New Trypanosoma evansi Type B Isolates from Ethiopian Dromedary Camels

Background Trypanosoma (T.) evansi is a dyskinetoplastic variant of T. brucei that has gained the ability to be transmitted by all sorts of biting flies. T. evansi can be divided into type A, which is the most abundant and found in Africa, Asia and Latin America and type B, which has so far been isolated only from Kenyan dromedary camels. This study aimed at the isolation and the genetic and phenotypic characterisation of type A and B T. evansi stocks from camels in Northern Ethiopia. Methodology/principal findings T. evansi was isolated in mice by inoculation with the cryopreserved buffy coat of parasitologically confirmed animals. Fourteen stocks were thus isolated and subject to genotyping with PCRs targeting type-specific variant surface glycoprotein genes, mitochondrial minicircles and maxicircles, minisatellite markers and the F1-ATP synthase γ subunit gene. Nine stocks corresponded to type A, two stocks were type B and three stocks represented mixed infections between A and B, but not hybrids. One T. evansi type A stock was completely akinetoplastic. Five stocks were adapted to in vitro culture and subjected to a drug sensitivity assay with melarsomine dihydrochloride, diminazene diaceturate, isometamidium chloride and suramin. In vitro adaptation induced some loss of kinetoplasts within 60 days. No correlation between drug sensitivity and absence of the kinetoplast was observed. Sequencing the full coding sequence of the F1-ATP synthase γ subunit revealed new type-specific single nucleotide polymorphisms and deletions. Conclusions/significance This study addresses some limitations of current molecular markers for T. evansi genotyping. Polymorphism within the F1-ATP synthase γ subunit gene may provide new markers to identify the T. evansi type that do not rely on variant surface glycoprotein genes or kinetoplast DNA.


Introduction
Surra, a wasting disease caused by Trypanosoma (T.) evansi, is one of the non tsetse-transmitted Animal African Trypanosomoses (AAT) occurring in Ethiopia. The disease imposes significant financial losses due to reduced fertility and mortality and is prohibiting the import of highly productive yet trypanosusceptible cattle breeds [1][2][3]. T. evansi belongs to the subgenus Trypanozoon, that also comprises T. brucei and T. equiperdum [4][5][6]. In terms of geographical distribution, Trypanosoma equiperdum and T. evansi, causing respectively dourine in horses and surra in livestock in Africa, Asia, and South America, have been far more successful than T. brucei, a parasite confined to sub-Saharan Africa where its vector, the tsetse fly, is present [7]. Recent phylogenetic studies suggest that T. evansi and T. equiperdum evolved from T. brucei on several occasions and from genetically distinct T. brucei strains and therefore could be considered as subspecies of T. brucei [8,9].
Trypanosomes are characterised by the presence of a structure called kinetoplast that corresponds with the DNA (kDNA) of their unique mitochondrion. T. brucei kDNA contains 20-50 copies of maxicircles (about 23 kb) and a highly diverse set of thousands of minicircles (about 1 kb). Maxicircles contain rRNA coding regions and genes coding for subunits of the respiratory chain complexes while minicircles code for guide RNAs required for editing [10].
T. evansi type A is the most abundant and is found in Africa, South America and Asia. It is characterised by the presence of the gene for the variant surface glycoprotein (VSG) RoTat 1.2. This RoTat 1.2 VSG is expressed early during infections resulting in the detectability of anti-RoTat 1.2 antibodies in animals infected with T. evansi type A [17,18]. In contrast, T. evansi type B is far less common and has so far been isolated only from camels in Kenya [16,19]. More recently, serological and molecular evidence for the presence of T. evansi type B in Sudan, Ethiopia and Chad was published [20][21][22][23][24]. T. evansi type B lacks the RoTat 1.2 gene and as a consequence, infections with this type are not detected with serological and molecular tests based on RoTat 1.2 VSG, such as the CATT/T. evansi and RoTat 1.2 PCR [15,18,19,25]. So far, three molecular tests have been developed for the identification of T. evansi type B: the EVAB PCR, targeting a type B-specific minicircle DNA sequence, and a PCR and a LAMP targeting a type B-specific VSG JN 2118Hu [15,19,26]. T. equiperdum is the least known parasite of the Trypanozoon group, with very few isolates available for research, albeit new stocks were isolated from Ethiopian and Venezuelan horses recently [24,27].
