Molecular Analysis Revealed that Namibian Cheetahs (Acinonyx Jubatus) are denitive hosts of a so far undescribed Besnoitia species

Background: Besnoitia darlingi, B. neotomofelis and B. oryctofelisi are closely related coccidian parasites with felids as denitive hosts. These parasites use a variety of animal species as intermediate hosts. North American opossums (Didelphis virginiana), North American southern plains woodrats (Neotoma micropus) and South American domestic rabbits (Oryctolagus cuniculus) are intermediate hosts of B. darlingi, B. neotomofelis and B. oryctofelisi, respectively. Based on conserved regions in the Internal Transcribed Spacer-1 (ITS-1) sequence of the ribosomal DNA (rDNA), a real-time PCR for a sensitive detection of these Besnoitia spp. in tissues of intermediate hosts and faeces of denitive hosts has recently been established. Available sequence data suggest that species such as B. akodoni and B. jellisoni are also covered by this real-time PCR. It has been hypothesised that additional Besnoitia spp. exist worldwide, which are closely related to B. darlingi or B. darlingi-like parasites (B. neotomofelis, B. oryctofelisi, B. akodoni or B. jellisoni). Related but not closely related to these species is B. besnoiti, the cause of bovine besnoitiosis. Methods: Faecal samples from two free-ranging cheetahs (Acinonyx jubatus) from Namibia that had previously tested positive for coccidian parasites by coproscopy, were used for this study. A conventional PCR veried the presence of coccidian parasite DNA. To clarify the identity of these coccidia, the faecal DNA samples were further characterised by species-specic PCRs and Sanger sequencing. Results: One of the samples tested positive for B. darlingi or B. darlingi-like parasites by real-time PCR, while no other coccidian parasites including Toxoplasma gondii, Hammondia hammondi, H. heydorni, B. besnoiti, and Neospora caninum were detected in the two samples. The rDNA of the B. darlingi-like parasite was amplied and partially sequenced. Comparison with existing sequences in GenBank revealed a close relationship to other Besnoitia spp., but showed also clear divergences. Conclusions: Our results suggest that a so far unknown Besnoitia species exists in Namibian wildlife, which is closely related to B. darlingi, B. neotomofelis, B. oryctofelisi, B. akodoni or B. jellisoni. The cheetah appears to be the denitive host of this newly discovered parasite, while a prey species of the cheetah may act as intermediate hosts.

It was therefore recently argued that other, so far unknown, Besnoitia species may exist in other parts of the world [16]. The high level of conservation in the Internal Transcribed Spacer-1 (ITS-1) sequence of the ribosomal DNA (rDNA) among related species of B. darlingi and B. darlingi-like species (B. neotomofelis, B. oryctofelisi B. akodoni, B. jellisoni) was utilised to establish primers and a probe for the detection of such Besnoitia spp. parasites in their intermediate (e.g. rodents, lagomorphs and marsupials) or in de nitive hosts (e.g. wild felids or canids) [16]. This real-time PCR is not able to detect Besnoitia spp. of cattle, goats, donkeys and horses, and caribou and reindeer, such as B. besnoiti, B. caprae, B. bennetti and B. tarandi, respectively [16].
Our previous work suggested the presence of Toxoplasma gondii in most and B. besnoiti and Neospora caninum in few of 12 Namibian wildlife species, including six Feliformia, four Caniforma and two Ruminantia [17]. Felids, including cheetahs (Acinonyx jubatus), are known as de nitive hosts of T. gondii, but for B. besnoiti, the causative agent of bovine besnoitiosis, the de nitive host is unknown, although wild felids have been discussed as candidates [17,18]. Since morphological identi cation of coccidian parasites is challenging, we examined with molecular methods the faeces collected from the ampulla recti of two free-ranging cheetahs for B. besnoiti. In a previous study, these two cheetahs had been identi ed positive for coccidian oocysts by coproscopy [19].

