A new piroplasmid species infecting dogs: morphological and molecular characterization and pathogeny of Babesia negevi n. sp.

Introduction Babesiosis is a protozoan tick-borne infection associated with anemia and life-threatening disease in humans, domestic and wildlife animals. Dogs are infected by at least six well-characterized Babesia spp. that cause clinical disease. Infection with a piroplasmid species was detected by light microscopy of stained blood smears from five sick dogs from Israel and prompted an investigation on the parasite’s identity. Methods Genetic characterization of the piroplasmid was performed by PCR amplification of the 18S rRNA and the cytochrome c oxidase subunit 1 (cox1) genes, DNA sequencing and phylogenetic analysis. Four of the dogs were co-infected with Borrelia persica (Dschunkowsky, 1913), a relapsing fever spirochete transmitted by the argasid tick Ornithodoros tholozani Laboulbène & Mégnin. Co-infection of dogs with B. persica raised the possibility of transmission by O. tholozani and therefore, a piroplasmid PCR survey of ticks from this species was performed. Results The infected dogs presented with fever (4/5), anemia, thrombocytopenia (4/5) and icterus (3/5). Comparison of the 18S rRNA and cox1 piroplasmid gene sequences revealed 99–100% identity between sequences amplified from different dogs and ticks. Phylogenetic trees demonstrated a previously undescribed species of Babesia belonging to the western group of Babesia (sensu lato) and closely related to the human pathogen Babesia duncani Conrad, Kjemtrup, Carreno, Thomford, Wainwright, Eberhard, Quick, Telfrom & Herwalt, 2006 while more moderately related to Babesia conradae Kjemtrup, Wainwright, Miller, Penzhorn & Carreno, 2006 which infects dogs. The piroplasm forms detected included tetrads (Maltese cross), merozoite and trophozoite stages whose average size was larger than stages of other canine Babesia spp. belonging to the Babesia (s.l.) and B. gibsoni Patton, 1910, and smaller than other canine Babesia (sensu stricto) spp. Of 212 O. tholozani ticks surveyed, 11 (5.2%) harbored DNA of the new species of Babesia. Conclusions Babesia negevi n. sp. is described based on morphological and genetic characterization and phylogenetic analyses. The species is named after the Negev desert of southern Israel, where the first infected dog originated from. Despite co-infection in four dogs, the fifth dog had fatal disease attesting that B. negevi n. sp. infection requires clinical attention. Incriminating O. tholozani or another tick species as the vector of Babesia negevi n. sp., would require additional studies.


Background
Babesia Starcovici, 1893 is a tick-borne protozoan genus classified in the phylum Apicomplexa, class Piroplasmea and order Piroplasmida. Babesia spp. infect domestic and wildlife animals, and humans, and cause severe hemolytic disease [1]. Molecular phylogeny corroborates the taxonomic entities of Babesia (sensu stricto) as a monophyletic group referred to as Clade VI while in contrast, Babesia (sensu lato) parasites represent a complex of species that can be assigned to at least two other monophyletic piroplasmid groups, designated as Clade I ("Babesia microti-like piroplasmids") and Clade II (western group) [2].
Domestic dogs are infected with several Babesia spp. that cause severe disease and have been characterized genetically and described in detail. These include: (i) Babesia rossi Nuttall, 1910 [3][4][5]. The first three Babesia spp. present large intraerythrocytic merozoites, which are morphologically identical when examined by light microscopy while the other three species are smaller and differ in sizes and shapes among themselves. These six species vary in the severity of clinical manifestations which they cause, their tick vectors, genetic and antigenic characteristics and geographical distributions. Most of these species are known or presumed to be transmitted by hard ticks of the family Ixodidae [3][4][5].
Babesia sp. infection was detected by light microscopy of stained blood smears from five dogs in Israel. PCR of marker genes and phylogenetic analysis inferred the presence of a previously unknown species of Babesia in the blood of all dogs. Four of the dogs were co-infected with Borrelia persica (Dschunkowsky, 1913) Steinhaus, 1946, a relapsing fever agent which causes disease in humans, dogs and cats, and is transmitted by the argasid tick Ornithodoros tholozani Laboulbène & Mégnin [6][7][8]. This led to the hypothesis that this previously unknown Babesia sp., which was not identified in earlier molecular studies of pathogens transmitted by hard ticks in Israel and its vicinity [9,10], might be transmitted by O. tholozani. We therefore surveyed soft ticks in Israel for the presence of piroplasms.
The aim of this study was to describe morphologically and characterize genetically the previously unknown piroplasmid species found to infect dogs in Israel.

