Diminazene resistance in Trypanosoma congolense is not caused by reduced transport capacity but associated with reduced mitochondrial membrane potential

Trypanosoma congolense is a principal agent causing livestock trypanosomiasis in Africa, costing developing economies billions of dollars and undermining food security. Only the diamidine diminazene and the phenanthridine isometamidium are regularly used, and resistance is widespread but poorly understood. We induced stable diminazene resistance in T. congolense strain IL3000 in vitro. There was no cross‐resistance with the phenanthridine drugs, melaminophenyl arsenicals, oxaborole trypanocides, or with diamidine trypanocides, except the close analogs DB829 and DB75. Fluorescence microscopy showed that accumulation of DB75 was inhibited by folate. Uptake of [3H]‐diminazene was slow with low affinity and partly but reciprocally inhibited by folate and by competing diamidines. Expression of T. congolense folate transporters in diminazene‐resistant Trypanosoma brucei brucei significantly sensitized the cells to diminazene and DB829, but not to oxaborole AN7973. However, [3H]‐diminazene transport studies, whole‐genome sequencing, and RNA‐seq found no major changes in diminazene uptake, folate transporter sequence, or expression. Instead, all resistant clones displayed a moderate reduction in the mitochondrial membrane potential Ψm. We conclude that diminazene uptake in T. congolense proceed via multiple low affinity mechanisms including folate transporters; while resistance is associated with a reduction in Ψm it is unclear whether this is the primary cause of the resistance.

has been shown to cause a loss of sensitivity to these drugs (Matovu et al., 2003). Subsequently, it was shown that a second and more important determinant for pentamidine-melarsoprol cross-resistance is the High Affinity Pentamidine Transporter (HAPT1) De Koning & Jarvis, 2001), which was more recently shown to be the aquaglyceroporin TbAQP2 Munday, Settimo, et al., 2015). Rearrangements in the TbAQP2 locus were clearly linked to treatment failures, particularly with melarsoprol (De Koning, 2020;Graf et al., 2013). Similarly, resistance to eflornithine, used against late-stage gambiense HAT, is associated with the loss of an amino acid transporter, TbAAT6 (Alsford et al., 2012;Vincent et al., 2010), and resistance to suramin, the first-line drug against early-stage rhodesiense HAT, is the result of mutations that interfere with its uptake by receptor-mediated endocytosis (Alsford et al., 2012;Zoltner et al., 2016Zoltner et al., , 2020. In T. congolense, resistance to isometamidium has also been attributed to reduced accumulation of the drug (Sutherland & Holmes, 1993;Sutherland et al., 1991Sutherland et al., , 1992Tihon et al., 2017), possibly in part linked to reduced mitochondrial membrane potential (Eze et al., 2016;Wilkes et al., 1997), which diminishes sequestration of cationic drugs into the mitochondrion (Alkhaldi et al., 2016;Lanteri et al., 2008;Stewart et al., 2005)-the likely site of drug action. For DA, it has been shown in the T. b. brucei model that it is almost exclusively taken up by the P2/TbAT1 transporter (De Koning et al., 2004) as it is a very poor substrate for the HAPT1/TbAQP2 transporter (Munday et al., 2014;Teka et al., 2011). It was subsequently proposed that the same mechanism would apply in T. congolense, with an adenosine transporter designated TcoAT1 mediating DA uptake, and DA resistance being the result of specific singlenucleotide polymorphisms (SNPs) that could be used to screen for resistance (Delespaux et al., 2006;Delespaux & De Koning, 2013;Vitouley et al., 2011). However, expression of the putative TcoAT1 in a multi-drug resistant T. b. brucei clone, B48, showed that this gene encoded a P1-type broad specificity purine nucleoside transporter with no ability to transport diminazene (Munday et al., 2013), and genomic analysis has shown that there is no direct ortholog of the P2/TbAT1 transporter in the genome of T. congolense. Regrettably, this has left us still without any insights into the urgent problem of DA resistance in animal African trypanosomiasis.
One urgent question beyond that of the resistance mechanism is that of cross-resistance with isometamidium, which has been reported repeatedly from the field (Mamoudou et al., 2008;Sinyangwe et al., 2004), but inconsistently, as others report no cross-resistance and it used to be a rare occurrence (Giordani et al., 2016;Joshua et al., 1995;Sow et al., 2012). Clearly, it is possible that trypanosomes may have developed resistance to both drugs independently over the decades and are now resistant to both by separate mechanisms. Thus, it is currently unclear whether other diamidine drugs should be considered for novel drug development for AAT because of the risk of cross-resistance with DA. This adds to the many reasons why it is important to understand the mechanism of DA resistance in animal-infective trypanosomes, alongside the need to identify screening tools for AAT drug resistance to assess the real levels of resistance in the field.