DNA extraction
In this study, we used faecal samples from two free-ranging female cheetahs, one sub-adult and one adult, from farmland in central Namibia. The animals had tested positive for coccidian parasites by coproscopy [19]. Capture and handling of the animals, sample collection, transport and storage has previously been described [17,19,20]. The Quick-DNA Fecal/Soil Microbe DNA Miniprep Kit (Zymo Research Europe GmbH, Freiburg, Germany) was used to extract DNA from ~200 mg aliquots according to the manufacturer's recommendations. From the faecal sample of the sub-adult female, two aliquots were available, which were independently extracted. Extraction typically yields 100 μl DNA per faecal sample [16].

Endpoint PCR
To test for coccidian parasites, a PCR was performed using the common apicomplexan small subunit ribosomal DNA (18S rDNA) primers COC-1 and COC-2 [21,22]. Hammondia heydorni DNA was tested using the primers JS4/JS5 as described [23,24]. Due to the high level of sequence identity in the rDNA target, the primer pair JS4/JS5 was expected to amplify also DNA of H. tri tae, a coccidian parasite using foxes as de nitive hosts [25,26,27].
For the identi cation of coccidian parasites by Sanger sequencing, rDNA was ampli ed by end-point PCR using primer pairs ( Fig.1) as previously published [23,28,29] and listed in Additional le 1: Table S1.
For all PCRs, primers were used at a nal concentration of 0.5 mM and dNTPs at a nal concentration of 250 mM each (Stratec Molecular GmbH, Berlin, Germany). Taq polymerase (Stratec Molecular GmbH, Berlin, Germany) had a nal concentration of 1 U/25 µl applying the buffer system supplied with the enzyme. The PCR cycling conditions were 94 °C for 5 min, followed by 10 cycles of 56 °C (with 0.5 °C decrement per cycle after the 1 st cycle) for 1 min, 72 °C for 1 min, and 94 °C for 1 min, followed by 40 cycles of 51 °C for 1 min, 72 °C for 1 min, and 94 °C for 1 min. The PCR ended with an incubation at 51 °C for 1 min and a nal extension at 72 °C for 5 min.

Real-time PCRs
Real-time PCRs were used to test for T. gondii, H. hammondi, B. besnoiti, B. darlingi and B. darlingi-like parasites, or N. caninum. T. gondii was examined as previously reported targeting the TgREP-529 [30,31]. H. hammondi was diagnosed using a recently published real-time PCR targeting HhamREP-529 [22]. N. caninum DNA was examined using a previously published real-time PCR targeting Nc5 [31]. In case of B. besnoiti, a fragment of the ITS-1 region in the rDNA was ampli ed as described (BbRT1; [32]). For B. darlingi and B. darlingi-like parasite, a recently published real-time PCR designated BdanjoRT1 PCR was applied [16]. For the detection of N. caninum DNA a previously published real-time PCR [31,33] was used.
To monitor inhibition of the real-time PCRs, a heterologous plasmid with DNA sequences resembling the enhanced green uorescent protein (EGFP) gene [34] was added to the reaction mix in all real-time PCRs except for N. caninum. The internal control PCR included the primers EGFP1-F, EGFP2-R, and the probe EGFP1 [22]. A 712 bp fragment of the EGFP gene was ampli ed and cloned into the pGEM-Teasy standard cloning vector (Promega, Walldorf, Germany) in reverse orientation to obtain the internal control (IC) DNA (pGEM-EGFP2-rev). The amount of the IC DNA added to each reaction was adjusted so that it resulted in a cycle of transition (Ct) value of approximately 32 in the real-time PCR.
Reactions were performed in a nal volume of 20 µl using a commercial master mix (PerfeCTa MultiPlex qPCR ToughMix, Quantabio, VWR International, Darmstadt, Germany) and a CFX384 instrument (Biorad Laboratories GmbH, Munich, Germany). Primers and probes were purchased from MWG-Biotech (Ebersberg, Germany). Standard concentrations for primers (500 nM) and probes (100 nM, target speci c primers; 160 nM, EGFP1) were applied. The cycling conditions in real-time PCR were 95.0 °C (5 min, initial denaturation), followed by 45 cycles, during which the samples were rst incubated at 95.0 °C for 10 s and then at 60.0 °C for 30 s. After each cycle, the light emission by the uorophore was measured. Real-time PCR results were analysed using the CFX manager software Version1.6 (Biorad Laboratories GmbH, Munich, Germany).

