Chimpanzee Malaria Parasites Related to Plasmodium ovale in Africa

Since the 1970's, the diversity of Plasmodium parasites in African great apes has been neglected. Surprisingly, P. reichenowi, a chimpanzee parasite, is the only such parasite to have been molecularly characterized. This parasite is closely phylogenetically related to P. falciparum, the principal cause of the greatest malaria burden in humans. Studies of malaria parasites from anthropoid primates may provide relevant phylogenetic information, improving our understanding of the origin and evolutionary history of human malaria species. In this study, we screened 130 DNA samples from chimpanzees (Pan troglodytes) and gorillas (Gorilla gorilla) from Cameroon for Plasmodium infection, using cytochrome b molecular tools. Two chimpanzees from the subspecies Pan t. troglodytes presented single infections with Plasmodium strains molecularly related to the human malaria parasite P. ovale. These chimpanzee parasites and 13 human strains of P. ovale originated from a various sites in Africa and Asia were characterized using cytochrome b and cytochrome c oxidase 1 mitochondrial partial genes and nuclear ldh partial gene. Consistent with previous findings, two genetically distinct types of P. ovale, classical and variant, were observed in the human population from a variety of geographical locations. One chimpanzee Plasmodium strain was genetically identical, on all three markers tested, to variant P. ovale type. The other chimpanzee Plasmodium strain was different from P. ovale strains isolated from humans. This study provides the first evidence of possibility of natural cross-species exchange of P. ovale between humans and chimpanzees of the subspecies Pan t. troglodytes.


Introduction
Plasmodium ovale, P. falciparum, P. vivax and P. malariae belong to phylum Apicomplexa, order Haemosporidia and family Plasmodiidae. Haemosporidia are intracellular parasites transmitted by haematophagous dipterans. They infect a large variety of vertebrate amniotes, such as mammals (including humans), birds, chelonians, squamates, and crocodilians, [1]. Some are highly pathogenic and may have important implications for human public health, domestic animal health and wildlife biodiversity conservation [2,3].
P. ovale, the last of the human malaria parasites to be identified, was described in the blood of an East African patient, by Stephens in 1922. It is a relapse parasite, generating secondary infections that are usually asymptomatic [4]. However, P. ovale may interact with other species of Plasmodium infecting humans, such as P. falciparum and P. vivax, and may have a major influence on the epidemiological features of malaria [5].
Few epidemiological data are available for P. ovale. Its reported prevalence is generally low (,5%), except in West Africa, where prevalences above 10% have been observed in humans [6,7]. P. ovale is often present in mixed infections and parasitaemia is usually low. P. ovale was previously thought to be present only in sub-Saharan Africa, Papua New Guinea, Irian Jaya in Indonesia and the Philippines [4]. However, it appears to be more widely distributed, having been reported in the Middle East, the Indian Subcontinent and various parts of Southeast Asia [8][9][10][11]. P. ovale has not been yet reported in South America. However, no global map of the geographical distribution of P. ovale has been produced since that of Lysenko and Beljaev in 1969 [12].
Few studies document the molecular diversity, geographical origin, evolutionary history and age of P. ovale populations. Based on complete DNA sequences of the small subunit ribosomal RNA (SSUrRNA) gene, partial sequences of cysteine protease, ookinete surface protein and cytochrome b genes, Win et al. (2004) compared P. ovale isolates from Myanmar, Indonesia and sequences available from GenBank. The result obtained supported the division of P. ovale into at least two types, but the classical and variant types identified did not differ morphologically and occurred in sympatry [13,14].
Phylogenetically, P. ovale clusters with Plasmodium species affecting simian primates (as do P. malariae and P. vivax, but not P. falciparum), but its phylogenetic relationships to other Plasmodium species or haemosporidian parasite genera remain unclear [4].
Three Plasmodium species, P. reichenowi, P. schwetzi and P. rodhaini, have already been reported in African great apes (chimpanzees and gorillas) and have been described as morphologically similar to P. falciparum, P. ovale or P. vivax (there are differing opinions) and P. malariae, respectively [15]. Like humans, the African great apes belong to the Hominidae family. Despite the close phylogenetic relationships between these non human primates and human hosts, the diversity of Plasmodium parasites in African great apes has been little studied and few molecular data for these parasites are available. Indeed, only one strain of P. reichenowi, originally isolated from a naturally infected chimpanzee (Pan troglodytes) in Central Africa (East of the Democratic Republic of the Congo) and adapted to a laboratory splenectomized chimpanzee, has been molecularly characterized [15]. This parasite is closely phylogenetically related to P. falciparum, the principal cause of human malaria. Data for other taxa, including genetically characterized non human primate malaria parasites, are required to provide insight into the evolutionary history of P. ovale [16].
In order to investigate the diversity of Plasmodium parasites in African great apes, we screened 130 DNA samples from chimpanzees and gorillas in Cameroon. We found three chimpanzees infected by Plasmodium related to the human P. ovale. We present here the diversity of these chimpanzee parasites using two mitochondrial and one nuclear partial gene sequences and compared them to human P. ovale strains.