| In vitro induction of diminazene aceturate resistance in T. congolense
Four cultures of T. congolense TcoIL3000 were grown and passaged in parallel; a control without added drug and three separate, independent cultures in the presence of DA for adaptation to increasing concentrations, from which ultimately six clonal lines were obtained.
Resistance was very slow to develop, as documented in Figure S1.
At the end of the adaptation, after 227 passages, the control culture tolerated 50 nM DA, whereas all the adapted cultures tolerated a maximum level of 800 nM DA, in what appeared to be at least two steps, with an initial plateau at 550 nM, where the strains were assessed for their level of DA resistance and cross-resistance pattern.
At the 550 nM point, final clones were obtained from each of the parallel cultures, by limiting dilution. In this study we used one clone obtained from original line 4 (4C2), and two clones each from lines 5 (5C1, 5C2) and 6 (6C1, 6C3).
At the point of adaptation to 550 nM, the level of DA resistance and the extent of any cross-resistance with the diamidine drugs pentamidine and DB829, and with oxaboroles AN7973 and SCYX7158, were tested using an assay based on the viability dye Alamar blue (resazurin), in which live trypanosomes (but not dead ones) reduce blue and non-fluorescent resazurin to pink, fluorescent resorufin. This assay yields a fluorescent output that is proportional to the number of cells in the well (Gould et al., 2008). The results, summarized in Figure 1a, show that all the DA-Res strains display highly similar drug sensitivity patterns. Resistance to DA, as measured in this assay, was ~2.3-3-fold, and highly significant ( Figure 1a for fold differences and p values, Table 1 for EC 50 values).
No cross-resistance with pentamidine was observed, but the level of resistance to DB829, a structurally closer analogof DA, mirrored the resistance profile of DA very closely, demonstrating that crossresistance to diamidines is not inevitable, but observed only with close structural analogs. Importantly, there was no cross-resistance with isometamidium, nor with either of the oxaboroles tested, AN7973 and SCYX7158.
To further investigate the limits of cross-resistance with other diamidines, a series of bis-benzofuran analogs (Bakunova et al., 2007) were tested (Table 1). These analogs were chosen to explore the effect of variable linker length between the benzofuramidine end groups, varying inter-amidine distance as well as the flexibility of the molecule. All the benzofuramidines displayed low micromolar activity (Table 1), and there was no cross-resistance with DA ( Figure 1b).
These results clearly show that the cross-resistance phenotype is F I G U R E 1 In vitro cross-resistance pattern of clones from three independent diminazene-adapted cultures with EC 50 values given relative to that for the control strain, IL3000, adapted to growth in 550 nM DA. (a) Cross-resistance to DB829 but not pentamidine, isometamidium, AN7973 or SCYX7158 was observed (n = 5-9). (b) Cross-resistance to a series of bisbenzofuramidines was investigatednone of the strains were more than marginally resistant or sensitized to these compounds and the average EC 50 of the 5 resistant strains was in all cases very close to the value for the IL3000 control. All structures are displayed in Table 1. Statistical significance between EC 50 s of a resistant line and the IL3000 control was determined using Student's unpaired, two-tailed t-test; *p < .05; **p < .01; ***p < .001; ****p < .0001 The DA-resistance of the five DA-resistant clones was then increased via further in vitro exposure until they could tolerate growth at 800 nM DA, and attempts to raise the resistance level further were not immediately successful; all subsequent experiments were done with the clones adapted to 800 nM DA. The in vitro growth rates of these final strains consistently appeared to be slightly slower than that of the drug-sensitive parental strain (Figure 2), although this reached statistical significance only for the 48 hr and 64 hr points of clones 5C2 and 6C1 (p < .05, unpaired student's t-test). The EC 50 values, at ~2.2 µM, were similar for all clones but significantly higher than when they were measured before the cloning, at 550 nM DA medium concentration. This level of resistance, approximately ninefold compared with IL3000 WT (0.244 ± 0.004 µM in parallel experiment, n = 4, p < .00001, t-test, for all DA-RES strains versus IL3000 WT), was stable for at least 3 months of in vitro culturing in the absence of drug pressure ( Figure 3a). Cross-resistance to DB829 was again observed in all clones (ca. fivefold), whereas sensitivity to pentamidine was identical in all clones including wildtype IL3000. Interestingly, sensitivity to AN7973 was slightly but significantly higher in most DA-Res clones ( Figure 3d). Cross-resistance with phenanthridines isometamidium, and ethidium bromide, as well as with the melaminophenyl arsenical cymelarsan, was also tested for selected clonal lines. A slight loss of sensitivity to phenanthridines was consistently observed and reached significant (but lesser than twofold) levels in some clones. No change in cymelarsan sensitivity was observed ( Figure 3e).

| In vivo confirmation of cross-resistance profile
In order to confirm that the in vitro pattern of cross-resistance held up for in vivo infections, including for a field-derived DA-resistant strain, we infected groups of six mice that were treated i.p. with either DB829 (Figure 4a) or AN7973 (Figure 4b), using single doses of 5, 10, or 20 mg/ kg body weight (two separate independent experiments with each drug, each experiment n = 6). For the resistant field isolate, we used stain KONT2/151 (Mamoudou et al., 2006). IL3000 was more virulent than KONT2/151, as is often observed with cultured trypanosome strains, and was sensitive to both DB829 and AN7973. DB829 delayed parasitemia and death with a dose of 10 mg/kg and was F I G U R E 3 Stability of resistance phenotype in diminazene-adapted T. congolense clones. EC 50 values were obtained using the Alamar blue assay. (a) Diminazene aceturate; (b) Pentamidine; (c) DB829; (d) AN7973; (e) Cross-resistance with phenanthridines ethidium and isometamidium and melaminophenyl arsenical cymelarsan. All bars represent the average and SEM of four independent determinations. Statistical significance between EC 50 s of a resistant line and the IL3000 control was determined using Student's unpaired, two-tailed t-test; *p < .05; **p < .01; ***p < .001; ****p < .0001. IL3000 = wild-type control; "Clone A" = grown in the presence of 800 nM diminazene; "Clone B" = passaged for 3 months in the absence of diminazene drug pressure but previously grown in 800 nM DA. In panel E, all clones had been grown >3 months in the absence of diminazene drug pressure.  curative at 15 and 20 mg/kg, whereas AN7973 delayed parasitemia at 5 mg/kg and cured with 10 or 20 mg/kg. The DA-resistant field isolate was equally sensitive to the oxaborole but resistant to DB829, as even the highest dose, 20 mg/kg, merely delayed the onset of parasitemia.

| Screen for potential diminazene transport inhibitors by DB75 fluorescence microscopy
In T. b. brucei, fluorescent diamidines such as DB75, a close structural analog of DA and DB829, have previously been used to monitor uptake and cellular distribution of such compounds (Stewart et al., 2005). In those experiments, the rate at which the kinetoplast (first) and the nucleus (second) became visible in the blue wavelengths (λ = 405/435 nm for excitation and emission, respectively) was much delayed in resistant strains, and in the presence of transport inhibitors (Stewart et al., 2005). Here, the technique was used to screen for potential transport substrates that competitively in- parasite outlines would become visible after 4 min, followed by kinetoplasts at 5-6 min and finally nuclei were faintly stained by 8 min and bright by 15 min. These stages were all observed and scored in the presence of potential inhibitors. As expected, based on current T.
Because of the involvement of the TbAT1/P2 purine transporter in DA uptake by T. b. brucei, we tested a range of purine and pyrimidine nucleosides and nucleobases for delaying DB75 cell entry in T.
congolense, using the IL3000 WT cell line. Adenine and adenosine did not appear to slow DB75 fluorescence up to 1 mM, but were toxic to the cells at the concentrations of 200 µM and above. Other purines (inosine, guanosine, hypoxanthine, guanine, xanthine) or pyrimidines (uridine, cytidine, thymidine, uracil, cytosine) likewise failed to inhibit DB75 uptake; moreover, polyamines and amino acids (500 µM) were similarly ineffective in this assay (Table S1). Glycerol was also tested at up to 20 mM because some diamidines including pentamidine are taken up by an aquaglyceroporin in T. brucei (Alghamdi et al., 2020;Baker et al., 2012), but the data indicated that glycerol also does not compete for the T. congolense DB75 transporter. However, the diamidines DA and, particularly, propamidine and pentamidine dosedependently (100-500 µM) delayed DB75 fluorescence, and so did folic acid, from 200 to 500 µM ( Figure 5). T. vivax was also assayed by this method and similar results were found, with no effect from a series of purines and pyrimidines, glycerol, and the cationic amino acids lysine and arginine, but dose-dependent inhibition by pentamidine and DA. For this parasite, there appeared to be some minor fluorescence delay in the presence of 500 µM folic acid or biopterin, but not at lower concentrations, and possibly in the presence of the polyamines spermidine and putrescine (Table S1).