Cloning
For Sanger sequencing of ampli cation products, bands of the expected size were excised from agarose gels and puri ed with a commercial kit (NucleoSpin® Gel and PCR Clean-up, Macherey-Nagel, Düren, Germany), following the manufacturer's instructions. Puri ed ampli cation products were then cloned into a commercially available vector (pGEM®-T Easy Vector System I, Promega, Mannheim, Germany) and used to transform chemically competent Escherichia coli (OneShot TOP10, Thermo Fisher Scienti c, Langenselbold, Germany). The transformed E. coli were cultivated and the plasmid DNA was subsequently collected using a commercial kit (QIAprep Spin Miniprep Kit, Qiagen, Hilden, Germany) according to the manufacturer's instructions. Sequencing was performed using the BigDye Terminator v1.1 Cycle Seq. Kit (Thermo Fisher Scienti c, Langenselbold, Germany) and passage through NucleoSEQ Columns (Macherey-Nagel, Düren, Germany) for cleaning up nucleic acids, in an ABI 3130 capillary sequencer (Thermo Fisher Scienti c, Langenselbold, Germany).
The forward and reverse sequences were aligned, if necessary trimmed based on primer sequence information and the consensus sequences for the individual cloned ampli cation products compared to sequences stored in GenBank, EMBL, DDBJ, or RefSeq using BLASTn with standard conditions.

Phylogenetic analysis
The evolutionary history based on ITS-1 rDNA sequences was inferred using the Maximum Parsimony (MP) method. The number of base substitutions per site between sequences included into the analysis was termed "pairwise distance" or "evolutionary divergence" in the following. The MP tree was obtained using the Subtree-Pruning-Regrafting (SPR) algorithm [35] with search level 0, in which the initial trees were obtained by the random addition of sequences (10 replicates). All codon positions (1 st , 2 nd , 3 rd , Noncoding) were included. Evolutionary analyses were conducted in MEGA X [36].

Results
3.1. Polymerase chain reactions (PCRs) for coccidian parasites hammondi, and B. besnoiti. The unlikely presence of N. caninum or H. heydorni was also excluded by real-time or endpoint PCR. Using a real-time PCR established to detect B. darlingi and B. darlingi-like parasites [16], we observed in both sample aliquots of the sub-adult female cheetah a positive signal (Ct 28.3 or 31.9). This suggests that genomes equivalent to 10-100 tachyzoites were present in 10 µl of the 100 µl DNA extracted from this faecal sample. In the sample of the adult female cheetah no signal was observed, although the IC real-time PCR revealed no inhibition. To identify this parasite, we partially characterised its rDNA using overlapping ampli cation products for parts of the 18S rDNA, the ITS-1 rDNA and parts of the 5.8S rDNA ( Fig. 1; Table 1; Additional le 1: Table S1). The sequences of cloned amplicons were analysed by BLASTn with recording the ve top species hits using BLASTn suite (Max Score). The amplicons of four different targets, using the primer pairs JS4-TIM11, 18S-5F-TIM11, BbGS4F-BdanjoRev, and BbGS5F-BbGS5R revealed sequences related to the candidate of a Besnoitia species (Table 1) (Table 1), which suggests that a parasite closely related to B. darlingi or B. darlingi-like parasites had been excreted as oocysts by the Namibian cheetah. The sequences of JS4-TIM11 (n=6), 18S-5F-TIM11 (n=3) and BbGS4F-BdanjoRev (n=3) clones, i.e. clones covering the ITS-1 sequence were aligned and the consensus sequence stored at GenBank (MW468050) using the parasite isolate designation "Besnoitia-acinonyx". The B. darlingi-like sequences ampli ed by BbGS5F-BbGS5R (i.e. a part of the 18S rDNA) was also stored in GenBank (MW559556).