Unlike T. brucei, T. evansi and T. equiperdum cannot develop in tsetse flies due to their inability to transform into the procyclic life stage. They can only survive in a mammalian host where they produce ATP exclusively through glycolysis. In contrast to bloodstream forms, ATP production in procyclic trypanosomes relies on oxidative phosphorylation and, therefore, on the capacity to express the full set of corresponding mitochondrial genes, including some which are encoded by the kDNA [10,28]. Bloodstream forms of T. evansi, T. equiperdum and laboratory-generated T. brucei strains that have lost all or critical parts of their kDNA, can survive without kDNA due to specific single amino acid mutations in the gamma (γ) subunit of the mitochondrial F1-ATP synthase [28]. Interestingly, the specific mutations/deletions in the C-terminal region of F1-ATP synthase γ subunit enable differentiation among the Trypanozoon strains [8]. Furthermore, when the F1-ATP synthase γ subunits of T. evansi type A (A281del), T. equiperdum (A273P) and the laboratory-generated T. brucei (L262P) strains are overexpressed in a T. brucei γ subunit knock out strain, the latter can survive after loss of its kinetoplast after treatment with DNA intercalating drugs such as acriflavin or ethidium bromide [28,29]. Once the genetically modified T. brucei are independent from kDNA maintenance and expression, they become multidrug resistant to the diamidine and phenanthridine class of drugs [30].
In T. evansi, drug resistance has been reported in several type A strains originating from Africa, Asia and Latin America [31][32][33][34]. Some Chinese strains appear to be innately resistant to the phenanthridine class of drugs [35]. In contrast, nothing is known on the drug susceptibly of the T. evansi type B strains. In a previous study, we reported that T. evansi infections are very common in camels, equines, cattle and small ruminants in Tigray and Afar provinces in Northern Ethiopia [20]. We also provided molecular and serological evidence that both T. evansi type A and type B occur in these provinces. In that study, of those dromedary camels that were parasitologically positive, buffy coat samples were collected and cryopreserved in liquid nitrogen for later isolation of the parasite. We here report on the isolation, adaptation to in vitro culture, genetic and phenotypic characterisation and in vitro drug sensitivity of T. evansi type A and B from Northern Ethiopia.

Ethics statement
The Animal Experimentation Ethics Committee (AEEC) of the Institute of Tropical Medicine (ITM) advised on the protocol for collection of blood samples from dromedary camels (EXT2012-1) and for the isolation of trypanosomes via inoculation of mice (EXT2012-2) at the College of Veterinary Medicine, Mekelle University. The study protocol for in vivo expansion of trypanosomes at ITM was approved by the AEEC (BM2013-1). Collecting blood from camels and experiments on mice were conducted according to the national guidelines of the Ethiopian Ministry of Livestock and Fishery Development and the Institutional Review Board of the Ministry of Science and Technology.
In vivo isolation of parasites from cryopreserved buffy coat in mice Details on the collection and cryopreservation of buffy coat samples from dromedary camels that were parasitologically confirmed in the micro haematocrit centrifugation technique have been fully described elsewhere [20]. Two hundred μl of thawed buffy coat were inoculated intraperitoneally (IP) in two 25-30 g Swiss albino mice that were immunosuppressed with 0.16 μg kg -1 body weight dexamethasone (Shanghai Central Pharmaceutical, China) one day prior to inoculation [36]. Parasitaemia was checked in 5 μl of tail blood using the matching method [37], starting from day 7 post-infection and subsequently on every third day. As soon as trypanosomes were detected in at least one mouse, the animal was anaesthetised (the other kept as a backup), its blood was collected on heparin by heart puncture, diluted in an equal volume of phosphate buffered saline glucose (PSG; 7.5 g/l Na 2 HPO 4 2H 2 O, 0.34 g/l NaH 2 PO 4 H 2 O, 2.12 g/l NaCl, 10 g/l D-glucose, pH 8) and subinoculated into four naïve mice (200 μl each) which were monitored for parasitaemia as described above. Mice used as backup were euthanised when the newly infected mice became positive. When parasitemia reached about ± 10 7.8 cells ml −1 of blood, two of these parasitaemic mice were euthanised (the other two were kept as back up) and blood was taken for subinoculation into four other naïve mice. This protocol was repeated until the parasitaemia reached about 10 8.4 cells ml −1 . At this stage the stock was considered in vivo adapted. All four mice were anaesthetised and exsanguinated by heart puncture in an equal volume of Triladyl-egg yolk-phosphate buffered saline glucose (TEP) cryomedium [38] for cryopreservation in 1 ml aliquots.