| Uptake of [ 3 H]-Diminazene by DA-sensitive and -resistant T. congolense clones
In T. brucei, resistance to DA is linked to loss of the TbAT1/P2 transporter (De Koning et al., 2004;Graf et al., 2013;Matovu et al., 2003). While T. congolense does not have an equivalent transporter (Munday, Tagoe, et al., 2015), drug resistance in trypanosomatids has very often been associated with the loss of transporters at the plasma membrane (De Koning, 2020;Landfear, 2008;Munday, Settimo, et al., 2015) and we thus investigated whether uptake of congolense IL3000 from culture, and the effects of 1 mM folic acid or 1 mM DA. One representative experiment in triplicate is shown. All experiments were performed in triplicate and the average and SEM are shown. When error bars are not shown, they fall inside the symbol. Statistical significance was calculated using Student's unpaired t-test: *p < .05; **p < .01; ***p < .001 microscopy, also showed no difference between IL3000 and DAresistant clone 6C3, at either 1 or 10 µM DB75; lower concentrations did not produce detectable staining in either cell line.
Nor was there significant efflux from the cells, after preincubation with [ 3 H]-DA for 30 min followed by incubation of up to 30 min in fresh medium without label or diminazene. Although the slope of both the IL3000 WT and the 4C2 clone trended downwards, in each of four such experiments the slopes were not significantly different from zero, nor different from each other (F-test, p > .05). Figure 6b shows the average of four experiments, which still shows a nonsignificant downward trend and no difference between the slope of the sensitive and resistant lines. In order to better understand why DA-resistance in T. congolense, in contrast to T. b. brucei, is not associated with loss of import or gain of efflux function, the mechanism by which DA is internalized by T. congolense was further investigated.
Uptake of 0.1 µM [ 3 H]-DA was poorly saturable by even a large excess of unlabeled DA (Figure 6c), indicating that DA enters T. congolense by a low affinity and/or non-saturable process. Indeed, out of several potential inhibitors or substrates of the DA uptake, the process was consistently (but always partially) inhibited by 100 µM pentamidine, 1 mM folate, and 100 µM DB829 ( Figure 6d). This is in sharp contrast to DA uptake in T. brucei which is completely saturable and 100% inhibited by a competitive substrate of the TbAT1 transporter ( Figure 6e). The inhibition with folate was confirmed and quantified in a series of experiments with added time points, and found to be highly significant ( Figure 6f). Moreover, in a separate experiment it was tested whether the inhibition by folate and pentamidine was additive and the data indicated that this was the case (Figure 6g), suggesting that in T. congolense DA is taken up by at least two separate low affinity transporters, one sensitive to folate and another that is sensitive to pentamidine. However, it must be noted that the inhibition by both compounds was modest in this experiment and that the joint administration still did not inhibit [ 3 H]-DA uptake by even 50%. The complete lack of cross-resistance with pentamidine, despite high concentrations of pentamidine partially inhibiting DA uptake, is consistent with DA resistance not being the result of changes in its uptake rate. Indeed, the overlap in transport mechanisms of DA and pentamidine was even more clear from the inhibition of [ 3 H]-pentamidine uptake by DA in wild-type IL3000 ( Figure 6h). To further investigate whether (a) folate transporter(s) might contribute to DA uptake in T. congolense, we measured the uptake of [ 3 H]-folate and found this to be partially inhibited by 1 mM DA ( Figure 6i).

| T. congolense folate transporters are engaged in diminazene uptake
Three genes encoding putative folate transporters of T. congolense were identified via homology searches in TriTrypDB in the original T.
congolense IL3000 genome and amplified by PCR with Phusion proofreading polymerase from IL3000 and from the DA-resistant isolates KONT2/133 and KONT2/151 (sequences, GeneIDs, and GenBank accession numbers in Table 2); all three strains are Savannah-type T. congolense. The genes were cloned into pGEM-T Easy for amplification in E. coli and for each gene at least four independent clones were sequenced. The gene sequences from the two KONT strains, both from Cameroon, were almost identical to each other, but displayed some SNPs relative to IL3000 (  Munday et al., 2014;Ward et al., 2011). As the sensitivity to these drugs is completely restored upon re-introduction of the transporters (Alghamdi et al., 2020;Munday et al., 2014;Munday, Tagoe, et al., 2015) this is an ideal system to test potential diamidine transporters from other kinetoplastids. Clonal lines were obtained by limiting dilution; the correct integration of the expression constructs was verified by PCR.
Expression of the T. congolense folate transporters in T. b. brucei B48 confirmed that these transporters have some capacity to take up diamidines, but not the oxaborole AN7973 (Figure 7). Only FT2 of IL3000 did not significantly sensitize to any of the drugs tested, although its diamidine EC 50 values also trended to be lower than the control (B48 transfected with the "empty vector" and pentamidine instead of twofold (Munday, Tagoe, et al., 2015).
The data presented in Figure 7 do not suggest that the two KONT strains are DA resistant because of changes in DA uptake-or at least not by their folate transporters. With respect to DA and DB829, the IL3000 and KONT FT1 and FT3 performed identically, but the KONT FT2 clearly sensitized stronger than the IL3000 FT2.

