Mitochondrial genome of Plasmodium vivax/simium detected in an endemic region for malaria in the Atlantic Forest of Espírito Santo state, Brazil: do mosquitoes, simians and humans harbour the same parasite?

Background The transmission of malaria in the extra-Amazonian regions of Brazil, although interrupted in the 1960s, has persisted to the present time in some areas of dense Atlantic Forest, with reports of cases characterized by particular transmission cycles and clinical presentations. Bromeliad-malaria, as it is named, is particularly frequent in the state of Espírito Santo, with Plasmodium vivax being the parasite commonly recognized as the aetiologic agent of human infections. With regard to the spatial and temporal distances between cases reported in this region, the transmission cycle does not fit the traditional malaria cycle. The existence of a zoonosis, with infected simians participating in the epidemiology, is therefore hypothesized. In the present study, transmission of bromeliad-malaria in Espírito Santo is investigated, based on the complete mitochondrial genome of DNA extracted from isolates of Plasmodium species, which had infected humans, a simian from the genus Allouata, and Anopheles mosquitoes. Plasmodium vivax/simium was identified in the samples by both nested PCR and real-time PCR. After amplification, the mitochondrial genome was completely sequenced and compared with a haplotype network which included all sequences of P. vivax/simium mitochondrial genomes sampled from humans and simians from all regions in Brazil. Results The haplotype network indicates that humans and simians from the Atlantic Forest become infected by the same haplotype, but some isolates from humans are not identical to the simian isolate. In addition, the plasmodial DNA extracted from mosquitoes revealed sequences different from those obtained from simians, but similar to two isolates from humans. Conclusions These findings strengthen support for the hypothesis that in the Atlantic Forest, and especially in the state with the highest frequency of bromeliad-malaria in Brazil, parasites with similar molecular backgrounds are shared by humans and simians. The recognized identity between P. vivax and P. simium at the species level, the sharing of haplotypes, and the participation of the same vector in transmitting the infection to both host species indicate interspecies transference of the parasites. However, the intensity, frequency and direction of this transfer remain to be clarified. Electronic supplementary material The online version of this article (10.1186/s12936-017-2080-9) contains supplementary material, which is available to authorized users.


Background
In Brazil, malaria occurs originally across the entire national territory. However, the Amazon region reports 99% of all the cases in the country [1]. Since the 1940s, a national control programme has kept malaria transmission restricted to the northern area. Consequently, in the 1960s and 1970s, the extra-Amazonian region came close to a complete elimination of the disease. Nevertheless, residual transmission persisted in certain areas of dense Atlantic Forest [2]. In the Atlantic Forest, malaria presents at a very low incidence, with cases being mainly related to Plasmodium vivax and presenting few clinical symptoms [1][2][3][4]. The low incidence and the territorial dispersion of the reported cases provide evidence in favor of the existence of an unrecognized reservoir of parasites. This in turn raises questions regarding the participation of asymptomatic carriers or local simians in the transmission [5]. Genetic similarity between P. vivax, which infects humans, and the parasites which infect simians in the Atlantic Forest, named Plasmodium simium, further supports the hypothesis of zoonosis. In fact, following the first report of a natural infection of P. simium in a human being [5], the identity between P. simium and P. vivax has been established many times: (1) by studies regarding the CSP protein in the early 1990s [6], (2) by phylogenetic analyses based on sequencing of the cytochrome b gene of the mitochondrial genome [7], (3) by microsatellite polymorphisms [8], (4) again by CSP variations [9], and (5) by Msp-1 gene sequencing [10]. In this context, the recent evidence presented by Brasil et al. [11] of some single nucleotide polymorphisms differentiating P. simium from P. vivax do not imply their separation in two different species. In the extra-Amazonian region, the term bromeliad-malaria refers to the disease whose vector, recognized as Anopheles (Kerteszia) cruzii [12], depends on bromeliads as breeding sites. Molecular and serological evidence presented by different studies has suggested that bromeliad-malaria is highly dependent on human activities carried out close to the forest environment [13][14][15]. In addition, the occurrence of the disease is sparse, and the outbreaks are rare [16]. Considering the characteristics presented above and the fact that the parasites harboured by local simians are genetically indistinguishable from those found in human blood, the hypothesis of a zoonotic scenario for bromeliad-malaria is strongly supported [17][18][19][20][21][22][23]. However, even with a variety of scientific investigations corroborating the zoonosis hypothesis, considerable debate remains regarding the direction of parasite transference. For instance, by comparing the genetic variability in studies based on the Duffy binding protein of erythrocytes collected from simians of the species Alouatta guariba, Costa [24] has suggested that the simian parasite originated from its human counterpart. This hypothesis is additionally supported by Rodrigues et al. [25] based on limited genetic variability between P. simium and P. vivax.
In their study, Brasil et al. [11] suggested the possibility of using particular single nucleotide polymorphisms (SNPs) in order to differentiate between P. simium and P. vivax, focusing on whole mitochondrial genome sequences. However, differentiating between two variants of the same species is not an easy task. In order to better ascertain the distinctive genetic characteristics of these variants, this study presents the molecular characterization of P. vivax/simium based on the sequencing of the mitochondrial genomes of parasites isolated from both human and simian hosts, and, unprecedentedly, from Anopheles mosquitoes in an endemic area of the Brazilian Atlantic Forest.