In vivo expansion and purification of parasite populations
Cryostabilates were thawed in a water bath at 37°C and diluted in PSG to 1 trypanosome per field (± 10 5.7 cells ml −1 ). Two-hundred μl volumes were injected IP in two naïve 20-30 g female OF-1 mice (Charles River, Belgium). Starting from three days post infection (DPI), parasitaemia was monitored daily and harvested at first peak parasitemia, typically at day 4 to 5 postinfection, as described above. Volumes of 0.5 ml of the blood were run over a mini Anion Exchange Centrifugation Technique (mAECT) column to separate the trypanosomes from the blood [39]. The trypanosomes eluted from the column were washed twice with 5 ml ice-cold PSG by centrifugation at 1500 g for 15 min. After the last centrifugation, the supernatant PSG was discarded and the trypanosome sediment was re-suspended in 100 μl of PSG. Part of this suspension was used for in vitro culture adaptation. The remainder was centrifuged at 1500 g for 5 min and the sediment was frozen at -80°C until DNA extraction. The isolates used for in vivo isolation and expansion and the corresponding T. evansi type A and B specific PCR result on their corresponding buffy coat DNA are indicated in Table 1.
In vitro adaptation in HMI-9 medium with horse serum The highly concentrated trypanosome suspension in PSG was diluted to 2 x 10 5 cells ml −1 in Hirumi's modified Iscove's medium 9 (HMI-9), complemented with 15% (v/v) heat-inactivated foetal bovine serum (Gibco, Belgium) and 5% (v/v) heat-inactivated horse serum (Gibco, Belgium) (abbreviated as HMI-9 (HS)) [40,41]. Parasites were seeded at 2 x 10 4 , 2 x 10 3 and 2 x 10 2 cells ml −1 , in a total volume of 500 μl in a 48-well plate (Nunc, Denmark) and incubated at 37°C and 5% CO 2 . After 72 hours, a well, where trypanosome density had increased above 2 x 10 5 cells ml −1 , was used for further subpassage in 500 μl of HMI-9 (HS). The well with the highest density of viable parasites was then further maintained in HMI-9 without horse serum [40]. When possible, log phase growing in vitro cultures were scaled up in flasks (Nunc, Denmark) to obtain larger numbers of parasites for cryostabilisation, DNA extraction and in vitro drug sensitivity testing [42]. The in vitro growth curves of the different stocks were generated by seeding cells at 1 x 10 4 cells ml −1 in 500 μl of HMI-9 in three replicate wells that were counted every 24 h. The doubling times (T d ) were calculated from the exponential part of the curve using non-linear regression fitted with an exponential equation in GraphPad Prism 6 (Graph-Pad, version 6, USA).

Molecular characterisation of parasite populations
DNA extraction of trypanosome sediments prepared from the in vivo expanded and the in vitro adapted populations was performed with DNA Isolation Kit (Roche Diagnostics, Germany) following the protocol recommended for isolation of DNA from mammalian tissue. From T.b. brucei AnTat 1.1 E , T.b. gambiense LiTat 1.3, T.b. gambiense type II ABBA and T. equiperdum Dodola 940, DNA was extracted using the Maxwell 16 Tissue DNA Purification kit on a Maxwell 16 instrument according to the manufacturer's instructions (Promega, Belgium). DNA concentrations were measured using the Nanodrop ND-1000 UV-Vis spectrophotometer (NanoDrop Technologies, USA) and adjusted to 10 ng μl -1 . A set of PCRs targeting VSG genes (RoTat 1.2 and JN 2118Hu), maxicircle genes (ND4, ND5, ND7 and A6), class A minicircles (miniA PCR) and class B minicircles (EVAB PCR) minisatellites (MORF-2REP), P2 adenosine transporter (AT1) and the F1-ATP synthase γ subunit were adopted to characterise the studied parasite populations [4,15,19,28,[43][44][45]. Where applicable, the published PCR protocols were adjusted to the requirements of the HotStarTaq Plus DNA polymerase (Qiagen, Germany). Primer sequences, reaction mixture contents, cycling conditions and expected amplicon size are described and referenced in Table 2. All PCR amplifications were carried out in 200 μl thin-wall PCR tubes (ABgene, UK) in a T3 thermocycler 48 (Biometra, Germany). Ten μl of amplified products were electrophoresed in 1 to 2% agarose gel at 135 V for 30 min and afterwards stained with ethidium bromide for visualization under UV light. For direct sequencing, PCR was performed in 50-100 μl volumes and amplicons were cleaned up and concentrated using a PCR cleanup kit (QIAquick PCR Purification Kit, Qiagen, Germany) and sent out for bidirectional direct sequencing at the Genetic Sequencing Facility (VIB, Belgium) using the described PCR primers.