| The mitochondrial membrane potential is diminished in DA-resistant T. congolense clones
As DA resistance is apparently not linked to changes in DA uptake, it was hypothesized that an intracellular rather than a cell surface change must be responsible for the resistance phenotype.
Moreover, DA is a DNA minor grove binder, and resistance arising due to mutations in a target protein is therefore not expected. Our observations with DB75 fluorescence microscopy showed that such drugs accumulate first in the kinetoplast, as reported for other species (Basselin et al., 2002;Stewart et al., 2005). This requires rapid uptake into the mitochondrion, which is driven by the mitochondrial membrane potential Ψm as DA, DB75, and other diamidines are dications. We therefore investigated whether Ψm was altered in the resistant cell lines. Fluorescence microscopy after DAPI staining confirmed that the kinetoplast was present and did not appear altered or damaged in the DA-resistant clones (Fig. S3).
It is already well established that (di)cationic trypanocides that accumulate in the trypanosome's mitochondrion cause a reduction in Ψm, both through the very fact that they are cations, and through disruption of mitochondrial processes (e.g. Alkhaldi et al., 2016;Fueyo Gonzalez et al., 2017;Ibrahim et al., 2011;Lanteri et al., 2008).
For instance, pentamidine has been shown to act as a cationic uncoupler of oxidative phosphorylation in isolated rat liver mitochondria (Moreno, 1996). However, the more important question here was whether, as (part of) the adaptation to DA, the cells had F I G U R E 7 Drug sensitivity profile of T. b. brucei strain B48 (Tbb Control) and the same strain transfected with folate transporters 1-3 (FT1-3) of IL3000 (S), KONT2/133 (R1) or KONT2/151 (R2). EC 50 values were determined using the Alamar blue assay. Bars represent the average and SEM of three independent experiments. Significant differences relative to the Tbb Control were calculated using Student's unpaired t-test: *p < .05; **p < .01. PMD, pentamidine permanently lowered their Ψm, as previously reported for isometamidium (Eze et al., 2016;Wilkes et al., 1997). The DA-Res parasites did present flow cytometry profiles that had shifted significantly to lower fluorescence and were broader than the control IL3000 cells ( Figure 8a). For quantification, the peak of the IL3000 profile was set at 80 Arbitrary Units (AU). Taking the percentage of cells with fluorescence >80 AU as a measure for Ψm (Fueyo Gonzalez et al., 2017;Ibrahim et al., 2011), a highly significant decrease was found in all four DA-Res cell lines tested, although the extent of the Ψm decrease was not identical in all cell lines (Figure 8b).

| Diminazene causes mitochondrial lesions
In order to further assess the T. congolense mitochondrion as the target for diminazene, and as a potential contributor to diminazene resistance, culture of IL3000 and of 6C3 were grown for 4h or 8 hr  (Table S3, worksheet "SNPs-Hi-filtered") and  (Vickerman, 1969).
Due to the previously described correlation between resistance to the related compound isometamidium chloride (a fusion molecule of ethidium and diminazene) and reduced mitochondrial membrane potential , as well as reduced potential observed in the DA-resistant cell lines, data were mined to identify genes that could be involved in maintaining or generating this membrane potential. Two copies of a vacuolar-type Ca 2+ -ATPase (TcIL3000.A.H_000569200 and TcIL3000.A.H_000569400) harbored missense mutations in all three resistant lines (TcIL3000.A.H_000569200: A109T in clones 4C2 and F I G U R E 9 (Continued) 6C3; TcIL3000.A.H_000569400: G855E in clones 4C2 and 5C1). No further candidates were found. Two mis-sense mutations were found in folate transporter TcIL3000.A.H_000597400 (A132S and N131Y, both in clone 4C2 only, both heterozygous) and another two in folate transporter TcIL3000.A.H_000597700 (L122R in 4C2 and F371C in clone 6C3; both heterozygous). No mutations were found in the other folate or pteridine transporters. Mis-sense mutations were also found in several copies of an amino acid transporter (TcIL3000.A.H_000388400, TcIL3000.A.H_000388500, & TcIL3000.A.H_000388600), a putative plasma-membrane choline transporter (TcIL3000.A.H_000774600) and two copies of a mitochondrial chaperonin HSP60 gene (TcIL3000.A.H_000771700 & TcIL3000.A.H_000772600) (Table S3, worksheet "SNPs-HIGH-MODERATE-LOW").

| Transcriptomics analysis of diminazene resistance
To ascertain whether differential gene expression could explain increased levels of resistance to diminazene, three replicates of the wild-type parental line, as well as three replicates of each independent resistant clone, were subjected to transcriptomic analysis by Illumina paired-end sequencing. Data were processed as described in the methods, resulting in transcript abundances for 9,360 T. congolense genes (Table S3, worksheet "RNAseq_all"). Data were filtered to obtain differentially expressed genes relative to wild-type (q value ≤.05), resulting in 274, 247, and 265 genes identified as being significantly differentially expressed in DA-resistant clones 4C2, 5C1, and 6C3, respectively, with 120 of these genes being common to all three clones ( Figure 10). Genes were then filtered to remove VSGs, ESAGs, and RHS proteins, resulting in a final list of 61 genes, of which 15 were annotated as hypothetical (Table S3, worksheet "RNAseq_filtered").
One the most significant changes was downregulation of an array of H4 histone encoding genes (mean Log 2 fold change: −7.842) (Table S3, worksheet "RNAseq_DOWN"). Furthermore, there was significant downregulation of a putative protein involved in elongation of very long fatty acids (TcIL3000.A.H_000436100.1; mean Log 2 fold change: −6.526) and a putative ZIP Zinc transporter that is, however, indicated as a possible pseudogene (TcIL3000.A.H_000205900.1; mean Log 2 fold change: −4.099).
Other downregulated genes of note included a cysteine peptidase, as well as four hypothetical proteins (TcIL3000.A.H_000205700.1, 000206000.1, 000429100.1 and 000429200.1) for which ex- There were only 13 significantly differentially expressed genes exhibiting a Log 2 fold change of >1.0 in all three resistant clones, and these were identified as primarily RNA helicases, in addition to copies of cysteine peptidase and several hypothetical proteins.
As the data for the RNA helicases have different significance and expression values, it is unlikely that the identification of this gene family is due to multimapping as in the Histone H4 genes, in this case leading to a firmer conclusion that all individual genes identified are differentially expressed (Table S3, worksheet "RNAseq_UP"). In F I G U R E 1 0 Differential gene expression in three independent DA-resistant clones (4C2, 5C1 and 6C3) of T. congolense IL3000. For each DA-resistant clone, Log 2 fold change was calculated compared with wild-type parasites, as well as significance (q-value), and plotted as a volcano plot. Differential gene expression was deemed significant if the gene exhibited a Log 2 fold change of >1 or <−1 in all three clones, as well as a q-value of <.05 (-Log 10 q-value of 1.30103) (genes shown as asterisks). The most significantly upregulated (Helicase associated domain (HA2) putative: TcIL3000.A.H_000255300.1) and downregulated (Histone H4 putative: TcIL3000.A.H_000205600.1) genes are shown as circled asterisks to illustrate reproducibility across clones some cases, there was significant differential expression in only two of the DA-resistant clones, for example, a predicted long-chain fatty acid-CoA ligase (TcIL3000.A.H_00065500.1; mean q-value: .0356; mean Log 2 fold change: −0.297).