Animal samples
Piroplasms were detected by light microscopy in blood smears stained by Romanowsky staining solutions in five dogs from Israel whose blood was tested at the Koret School of Veterinary Medicine Laboratory for vectorborne diseases in Rehovot, Israel, during 2012-2016. The medical history, physical examination findings and results of complete blood count (CBC), as well as other diagnostic and laboratory test results were extracted from the medical records of all dogs ( Table 1).

Collection of blood and examination of parasite morphology
Blood was collected by venipuncture of the jugular or cephalic veins of dogs into EDTA and clot tubes for hematology and serum biochemistry, respectively. Stained blood smears were examined by oil immersion light microscopy at 1000× magnification using the Nikon Eclipse N-U microscope, fitted with the Nikon DS-Ri1 camera (Nikon Corporation, Tokyo, Japan) operated by the NIS-Elements F software package (Nikon Corporation, Tokyo, Japan). Parasites were measured and photographed. All measurements are in micrometres and are given as the range followed by the mean ± standard deviation (SD) in parentheses.
Additionally, PCR was also performed to test the dogs for co-infection with relapsing fever Borrelia spp. [14], Ehrlichia spp. [15] and Hepatozoon spp. [16] (Table 2). DNA from dogs infected with B. vogeli, B. persica, E. canis and H. canis were used as positive control for the respective PCRs. DNA from a laboratory-bred dog PCR negative for Babesia, Borrelia, Ehrlichia and Hepatozoon spp. was used as a negative control and a non-template negative control (NTC) was also included in each PCR run. Positive DNA amplicons were purified (EXO-Sap, New England Biolabs Inc., Ipswich, MA, USA) and sequenced in the Center for Genomic Analyses at the Hebrew University (Jerusalem, Israel) using the BigDye Terminator cycle from Applied Biosystems ABI3700 DNA Analyzer. The ABI Data Collection and Sequence Analysis software (ABI, Carlsbad, CA, USA) was used for analysis. DNA sequences were compared to other sequences deposited on GenBank using the BLASTn website hosted by NCBI,

Soft ticks
Argasid ticks were trapped in four caves in Israel (Beit Guvrin, Lavi, Canada Park and Nitzana) where O. tholozani ticks had been previously detected. Carbon dioxide traps were used to collect O. tholozani ticks as previously described [6]. Briefly, three collector traps connected to a cool box emitting CO 2 from dry ice were buried in the soil in each sampling site and left overnight. On the next morning, all trapped ticks were collected with tweezers, kept in vials with 70% ethanol and brought to the laboratory for analysis. All ticks were identified morphologically as O. tholozani and subsequently confirmed by PCR targeting a 460-bp segment of the tick mitochondrial 16S rRNA gene followed by DNA sequencing [17] ( Table 2). DNA was extracted from ticks using a commercial kit (DNeasy Blood & Tissue Kit, Qiagen, Hilden, Germany) following the manufacturerʼs protocol. PCR to detect the presence of Babesia spp. DNA in ticks was done using the piroplasmid PCR and primers ( Table 2). All positive DNA amplicons were sequenced as described above and identified using BLASTn.

Phylogenetic analyses
Phylogenetic trees were constructed based on marker gene sequences determined in this study and corresponding relevant sequences of other Babesia spp. deposited previously in the GenBank database. Following nucleotide sequence alignment using MUSCLE, maximum likelihood (ML) and neighbor-joining trees were inferred using MEGA version X [18]. Percentages of replicate trees as determined by 1000 bootstrap replicates (bs) are shown next to branches. A bs ≥ 85 was considered to provide strong support. Additional details of alignment and tree construction are described herein for the two ML phylograms created. The first tree was composed of 41 nearly complete 18S rRNA gene sequences of analyzed and relevant piroplasmid species, including Cardiosporidium cionae 18S rRNA sequence as the outgroup. After creation of an alignment of 1592 bp in length, a ML tree was constructed based on the TN93+G+I model as estimated by using Aikaikeʼs information criterion (AIC). Based on a discrete gamma distribution consisting of five categories, the shape parameter (G = 0.49) was determined allowing for the existence of invariant sites (I = 50.5%), and the complete deletion option resulted in 1282 positions in the final dataset [19].
A piroplasmid cox1 tree which included 23 partial cox1 gene sequences of the present and other relevant piroplasmid species, with Plasmodium falciparum cox1 sequence as the outgroup was also inferred. After estimation of the GTR+G+I model using AIC, a discrete shape parameter was estimated based on five categories (G = 1.40) allowing for the existence of invariant sites (I = 31.6%), and the complete deletion option resulted in 463 positions in the final dataset [20]. Pairwise distance matrices were generated with multiple global alignment (Needleman-Wunsch algorithm) using the Geneiuos software, version 7.1.9 (Biomatters Ltd., Auckland, New Zealand). Results were calculated as percent identity (p-distance: identical pairs of bases/total number of pairs) [21].

