Novel piroplasmid and Hepatozoon organisms infecting the wildlife of two regions of the Brazilian Amazon.

During 2009-2012, wild animals were sampled in two areas within the Amazon biome of Brazil, in the states of Mato Grosso and Pará. Animal tissues and blood were molecularly tested for the presence of Piroplasmida (genera Babesia, Theileria, Cytauxzoon) or Hepatozoon DNA. Overall, 181 wild animals comprising 36 different species (2 reptiles, 5 birds, and 29 mammals) were sampled. The following Piroplasmida agents were detected: Cytauxzoon felis in one ocelot (Leopardus pardalis), Theileria cervi in two red brocket deer (Mazama americana), Theileria spp. in three nine-banded-armadillos (Dasypus novemcinctus), one agouti (Dasyprocta sp.), and four lowland pacas (Cuniculus paca), Babesia spp. in one common opossum (Didelphis marsupialis) and one white-lipped peccary (Tayassu pecari). The following Hepatozoon agents were detected: Hepatozoon sp. (possibly Hepatozoon caimani) in three spectacled caimans (Caiman crocodilus), Hepatozoon felis in an ocelot (Leopardus pardalis), and Hepatozoon spp. in one scorpion mud turtle (Kinosternon scorpioides) and one lowland paca (Cuniculus paca). Phylogenetic analyses inferred by the 18S rRNA gene partial sequences supported these results, highlighting at least five novel Piroplasmida agents, and two novel Hepatozoon agents. This study screened the presence of tick-borne protozoa in a number of wildlife species from the Amazon for the first time. Our results indicate that a variety of genetically distinct Piroplasmida and Hepatozoon organisms circulate under natural conditions in the Amazonian wildlife.

Introduction

The Amazon is the largest and most diverse tropical forest of the world, covering over 6.6 million km2 in South America, with more than 50% of its original distribution occurring in Brazil (Mittermeier et al., 2003). Despite of a rich vertebrate fauna, although yet poorly explored and with a great potential for discovery of many new species, almost nothing is known about infection of the Amazonian wildlife by vector-borne protozoa of the order Piroplasmida (e.g., genera Babesia, Theileria, Cytauxzoon) or hemogregarines of the genus Hepatozoon. While there have been scarce reports of the felid Panthera onca (jaguar) infected by both Hepatozoon felis and Cytauxzoon felis (Furtado et al., 2017a, Furtado et al., 2017b), and Hepatozoon caimani infecting the caiman Caiman crocodilus (Lainson et al., 2003) in the Amazon, to our knowledge, there have been no additional reports of Piroplasmida or Hepatozoon agents infecting free-ranging wildlife from the Amazon biome. This scenario motivated the present study, which performed a preliminary investigation of the infection of these protozoan agents in different free-ranging vertebrate wild species (reptiles, birds, and mammals) that were conveniently obtained in two large areas of the Brazilian Amazon.

Materials and methods

Wild animals were sampled in two areas within the Amazon biome of Brazil. In one area in Mato Grosso state (central-western Brazil), samples comprised animals that were hunted by Indians during September 2010 to June 2012 in the Tapirapé Indian Reserve, within Confresa Municipality (10°36′ to 10°52′S; 51°10′ to 51°21′W). In another area in Pará state (northern Brazil), samples comprised road-killed animals during February 2009 to November 2011, alongside the BR163 highway, between Km 50 and 217 within Santarém (02°24′S; 54°42′W) and Rurópolis (04°05′S; 54°54′W) municipalities. Further details on these animal samplings have been reported elsewhere (Soares et al., 2015).

From each animal, fragments of internal organs or blood samples were collected and kept frozen at −20 °C until being processed in the laboratory. DNA extraction of blood or fragments of lung or liver was conducted with the DNeasy Tissue and Blood Kit (Qiagen, Chatsworth, CA, USA) according to manufacturer's instructions. Blank tubes containing water were always included as a contamination control during DNA extraction. The concentration of extracted DNA was measured in a spectrophotometer UV (Bio Photometer plus, Eppendorf, Hamburg, Germany). Only samples with at least 20 ng/μl of DNA were subjected to PCR assays.