| D ISCUSS I ON
In this manuscript, we have studied the phenomenon of diminazene resistance in T. congolense. This veterinary parasite does not have an equivalent of the T. brucei diminazene transporter TbAT1 (Munday et al., 2013) and thus it is clear that the T. brucei model for diminazene resistance, based on loss of TbAT1 function (De Koning et al., 2004;Graf et al., 2013), is not applicable to T. congolense. We found that it was possible to induce a substantial level of diminazene resistance in T. congolense by slowly increasing the drug concentration in the culture medium. The level of DA resistance was entirely stable in all clones, after months of in vitro passage in the absence of drug pressure, and repeated cycles of stabilating and reculturing. Remarkably, all the independently generated DA-resistant clones displayed the same level of resistance, cross-resistance patterns, and growth rates, which could indicate that they had developed very similar adaptations.
One of the main aims of this study was to establish whether DA resistance in T. congolense necessarily leads to cross-resistance to other (potential) AAT drugs. Co-resistance with isometamidium chloride has been reported in the field (Afewerk et al., 2000;Ainanshe et al., 1992;Mulugeta et al., 1997), but it is not clear whether this constitutes genuine cross-resistance. This conclusion can be hard to reach in the field as cattle are often treated with both drugs and independently developed resistance to both drugs is therefore a real possibility.
However, there have been reports of single resistance to diminazene and to isometamidium, and more recently to both drugs (Geerts et al., 2001;Giordani et al., 2016;Sinyangwe et al., 2004). This clearly shows that multiple-resistance is not automatic, as it seems to be for pentamidine and melarsoprol in T. brucei sspp. (Baker et al., 2013).
Our observation that induced DA resistance in T. congolense did not change isometamidium sensitivity in any of the independent clones is certainly a strong confirmation of that hypothesis and to the best of our knowledge the first such evidence from a controlled laboratory study since experiments in 1963 by Frank Hawking with multiple T. congolense strains in guinea pigs and mice, which also found no evidence of cross-resistance between DA and phenanthridines (Hawking, 1963). We have also attempted to similarly induce isometamidium resistance in T. congolense, but this was not successful in vitro although it has been achieved in vivo (Tihon et al., 2017).
Another important observation on the nature of DA resistance in T. congolense is that it was remarkably stable, which, considering the many reports of DA resistance from all over Africa, has implications for the future utility of the drug.
As mentioned earlier, it has been argued that the further development of diamidine drugs against AAT would be unwelcome and only result in more drug pressure on populations that already harbor DA resistance. The strong cross-resistance with furamidines such as DB75 and DB829, observed both in vitro and in vivo and with field isolates and in vitro-adapted clones, would certainly strengthen this argument. However, the complete lack of cross-resistance with pentamidine and the tested series of bis-benzofuramidines demonstrates that only the closest structural analogs display crossresistance with DA, and that therefore not all diamidines should be ruled out from consideration for further development against AAT.
The caveat, however, would be that any new diamidine drug must be active, at a minimum, against all the major species causing nagana in sub-Saharan Africa, including DA-resistant strains of each of those species. Given the dearth of knowledge about DA-resistant T.
vivax, in particular, this would still constitute a significant hurdle at the moment. In this context, the observation that there is no crossresistance between DA and oxaboroles, promising new agents in development for HAT and AAT (Akama et al., 2018;Begolo et al., 2018;Jacobs et al., 2011), is highly significant.
In T. brucei, drug resistance is mostly linked to changes in drug accumulation, particularly through mutations in transporter genes (De Koning, 2020). In T. congolense, isometamidium resistance has also been linked to reduced accumulation Sutherland et al., 1991Sutherland et al., , 1992Tihon et al., 2017), but no drug transporters have been identified in this species. We therefore investigated the uptake of [ 3 H]-DA in wild-type IL3000 and found it to be very slow and, apparently, low affinity. This indicates that uptake of diminazene could be limiting to its efficacy, whereas in T. brucei sspp uptake of diminazene, the furamidines, and pentamidine is very fast with high affinity (De Koning et al., 2004;De Koning & Jarvis, 2001;Teka et al., 2011;Ward et al., 2011). We found no consistent indication that DA resistance in T. congolense is linked to reduced cellular accumulation of the drug, as the accumulation rate was similar in the parental and resistant clones.