The full length sequence of the F1-ATP synthase γ subunit was cloned into a BamHI and HindIII double digested pHD309 vector using the In-Fusion Cloning kit (Clontech, Japan). Primers contained a F1-ATP synthase γ subunit specific sequence based on the T. evansi sequence of STIB 810 (EU185797) and a 5 0 extension of 15 bp specific to the place of integration in pHD309, containing the restriction sites and sequence overlap with the vector, as required for the In-Fusion Cloning reaction. Proofreading-PCR was performed using the Clone-Amp HiFi PCR premix (Clontech, Japan). Amplicons were cleaned up (QIAquick PCR Purification Kit, Qiagen, Germany) before use in the In-Fusion protocol. The reaction products were transformed in Stellar competent cells according to the manufacturer's recommendations (Clontech, Japan). Transformant clones were checked for the presence of insert using colony PCR, cultured in LB medium, plasmid purified (QIAprep Spin Miniprep Kit, Qiagen, Germany) and at least 7 to 12 clones per transformation were bidirectionally sequenced at the Genetic Sequencing Facility (VIB, Belgium) using primers binding to pHD309.

In vitro drug sensitivity testing
Melarsomine dihydrochloride (Cymelarsan, Sanofi Aventis, France) and isometamidium hydrochloride (Veridium, Ceva Santé Animale, Belgium) were prepared as 10 mg ml −1 stock solutions in distilled water. Dophanil powder (Dophanil, Docpharma, Belgium), containing 445 mg diminazene diaceturate and 555 mg antipyrine per gram, was concentrated to a 10 mg ml −1 diminazene diaceturate solution in DMSO (Sigma, Belgium). Suramin (Germanin, Bayer, Germany) was prepared as a 100 mg ml −1 in DMSO. A method to measure the IC 50 values of compounds in 96-well plates was performed as described elsewhere [46]. Briefly, 2 × 10 4 cells ml −1 from in vitro adapted stocks were exposed to seven threefold drug dilutions, ranging from 5000 to 7 ng ml −1 for suramin, 500 to 0.7 ng ml −1 for diminazene diaceturate and from 250 to 0.35 ng ml −1 for melarsomine dihydrochloride and isometamidium hydrochloride, in a total volume of 200 μl of HMI-9 medium. Next, the plate was incubated for 72 hours at 37°C with 5% CO 2 followed by addition of 20 μl of resazurin (Sigma, Belgium; 12.5 mg in 100 ml PBS) for measuring trypanosomes viability. After a further 24 h incubation at 37°C and 5% CO 2 , fluorescence was measured (excitation λ = 560 nm; emission λ = 590 nm) with a VictorX3 multimodal plate reader using top reading (Perkin Elmer, Belgium) [42]. The results were expressed as the percent reduction in parasite viability compared to the parasite viability in control wells without drugs. The 50% inhibitory concentration (IC 50 ) was calculated using non-linear regression fitted with a (log) inhibitor versus normalised response (variable slope) equation (GraphPad, version 6, USA). The IC 50 values obtained from day 30 and day 60 in vitro cultures were compared using t-tests corrected for multiple testing according to the Holm-Sidak method (α = 0.05) (GraphPad, version 6, USA).

Microscopic examination for presence of a kinetoplast in trypanosomes
Trypanosome populations at different stages of in vivo and in vitro expansion were examined for the presence of the kinetoplast using 4',6-diamidino-2-phenylindole (DAPI) staining. Briefly, live trypanosomes in PSG or in vitro culture medium were washed in PBS by centrifugation, deposited onto microscope slides, air dried and fixed with methanol for 30 min. Subsequently, the slides were rehydrated in PBS and mounted in 87% glycerol containing 1 μg ml -1 DAPI (Sigma, Belgium) [28]. Images were captured with an epifluorescence microscope (Olympus BX41, Olympus, Japan) equipped with a NU fluorescent cube (excitation: 360-370 nm and emission > 420 nm)) and Cell^D software (Olympus, Japan). DAPI stained trypanosomes were grouped according to the number of kinetoplasts (K) and nuclei (N) present within each cell. The percentage of kinetoplastic cells in a DAPI stained slide was calculated on the basis of on average 300 examined trypanosomes, by dividing the sum of 1K1N + 2K1N + 2K2N cells by the sum of 1K1N + 2K1N + 2K2N + 0K1N + 0K2N cells. A two-tailed Spearman correlation matrix (using a confidence interval of 95%) was used to find the correlation between the percentage of kinetoplastic cells at day 30 and day 60 of in vitro culture and the respective IC 50 value for a particular drug (GraphPad, version 6, USA).

In vivo infectivity check
To check the in vivo infectivity of trypanosome populations that were cryostabilised after continuous propagation in vitro for 60 days, 5 x 10 6 cells in 300 μl were inoculated in a single OF-1 mouse where after parasitaemia was checked as described above.