Tissue or blood DNA samples were tested by two polymerase chain reaction (PCR) protocols. One protocol employed primers BAB2 143–167 (5′-CCG TGC TAA TTG TAG GGC TAA TAC A-3′) and BAB2 694-667 (5′-GCT TGA AAC ACT CTA RTT TTC TCA AAG-3′), targeting a ≈551-bp of the 18S rRNA gene of tick-borne Piroplasmida (genera Babesia, Theileria, Cytauxzoon). The second protocol employed primers HEP2 144–169 (5′- GGT AAT TCT AGA GCT AAT ACA TGA GC-3′) and HEP2 743-718 (5′-ACA ATA AAG TAA AAA ACA YTT CAA AG-3′), targeting a 574-bp of the 18S rRNA gene of Hepatozoon spp. Reactions were performed as previously reported (Almeida et al., 2012).

For each reaction, both positive (DNA of Babesia vogeli or Hepatozoon canis) and negative (water) controls were included. Amplified products were analyzed after electrophoresis in 1.5% agarose gels stained by SYBR Safe DNA Gel Stain (Life Technologies, Grand Island, NY, USA) and visualized under U.V. transilluminator. All PCR products of the expected size were purified with ExoSap (USB, Cleveland, OH, USA) and DNA-sequenced (bi-directional Sanger sequencing) in an ABI automated sequencer (Applied Biosystems/Thermo Fisher Scientific, model ABI 3500 Genetic Analyser, Foster City, CA, USA) with the same primers used for PCR. Generated sequences were compared to each other and submitted to BLAST analysis (www.ncbi.nlm.nih.gov/blast) to infer closest similarities available in GenBank.

Partial sequences of the 18S rRNA gene generated in this study by the Piroplasmida-targeted PCR were aligned with corresponding sequences of related genotypes and Piroplasmida species available in GenBank. Similarly, partial sequences of the 18S rRNA gene generated in this study by the Hepatozoon-targeted PCR were aligned with corresponding sequences of related genotypes and Hepatozoon species available in GenBank. Sequences were aligned using ClustalX (Thompson et al., 1997) and secondary structure comparative analysis, and were adjusted manually using GeneDoc (Nicholas and Nicholas, 1997). The 18S rRNA sequences were used to construct a phylogenetic tree using maximum parsimony, as implemented in PAUP version 4.0b10 (Swofford, 2002) with heuristic search in 1000 replicates and 500 bootstrap replicates, random stepwise addition starting trees (with random addition sequences) and TBR branch swapping.

This work was authorized by the Brazilian Institute of Environment and Natural Resources (IBAMA authorization no. 23225-1 and 21526-1) and the Indian National Foundation (FUNAI authorization no. 45/AAEP/10 – Process no. 2433/07), and was approved by the Ethical Committee of Animal Use of the Faculty of Veterinary Medicine of the University of São Paulo (protocol no. 1747/2009).

Results

In Mato Grosso state, 49 wild animals comprising 29 different species (2 reptiles, 5 birds, and at least 22 mammals) were sampled (Table 1). In Pará state, 132 wild animals comprising 18 mammal species were sampled (Table 2). The Piroplasmida 18S rRNA gene-targeted PCR showed that among 49 evaluated animals from Mato Grosso, and among 132 evaluated animals from Pará, 5 (10.2%) and 11 (8.3%), respectively, contained Piroplasmida DNA (Table 1, Table 2). All PCR-positive animals were mammals. Among Mato Grosso samples, only the lung yielded Piroplasmida DNA; for Pará samples, both lung and liver yielded Piroplasmida DNA. In only two animal species, the detected Piroplasmida agent was a known agent since the resulting 18S rRNA gene sequences of the PCR amplicons were 100% identical to GenBank available sequences; these include Cytauxzoon felis detected in one ocelot (Leopardus pardalis), and Theileria cervi detected in two red brocket deer (Mazama americana) (Table 2). All remaining piroplasmid sequences are new haplotypes, with closest matches in GenBank varying from 95 to 98%, as shown in Table 1, Table 2. These novel agents were detected in three nine-banded-armadillos (Dasypus novemcinctus), one agouti (Dasyprocta sp.), four lowland pacas (Cuniculus paca), one common opossum (Didelphis marsupialis), and one white-lipped peccary (Tayassu pecari). No piroplasmida haplotype was shared by different vertebrate species. On the other hand, armadillos (D. novemcinctus) from Mato Grosso and Pará states had identical Piroplasmida haplotypes. The phylogenetic tree inferred by Piroplasmida 18S rRNA gene partial sequences showed that these two armadillo (D. novemcinctus) haplotypes grouped with the agouti (Dasyprocta sp.) haplotype under 100% branch support and 2.40% sequence divergence; the clade composed by Theileria parasites included the deer haplotype (identical to T. cervi; 70% bootstrap and 1.02% sequence divergence) detected in this study; the paca (C. paca) haplotype grouped with a piroplasmid sequence from capybara (a rodent species closely related to paca) from southern Brazil under 97% bootstrap support and 1.28% sequence divergence (Fig. 1). The opossum (D. marsupialis) haplotype formed a clade with a Babesia sp. recently reported in another opossum species (Monodelphis domestica) from the Brazilian Pantanal (76% bootstrap, 0.96% sequence divergence). Finally, the peccary (T. pecari) haplotypes grouped within a clade (97% bootstrap support) composed by several piroplasmid agents phylogenetically distinct from the classic Babesia or Thelieria organisms (70% bootstrap and 0.83% sequence divergence).