Several compounds were identified as inhibitors of both [ 3 H]-
diminazene and fluorescent DB75 uptake in T. congolense, particularly pentamidine, propamidine, and folate. The latter observation indicates the involvement of folate transporters in diminazene uptake, although high concentrations were required for significant inhibition and even then diminazene uptake was only partially inhibited. Clearly, a so-far unidentified, folate-insensitive uptake mechanism is additionally involved in diminazene uptake. The notion of multiple low-affinity entry routes was further strengthened by the demonstration that inhibition by folate and pentamidine was additive, and yet still incomplete. To further investigate whether T. congolense folate transporters may contribute to diminazene uptake, all three then-known FTs were cloned and expressed in the diamidine-and melaminophenyl-resistant clone T. brucei B48 . The expression of IL3000 transporters FT1 and FT3 appeared to induce a small but significant sensitization to diminazene and its structural analog, DB829, but not to pentamidine or oxaborole AN7973; IL3000 FT2 trended in the same direction as the other two transporters, but the effect did not reach statistical significance. The folate transporters of two DA-resistant isolates from Cameroon displayed several SNPs relative to the IL3000 sequences, but the levels of sensitization induced by those FTs were not significantly different from those of IL3000. However, FT2 from both resistant isolates significantly sensitized to all three diamidines including pentamidine, but FT2 from IL3000 did not sensitize to any of the diamidines tested. None of the FTs changed the EC 50 value for AN7973. These observations show that the folate transporters have a minor capacity for diamidine transport, but also that SNPs observed in these genes are not the cause of DA resistance, either in the field or in our laboratory-adapted clones. The lack of genuine resistance mutations in the folate transporters is presumably linked to the observation that at least two of those transporters have this capacity, and that the FTs mediate only a proportion of DA uptake, considering the modest effect of 1 mM folate on [ 3 H]-diminazene uptake.
Based on the above reasoning, we must conclude that diminazene resistance in T. congolense is not principally the result of a reduced rate of cellular diminazene accumulation. The alternative cause of resistance would be in changes of the target. In the case of diminazene and other cationic drugs this target is the mitochondrion (Alkhaldi et al., 2016;Basselin et al., 2002;Fueyo Gonzalez et al., 2017;Lanteri et al., 2008), and the fluorescent DB75 has been shown to accumulate in the kinetoplast (Mathis et al., 2006(Mathis et al., , 2007Stewart et al., 2005), binding to the kDNA or interfering with enzymes involved in the functioning and/or replication of the kinetoplast. We confirm the mitochondrial targeting of DA in T. congolense with TEM, showing mitochondrial lesions appearing after a 4-hr incubation with 1.5 µM DA (~5 × EC 50 ), whereas no other cellular changes were apparent. While dyskinetoplastic T. brucei have been described, including as part of the adaptation to isometamidium (Eze et al., 2016), and T. evansi is considered a dyskinetoplastic T. brucei (Lun et al., 2010), dyskinetoplastic T. congolense have not been described, and our DA-resistant clones displayed normal kinetoplasts as observed by DAPI staining and TEM. Thus, unlike T. brucei sspp (Dean et al., 2013;Eze et al., 2016), T. congolense does not appear to have the ability to become resistant to kinetoplast-targeting drugs by losing dependence on genes encoded by kDNA. T. brucei bloodstream forms have a unique mitochondrial function and lack the usual oxidative phosphorylation pathways, instead employing a trypanosome alternative oxidase (TAO) as the sole terminal oxidase (Chaudhuri et al., 1998;Ebiloma et al., 2018Ebiloma et al., , 2019 and using the F 1 F o -ATPase to pump H + from the mitochondrial matrix and so generate the necessary Ψm (Schnaufer et al., 2005). Bloodstream T. congolense may have a similar energy metabolism and mitochondrion, seeing that they have similar sensitivity to inhibitors of TAO (Fueyo Gonzalez et al., 2017) and no sensitivity to cyanide (Bienen et al., 1991).
However, in order to reach and disable the kinetoplast and/or other mitochondrial targets, the drug must accumulate inside the mitochondrion-a process that is dependent on the mitochondrial membrane potential Ψm. We observed, in all four DA-resistant clones investigated, a statistically significant reduction in Ψm, which might serve to diminish the entry of diminazene into the mitochondrion, where the drug would otherwise be strongly accumulated. Although DA and structurally related diamidines bind to kDNA, it should be mentioned that the lesions observed with TEM were in the mitochondrial matrix rather than any observable alterations to the kinetoplast structure, leaving open the possibility of mitochondrial targets other than the kinetoplast. This is particularly important as the entry of the drug into T. congolense is low affinity and inefficient, as opposed to the energy-dependent, high affinity process in T. brucei. Diminazene uptake in T. congolense is thus very likely equilibrative rather than concentrative, implying that a failure of the drug to accumulate in the mitochondrion will eventually result in a reduced cellular uptake as well. A reduced (rate of) diminazene accumulation in the mitochondrion of the resistant cells would be consistent with the observed of lack of diminazene-induced mitochondrial lesions at 4 hr and relatively few alterations even at 8 hr. However, we acknowledge that the reduced mitochondrial membrane potential is likely not the only, or main, factor contributing to DA resistance, particularly as we did not observe a reduced time-to-fluorescence for kDNA on incubation with DB75. Moreover, the lack of cross-resistance to pentamidine and isometamidium also indicates that a more specific adaptation is likely to be at least partly responsible for the narrow resistance phenotype. In light of these observations, it is likely that the reduced Ψm is the result of that adaptation, rather than the primary cause of the DA resistance.
Genomic and transcriptomic analyses of the 4C2, 5C1, and 6C3 DA-resistant clones were performed to further investigate the causes of resistance. For each clonal line, several hundred genes were differentially regulated, relative to the parental clone IL3000.
After filtering-out inconsistent and functionally irrelevant returns relatively few genes were significantly downregulated (including H4 histones, a cathepsin-like cysteine protease, and several hypothetical proteins with no known domains) or upregulated (mainly RNA helicases and a few hypothetical proteins). However, the highly stable level of resistance would be consistent with permanent mutations rather than the more transient changes in transcription. The genomic sequencing revealed 19 high impact SNPs and 192 high impact indels. SNPs were observed in several transporter genes, including amino acid transporters, some of the folate transporter genes and a putative choline transporter, as well as mitochondrial HSP60, but mostly in only 1 of the resistant clones. Most interestingly, however, in all three DA-resistant clones sequenced, missense mutations were observed in at least one of two copies of a vacuolartype Ca 2+ -ATPase.
The mutations in the vacuolar Ca 2+ -ATPases could be relevant to a resistance mechanism that is linked to mitochondrial function, as Trypanosoma mitochondria are known to take up Ca 2+ (Ramakrishnan & Docampo, 2018), which appears to be released to the cytosol upon membrane depolarization (Alkhaldi et al., 2016) through a Ca 2+ /H + exchanger (Ramakrishnan & Docampo, 2018), and the bioenergetics of trypanosomes is regulated by inositol 1,4,5-trisphosphate receptormediated Ca 2+ -release from acidocalcisomes (Chiurillo et al., 2020;Huang, Bartlett, et al., 2013) and a mitochondrial Ca 2+ uniporter (Huang, Vercesi, et al., 2013). The T. congolense Ca 2+ -ATPases are syntenic with T. brucei vacuolar Ca 2+ /H + antiporters PMC1 and PMC2, which are essential genes and localized to the acidocalcisomes (Huang & Docampo, 2015;Luo et al., 2004). RNAi knockdown of TbPMC1 expression increased cytosolic calcium levels but reduced the amount of Ca 2+ in intracellular stores. Interestingly, diamidines including DB75 and DB820 have been shown to accumulate not just in the nucleus and kinetoplast of T. brucei, but also in the acidocalcisomes, as evidenced by a slow-onset yellow-orange fluorescence (Mathis et al., 2006).
Acidocalcisomes have been described as drug targets and have essential regulatory roles in trypanosomes, particularly concerning ion distribution, pH, and Ca 2+ -signaling (Docampo & Moreno, 2008). As such, the two missense mutations in the vacuolar-type Ca 2+ -ATPase, each of which occurred in two independent clones, should certainly be followed up as potential contributing factors in T. congolense DA resistance, but this will clearly require extensive further validation. The very high level of DA-induced SNPs and indels is consistent with the DNA-targeting nature of this minor grove binder.
Altogether, we conclude that a clinically relevant level of DAresistance in T. congolense can be induced in vitro and is stable, and not primarily due to reduced DA accumulation but instead associated with a reduced mitochondrial membrane potential, although this is not likely to be the main cause of the resistance phenotype. DA uptake is low affinity and partially mediated not only by the T. congolense folate transporters, but also by separate transporters that are sensitive to pentamidine.
These results constitute the first systematic analysis of DA-transport and resistance mechanisms in T. congolense and provide some insights important for the development of new drugs against AAT.