Isolation of Ethiopian T. evansi stocks
Thirty cryopreserved buffy coat specimens from parasitologically positive dromedary camels were inoculated in immunosuppressed Swiss albino mice. In total, 22 parasite stocks originating from 22 different animals could be isolated and cryopreserved after 2 to 5 subpassages in mice.

Morphological and genotypic kDNA status of the in vivo expanded stocks
Four PCRs that target maxicircle DNAs, of which three NADH-dehydrogenase subunits (ND4, ND5, ND7) and the ATPase subunit 6 (A6), and two PCRs that target class-specific minicircle sequences (miniA PCR and EVAB PCR) were run on DNA extracts of the purified trypanosomes (Table 3) (Table 1). Additionally, EVAB PCR amplicons were detected in 3 stocks that were also positive for RoTat 1.2 VSG PCR suggesting a mixed infection with type A and B: a strong amplification was present in MCAM/ET/ 2013/MU/15, while a weak amplification was visible in MCAM/ET/2013/MU/11 and 17. The presence of kinetoplasts in the trypanosome cells was demonstrated using fluorescence microscopy with DAPI staining on ex vivo isolated trypanosomes (Table 3). T. evansi RoTat 1.2, T. evansi KETRI 2479 and all but one Ethiopian T. evansi stocks show a kinetoplast in > 96% of the cells. Stock MCAM/ET/2013/MU/09 was found to be akinetoplastic since only the nucleus of the trypanosomes was visible with DAPI.

MORF2-REP minisatellite profile of the in vivo expanded stocks
In T. evansi RoTat 1.2, the MORF2-REP locus consists of 4 and 6 repeats, while in T. evansi KETRI 2479, 3 and 5 repeats were found (Table 3). In vivo expanded Ethiopian stocks of type A had either 1 allele (7 repeats) or 2 alleles (6 and 7 repeats), thus displaying a different pattern than T. evansi type A RoTat 1.   Table 4).
The F1-ATP synthase γ subunit of T.b. gambiense LiTat 1.3 (KT934830) appeared homozygous and identical to the T.b. gambiense DAL972 sequence (Tbg972.10.90). T.b. gambiense type II ABBA (KT934831) appeared homozygous and differed in only 2 SNPs (G801T and A882G) from the T.b. gambiense sequence. T. evansi RoTat 1.2 and the Ethiopian stocks MCAM/ET/2013/MU/04, 06, 09,11 and 13 were heterozygous and revealed in one allele (KT934833), identical to the published full length T. evansi STIB 810 (EU185798) sequence, the deletion of nucleotides A841-843del. The second allele contained a C142T polymorphism (KT934832), that is not present in the wild-type T. evansi STIB 810 sequence (EU185797), but that could be identified in the genome sequence of the Chinese akinetoplastic T. evansi STIB 805 strain [9]. For T. evansi KETRI 2479 and the Ethiopian stocks MCAM/ET/2013/MU/10 and 14 we obtained heterozygous alleles, different from the partial sequence of T. evansi KETRI 2479 (EU185794). The first allele had the unique A844T polymorphism (KT934835), and differed from the second allele in 3 additional SNPs (T321C, T807C, T867G) that were also found in some T.b. brucei and T. equiperdum. Interestingly, the in vivo expanded stock of MCAM/ET/2013/MU/15 revealed alleles that belonged to T. evansi type A and type B. In contrast, when the original buffy coat of this stock was tested, only alleles of T. evansi type A were found. Finally, T. equiperdum Dodola 940 (KT934836) appeared homozygous and its single    allele was identical to one of the two alleles found in T.b. brucei AnTat 1.1 E (KT934837), but differed in 5 SNPs with the sequence from T. equiperdum BoTat 1.1 (EU185793) and in 6 SNPs with T. equiperdum STIB 841 (EU185792). However, for the T. equiperdum STIB 841 strain, 5 of the 6 SNPs were ambiguous polymorphisms that do not rule out similarity to T. equiperdum Dodola 940.