Table 1

Molecular detection of Piroplasmida and Hepatozoon spp. in tissue samples of wild animals in the Amazon forest within Confresa municipality, Mato Grosso state, Brazil, during 2010–2012.

Wild animal(No. sampled)	Tissue testeda	PCR for Piroplasmida		PCR for Hepatozoon
		No. positive (%)	CSGBb	No. positive (%)	CSGBb
Reptilia
Kinosternon scorpioides (2)	B	0		2 (100)	96% (522/545) H. ayorgbor (EF157822)
Caiman crocodilus (3)	Lu, Li	0		3 (100)	100% (544/544) Hepatozoon sp. (KJ413113)
Birds
Ardea cocoi (1)	Lu	0		0
Crax fasciolata (3)	Lu	0		0
Megascops choliba (1)	Lu	0		0
Pauxi tuberosa (1)	Lu	0		0
Penelope superciliaris (2)	Lu	0		0
Mammals
Akodon sp. (1)	Lu, Li	0		0
Cabassous unicinctus (1)	Lu, Li	0		0
Cavia sp. (3)	Lu, Li	0		0
Sapajus sp. (2)	Lu, Li	0		0
Cerdocyon thous (2)	Lu, Li	0		0
Chiropotes sp. (3)	Lu, Li	0		0
Coendou prehensilis (1)	Lu, Li	0		0
Cuniculus paca (2)	Lu	0		0
Dasyprocta sp. (1)	Lu, Li	1 (100)	98% (307/313) Theileria capreoli (KJ188220) c	0
Dasypus novemcinctus (2)	Lu, Li	1 (50.0)	96% (384/399) Theileria equi (EU888903)	0
Didelphis marsupialis (2)	Lu, Li	0		0
Euphractus sexcinctus (2)	Lu, Li	0		0
Gracilinanus microtarsus (1)	Lu, Li	0		0
Mazama americana (1)	Lu, Li	0		0
Micoureus demerarae (1)	Lu, Li	0		0
Monodelphis brevicaudata (1)	Lu, Li	0		0
Monodelphis sp. (1)	Lu, Li	0		0
Nasua nasua (1)	Lu, Li	0		0
Pecari tajacu (1)	Lu, Li	0		0
Sylvilagus brasiliensis (3)	Lu, Li	0		0
Tamandua tetradactyla (2)	Lu, Li	0		0
Tayassu pecari (2)	Lu, Li	0		0
Total (49)	Lu, Li	2 (4.1)		5 (10.2)

B: blood; Lu: lung; Li: liver.

CSGB: closest similarity in GenBank (accession number) by BLAST analysis; in all cases, E value was 0.0, unless when stated.

E value: 7e-151.

Table 2

Molecular detection of Piroplasmida and Hepatozoon spp. in tissue samples of wild animals in the Amazon forest within Santarém and Rurópolis municipalities, Pará state, Brazil, during 2009–2011.