Study area
Espírito Santo is a Brazilian state located in the southeast region, with large areas of dense Atlantic Forest. The fieldwork for collecting samples of anopheline mosquitoes and monkeys was concentrated in Valsugana Velha, district of Santa Teresa, and the main area with reports of malaria in this municipality. Santa Teresa is located 78 km from the capital of Espírito Santo, Vitória ( Fig. 1). The landscape in the region is irregular, with a mountainous relief reaching an altitude of 655 m above sea level, and average temperatures that vary between 15.3 and 21.0 °C. Four human blood samples were collected from the inhabitants of Santa Teresa, and 18 from other municipalities of Espírito Santo, also covered by the Atlantic Forest.

Samples origin
Human blood samples were collected from the 22 cases of malaria caused by P. vivax between 2001 and 2004, and previously detected by thick-stained blood smears in the communities of the endemic area [16]. The simian blood sample was obtained from a monkey of the genus Alouatta captured alive in Valsugana Velha in 2009. Six specimens of anopheline mosquitoes infected by P. vivax and captured in the same area, between 2014 and 2015, Keywords: Anopheles, Malaria, Epidemiology, Real-time polymerase chain reaction, DNA, mitochondrial, Sequence analyses, DNA, Zoonoses were also included (one Anopheles lutzi, one Anopheles strodei and four Anopheles cruzii) [26].

DNA extraction and confirmation of the infection
Plasmodial DNA from human and simian blood samples was extracted by the QIAamp Blood DNA Mini Kit, while the plasmodial DNA from the mosquitoes was extracted by the DNAeasy Blood and Tissue Kit, both following the instructions of the manufacturer (Qiagen). Infection was confirmed in all samples by nested PCR [27,28] and realtime PCR (adapted from Rubio et al. [29]) with primers designed to amplify the 18S RNA subunit gene. Positive and negative controls were used in all reactions.