In vitro adaptation of Ethiopian T. evansi stocks Fourteen Ethiopian T. evansi stocks, T. evansi RoTat 1.2 and T. evansi KETRI 2479 were expanded in mice and purified from blood at peak parasitaemia to initiate primary in vitro cultures in HMI-9 (HS) medium. After 96 hours, the initial 2x10 4 cells ml −1 inoculum reached concentrations above 2x10 5  Because the HMI-9 (HS) medium did not support sufficient in vitro culture growth for most of the Ethiopian T. evansi stocks, it was abandoned and replaced with HMI-9 without horse serum. In vitro adapted strains of T.b. brucei AnTat 1.1 E and T.b. gambiense LiTat 1.3 were cultured in HMI-9 in parallel. In vitro cultures were only considered adapted to HMI-9 medium when it was possible to maintain the parasites in continuous proliferation. To this extent, dense parasite cultures, containing 2-5 x 10 5 cells ml −1 , were subpassaged into new wells using serial fivefold dilutions in fresh medium. When these subpassages reached densities above 2 x 10 5 cells ml −1 within a 48-96 hours period, the stock was considered adapted. The five stocks that already grew well in the HMI-9 (HS) medium continued proliferating when inoculated from the dense cultures at serial fivefold dilutions in HMI-9. These five stocks were considered to be in vitro adapted after 15 days of in vitro culture. Out of the four remaining stocks, only MCAM/ET/2013/MU/04 and 10 slowly regained the ability to proliferate in HMI-9 at a reduced subpassaging scheme using serial twofold dilutions. MCAM/ET/2013/MU/04 required 25 days to adapt, while MCAM/ET/2013/MU/10 was only fully adapted after day 35 of in vitro culture. Gradually increasing the culture volume allowed to obtain sufficient parasites from the adapted cultures for in vitro drug testing, DNA extraction, and cryostabilisation at day 30 (all, except MCAM/ET/2013/MU/10) and at day 60 of in vitro culture (all stocks).
DNA of the in vitro adapted stocks was subjected to RoTat 1.2 PCR, EVAB PCR and MOR-F2-REP PCR. All in vitro stocks had similar molecular profiles as their corresponding in vivo expanded parental stocks, except MCAM/ET/2013/MU/15. While the in vivo expanded stock of the latter was identified as a mixed infection of T. evansi type A and type B, the in vitro adapted stock (at day 30 and day 60 in vitro culture) was identified as pure T. evansi type B with the above mentioned PCRs and confirmed by cloning and sequencing of the F1-ATP synthase γ subunit. Thus, beside T. evansi RoTat 1.2 and T. evansi KETRI 2479, we achieved the in vitro adaptation of 2 Ethiopian type A stocks, 2 Ethiopian type B stocks and additionally ended up with a pure T. evansi type B in vitro adapted stock originating from a mixed type A and type B in vivo adapted stock. Growth curves were generated for T.b. brucei AnTat 1.1 E and all seven in vitro adapted stocks (Fig 2). T.b. brucei AnTat 1.1 E and T. evansi RoTat 1.2 had the shortest T d , 7.5 ± 0.3 h -1 and 7.7 ± 0.2 h -1 respectively, and reached the highest maximum population density (MPD) of ± 3-4 x 10 6 cells ml -1 , while T. evansi KETRI 2479 had a longer T d , 10.8 ± 0.2 h -1 , and a lower MPD of ± 1 x 10 6 cells ml -1 . The Ethiopian type A stocks MCAM/ ET/2013/MU04 and MU09 had a T d of 11.2 ± 0.4 and 11.3 ± 0.4 respectively, and a MPD of ± 1 x 10 6 cells ml -1 . Similarly, the Ethiopian type B stocks MCAM/ET/2013/MU10, 14 and 15 had a T d of 12.9 ± 0.5, 11.3 0.5 and 12.1 ± 0.6 respectively, and a MPD of ± 0.7-1 x 10 6 cells ml -1 (Fig 2).

Discussion
Previous molecular and serological studies revealed that trypanosome infections in camels from Northern Ethiopia are caused by either RoTat 1.2 PCR or EVAB PCR positive parasites. In some instances amplicons of both PCRs were detected within the same buffy coat extract, suggesting the occurrence of mixed infections [20]. The present study was undertaken to isolate the trypanosomes from camels carrying apparent single infections through inoculation of their buffy coats in immunosuppressed mice. The in vivo inoculation led to the successful isolation of 22 stocks, out of which 14 were selected on the basis of their geographical origins for further investigations (5 stocks from Tigray and 9 stocks from Afar). Next, we performed an in-depth comparative molecular analysis on DNA extracts from the isolated parasite stocks using diverse PCRs. Furthermore, we analysed the specificity of each of these PCRs on a collection of Trypanozoon strains.
The RoTat 1.2 VSG sequence can be used to characterise T. evansi type A [25,43]. In our collection, all buffy coats positive in RoTat 1.2 PCR yielded in vivo isolated stocks that were RoTat 1.2 PCR positive but that were negative in the maxicircle gene targeting PCRs. Furthermore, with the exception of the akinetoplastic stock MCAM/ET/2013/MU/09, all these strains had type A minicircles. MCAM/ET/2013/MU/09 may be naturally akinetoplastic since the DNA extracted from the original buffy coat was negative in all PCRs targeting kinetoplast DNA. The occurrence of naturally akinetoplastic strains was previously documented in Latin America and China [12][13][14]47]. One stock (MCAM/ET/2013/MU/04) contained a SNP in its RoTat 1.2 VSG PCR amplicon. SNPs in RoTat 1.2 amplicons were previously reported in Egypt but do not necessarily lead to a negative result in RoTat 1.2 based antibody detection tests. This was also the case for the camel from which MCAM/ET/2013/MU/04 was isolated [48,49].
