Characterization of a Melamino Nitroheterocycle as a Potential Lead for the Treatment of Human African Trypanosomiasis

Abstract

This paper reports an evaluation of a melamino nitroheterocycle, a potential lead for further development as an agent against human African trypanosomiasis (HAT). Studies on its efficacy, physicochemical and biopharmaceutical properties, and potential for toxicity are described. The compound previously had been shown to possess exceptional activity against Trypanosoma brucei in in vitro assays comparable to that of melarsoprol. Here, we demonstrate that the compound also was curative in the stringent acute mouse model T. brucei rhodesiense STIB 900 when given intraperitoneally at 40 mg/kg of body weight. Nevertheless, activity was only moderate when the oral route was used, and no cure was obtained when the compound was tested in a stage 2 rodent model of infection. Genotoxic profiling revealed that the compound induces DNA damage by a mechanism apparently independent from nitroreduction and involving the introduction of base pair substitutions (Ames test), possibly caused by oxidative damage of the DNA (comet test). No significant genotoxicity was observed at the chromosome level (micronucleus assay). The lack of suitable properties for oral and central nervous system uptake and the genotoxic liabilities prevent the progression of this melamine nitroheterocycle as a drug candidate for HAT. Further modification of the compound is required to improve the pharmacokinetic properties of the molecule and to separate the trypanocidal activity from the toxic potential.

INTRODUCTION

Human African trypanosomiasis (HAT) remains an unresolved health problem in sub-Saharan Africa, where people in 36 countries are at risk of disease. HAT is caused by two species of parasite, Trypanosoma brucei rhodesiense and T. brucei gambiense. Once the parasite is injected into the human host following the bite of an infected tsetse fly, it enters the bloodstream (stage 1) before eventually invading the central nervous system (CNS; stage 2), giving rise to the classical neurological symptoms of HAT. The disease is fatal unless treated (http://www.who.int/tdr). Chemotherapy is central to efforts to control HAT (1), but the drugs available to treat the disease, especially stage 2, are inadequate due to problems with clinical efficacy, toxicity, resistance, and inappropriateness for a rural African setting. The mainstay of treatment for both forms of the disease once there is CNS involvement has, until recently, been melarsoprol (Fig. 1). This arsenic-based drug is highly toxic and gives rise to a potentially fatal reactive encephalopathy in a significant number of patients. Since its introduction in 1990, eflornithine has been increasingly used as a first-line drug for T. brucei gambiense stage 2 HAT instead of melarsoprol due to its reduced side effects. However, this drug is far from ideal, requiring intravenous infusion of large quantities of the compound over a number of days, a treatment that is problematic for a remote rural setting. NECT, a nifurtimox-eflornithine combination therapy, today is preferred to the eflornithine monotherapy to treat stage 2 HAT due to T. brucei gambiense, as it is easier to administer (2). Nevertheless, neither NECT nor the eflornithine monotherapy are efficacious on T. brucei rhodesiense infections.

FIG 1.

Structures of melarsoprol, nifurtimox, and WSP934.

Trypanosomes live in the bloodstream and are dependent on the human host for uptake of various nutrients, among which are the purine nucleosides. In order to salvage these nucleosides, trypanosomes have a number of transporters. Carter and Fairlamb characterized an unusual transporter, called P2 (3). They found that in addition to the uptake of adenosine and adenine, the P2 transporter also was involved in the uptake of melaminoarsenicals, such as melarsen oxide, and diamidines, such as pentamidine (4). It was established that melamines and benzamidines are substrates for the P2 transporter but not significantly so for mammalian transporters. It has emerged subsequently that the situation with regard to the uptake of benzamidine and melamine moieties in T. brucei is more complex and involves other transporters, such as the high- and low-affinity pentamidine transporters, HAPT1 and LAPT1 (5, 6), the former now being identified as an aquaglyceroporin (7, 8). However, it is clear that these moieties are selectively concentrated in T. brucei, unlike the case in mammalian (human) cells. This has led us to couple the melamine motifs to (potentially) trypanotoxic agents, with the aim of selectively targeting these trypanocides to the parasites (9–14). One of the toxic moieties that we attached to the melamine motif was a nitrofuran (11, 12). The idea behind this was that the 5-nitrofuran nifurtimox (Fig. 1) had been used previously against melarsoprol-refractory HAT (15) and, when administered in combination with eflornithine, had proven greatly efficacious (16). We reasoned that attaching a melamine to the bioactive nitrofuran increases potency, due to such compounds being selectively accumulated into trypanosomes. This led to some very potent molecules. We have described the design process and data in previous publications (11, 12, 17). One of these molecules, our initial lead, WSP934 (Fig. 1) (compound 3a in Stewart et al. [17]), was found to be curative of a rodent model of HAT (11, 12, 17).

