variants (results not shown). This indicated that no new genotype/species that amplify with the hybridization assay primer set was identified. In the case of some of the T. parva variants, the absence of a melting peak at 64°C (640 nm) is abrogated by the presence of the shoulder peak, making identification of T. parva possible. Therefore, while numerous new genotypes/species were found in buffalo and cattle, those which affect the hybridization assay in terms of co-amplification during mixed-infections, remain T. sp. (buffalo) and T. sp. (bougasvlei). The within-species variation in the 18S SSU RNA gene observed for T. parva is also limited to single or double nucleotide polymorphisms that are detected by the current hybridization test. The use of this test in regions outside southern Africa should be useful for routine screening purposes. However, determination of 18S sequence variability should be considered when used as a diagnostic assay since it is possible that additional genotypes that amplify with the current primer set will be discovered in the future.
ACKNOWLEDGEMENTS
We thank Dr Roy Bengis, Kruger National Park for buffalo samples. The OVI sequencing facility is thanked for their support.
FINANCIAL SUPPORT
This research was funded by OVI, Project numbers OV15/08/1P01 and OV21/03/C148.