Wild animal(No. sampled)	Tissue testeda	PCR for Piroplasmida		PCR for Hepatozoon
		No. positive (%)	CSGBb	No. positive (%)	CSGBb
Mammals
Alouatta nigerrima (11)	Lu, Li	0		0
Bradypus tridactylus (1)	B	0		0
Cabassous unicinctus (1)	Lu, Li	0		0
Callicebus moloch (1)	Lu, Li	0		0
Sapajus sp. (1)	Lu, Li	0		0
Choloepus didactylus (1)	B	0		0
Coendou prehensilis (6)	Lu, Li	0		0
Cuniculus paca (33)	Lu, Li	4 (12.1)	95% (436/459) Babesia sp. capybara (EF222255)	1 (3.0)	95% (494/520) H. felis (GQ377216)
Dasyprocta sp. (1)	Lu, Li	0		0
Dasypus novemcinctus (30)	Lu, Li	2 (6.7)	96% (387/402) Theileria equi (EU888903)	0
Didelphis marsupialis (17)	Lu, Li	1 (5.9)	97% (509/524) Babesia sp. Marsupial (KP757839)	0
Leopardus pardalis (1)	Lu, Li	1 (100)	100% (508/508) Cytauxzoon felis (AF399930)	1 (100)	100% (477/477) H. felis (AB771547)
Mazama americana (2)	Lu, Li	2 (100)	100% (434/434) Theileria cervi (AY735134)	0
Metachirus nudicaudatus (2)	Lu, Li	0
Pecari tajacu (2)	Lu, Li	0		0
Proechimys sp. (5)	Lu, Li	0		0
Tamandua tetradactyla (8)	Lu, Li	0		0
Tayassu pecari (9)	Lu, Li	1 (11.1)	94% (483/512) Babesia duncani (HQ289870)	0
Total (132)		11 (8.3)		2 (1.5)

B: blood; Lu: lung; Li: liver.

CSGB: closest similarity in GenBank (accession number); in all cases, E value was 0.0.

Fig. 1

Maximum parsimony tree inferred from 18S rRNA gene sequences of piroplasmids (Babesia spp., Theileria spp., Cytauxzoon spp.), with Plasmodium ovale as outgroup (316 characters; 65 parsimony-informative sites). Numbers at nodes are the support values for the major branches (bootstrap over 500 replicates). The sequences obtained in this study are in bold. Numbers in brackets are GenBank accession numbers.

The Hepatozoon 18S rRNA gene-targeted PCR showed that among 49 evaluated animals from Mato Grosso, and among 132 evaluated animals from Pará, 5 (10.2%) and 2 (1.5%), respectively, contained Hepatozoon DNA (Table 1, Table 2). PCR-positive animals included 2 reptile species and 2 mammal species. From the lung of three spectacled caimans (Caiman crocodilus), the Hepatozoon 18S rRNA partial DNA sequence was 100% identical to GenBank Hepatozoon sp. MRA-2014b clone C23 (KJ413113), previously detected in Caiman yacare from the Brazilian Pantanal. From the liver of an ocelot (L. pardalis), the Hepatozoon 18S rRNA partial DNA sequence was 100% identical to GenBank Hepatozoon felis sequence (Table 2). The remaining Hepatozoon sequences generated in this study are new haplotypes, with closest matches in GenBank varying from 95 to 96% (Table 1, Table 2). These novel agents were detected in the blood of one scorpion mud turtle (Kinosternon scorpioides), and in the lung of one lowland paca (C. paca). The phylogenetic tree inferred by Hepatozoon 18S rRNA gene partial sequences showed that the mud turtle haplotype grouped with Hepatozoon catesbianae (AF130361) in a large clade composed by either rodent- or reptile-associated Hepatozoon organisms, including the caiman Hepatozoon haplotype generated in this study (Fig. 2). The paca (C. paca) haplotype branched separately, basal to a large clade of Hepatozoon organisms associated with reptiles and marsupials.

Fig. 2

Maximum parsimony tree inferred from 18S rRNA gene sequences of Hepatozoon spp., with Babesia sp. as outgroup (488 characters; 52 parsimony-informative sites). Numbers at nodes are the support values for the major branches (bootstrap over 500 replicates). The sequences obtained in this study are in bold. Numbers in brackets are GenBank accession numbers.

The GenBank nucleotide sequence accession numbers for the 18S rRNA gene partial sequences generated in the present study are KY683997, KY683998, KY683999, KY684000, KY684001, KY684002, and KY684003 for the Piroplasmida haplotypes of armadillo, deer, paca, agouti, peccary, opossum, and ocelot, respectively; KY684004, KY684005, KY684006, and KY684007 for the Hepatozoon haplotypes of paca, ocelot, turtle, and caiman, respectively.