Amplification and sequencing of the plasmodial complete mitochondrial genome
The complete mitochondrial genome (6 kb) of P. vivax/simium from the 22 samples of human blood was amplified and sequenced following the protocol proposed by Rodrigues et al. [30]. A new protocol had to be developed in order to perform the amplification of the plasmodial DNA extracted from simian and mosquito samples. Fourteen primers (Pvm1F/Pvm1R to Pvm14F/ Pvm14R- Table 1) designed by the software Primer3 were used in a conventional PCR. The procedure for each sample included 0.5 µl of the enzyme Taq DNA polymerase (5.0 U/µl) (Fermentas), 2.0 µl of the extracted DNA, 0.5 µl of each oligonucleotide primer (5.0 µM), 2.0 µl of 10 × Buffer for Taq DNA polymerase (with KCl), 0.6 µl of dNTP mix (2.0 mM each) and 1.6 µl of MgCl 2 (25.0 mM), in a final volume of 20 µl. The amplification was run in the GeneAmp PCR 9700 thermocycler (Applied Biosystems), with initial denaturation at 95 °C for 1 min, followed by 40 cycles of denaturation at 95 °C for 15 s, annealing at 60 °C for 30 s, and extension at 72 °C for 30 min. The final step of the extension was performed at 72 °C for 5 min. PCR products were purified by the Illustra GFX PCR and the Gel Band Purification Kit (GE Healthcare Biosciences), and sequenced using the BigDye kit (Applied Biosystems) in the DNA sequencer ABI 3100 (Applied Biosystems). Complete mitochondrial genome assemblies were generated using the software DNASTAR (version 8.1.13, Madison). The sequences were deposited in GenBank [31]. (Additional file 1: Table S1).

Data analysis
The sequences of the complete mitochondrial genome were aligned by the program ClustalX (version 2.1) and edited manually in the program MEGA (version 7.0). The genetic p-distance between sequences was also calculated in MEGA. Number of haplotype, haplotype diversity (Hd) and nucleotide diversity (Pi) were calculated using DNAsp (version 5). The haplotype network was generated by median-joining [32] in the program Network, version 4.6 [33], with standard parameters. Two different datasets were used for the haplotype network analysis: one including all 29 P. vivax/simium mitochondrial genome sequences from the Atlantic Forest processed in the present study (n = 29); and another also including all 149 P. vivax/simium mitochondrial genome sequences from Brazilian Amazonian and extra-Amazonian samples deposited on GenBank, with 10 of the sequences sampled from simian, and 139 from human mitochondrial DNA (mtDNA) (n = 178).
Bayesian phylogenetic analysis was carried out for P. vivax/simium complete dataset (n = 178) using MrBayes (version 3.2.1), with two runs of four chains, three heated and one cold, for 7·10 6 generations. Only groups with Bayesian posterior probability (BPP) ≥ 95% were considered significant. The consensus tree was visualized using FigTree (version 1.4.2).
Taking the GenBank accession number NC_007243.1 as the reference, the SNPs at positions 4134 and 4468 were observed in all sequences sampled from human, simian and mosquito mtDNA, in order to verify if they were useful in distinguishing between P. simium and P. vivax. As suggested by Brasil et al. [11], P. vivax should present T/A, and P. simium C/G in positions 4134 and 4468, respectively.

Results
The network comprising only samples from this study (n = 29) is shown in Fig. 2, while the haplotype network and the phylogenetic tree built using the entire dataset (n = 178) are shown in Fig. 3 Fig. 3a. Hap1 contained SNPs identical to the sequence of the isolate obtained from the simian captured in the study area. Two other samples obtained from the human isolates ( Fig. 3a:  Hap8) contained SNPs identical to those found in the isolates from human infections acquired in the Amazonian region (P. vivax). The remaining four haplotypes (Hap2, Hap4, Hap5 and Hap6) contained SNPs exclusive to the area of the present study (Fig. 3a, Table 2).
As demonstrated in Fig. 2, three haplotypes were identified in the isolates obtained from Anopheles mosquitoes (Hap7, Hap7a and Hap8): of those, two were exclusive to the vector (Figs. 2, 3a: Hap7 and Hap7a), while the third (Fig. 3a: Hap8) was identical to the haplotype identified in the isolates from the Amazonian region, as well as to two human isolates from the study area ( Table 2).
The genetic divergence within the haplotypes sampled from the Atlantic Forest was very low, with only seven SNPs identified from a stretch of DNA 5590 bp long (maximum p-distance 0.1%) ( Table 2). Among these, three SNPs were found in the non-coding region of the mitochondrial genome, one SNP within the CYTB gene sequence (synonymous mutation), and three SNPs