Differential diagnosis
Intraerythrocytic parasites presented in several shapes. Merozoites of B. negevi n. sp. are round to oval ringshaped and appear to be encircled by a basophilic-staining layer of variable width around a less densely stained pale center (Fig. 1b, c). This peripheral basophilic staining layer has single or multiple wider and denser staining chromatin nuclei with larger more mature merozoites often having more than a single dense chromatin nucleus. Mature merozoites divide by binary fission forming pairs of merozoites in the same erythrocyte (Fig. 2f ). The presumptive development of mature merozoites and their division is outlined in Figs. 1 and 2. Small slender zoites such as those arising from disintegrated tetrads invade the erythrocyte (Fig. 2a) and enlarge to develop into early trophozoites (Fig. 2b) which develop further into larger trophozoites with a pointed posterior end (Fig. 1d, e). Trophozoites become round to oval ring-shaped merozoites (Fig. 2c, d) and eventually begin to divide by binary fission (Fig. 2e) until they form two individual rounded merozoites within the erythrocyte (Fig. 2f ). The presumptive development of tetrads is shown in Fig. 3. An early form develops from a rounded parasite which develops four dark-staining projections (Fig. 3a). The rounded form shrinks, leaving four interconnected zoites with prominent nuclei (Fig. 3b) which elongate to form a full Maltese cross tetrad shape (Fig. 3c). The zoites may deform and bend (Fig. 3d, e) until they dissociate from each other (Fig. 3f ) and are ready to break free from their host erythrocyte and infect additional erythrocytes (Fig. 2a).
The forms of B. negevi n. sp. described here from dog erythrocytes are morphologically similar to the forms described for B. conradae described from dogs in California [23], and also bear resemblance to other small canine-infecting Babesia spp. such as B. gibsoni [24] and B. vulpes [5]. Nevertheless, the merozoites of B. negevi n. sp. measuring on average 2.66 × 2.03 µm are larger than the ring forms described for B. conradae which measure 2.2 × 1.85 µm, and also larger than the pyriform shapes of B. conradae which measure 1.38 × 0.66 µm [23]. They are also larger than the merozoites of B. vulpes which measure on average 1.33 × 0.98 µm [5] and the forms of B. gibsoni, measuring 1.9 × 1.2 µm [25]. Conversely, the merozoites of B. negevi n. sp. are distinctly smaller than the merozoites of canine Babesia spp. producing large merozoites including Babesia vogeli, Babesia canis and Babesia rossi, which typically measure 4.5-5.0 × 2.0-2.5 µm. The tetrad form of B. negevi n. sp. measured 4.71 × 3.8 µm which is larger than that reported for B. conradae (2.5 × 2.0 µm) [23]. Importantly, tetrad forms such as those found in B. negevi sp. nov, were observed only in B. conradae [23] and in none of the other species of Babesia spp. infecting dogs.
The above comparisons indicate that B. negevi n. sp. is a distinct form consistent with the small-form piroplasms infecting canines. Babesia negevi n. sp. forms tend to be larger than their respective life stages in B. conradae, B. vulpes and B. gibsoni, and smaller than the large-form Babesia spp. infecting canines including B. canis, B. vogeli and B. rossi.