Results
DNA samples from 130 chimpanzees and gorillas were tested for Plasmodium infection, using cytochrome b molecular tools. Two chimpanzees, CPZcam89 (225) and CPZcam91 (451), both belonging to subspecies Pan t. troglodytes, presented a single infection with Plasmodium parasites phylogenetically related to P. ovale. Both Plasmodium isolates were characterized by a unique DNA sequence for each of the cox1, cyt b and ldh markers, differing between the two isolates. A third chimpanzee (CPZcam63 (2360)), belonging to subspecies, Pan t. vellerosus, had a mixed infection composed of P. reichenowi and P. ovale related parasites. The latter has an identical cyt b sequence to Plasmodium found in CPZcam89 (451) chimpanzee; this isolate was discarded from the phylogenetic construction. The prevalence of P. ovale related Plasmodium species was found to be 2.3% (3/130) in the Cameroonian great apes tested. This prevalence is comparable to the prevalence of P. ovale in human populations from most endemic areas (,5%).
The 708 bp cyt b and the 964 bp cox1 sequences as well as the 350 bp ldh sequence of the CPZcam89 (225) chimpanzee parasite strain are all identical to the human P. ovale variant type sequences (Tables 1, 2 and 3). Based on this genetic homology, this chimpanzee parasite strain was identified as being of the P. ovale variant type. The cyt b, cox1 and ldh nucleotide sequences of the CPZcam91 (451) chimpanzee parasite diverged from the reported classical and variant P. ovale type nucleotide sequences (Tables 1, 2 and 3). For the cyt b marker, this chimpanzee Plasmodium sequence presented four synonymous mutations with respect to the classical P. ovale type sequence and one non synonymous mutation, M248I, with respect to the variant P. ovale type sequence ( Table 1). The cox1 marker displayed two non synonymous mutations with respect to the classical P. ovale type and three with respect to the variant P. ovale type ( Table 2). The nuclear ldh sequence shows two non synonymous mutations compared to the classical P. ovale and four non synonymous mutations compared to the variant P. ovale ( Table 3).
Investigation of the mitochondrial cyt b, cox1 and nuclear ldh partial gene sequences in 13 P. ovale strains from humans from 12 different sites showed that P. ovale species could be divided into two distinct groups. Both classical and variant P. ovale (Table 4) were associated with a unique sequence for each marker, consistent with the finding of Win et al, 2004 on cyt b gene [13,17]. Comparisons of cyt b nucleotide sequences revealed 10 different substitutions between the variant and classical P. ovale types, one of which was a non synonymous mutation, M248I ( Table 1). Comparisons of the classical and variant cox1 nucleotide sequences, also revealed 10 different mutations, one of which was a non synonymous mutation M211I (Table 2). Comparisons of ldh classical and variant P. ovale nucleotide sequences showed 13 different substitutions, two of which were non synonymous mutations, S143P and K168N ( Table 3).
The sequences presented are derived from a single PCRsequencing event. The differences observed between these sequences, though likely to reflect reality, might be the result of PCR amplification artefacts.
Both of the methods used, maximum likelihood (ML) and Bayesian analyses, produced the same tree topology consistent with previous published Plasmodium phylogenetic analysis [18,19]. The phylogenetic relationships between the two Plasmodium strains isolated from chimpanzees to classical and variant P. ovale types, and the position of these strains within primate parasite group, are presented in Figure 1. The two chimpanzee parasites formed a monophyletic group with the two human P. ovale types. Monophyly was well supported by Bayesian posterior probabilities of 0.98 and a bootstrap value of 94%.