Hap1
PsimiumES within the COX1 gene sequence (one synonymous and two nonsynonymous mutations). COX3 sequences were conserved among all samples from the Atlantic Forest. Furthermore, both haplotype diversity (Hd) and nucleotide diversity (Pi) were low for samples from the Atlantic Forest (Table 3). Interestingly, even though simian isolates represent larger and more geographically widespread samples than mosquito isolates, they had lower Hd and lower Pi ( Table 3). The two SNPs suggested by Brasil et al. [11] could not be used to distinguish between P. simium and P. vivax, at least for the samples of the present study. This is because some haplotypes had sequences different from those proposed as distinctive between P. simium and P. vivax. More specifically, they showed a combination of these sequences ( Table 2: Hap4, Hap5, Hap6). The phylogenetic tree (Fig. 3b) also shows that isolates sampled from humans, mosquitoes and simians were not reciprocally monophyletic and some of these sequences clustered together with high statistical support (BPP ≥ 95%).
The haplotype network showed a reticulate relationship between haplotypes, with no evidence of isolation of any haplotype, and with only one or two mutation steps connecting all of the sequences from the samples of the Atlantic Forest region.

Discussion
Plasmodium vivax is a ubiquitous protozoan with a cosmopolitan distribution, causing infections in a number of populations across different continents. Its South American counterpart, P. simium, is the aetiological agent of malaria in simians inhabiting the Atlantic Forest [34]. Several studies have suggested that P. vivax and P. simium are the same species, based on their genetic similarities [5][6][7][8][9][10]. In some areas, including the one of the present study, both simians and humans are infected by this agent, making the hypothesis of zoonosis plausible. The results of the present study, uncovering a haplotype diversity in a situation of low genetic divergence in Espírito Santo, indicate a heterogeneity of the isolates obtained from different host species, and strengthen the understanding that P. vivax and P. simium are the same species with small genetic variations. What is more, these results corroborate the findings of Costa et al. [35] and Rodrigues et al. [25], whose phylogenetic analyses of samples from different world regions indicated a recent transfer of the parasite from humans to New World simians.
Among the eight haplotypes identified in the study area, two were common to humans and simians, based both on the sequences deposited on GenBank, as well as those obtained from the local simian. This finding represents evidence of parasite transmission from one species to another. At the same time, such a sharing could not be confirmed for the remaining haplotypes, as the sequences were clearly distinctive. Four of the haplotypes obtained from humans were exclusive to the study area, and the two remaining ones were compatible with those previously considered from the Amazonian region [25].
The inclusion of samples obtained from mosquito vectors for the comparison of mitochondrial sequences had never been performed before in Brazil, despite being previously suggested by Brasil et al. [11]. Interestingly, the results of the present study were not consistent with Rodrigues et al. [25], who suggested that two SNPs were distinct between malaria from the Atlantic Forest (C/G) and from the Amazonian region (A/T). Rather, plasmodial DNA extracted from mosquito vectors from the Atlantic Forest study area in Espírito Santo revealed the nucleotides A/T at these loci. These same SNPs were also proposed by Brasil et al. [11] as distinctive between P. vivax and P. simium. However, the present results demonstrate that said SNPs were not able to distinguish between the two lineages in Espírito Santo, as they were not fixed in at least three samples from humans in the study area (Table 2). Furthermore, it was shown that isolates from different hosts share haplotypes, and there is no evidence of monophyly among human and simian samples.