Trypanocidal activity alone is insufficient to render a new chemical entity a useful agent to proceed in the drug discovery process. Biopharmaceutical properties and toxicological issues must be considered as well. Since genotoxicity in particular has mitigated the development of nitroheterocycles, a thorough evaluation of possible genotoxic potential of such agents is essential. Therefore, we have made a complete investigation into the pharmacological and genotoxicological status of WSP934 with a view to the further development of this compound.

MATERIALS AND METHODS

Antitrypanosomal studies in vivo.

For studies of the acute stage of the disease, groups of four mice were infected intraperitoneally (i.p.) with 1 × 10⁴ bloodstream forms of T. brucei (STIB 795) or T. brucei rhodesiense (STIB 900) (18). T. brucei parasites were maintained as described previously (19). On days 3, 4, 5, and 6 postinfection, the experimental groups were treated with compounds (melarsoprol, nifurtimox, and WSP934) either by the intraperitoneal or the oral route, as stated in the text. Parasitemia of the mice was checked on day 7 postinfection and twice a week thereafter up to day 60. One group of mice was not treated and acted as the control. For late-stage studies, groups of five mice were inoculated similarly with T. brucei brucei (GVR35), but treatment was not initiated until 21 days postinfection (20). Parasitemia of the mice was checked twice a week up to day 50 and once a week thereafter until day 180. One group of mice was treated with 40 mg/kg of body weight diminazene diaceturate and acted as the control. For relapsing mice, the day of death was recorded and the survival time determined. In vivo efficacy studies in mice were conducted at the Swiss Tropical and Public Health Institute (Basel) according to the rules and regulations for the protection of animal rights (Tierschutzverordnung) of the Swiss Bundesamt für Veterinärwesen. Studies were approved by the veterinary office of Canton Basel-Stadt, Switzerland.

Pharmacokinetic studies.

Swiss white NMR1 mice were given an i.p. dose of WSP934 at 40 mg/kg. The compound was prepared in 10% dimethylsulfoxide (DMSO), 90% (vol/vol) water for injection at a concentration of 4 mg/ml. Blood samples were taken 5, 15, 30, 60, 120, 240, 480, 720, and 1,440 min postinjection, and levels of WSP934 in the blood were determined by liquid chromatography-tandem mass spectrometry (LC-MS/MS). Blood samples were taken from 3 mice at each time point. These experiments were carried out by outsource within the United Kingdom under UK Home Office regulations.

Ames Salmonella/microsome test.

The bacterial reversed mutation assay using Salmonella enterica serovar Typhimurium strains TA98, TA100, and TA1535 was used to evaluate the mutagenic properties of WSP934 (21). TA1535 and TA100 predominantly are sensitive to base pair substitution mutagens. TA98 is sensitive to frameshift mutagens. All strains are deficient in excision repair (uvrB mutation). The strains TA98 and TA100 contain the plasmid pKM101, which activates an error-prone DNA repair system, making these strains more responsive to a variety of mutagens. The tests were performed in the absence or presence of an exogenous metabolic activation system (Aroclor 1254-induced rat liver S9, prepared from adult Sprague Dawley rats and supplemented with cofactors glucose-6-phosphate and NADP-Na² to a final concentration of 2 mg/ml) to detect indirect and direct mutagenic activity.