Characterisation of 18S and
Some of the remaining sequences had only coccidian parasites among the rst ve species hits. However, these species hits were dominated by Cystoisospora spp. This may suggest that Cystoisospora spp. had been present in the faecal samples in addition to the B. darlingi-like parasites ( Table 1, coccidia-related).

Discussion
In this study, we examined archived faecal samples of two free-ranging cheetahs from farmland in central Namibia. Coccidian parasites had been identi ed by coproscopy in these samples previously [19]. Using a coccidia-speci c PCR, the microscopic observations were con rmed for both animals.
We originally expected T. gondii or H. hammondi in the cheetahs, as these parasites are known to use felids as de nitive hosts [37]. Antibodies against the tachyzoite stage of B. besnoiti had been detected in blue wildebeest (Connochaetes taurinus) and lions (Panthera leo) in Namibia by serology [17]. Thus, in addition, we suspected that B. besnoiti might be present in faecal samples of Namibian cheetahs because felids might be a de nitive (and/or intermediate) host of B. besnoiti [17,18]. In southern Africa, the existence of B. besnoiti, which uses cattle as intermediate hosts, has long been known (summarised in [38]). B. besnoiti-like parasites were previously isolated from or observed in prey animals of cheetah, such as blue wildebeest, impala (Aepyceros melampus), and kudu (Tragelaphus strepsiceros) in South Africa [39].
For reasons of completeness, DNA extracted from the faecal samples was also examined for N. caninum and H. heydorni, although these are parasites of dogs, dingoes, wolves or coyotes [27,40] and for B. darlingi, which uses marsupials as intermediate hosts, and related parasites [16].
Our previously developed real-time PCR, established to identify B. darlingi, B. neotomofelis, B. oryctofelisi, B. akodoni, and B. jellisoni in intermediate and de nitive hosts, tested positive with two sample aliquots of one cheetah (Ct 28.3 and Ct 31.9). When this newly developed real-time PCR was reported, we hypothesised that further B. darlingi-related parasites might exist worldwide and might be picked up by this PCR [16]. Since all B. darlingi-like parasites known so far were detected in North or South America, it was unlikely that the B. darlingi-like PCR signal that was observed in the faeces of the cheetah belonged to B. darlingi or one of the American B. darlingi-like parasites. Therefore, this study reports rst evidence for the existence of an additional Besnoitia sp. in southern Africa.
A part of the rDNA (18S rDNA) and in particular the ITS-1 rDNA sequence of this parasite was characterised in more detail. In analogy to other coccidian parasites such as T. gondii, we expected that the rDNA sequence was present more than 100-times in the genome of a single parasite organism [41], which makes the rDNA gene and particularly the ITS-1 region a sensitive target for species identi cation. Since there were no puri ed oocysts from the faecal samples available, it was di cult to identify or amplify Besnoitia sp. DNA selectively from the plethora of organisms (most likely bacteria and fungi) present in the faecal samples. The morphological description of the contents of this faecal sample had reported oocysts of approximately 18 x 18 µm in size (Seltmann et al., unpublished data). This suggested that in addition to oocysts of B. darlingi-like parasites with an expected size of 10 × 12 μm (B. darlingi, [1,5]), 11 x 12 µm (B. oryctofelisi, [1,2]), or 13 × 14 μm (B. neotomofelis, [2]), other coccidian parasites, probably Cystoisospora sp., were also present in the sample. Oocysts with a diameter of 10 -14 µm (expected for Besnoitia sp.) had not been reported in the previous coproscopy study [19] but may have been overlooked. Thus, future studies are necessary to isolate oocysts of this parasite and to determine the oocysts size of this B. darlingi-related parasite reliably. In addition to the oocyst sizes observed, the sequences of 18S rDNA fragments ampli ed suggest that Cystoisospora sp. were also present in these faeces. However, the observation of such sequences is not a nal prove of the existence of Cystoisospora sp., because 18S rDNA sequences are largely conserved and particular sequence fragments of the 18S rDNA can belong to many different coccidian parasites [42]. Further, a substantial number of sequences clearly showed the presence of fungi or bacteria, which need to be separated from those sequences, which were unambiguously in line with those of Besnoitia spp. or other coccidia.