Discussion
The characterization of 13 P. ovale human isolates, using mitochondrial cyt b and cox1 markers and nuclear ldh marker from 12 different geographical locations, confirmed the diversification of human strains of P. ovale into two types, classical and variant [13].
Non synonymous mutation is shown in brackets. doi:10.1371/journal.pone.0005520.t001 We reported here the first molecular finding of three chimpanzee Plasmodium isolates, one (CPZcam89 (225)) genetically identical to P. ovale variant type, one other (CPZcam91 (451)) closely related to human P. ovale types and a third one (CPZcam63 (2360)) showing mixed infection composed of P. reichenowi and P. ovale related parasite (the latter exhibits an cyt b sequence identical to CPZcam91 (451) cyt b sequence parasite). Phylogenetic analyses inferred from cyt b and cox1 concatenates are well supported and show a monophyletic group composed of human P. ovale types and related chimpanzee parasites. The monophyly of the group is confirmed using ldh nuclear partial gene sequences (data not shown).
P. schwetzi has been originally described by Reichenow in 1920 in blood apes in Cameroon [15]. P. schwetzi is morphologically similar to both P. vivax and P. ovale parasites that infect humans, and to date there are two equally convincing arguments to favour one or the other of these species as the most closely related to P. schwetzi [15]. Experimental infections by P. schwetzi in humans have also been reported [20] and in 1970, Contacos established its potential as a zoonosis for Africa [21]. At present, no isolate of this parasite from which molecular sequences can be obtained is available.
P. schwetzi often occurs as a mixed infection with P. reichenowi and P. rodhaini, the two other African great ape Plasmodium species described morphologically similar to P. falciparum and P. malariae respectively. In this study, we found one chimpanzee co-infected with P. reichenowi and a P. ovale related parasite molecularly identical to CPZcam91 (451) isolate. The CPZcam91 (451) chimpanzee parasite might be identified as being P. schwetzi regarding reports available on this species. Nevertheless, there is not enough evidence to support this. Morphological and other molecular information are needed to establish the identity of this parasite.
The identical sequences of CPZcam89 (225) chimpanzee parasite strain to the P. ovale variant type on both mitochondrial cyt b and cox1 and nuclear ldh markers suggest possible cross-species transmission between human and chimpanzee hosts in Cameroon. Interestingly, a prevalence of P. ovale higher than that usually reported in Africa (above 10%) has been reported in two villages in the Manyemen forest province in Cameroon, where humans and great apes live in sympatry [6]. Furthermore, earlier, Lysenko and Beljaev (1969) previously reported a close relationship between P. ovale prevalence in humans and proximity to great apes in Africa [12].
No direct evidence for human malaria parasite transmission between apes and humans was reported in Gabon [22], but natural transmissions of human malaria parasites to non human primates have been reported in South America. P. falciparum, P. vivax and P. malariae transmissions to wild monkeys of the rainforest in French Guyana [23] and to Brazilian wild monkeys [24] have also been documented. Experimental transmission of P. ovale to chimpanzees via sporozoite inoculation has been reported [25].
This study provides the first evidence of human P. ovale variant type in chimpanzees in Cameroon. A large molecular epidemiology study would be required to improve the documentation of potential natural bidirectional transmission between chimpanzee and human populations living in sympatry, making it possible to evaluate the potential role of African great apes as a reservoir for P. ovale in West Africa. The question raised by Haydon et al. (2002) concerning the possibility of human Plasmodium species being permanently maintained in chimpanzee populations, from which infection is transmitted to human, remains to be explored [26].

Chimpanzee and gorilla DNA specimens
Chimpanzees and gorillas, originated from different areas of Cameroon, were, for the most part, initially kept as pets for a Table 2.