Initially defined by minicircle class B, identification of T. evansi type B is possible with EVAB PCR that amplifies a fragment of this minicircle [15]. Additionally, it was proposed that the VSG JN 2118Hu, first described in a Kenyan T. evansi strain, is a specific marker for T. evansi type B [19].
In our collection, 2 buffy coat extracts that were positive in EVAB PCR yielded in vivo isolated stocks that were EVAB PCR positive as well. Interestingly, an EVAB PCR amplicon was also detected in three additional in vivo expanded stocks that were RoTat 1.2 PCR positive but for which the corresponding buffy coats were EVAB PCR negative. These three stocks might be mixed infections. JN 2118Hu VSG PCR appeared to be less sensitive because it detected only 3 out of 5 EVAB PCR positive isolated stocks. Furthermore, the JN 2118Hu VSG PCR appeared to be less specific since T.b. brucei AnTat 1.1 E and T.b. gambiense type II ABBA were also positive in this PCR. None of the EVAB PCR positive isolated stocks contained maxicircle DNA and they were all negative in miniA PCR, except for the three mixed infections. Therefore, we conclude that we isolated at least two "pure" T. evansi type B stocks from Ethiopian camels, decades after the initial isolation of T. evansi type B from camels in Kenya [15].
We used the minisatellite locus MORF2-REP to verify whether both putative mixed stocks, that were positive in RoTat 1.2 PCR and EVAB PCR, were real mixed infections or hybrids between T. evansi type A and B. The Ethiopian isolates clustered in two classes of T. evansi type A, of which one with a previously described heterozygous profile (6 and 7 repeats) and one with a homozygous profile (7 repeats). The Ethiopian T. evansi type B stocks had a heterozygous profile (3 and 4 repeats) differing from the only known profile described for Kenyan type B isolates (3 and 5 repeats) [50]. In one of the mixed infections we observed a profile that can be interpreted as a mixture of Ethiopian type A and type B, while the others only revealed the Ethiopian type A pattern. These results prove that we are dealing with mixed infections and not with hybrids between T. evansi type A and type B. To exclude that these apparent mixed infections represent cross-contamination with genetic material, we attempted in vitro cultivation of the in vivo expanded stocks.
Previously we have shown that addition of 1,1% methylcellulose to HMI-9 greatly helps the in vitro adaptation of Trypanozoon strains, including T.b. gambiense and T. evansi RoTat 1.2 [40]. However, to avoid the use of this highly viscous medium we preferred the use of horse serum to adapt T. evansi stocks as is suggested in previous reports [51][52][53]. While this approach proved to be successful for all type B stocks, only two out of nine Ethiopian T. evansi type A could be adapted. Interestingly, in the case of mixed stock MCAM/ET/2013/MU/15, this medium selected T. evansi type B out of the mixed population. While only the type A infection was detected in the buffy coat DNA extract, both types could be detected in the in vivo expanded stock DNA, but eventually only type B was detected in the in vitro adapted stock.