| In vitro culture of T. congolense
Cultures of bloodstream forms of Savannah-type strain IL3000 and the adapted clones derived from this strain were cultured exactly as described by Coustou et al. (2010), in a basal MEM-based medium supplemented with 20% goat serum 14 μl of β-mercaptoethanol, 800 μl of 200 mM glutamine solution, and 10 ml of 100× penicillin/streptomycin solution per liter of medium (pH = 7.3). T. congolense were cultured in 6-or 24-well plates at 34°C and 5% CO 2 . For adaptation to DA, cultures were serially passaged in the highest concentration of DA tolerated, which was increased as the cells adapted, starting at 50 nM DA (Sigma). Thus one 24-well plate contained 1 row of "no drug" control culture of IL3000 bloodstream forms and three rows of independent cultures in the presence of DA, in three drug concentrations per passage, one above and one below the concentration under test.

| Resazurin assay with T. congolense
This drug susceptibility assay was performed essentially as described previously (Cerone et al., 2019) and is based on only live cells reducing the blue and non-fluorescent viability indicator dye resazurin sodium salt ("Alamar blue"; Sigma) to pink, fluorescent metabolite resorufin (Gould et al., 2008). The fluorescence was quantified using a FLUOstar Optima (BMG Labtech, Durham, NC, USA) at wavelength of 540 nm (excitation), 590 nm (emission), and the data plotted to a sigmoidal curve with variable slope using Prism 5.0 (GraphPad Software Inc., San Diego, CA, USA). The assay was set up in 96-well plates with the test drug added at doubling dilutions in culture medium, usually starting at 50 µM, over one or two rows of the plate, with the last well receiving 100 µl culture medium as a drug-free control, that is either 11 or 23 doubling dilutions. To each well, 100 µl of culture medium containing 5 × 10 5 bloodstream T. congolense cells was added, and the plate was incubated for 48 hr at 34ºC/5% CO 2 , followed by the addition of 20 µl of a 125 mg/ ml solution of resazurin sodium salt in phosphate-buffered saline (PBS) and a further incubation period of 24 hr.

| Fluorescence microscopy screen for diamidine uptake inhibitors
NIH mice (Envigo) were infected with 100,000 trypanosomes of either T. b. brucei strain Lister 427 (i.p.), T. congolense strain IL3000 (i.p.) or T. vivax strain Y486 (Gibson, 2012) (i.v.) in PBS; each parasite species was injected in sterile saline solution. At peak parasitemia, the mouse was killed by CO 2 inhalation and the blood collected by aortic bleed. This was kept on ice if not being used immediately for an experiment.
Approximately, 1 ml of whole blood was spun at 1,000 × g, and supernatant, buffy coat and a visible layer of parasites were removed to a new tube and spun again. The resulting pellet was washed with 1 ml of CBSS buffer (25 mM HEPES, 120 mM NaCl, 5.4 mM KCl, 0.55 mM CaCl 2 , 0.4 mM MgSO 4 , 5.6 mM Na 2 HPO 4 , 11.1 mM glucose; pH adjusted to 7.4 with NaOH), and the pellet resuspended in CBSS buffer at an appropriate concentration to visualize the parasites (usually 1 ml). The duration of viability of parasites was assessed at room temperature, 34ºC and 4ºC prior to completion of experiments. T. congolense parasites performed similarly when kept at 34ºC and room temperature during experiments; therefore, experiments were conducted at room temperature; T. vivax parasites were motile for longer and took up DB75 at a faster rate at 34ºC, so were kept at this temperature as much as possible during experiments (but not while under the microscope for observation).
For non-inhibited cells, 10 µM of DB75 was added and fluorescence observed using a Zeiss Axioscope fluorescence microscope, using the DAPI filter set (λ exc = 330 nm, λ em = 400 nm) as well as brightfield. Images were obtained with Openlab imaging software (Improvision, Coventry, UK). For inhibited cells, the test inhibitor was added to the desired concentration (see Table S1) just before the addition of 10 µM DB75 and observation of fluorescence as before. Each potential inhibitor was tested at least twice on parasites from different mice, those which inhibited up to four times. Every experiment had cells with no inhibitor added visualized in parallel to control for slight variation in timing of appearance of fluorescence in kinetoplasts and nuclei between batches of ex vivo parasites.