Cyt b PCR amplification
We amplified 708 bp Cyt b gene fragments with two sets of primers, one for PCR reaction, PLAS1 (59-GAGAATTATG-GAGTGGATGGTG-39) and PLAS2a (59-GTGGTAATTGA-CATCCWATCC-39) and one for nested-PCR, PLAS3 (59 -Table 3. Substitutions and their positions in ldh nucleotide sequences (numbers correspond to base pair positions and were defined according to the complete sequence of the P. falciparum ldh gene PF13_0141).
Non synonymous mutations are shown in brackets. doi:10.1371/journal.pone.0005520.t003 GGTGTTTYAGATAYATGCAYGC-39) and PLAS4 (59-CATCCWATCCATARTAWAGCATAG-39) [29]. These primers are specifics for Haemosporidia parasites and do not amplify DNA from other Apicomplexa parasites or host DNA. PCR and nested-PCR were carried out in a final volume of 25 ml, under the following conditions: 2.5 ml of each primer (10 pmol/ml), 2 mM of each dNTP, 0.5 U of Taq polymerase (Solis), 2 mM MgCl 2 and 2 ml of DNA, heating for 5 minutes at 94uC, 30 s at 94uC, 30 s at 55uC and 1 min 30 s at 72uC for 40 cycles and a final extension phase for 10 minutes at 72uC. The PCR products were sequenced by Macrogen (Korea) using PLAS3 and PLAS4 primers.
The parasites isolated from African great apes were also characterized molecularly by another gene, the cytochrome c oxidase 1 gene (cox1). This mitochondrial gene has been chosen for the international barcoding programme for biodiversity identification [30]. Like cyt b, it is a conserved gene and is useful for resolving phylogenetic relationships between populations of parasite species that have diverged over tens or hundreds of millions of years [31,32].
These primers are specifics of Haemosporidia parasites and do not amplify DNA from other Apicomplexa parasites or host DNA. PCR and nested-PCR were carried out in a final volume of 25 ml, under the following conditions: 2.5 ml of each primer (10 pmol/ ml), 2 mM of each dNTP, 0.5 U of Taq polymerase (Solis), 1.5 mM MgCl 2 and 2 ml of DNA, 5 minutes at 94uC, 30 s at 94uC, 30 s at 53uC for PCR and 30 s at 58uC for nested-PCR, and 2 minutes at 72uC for 40 cycles, with a final extension period of 10 minutes at 72uC. The PCR products were sequenced by Macrogen (Korea) using cox1c and cox1d primers.
The nuclear lactate dehydrogenase (ldh) gene has also been used to characterize parasites isolated from chimpanzees.

P. ovale human strains
We also characterized P. ovale from 12 isolates collected from 11 different African locations and 1 isolate collected from South-East Asia, Cambodia (Table 4), in collaboration with the National Reference Center for Malaria (AP-HP, Hôpital Bichat-Claude Bernard, Paris, France) using the cyt b, cox1 and ldh partial gene sequences.

Phylogenetic analyses
The cyt b, cox1 and ldh sequences were checked using chromatograms and CLUSTALW alignment to ensure that none of the positions was ambiguous [33]. Mixed infection was discarded from the phylogenetic study. Phylogenetic analyses were based on the use of 708 bp cyt b and 964 bp cox1 concatenated sequences (Table 5). Reference sequences without ambiguous positions for either cyt b or cox1 were retrieved from GenBank.
Statistical analysis, based on the Xia and Xie method, was conducted to examine whether the number of substitutions was saturated or not [34]. In this method, both transitions and transversions were plotted against evolutionary distances calculated with the JC69 model. The relative rates at which transitions and transversions saturated at the third position were compared by counting substitutions in all pairwise comparisons between sequences. The analysis showed that the third base was saturated, and this base was therefore discarded for subsequent phylogenetic analyses.
We identified the most appropriate nucleotide substitution model, based on hierarchical likelihood ratio tests (hLRTs), Akaike Information criterion (AIC) and bayesian information criterion (BIC) values, using PHYML [35] in a similar way to Modeltest [36]. The Hasegawa, Kishino and Yano statistic HKY [37] was favoured by the hLRT and BIC tests. Rate variation between sites was allowed, with a gamma distribution for four rate categories for the nucleotide and amino acid data, allowing for invariant sites. Maximum likelihood and Bayesian trees were inferred using the previously described model. Maximum likelihood (ML) analysis was carried out with Phyml [38], with nodal robustness evaluated by non-parametric bootstrapping (1000 replicates). Bayesian analysis was performed with MrBayes [39], using two runs of 1 million generations sampled every 100 generations. Convergence was determined using the standard deviation of the split frequencies and runs were stopped when a value of less than 0.01 was reached. The burn in phase was defined as the first 250,000 generations.