Gillingwater and colleagues reported on the drug sensitivity profiles of a panel of T. evansi and T. equiperdum strains where they considered T. evansi STIB 806K to be a reference sensitive strain for suramin (IC 50 70.4 ng ml -1 ), diminazene diaceturate (IC 50 4.5 ng ml -1 ) and melarsomine dihydrochloride (IC 50 1.4 ng ml -1 ). They reported drug resistance in two T. evansi stocks with an IC 50 for suramin > 10000 ng ml -1 (STIB 780 and STIB 781), and in the T. equiperdum OVI strain, with an IC 50 for diminazene diaceturate of 302 ng ml -1 and an IC 50 for melarsomine dihydrochloride of 17.6 ng ml -1 [46]. The only strain that is shared between their panel and our collection is T. evansi RoTat 1.2, which despite different approaches in the experimental testing, yielded corresponding IC 50 values, especially for diminazene diaceturate and melarsomine dihydrochloride, thus facilitating comparison between both studies. In our Ethiopian T. evansi collection, no resistance against melarsomine dihydrochloride was found. However, some stocks appeared to have raised IC 50 values for suramin (> 200 ng ml -1 ) and diminazene diaceturate (> 50 ng ml -1 ). The IC 50 values that we observe for T.b. gambiense LiTat 1.3 and the Ethiopian T. evansi type B are similar to the in vitro IC 50 value of 0.82 ng ml -1 found by Sahin and coworkers for T. congolense IL3000 which is sensitive to isometamidium (Veridium) in vivo [54]. In the same study, an in vitro IC 50 of 11.06 ng ml -1 is reported for T.b. brucei AnTat 1.1 strain, which is slightly higher than the value that we obtained in experiments with our T.b. brucei AnTat 1.1 strain and the other T. evansi stocks [54]. Nevertheless, defining our T. evansi stocks as either sensitive or resistant based solely on the in vitro drug sensitivity results may be too audacious, given the fact that IC 50 values were determined in only one assay, the resazurin viability assay [55][56][57]. Therefore, an in vivo drug sensitivity profile of all our Trypanozoon strains against the commonly used trypanocides remains to be elucidated. Interestingly, both Ethiopian T. evansi type A stocks appear to be less susceptible to suramin, diminazene diaceturate and isometamidium hydrochloride than the three type B stocks. In T.b. brucei, resistance against suramin and isometamidum hydrochloride has been linked to several proteins [58,59], while resistance to diamidine and melaminophenyl classes of drugs is attributed to the transporter protein TbAT1 and the aquaporin AQP2 [60][61][62]. The lower sensitivity to diminazene diaceturate was not caused by mutations in the T. evansi TeAT1 [63].
Interestingly, DAPI staining of the trypanosomes indicated slight to severe loss of the kDNA in all in vitro adapted T. evansi stocks, when compared to in vivo adapted stocks. The loss of kDNA in in vitro cultured T. evansi is a phenomenon that has been known for a long time [10,15,55,64]. Non-vital loss of the kinetoplast is made possible by mutations in the F1-ATP synthase γ subunit of T. evansi allowing to uncouple from the Fo subunit and effectively circumventing the requirement for mitochondrial gene expression [65]. Furthermore, it has been shown that the expression of certain T. evansi F1-ATP synthase γ subunit coding sequences in T. brucei allows this species to survive loss of its kDNA after chemical treatment [28]. Moreover, in such genetically modified T. brucei, independence of kDNA maintenance and expression is associated with multidrug resistance [30]. In our collection of T. evansi stocks we did not observe differences in drug sensitivity between populations that were partially or completely akinetoplast confirming earlier evidence that the presence or absence of kDNA is irrelevant within this context [30,55].
Recently, Carnes et al. showed that SNPs in the F1-ATP γ subunit could be used to genotypically support the multiple origins of at least 4 dyskinetoplastic T. evansi/T. equiperdum lineages: one major group of RoTat 1.2 VSG positive T. evansi/T. equiperdum type A, and three very small groups each represented by only a single strain: T. evansi type B KETRI 2479, T. equiperdum BoTat and T. equiperdum OVI [9]. All Ethiopian T. evansi type A had the corresponding mutation of the type A group. The Ethiopian type B T. evansi shared a similar profile as KETRI 2479. Finally, the Ethiopian T. equiperdum strain Dodola, which had some maxicircle genes but was negative for both type A and type B markers revealed an F1-ATP synthase sequence similar to T.b. brucei AnTat 1.1 E strain, thus likely belongs to the same dyskinetoplastic group as T. equiperdum OVI [9,28].
In conclusion, our study shows that the apparent T. evansi type that is detected in a buffy coat of an infected camel does not necessarily represent the full diversity that is present in the infected animal. Moreover, the fact that 5 out of 22 new T. evansi isolates from camel in Ethiopia contain T. evansi type B may be an indication that is more widespread than currently known. The inoculation of the trypanosomes in immunosuppressed mice may allow the propagation of mixed populations. In contrast, in vitro cultivation seems to reduce the diversity by selecting for only one particular type, in our study T. evansi type B. Secondly, our study addresses some drawbacks of current molecular markers for T. evansi genotyping. To rely solely on VSG markers or kDNA markers for the molecular identification of T. evansi may be misleading due to possible recombinations occurring in VSG genes and to the presence of akinetoplastic T. evansi stocks. In this regard, we confirm that the F1-ATP synthase γ subunit gene, that is not related to the VSG repertoire nor to the presence of kDNA, may become an interesting target for genotyping T. evansi stocks in areas where both types overlap and where mixed infections can occur. Nevertheless, it is not possible to separate the Ethiopian T. equiperdum from T. brucei on the basis of this target gene. Thirdly, no evidence of in vitro drug resistance was found in our collection of T. evansi type A and type B stocks. The presence or partial absence of kDNA in the in vitro adapted T. evansi stocks did not correspond with the drug sensitivity phenotype.