Discussion

We report seven distinct piroplasmid haplotypes, each one corresponding to a different organism that is associated with a distinct mammal host. Two of the haplotypes (from ocelot and from deer) are related to known organisms (C. felis and T. cervi, respectively). In a previous study with captive felids in Brazil, a wild born ocelot (L. pardalis) that had lived at least nine years in captivity was found infected by C. felis (Filoni et al., 2012). This long period spent in captivity led the authors to suggest that the animal acquired the infection under captivity conditions. On the other hand, our findings of C. felis in a free-ranging ocelot support the hypothesis of natural infection of this felid under Amazonian conditions. This statement is corroborated by a recent report of C. felis infecting jaguars (P. onca) under natural conditions of three Brazilian biomes (Pantanal, Cerrado, and Amazon) (Furtado et al., 2017a). In a previous study from the Cerrado biome of Brazil, DNA of T. cervi was reported in the blood of gray brocket deer, Mazama gouazoubira (Silveira et al., 2011). This report, coupled with our findings of T. cervi in red brocket deer (M. americana) from the Amazon appoints to a possible widespread distribution of T. cervi among Mazama spp. deer in different Brazilian biomes. The five remaining piroplasmid haplotypes found in the present study are likely to represent new species of Babesia (from opossum) or Theileria (from paca, agouti, and armadillo) or a yet undetermined genus (from peccary). Our opossum (D. marsupialis) haplotype was phylogenetically very closely related to a Babesia sp. recently reported from another opossum species, M. domestica (Babesia sp. Monodelphis [KP757839]), in the Brazilian Pantanal (Wolf et al., 2016), suggesting that the two haplotypes could represent a single Babesia species. Interestingly, Babesia brasiliensis has been described as a piroplasmid of didelphid opossums in South America, including Didelphis spp. (Sampaio et al., 2001); unfortunately, there has been no molecular characterization of B. brasiliensis. It is possible that the opossum haplotypes detected in the present study (in the Amazon) and in the Pantanal (Wolf et al., 2016) represent the species B. brasiliensis, yet to be confirmed in further studies.

Piroplasmids are known to have ixodid ticks as biological vectors (Lack et al., 2012). The piroplasmid-infected mammals of the present study were also found to be infested by up to seven tick species of the genus Amblyomma, as reported elsewhere (Soares et al., 2015). Interestingly, despite of 130 Amblyomma species reported world widely (Guglielmone et al., 2014), the association of Amblyomma ticks with the transmission of piroplasmid agent has been only rarely reported (Reichard et al., 2010, Scoles et al., 2011). This scenario places the Amazon as an environment to break paradigms related to tick-borne piroplasmids, namely a diversity of piroplasmid agents potentially transmitted by Amblyomma ticks.

Among the four Hepatozoon haplotypes detected in this study, two were reported in reptiles (caiman and mud turtle). The caiman (C. crocodilus) haplotype was identical to an unpublished haplotype associated with a caiman species (Caiman yacare) from Pantanal. Several studies reported that the infection by the species H. caimani was very common among C. yacare populations in Pantanal (Lainson et al., 2003, Viana and Marques, 2005, Viana et al., 2010). One of these studies also reported H. caimani infection in C. crocodilus from the Brazilian Amazon (Lainson et al., 2003). While taxonomic identification of the parasite in these previous studies relied solely on morphological analyses (no molecular analyses were done), it is very likely that the Hepatozoon species found in C. crocodilus in the present study refers to H. caimani. Our finding of H. felis in a free-ranging ocelot (L. pardalis) is corroborated by several previous reports of this protozoan in wild felids, including free-ranging jaguars (P. onca) in the Brazilian Amazon (Furtado et al., 2017b) and ocelots and oncilla (Leopardus tigrinus) from northeastern Brazil (Metzger et al., 2008). Finally, our reports of Hepatozoon haplotypes in mud turtle and paca possibly refer to new species, yet to be properly characterized.

This study screened the presence of tick-borne protozoa in a number of wildlife species from the Amazon for the first time. Our results indicate that a variety of genetically distinct Piroplasmida and Hepatozoon organisms circulate under natural conditions in the Amazonian wildlife. Indeed, these results add relevant information for future studies concerning the wildlife conservation of the Amazonian region, and should also be important for understanding potentially emerging tick-borne pathogens that could be shared by the Amazonian wildlife, domestic animals, and humans. Further studies employing morphological and deeper molecular analyses are required to provide a more precise taxonomic identification of the protozoan agents that were detected in Amazonian wildlife during the present study.

Conflicts of interest

The authors declare no conflicts of interest.