Thirteen Chinese isolates of Theileria were used in this study, which comprised 9 isolates infective to cattle, i.e., T. orientalis Ningxian, T. orientalis Wenchuan, T. orientalis Lintan, T. sinensis Lintan, T. sinensis Weiyuan, T. annulata Ningxia, T. annulata Sanmenxia, T. annulata Inner Mongolia, and T. annulata Xinjiang, and 4 isolates infective to sheep, i.e., T. ovis Xinjiang, T. uilenbergi Longde, T. luwenshuni Weiyuan, and T. luwenshuni Ninxian. Detailed information on the 13 isolates is shown in Table 1.

Experimental animals (cattle and sheep) aged 6-12 months were purchased from an area where theileriosis had not been reported. One month before the experiments, all animals were splenectomized. Ten days before the experiment, blood films were taken from the ears of the animals, which were stained with Giemsa and examined to confirm the absence of hemoparasites [25]. The animal experiments were approved and performed according to the guidelines of Institutional Animal Care and Use Committee, and the number of IACUC is SYXK2010-0003.

Nine cattle and 4 sheep were inoculated with 10 ml of cryopreserved infectious blood containing different Theileria isolates. When the parasitemia level reached >5%, blood was drawn from the jugular vein and collected in tubes using heparin as an anticoagulant. Erythrocytes were isolated and the parasite DNA was obtained using a DNA MiniKit (QIAGEN GmbH, Hilden, Germany), according to the manufacturer's instructions. The amount of DNA isolated was assessed photometrically. Negative control DNA was obtained from the experimental animals prior to inoculation [26].

A pair of 28S rRNA sequence piroplasma universal primers were used based on the sequences of T. parva (GenBank no. AF218825 and no. AF013419) and a previous study [25]. The primers were as follows: forward, 5'(1011)-CTAGTAACG-GCGAGCGAAGA-3'(1030); reverse, 5'(4056)-AGGCGTTCAGTCATTATCCAA-3'(4036). The numbers in parentheses indicate the nucleotide positions of the consensus sequence. The PCR amplification conditions and the generation of sequences used the same method as a previous study [25]. The 18S rRNA sequence of each isolate was downloaded from the GenBank (Table 1).

Contigs were assembled using the Lasergene SeqMan, and the sequences produced were checked against GenBank to verify that the sequences were from the appropriate Theileria species. The sequences were initially compared with public sequences using the BLAST program at NCBI, USA (http://www.ncbi.nlm.nih.gov/). The distance matrices for the aligned sequences, where all gaps were ignored, were calculated using the Kimura's 2-parameter method in MEGA4. The D2-D3 segments with junction sequences from base pair positions 238-735 were selected based on the alignment results of 15 isolates.

Phylogenetic trees of the complete 28S rRNA, D2-D3 region, and 18S rRNA were constructed using the maximum parsimony (MP) and Bayesian inference (BI). The MP trees were constructed using the PAUP4.0 [27]. Gaps were treated as missing data, with equal weightings for transitions and transversions, and a heuristic search was made using the TBR branch-swapping algorithm [28]. BI analyses of each dataset were conducted separately using the MrBayes 3.1.2 [29]. The Bayesian analyses were initiated using a GTR+I+Γ model with no initial values assigned to these parameters and were performed with 4 chains of 1.0×10⁶ generations for the 28S rRNA dataset, 2.0×10⁵ generations for the D2-D3 dataset, and 1.0×10⁶ generations for the 18S rRNA dataset. The Markov chains were sampled at intervals of 100 generations. After discarding the burn-in samples and evaluating convergence, the remaining samples were retained for further analysis. The posterior probabilities (PP) were obtained for appropriate clades. Clades with PP ≥ 95% were considered to be highly supported. One Babesia bovis isolate was used as the outgroup. Finally, the phylogenetic trees constructed using the 2 methods were viewed using the TreeView program (version 32).

The lengths of the PCR products of the 28S rRNA gene ranged from 3,037 bp (T. annulata Inner Mongolia) to 3,061 bp (T. orientalis Wenchuan and T. orientalis Lushi). In the aligned sequences of the 28S rRNA gene, 526 sites were variable and 399 sites were parsimony-informative according to the criterion of the maximum parsimony. The nucleotide frequencies were 0.253 (A), 0.262 (T), 0.207 (C), and 0.278 (G). The transition/transversion rate ratios were 2.609 (purines) and 1.417 (pyrimidines). The percentage sequence divergence ranged from 0.1% to 5.8% at the intraspecies level, and from 1.2% to 11% at the interspecies level. The lengths of the selected D2-D3 segments varied from 467 bp (T. annulata Inner Mongolia) to 485 bp (T. orientalis Wenchuan). The percentage of sequence divergence ranged from 0.2% to 4.7% at the intraspecies level, and from 4.4% to 28% at the interspecies level. During this region, 192 sites were variable, and 141 sites were parsimony-informative. The nucleotide frequencies were 0.263 (A), 0.286 (T), 0.194 (C), and 0.256 (G). The transition/transversion rate ratios were 1.526 (purines) and 3.438 (pyrimidines). For 18S rRNA, the variation of sequences ranged from 0.1% to 1.2% at the intraspecies level, and from 1.6% to 4.0% at the interspecies level.

To compare the differentiation of different species of Theileria, phylogenetic trees were constructed for the 28S rRNA gene, D2-D3 region, and 18S rRNA gene (Fig. 2A-C). The Bayesian analysis produced essentially the same topology as the maximum parsimony analyses, so the 3 trees generated by PAUP are shown here with the bootstrap proportions (BP) and the posterior probabilities (PP) produced by the Bayesian analyses. Fig. 2A is based on the 28S rRNA gene and it shows that the 15 Theileria isolates were divided into 3 clades; 1 clade for T. annulata, another clade for T. parva, while the remaining Theileria spp. constituted the remaining clade. The phylogenetic tree generated for the D2-D3 region (Fig. 2B) showed that T. annulata and T. parva constituted 1 clade. However, the biggest difference in the phylogenetic tree based on the 18S rRNA gene was that T. uilenbergi represented 1 clade and the remaining Theileria spp. constituted another clade (Fig. 2C).