Molecular phylogeny
The nearly complete piroplasmid 18S rRNA gene (about 1700 bp) was amplified from the blood of dogs no. 1, 2 and 4 (GenBank: MN864546, MN864544 and MN864545, respectively), while a somewhat shorter sequence of 962 bp was amplified from dog no. 3 (Gen-Bank: MN864547). Pairwise comparisons showed a 99-100% identity (Additional file 1: Table S1). All five dogs also yielded shorter 18S rDNA sequences of about 330 bp with the piroplasmid primers PCR protocol ( Table 2) which showed 100% identity with respect to each other (GenBank: MN864539-MN864543). As determined by BLASTn, the closest matches of 94-97% identity to all Babesia spp. 18S rDNA sequences isolated from the dogs in this study included a sequence of a Babesia sp. from a wild meerkat (Suricata suricatta) in South Africa (GenBank: KM025199), and a similarly high identity of 97% was found between sequences for B. negevi n. sp. from dogs no. 1, 3, and 4, and a sequence  A phylogenetic tree based on nearly complete 18S rRNA gene sequences was inferred using the ML algorithm and including B. negevi n. sp. and other piroplasmids present on GenBank representing a wide range of parasites of the order Piroplasmida, including all those that have been found to infect dogs (Fig. 4). Sequences of B. negevi n. sp. were recovered within the western group of Babesia spp. into a distinct, strongly supported sister clade (bs: 100) sister to B. duncani (bs: 84) [26]. Furthermore, the B. negevi n. sp. clade is placed more distantly from the sequences representing the B. conradae clade (bs: 100) and from the clade (bs: 100) represented by B. lengau Bosman, Oosthuizen, Peirce, Venter & Penzhorn, 2010 [27]. Finally, within the western group, a Babesia sp. that has been reported from ruminants and humans (GenBank: AF158705-AF158708), represents a strongly supported sister clade (bs: 100) to all other species and is thus most distantly related to B. negevi n. sp. (Clade II, III) [2,28].   [2,28]. A neighbor-joining phylogram of the same sequences analyzed by ML showed corresponding topology and bootstraps to the ML analysis (Additional file 2: Figure S1).
18S rDNA Babesia sequences were amplified from 13 of 212 (6.1%) studied O. tholozani ticks and submitted to GenBank (GenBank: MN864547-MN864569). Of these, 11 (5.2%) were compatible with B. negevi n. sp., while the remaining two were different, with one sequence (Gen-Bank: MN864559) closely related (96% identity) to a sequence of a Babesia sp. amplified from a dog in Angola (GenBank: KX082919) and a second sequence (GenBank: MN864560) closely related (99% identity) to a Babesia sp. from a badger in China (GenBank: MG799846) (Additional file 1: Table S1).  Table S2). An additional pairwise comparison of a 332 bp sequence of the hypervariable 18S rRNA gene region of B. negevi n. sp., with other closely related Babesia spp. including all known to infect dogs showed that B. negevi n. sp. had a genetic distance of at least 5.42% from the most closely related species B. duncani (Additional file 4: Table S3).
The 18S rDNA sequence of this species exhibits an identity of only 95% with that of B. negevi n. sp. No cox1 sequence of this species was available for comparison.
Overall, the 18S rRNA gene phylogeny constructed using nearly complete gene sequences clearly demonstrated that B. negevi n. sp. represents a novel distinct species that is distinguished from other known piroplasmid species available on GenBank and particularly from those infecting canines. Babesia negevi n. sp. is placed within the western group as a sister clade with respect to B. duncani. Importantly, pairwise comparison of the 18S rRNA gene hypervariable region of B. negevi n. sp. from  [28,31] whereas the second Roman number corresponds to a recently revised novel clade designation [2]. The GenBank accession number, host and country of origin are included for each sequence. Only bootstrap values > 70 are indicated next to branches. The scale-bar represents the evolutionary distance in the units of the number of nucleotide substitutions per site dogs and O. tholozani ticks demonstrated the identity of these isolates. In addition, the phylogenetic tree based on cox1 sequences of B. negevi n. sp. amplified from infected dogs further corroborated the identity and distinct status of this species.
Interestingly, a study on hemoparasites of dogs from the Palestinian Authority, adjacent to Israel, has reported three dogs infected by unidentified Babesia spp. In a phylogenetic analysis of partial 18S rRNA gene sequences, their corresponding sequences segregated jointly with B. conradae into two additional distinct well supported clades within the western group of Babesia spp. [30]. Piroplasmid sequences amplified in the Palestinian study from two dogs and designated 33.5 and 24 clustered into a sister branch to that of B. conradae, whereas an additional piroplasmid sequence from the third dog designated 30.1 represented a more distantly related sister species to B. conradae and piroplasmid spp. 33.5 and 24. Although the piroplasmid sequences originating from these dogs have not been deposited in the GenBank database, we were able to receive these sequences by personal communication (Additional file 5: Table S4). A pairwise comparison of B. negevi n. sp. to the Palestinian dog piroplasmid sequences indicated that sequences 33.5 and 24 displayed a high identity of 99.7% with B. negevi n. sp. whereas sequence 30.1 showed a substantially lower identity of only 92.5% (Additional file 5: Table S4). This finding indicates that sequences 33.5 and 24 most probably represent B. negevi n. sp.