The mosquitoes responsible for the transmission of malaria in the Atlantic Forest system belong to the Kerteszia subgenus, with the species Anopheles cruzii [26,36,37] being the most prominent. Specimens of species of the subgenus Nyssorhynchus have also been captured in the region, occasionally infected by P. vivax/simium recovered from the blood contained in their abdomens [26]. Though it is possible that the mosquitoes of the subgenus Nyssorhynchus are being infected by feeding on human blood, their role as vectors is improbable. The fact that haplotype 8 was obtained from humans and from only one pool of Anopheles (Nyssorhynchus) strodei captured close to dwellings suggests that the mosquitoes were infected by the humans only incidentally. Haplotype 8 has sequences identical to those previously considered specific to the Amazonian region. Two other haplotypes (Hap7 and Hap7a), obtained from other mosquitoes, despite being closely related to those from the Amazonian region, also have distinctive SNPs, making them exclusive for the mosquitoes of the study area (Figs. 2, 3, Table 2). The study has some limitations. The parasite DNA was obtained from humans, the simian, and mosquitoes in different periods, precluding any conclusions regarding a possible circulation of all the haplotypes with the same magnitude at the same time. In addition, only a single simian sample was available, preventing determination of the diversity of the haplotypes infecting this host species (Fig. 2). The haplotype network constructed based on ten simians from the Atlantic Forest revealed two different haplotypes shared with humans in the study area (Fig. 3a:  Hap1 and Hap3). Araújo et al. [38] highlighted the apparent rareness of simian malaria in the Amazonian region, attributing it to the difficulties in capturing the nonhuman primates, and in obtaining samples of good quality. The same observational difficulties are applicable to the conditions of the Atlantic Forest.
The genetic diversity found in the present study is greater than that reported by Brasil et al. [11]. The presence of sequences identified in isolates obtained from mosquitoes and shared by two isolates from humans, but different from those obtained from simians in Espírito Santo indicates that the transmission cycle of this residual malaria is complex and cannot be adequately ascertained by only a few studies with small samples. It suggests that interspecies transference of the parasites has either occurred in the past or is still occurring. However, the intensity, frequency and direction of this transfer remain to be clarified.
The merit of the present study, similar to the one by Brasil et al. [11] is to document the presence of the same parasite in both human and simian hosts. Finding the same parasite in two host species is necessary, but not sufficient evidence to confirm a zoonosis. In order to verify a zoonotic cycle, one would have to show different genetic diversities of the parasites between hosts and estimate the time to the most recent common ancestor by phylogenetic analysis of specimens isolated from both hosts. Such an approach was performed to analyse the transmission of Plasmodium knowlesi, uncovering a higher number of genotypes per infection in simians than in humans [39]. Additionally, the time to the most recent common ancestor based on the analysis of mtDNA revealed that the species was derived from an ancestral parasite population that existed prior to human settlement in Southeast Asia [39]. Both findings were able to support the hypothesis of an actual zoonosis, pointing to a recent transference of the parasites to the human population. In the case of P. vivax/simium, however, the evidence points in the opposite direction. Here, the haplotype diversity is lower among the simians and the phylogenetic analyses indicate a recent transfer of the species from humans to simians [25,34]. Consequently, the present study, just like the study of Brasil et al. [11] is not sufficient to determine definitively how the transference occurs in the Atlantic Forest, precluding any conclusion regarding a zoonotic cycle.
By including more samples from simians and vectors, all obtained in the same period, future studies should facilitate a deeper understanding of the transmission cycle of this singular endemic disease.

Conclusions
Sequencing of the complete mitochondrial genome of P. vivax/simium in an area of the Atlantic Forest in Brazil uncovered eight haplotypes, two of which were shared by human and simian hosts. Interestingly, the other six haplotypes were distinctive, harboring sequences either unique to human infections in the Atlantic Forest or identical to those of the Amazonian region. Such results indicate the possibility of a zoonotic cycle, but given the observed diversity of the haplotypes, more studies are necessary to better ascertain the dynamics of the transference of parasites between humans and simians.  Table S1. GenBank accession numbers of published sequences used to construct the haplotype networks and the phylogenetic tree.