| Transport assays
Transport assays for T. congolense bloodstream forms were performed exactly as described for T. b. brucei bloodstream forms (Wallace et al., 2002) and Leishmania major promastigotes (Al-Salabi et al., 2003). Depending on the assay, parasites were either purified from infected blood using a DEAE-cellulose column and a 6:4 ratio PSG buffer as described (Lanham & Godfrey, 1970), or collected from culture. Briefly, cells were washed into assay buffer (33 mM HEPES, 98 mM NaCl, 4.6 mM KCl, 0.55 mM CaCl 2 , 0.07 mM MgSO 4 , 5.8 mM NaH 2 PO 4 , 0.3 mM MgCl 2 , 23 mM NaHCO 3 , 14 mM glucose, pH 7.3) and adjusted to a density of 10 8 /ml just before use. About,100 µl of the cell suspension was added to 100 µl of radiolabeled substrate ( 3 H-DA or 3 H-pentamidine) in assay buffer, sometimes also containing a competitive inhibitor at 2× final concentration, atop a layer of oil (1:7 of mineral oil and di-n-butyl phthalate (Sigma)) in a 1.5 ml microfuge tube.
After a predetermined incubation time, the incubation was stopped by the addition of 1 ml ice-cold "stop solution" (1 mM/250 µM unlabeled permeant in assay buffer), and cells were separated from the extracellular radiolabel by centrifugation through the oil layer (1 min at 13,000 rpm in a microfuge). The cell pellets were harvested into scintillation tubes by cutting off the tip of the microfuge tube after flashfreezing in liquid nitrogen, incubated with 2% SDS for at least 1 hr, and overnight with scintillation fluid (Optiphase HiSafe III, Perkin-Elmer, Waltham, MA, USA) before being agitated overnight in scintillation fluid (Optiphase HiSafe III, Perkin-Elmer, Waltham, MA, USA). Tubes were then read using a Hidex 300 SL scintillation counter (Lablogic, Sheffield, UK). Raw disintegrations per minute (dpm) reads were converted to units of pmol(10 7 cells) -1 and corrected for background radiation and non-specific radiolabel association, by subtracting dpm counts from no-label controls and parallel determinations in the presence of saturating levels of non-labeled permeant, respectively. Radiolabeled ring-[ 3 H]-DA was custom-made by PerkinElmer (CUST78468000MC; 60.7 Ci/mmol) and [ 3 H]-pentamidine isethionate was custom-made by Amersham (TRQ40084; 3.26 TBq/mmol). [3,5,7, H]-folic acid (ART 0125; 56.8 Ci/mmol) was from American Radiolabeled Chemicals.
Statistical analyses were performed with GraphPad Prism 8.1.

| Cloning and expression of T. congolense folate transporters
Three folate transporters were identified in the original T. congolense  Table 2    using an Amaxa Nucleofector, program X-01, and TbBSF buffer as described previously (Burkard et al., 2007;Munday et al., 2013). Transfectants were cloned out, by limiting dilution, in standard HMI-11 medium (Hirumi & Hirumi, 1989) containing 5 µg/ml blasticidin S and clones screened for correct integration of the cassette by PCR. All clones were maintained in HMI-11 using standard growth conditions, 37ºC and 5% CO 2 .
Drug sensitivity of the T. brucei cell lines was assessed using the standard resazurin assay, essentially as described for T. congolense, above and as described previously (Siheri et al., 2016). Briefly, the assay was performed in 96-well plates with 11 doubling test compounds dilutions, highest concentration 10 µM, and the 12th well containing cells in drug-free medium. Bloodstream form of 2 × 10 4 trypanosomes were added to each well, and the plates were incubated for 48 hr at 37°C/5% CO 2 after which 20 µl of resazurin solution was added to each well and the plates incubated for another 24 hr.

| DAPI staining of nuclei and kinetoplasts
Approximately, 1 × 10 6 cells were harvested in Eppendorf tube and centrifuged at 2,400 rpm for 10 min in a microfuge. The supernatant was discarded; the cells were resuspended in 1× PBS and washed by centrifugation (2,400 rpm for 5 min). The supernatant was discarded; the cells were resuspended in 50 µl 1× PBS pH 7.4, spread out on a microscope slide and allowed to dry for 30 min.
The cells were then fixed by flooding the slide with 4% formaldehyde for 10 min, followed by washing twice in 1× PBS. The slides were allowed to air dry before placing a drop of Vectashield mounting medium containing DAPI (4′,6-diamidino-2-phenylindole dihydrochloride) (Vector Laboratories) on them. The slides were then covered with a cover slip and the edges sealed with nail varnish. The slides were viewed under a DeltaVision microscope (GE Healthcare) using softWoRx software and the images were processed using ImageJ software.

| RNAseq
Total RNA isolations were performed on 1 × 10 7 cells of a parental strain of wild-type T. congolense IL3000, as well as three replicates of each of the three independent DA-resistant clones (4C2, 5C1 and 6C3), using a commercial kit (RNeasy, QIAgen gene abundances quantified, in both cases using Kallisto with default parameters (−b 100, −t 8) (Bray et al., 2016). Statistical analyses of differential expression between the wild-type parental strain and each of the three resistant clones were carried out using the R-based package Sleuth, with default parameters (for each transcript, a second fit was performed to a reduced model that presumes abundances are equal in the two conditions-wild-type and DA-resistant) as outlined in the manual (Bray et al., 2016), and figures were generated with Matlab (The MathWorks, Inc.). Genes exhibiting significant (q-value ≤.05) differential expression between wild-type and resistant parasites were tabulated and filtered to remove VSGs, ESAGs and retrotransposon hot spot proteins, and subsequently sorted to obtain genes with significance based on q-value (i.e., p-value adjusted for multiple testing by false discovery rate). The full Sleuth output is available in Table S3, worksheet "RNAseq_all". were filtered firstly to remove SNPs occurring in both wild-type and resistant lines, and subsequently to select for SNPs and indels predicted to have a high (a variant that is assumed to result in a highly disruptive impact on the protein such as truncation or loss of function), moderate (non-disruptive variant that is assumed to impact protein effectiveness such as non-synonymous SNPs or in frame deletions) or low (Mostly harmless variants such as synonymous SNPs) impact, in both cases using Microsoft Excel.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest regarding the publication of this research.

Sequence data have been deposited in the European Nucleotide
Archive (accession number PRJEB39051). All new sequence information on cloned folate transporters was deposited with GenBank and accession numbers are included in the manuscript text (