As several organisms (including other coccidia) may have been present in the faeces, we concentrated on sequences that belonged unambiguously to B. darlingi-like parasites. The BdanjoRev primer [16], which had been applied in the B. darlingi real-time PCR, played in combination with the BbGS4F primer [28] a central role in the identi cation of the correct Besnoitia-like rDNA sequences [16]. Using the previously published primer pair JS4 [23] and TIM11 [29], as well as the newly established primer 18S-5F in combination with TIM11, we exclusively observed B. darlingi-like sequences, which we aligned and made available as a provisional rDNA sequence of "Besnoitia-acinonyx" (MW468050).
The ITS-1 region of the rDNA Besnoitia spp. of New World marsupials, rodents, and domestic rabbits shows only a few differences [5,43]. The ITS-1 rDNA sequence of the Besnoitia sp. observed in the Namibian cheetah was similar to previously described ITS-1rDNA sequences, but differed from all B. darlingi and B. darlingi-like sequences described so far. Identities of the ITS-1 rDNA of "Besnoitia-acinonyx" with B. darlingi and B. darlingi-like parasites were ≤ 90% (i.e., 89.7-90.1% (B. darlingi), 89.7% (B. oryctofelisi), 88.9% (B. akodoni) and 86.9% (B. jellisoni and B. neotomofelis)). Thus, it appears to be justi ed to conclude that this sequence belongs to a so far unknown Besnoitia sp. that uses the cheetah as its de nitive host. Since no free-ranging marsupials exist in Africa, rodent or lagomorph species, which are prey for cheetahs, probably serve this parasite as intermediate hosts. In analogy to the South American B. oryctofelisi, lagomorphs such as the Cape hare (Lepus capensis), the Savanna hare (L. microtis), or the Scrub hare (L. saxatilis) may represent suitable intermediate hosts.
The ITS-1 rDNA sequence of "Besnoitia-acinonyx" suggested a closer relationship to the American B. darlingi and B. darlingi-like parasites then to B. besnoiti, which infects cattle and probably also antelopes in southern Africa. This nding suggests that all B. darlingi or B. darlingi-like parasites, regardless of their American or African origin and their ability to infect marsupials or placental mammals have a common ancestor, which evolved, when the South American and the African continents were not yet disconnected, i.e. 100-200 million years ago. Most likely, this common ancestor evolved together with marsupial and placental mammalian animals, which started to separate also around this time [44]. Marsupial mammals are the closest living relatives to placental mammals; they share a common ancestor that lived approximately 130 million years ago [44].
No species of the genus Besnoitia has been observed in Australian marsupials such as kangaroos (Family Macropodidae) and possums (suborder Phalangeriformes). These species have evolved from the South American marsupials and probably reached the Australian continent via Antarctica, when it was not yet separated from South America. Australia lacked felids as de nitive hosts for a long time [45], which may be the reason why Besnoitia spp. may have only recently (i.e. for less than the past 200 years) established on this continent and were reported in rats as intermediate and cats as de nitive hosts [13]. The presence of Besnoitia spp. in Australian marsupials might still be possible, as Besnoitia spp. infections might have been established, even in absence of felids, if they are able to use other unknown species as de nitive hosts.

Conclusion
Molecular analysis of a faecal sample revealed that Namibian cheetah (Acinonyx jubatus) is most likely a de nitive host of a newly described Besnoitia species. This species is closely related to B. darlingi and other related Besnoitia spp. parasites of rodents and lagomorphs. Future studies are needed to identify its natural intermediate host in southern Africa, which most likely is a common prey of the Namibian cheetah. Hares, rabbits, and rodents represent possible intermediate host candidates to be further examined.