All isolates were obtained during our recent field surveys in Myanmar and Indonesia (2,11,12). For molecular analysis of the variant and classic types, patients with single infections were selected. The variant isolates we analyzed were ST243 (Rakhine State) and MC53 (Tanintharyi Division), both from Myanmar, and M474 (Flores Island, eastern Indonesia). The three classic isolates of M3 (Shan State), M4 (Bago Division), and T134 (Mon State) were collected from Myanmar.

Parasite DNA templates were isolated from blood by using a DNA isolation Kit (High Pure PCR Template Preparation Kit, Boehringer Mannheim, Germany). Then the target sequences at the block 9 region used for PCR-based diagnosis were further analyzed to confirm the presence of the variant- or classic-type in P. ovale–positive samples. Amplified DNA products using the P1F-Up and specific reverse (PoR2) primers (Table 1) underwent direct sequencing, whereas the first PCR products were cloned into the pCR II plasmid from a TA Cloning Kit (Invitrogen, San Diego, CA). The target fragments of 12 positive clones from each sample were sequenced by using Big Dye Terminator sequencing kit on an ABI 310 sequencer (PE Applied Biosystems, Foster City, CA.

Primer sets used were shown in Table 1. For analysis of the SSUrRNA gene, PCR amplification was performed by using AmpliTaq Gold polymerase (PE Applied Biosystems) at 96°C for 10 min, 36 cycles at 94°C for 30 sec, 55°C for 1 min, and 72°C for 2 min, followed by one cycle at 72°C for 10 min.

For analysis of the cysteine protease and cyt b genes, PCR conditions were slightly modified from the original methods (15,16). The conditions used were one cycle at 96°C for 10 min, 36 cycles at 94°C for 30 s, 50°C for 1 min, and 72°C for 90 s, followed by one cycle at 72°C for 10 min. The amplified PCR products were cloned into the pCR II vector. Plasmid DNA was purified from the positive colonies and sequenced in both directions by using the primers described in Table 1 in combination with M13 primers. Sequencing was performed with an ABI 377 sequencer. Any ambiguity and putative polymorphism was checked by additional amplification and sequencing. Sequences obtained were compared with those reported in databases.

Gene sequences used for the SSUrRNA genes were clone 9 and 26 of the Nigerian I/CDC strain (6), isolates from Cameroon (8), MAL/MAI (X99790), Papua New Guinea (14), Ghana (AJ250701), and two strains of the Nigerian I/CDC and the London School of Hygiene and Tropical Medicine (13). For the cysteine protease, P. ovale Nigerian I/CDC, P. malariae WR314, P. cynomolgi, P. reichenowi (15), P. vivax Salvador-1 (19), and P. falciparum (20) were used. For the cyt b. P. ovale Harding strain, P. malariae Uganda-1, P. falciparum Kenya, Santa Lucia, Malaysian-4, 3D7, P. simiovale, P. knowlesi, P. cynomolgi (16), P. falciparum Malay Camp (21) and C10 (22), P. reichenowi, P. falciparum NF54, K1, T9/96, 7G8 (23), Indian isolate 317 (24), P. vivax Salvador-I (25), and Indian PH 10 (26) were retrieved as well as P. berghei (27) and P. yoelii (28). Dendrograms were obtained with PHYLIP (Version 3.5c, University of Washington, Seattle, WA) by using the neighbor-joining method with a Kimura’s two-parameter distance and the maximum likelihood method.

The procedures for first and nested PCR amplifications with primers (Table 1) were described previously (17,18). Nucleotide sequences were determined by direct sequencing with nested PCR products. Then, sequences obtained were compared with those reported previously (AB051631-3, AB074973-6).

Three different sequences were obtained for the SSUrRNA gene from each variant-type isolate, while two different sequences were detected from each classic-type isolate (Figure 1). However, whether all of them were A (asexual)-type genes or sequences included S (sexual)- or O (ookinete/oocyst)-type (5,29,30) genes was unknown. Hereafter, these sequences are referred to as Pov 1–3 for the variant-type and as Poc 1–2 for the classic-type.

The differences among Pov 1–3 were seen at the 5´ half (Figure 1). When compared with the four complete sequences in GenBank, the Cameroon and MAL/MAI isolates were grouped as variant-type (>99% identity with Pov 1–3 and <97% with Poc 1–2). Both African isolates also shared the same mutation at the block 9 region (nucleotide positions of 1158–1160). Particularly, the sequence found in Cameroon isolate resembled that of Pov 1 (only 4-bp difference).

The alignment of four partial sequences showed that, despite their same origin, the partial sequence of the Nigerian I/CDC (13) also showed 9-bp and 5-bp differences from those of clones 9 and 26, respectively. Among these isolates, the LS strain possessed the same sequence as the 3´ half of Pov 1–3, and thus it was grouped as variant type (<96% identity with Poc 1–2). The sequence of the Papua New Guinea isolate was more similar to that of clone 9 of the Nigerian I/CDC (>98.8% identity with Poc 1 >and <97% with Pov 1–3) than to that of clone 26 or Poc 2 (98.2%–98.4% identity). The Ghana isolate was also grouped as classic-type (>97% identity with the Nigerian I/CDC or Poc 1–2 and <92% with Pov 1–3). These results suggest that the Papua New Guinea, Ghana, and our classic isolates are members of the classic- (Nigerian I/CDC-) type group, whereas the Cameroon and MAL/MAI isolates, as well as our variant isolates, are all members of the variant- (LS-) type group.

The analysis of 531 bp of the cysteine protease genes, when compared with the reported sequence of the Nigerian I/CDC, showed that variant isolates differed at 19 bp (3.6%) with eight nonsilent mutations and that classic-type isolates had an almost identical sequence, except for a single base at position 700 (nonsilent substitution from Pro to Ala) (Table 2). Because the same substitution is also seen in the variant P. ovale, P. falciparum, P. vivax, P. malariae, P. reichenowi, and P. cynomolgi (data not shown), this nucleotide may have been misread in the sequence of the Nigerian I/CDC strain; if so, sequences of classic-type isolates are identical to those of the Nigerian I/CDC strain. At the amino acid level, the variant isolates showed 96.0% sequence identity with the classic isolates.

Tachibana et al. (17) have analyzed the ookinete surface protein genes in Thai isolates and reported that two (Nigerian I/CDC and LS) types of P. ovale, defined by the SSUrRNA genes, have distinct sequences. Our nearly complete sequences of Pos 25, Pos 28-1, and Pos 28-2 in variant isolates were identical to those found in LS-type Thai isolates, while sequences in the classic isolates were identical to those of the Nigerian I/CDC-type (data not shown).

The analysis of 1035 bp of the cyt b genes showed that variant- and classic-type P. ovale isolates differed from each other at 12 bp, with one nonsilent substitution (Table 3). Sequences of variant isolates differed from those reported for the Harding strain at 15 bp (1.4%), with two nonsilent substitutions. The SSUrRNA gene of the Harding strain was not reported yet, and whether this strain is of the classic or variant type is not known. From its cyt b sequence, it was expected that this strain may belong to the classic group, despite the differences between the respective sequences (3 bp, including one nonsynonymous replacement). The dendrogram based on the cyt b genes shown in Figure 2 suggests that P. ovale may be separated into three types. A similar branching pattern was obtained with the maximum likelihood method (data not shown). However, some sequence mistakes cannot be ruled out in GenBank-available sequences, such as that of Harding strain (for example, nt 202–221 are conserved in all reported Plasmodium spp. so far studied, except for P. ovale Harding strain and P. malariae Uganda-1). As a result, it seems more prudent to propose the separation of P. ovale into only two types.