Clinical findings in infected dogs and tick survey
The infected dogs included three males and two females ranging in age from three months to twelve years, including two Labrador Retrievers and three mixed breed dogs (Table 1) and one with normocytic hypochromic anemia (dog no. 5). Four of the five dogs were thrombocytopenic and the fifth dog had a normal thrombocyte count which was close to the low reference interval. The leukocyte count was within the reference interval in four of the five dogs, while moderate leukocytosis was noted in one. Co-infection with B. persica was evident on blood smear (Fig. 1d, e) and confirmed by PCR and DNA sequencing in four of the five dogs, while dog no. 4 was negative for borreliosis by both diagnostic techniques. All dogs were negative by PCR for E. canis, which is a common cause of anemia and thrombocytopenia in Israel, and dog no. 1 was also co-infected with H. canis as detected by PCR and DNA sequencing. All dogs were treated with the antiprotozoal imidocarb dipropionate and with an antibiotic against relapsing fever borreliosis, either doxycycline (dogs no.

Discussion
This study describes B. negevi n. sp. as a new taxon fulfilling the requirements of the ICZN guidelines for a new species [22]. The placement of B. negevi n. sp. in the genus Babesia and its segregation in the western clade of Babesia spp. within this genus is derived from the molecular phylogenetic analysis of the 18S rRNA and cox1 gene sequences. These phylogenetic findings clearly demonstrate the distinct species status of B. negevi n. sp. as corroborated by the demonstration of a single strongly supported clade in each of the constructed phylogenetic trees. The nearly identical 18S rRNA gene fragment sequences amplified from O. tholozani ticks and dogs from different locations indicate that B. negevi n. sp. is spread geographically in several areas in Israel. The description of two dogs from the Palestinian study [30] infected with previously unidentified Babesia sp. which upon analysis performed in this study showed a similar identity of 99.7% to B. negevi n. sp. further supports the information on the spread and geographical distribution of this species. The placement of B. negevi n. sp. as a sister clade of B. duncani within the western group of Babesia as demonstrated by the nearly complete 18S rRNA gene tree provides an interesting insight on the global spread of the western group babesiae. Babesia duncani, a species that infects humans in western North America [26], appears to be the closest known relative of B. negevi n. sp. by both 18S rRNA and cox1 phylogenetic analyses. Babesia conradae, which infects dogs in California, is another western group species found to be somewhat less closely related to B. negevi n. sp. [23]. Babesia duncani and B. conradae segregate into moderately related independent clades which are both strongly supported (bs: 100). Importantly, both of these species have been originally identified in the western part of the North American continent. Thus, these two species have not yet been reported in Israel nor in any other geographical region of the Middle East. As the understanding of the different clades within the Piroplasmida is evolving with more research on the genetics and life-cycles of different parasites [2,13,28,31], it is possible that the current genus Babesia may be split in the future to reflect clade differences.
Babesia negevi n. sp., B. conradae and B. duncani all produce tetrads. This distinguishes B. negevi n. sp. and B. conradae from the other described Babesia spp. that infect dogs. Nevertheless, the average size of B. conradae's tetrads is smaller than the average size described for the tetrads of B. negevi n. sp., which is also in agreement with the size differences of other comparable life stages of these two species such as merozoites.
The vector of B. negevi n. sp. is currently unknown and the detection of its DNA in O. tholozani in the present study is not sufficient to prove that this tick species is its vector. However, the fact that four of five dogs with B. negevi n. sp. infection were co-infected with B. persica, for which O. tholozani is a vector [6], and that previous molecular surveys of Babesia spp. in hard ticks in Israel and the adjacent Palestinian Authority have not reported the presence of DNA compatible with B. negevi n. sp. strengthen the idea that O. tholozani could be its vector [9,10].
Although ixodid ticks are the known vectors of Babesia spp. [1], the bat soft tick Argas vespertilionis is a suspected vector of Babesia vesperuginis [32], the only piroplasm currently known to infect bats, which is genetically closely related to B. conradae [33]. Furthermore, Ornithodoros moubata soft ticks infected by B. gibsoni by injection of the parasite into the tick's hemocoel have been shown to transmit B. gibsoni to dogs after attaching to their skin and feeding on them [34]. An earlier attempt to infect O. moubata by feeding through a parafilm membrane on horse blood infected with T. equi in that same