Since nitroheterocycles are frequently hypermutagenic in Salmonella due to the presence of particular nitroreductase enzymes in these microorganisms, we also investigated the genotoxic potential of the compounds in the nitroreductase-deficient (NR) strains TA98NR and TA100NR (22, 23).

A positive result was defined as a reproducible dose-related increase in the number of His⁺ revertants, and the threshold was set as an at least 2-fold increase in the number of control revertants. The activity of the S9 mix and the responsiveness of the tester strains were verified by including appropriate controls: DMSO (80 μl/plate), used to dissolve the compounds, was the negative control, whereas hycantone (HYC; 75 μg/plate) and 2-aminofluorene (2-AF; 2.5 μg/plate) were used as positive controls (all experiments were performed both with and without incubation with S9 mix). In order to characterize the mutagenic response of nitroreductase-deficient (NR) strains with respect to parental strains TA98 and TA100, we examined their sensitivity to furylfuramide (20 ng/plate), whose mutagenicity is substantially lower in nitroreductase-deficient than in nitroreductase-proficient strains (24).

Cytotoxicity assays in human cells. (i) Short-term exposure.

Heparin-anticoagulated peripheral blood was obtained by venipuncture from consenting healthy nonsmoking donors. The isolation of leukocytes and cytotoxicity induced by WSP934 and nifurtimox (following a 1-h exposure) was performed as reported in Buschini et al. (25).

(ii) Long-term exposure.

Two different types of cultured cells were used for this assay: (i) immortalized lymphocytes (LYMPH), i.e., peripheral blood cells transformed by Epstein-Barr virus in a lymphoblastoid cell line able to actively proliferate, kindly provided by R. Dolcetti (Centro di Riferimento Oncologico, Aviano, Italy), and (ii) histiocytic lymphoma cell line U937. Both cell lines were cultured in RPMI 1640 supplemented with 10% (vol/vol) fetal bovine serum, 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mM l-glutamine and were maintained at 37°C in a humidified atmosphere with 5% CO₂. The cells (LYMPH, 2 × 10⁵ cell/ml; U937, 5 × 10⁴ cell/ml) were incubated at 37°C in 5% CO₂ for 24 or 48 h in a 96-well plate in the presence of different concentrations of test compound. The cellular suspension then was mixed with 20 μl of CellTiter 96 aqueous one solution cell proliferation assay (Promega Corporation, Madison, WI, USA). After a further 4 h, the absorbance of produced formazan at 490 nm was recorded with a 96-well plate reader (Multiskan EX; Thermo Electron Corporation, Vantaa, Finland) (26, 27). The mitotic inhibitor demecolcine and the test compound solvent DMSO were used as controls.

Comet assay.

The comet assay was performed according to Singh et al. (28), with minor modifications, on fresh human leukocytes that showed a mortality of ≤30% (29) immediately after the treatment with WSP934 (1 h, 37°C). Cell lysis was carried out at 4°C overnight by exposing cells to a buffer containing 2.5 M NaCl, 10 mM Na₂EDTA, 10 mM Tris-HCl, 1% (vol/vol) Triton X-100, and 10% (vol/vol) DMSO, pH 10. DNA unwinding was achieved over 20 min in an electrophoretic alkaline buffer (1 mM Na₂EDTA, 300 mM NaOH, 0°C, pH >13). Electrophoresis then was carried out for 20 min (0.78 V/cm, 300 mA) at 0°C in the same buffer, followed by neutralization in 0.4 M Tris-HCl, pH 7.5. DNA was stained with 100 μl ethidium bromide (2 μg/ml) before the examination at 400× magnification under a Leica DMLB fluorescence microscope (excitation filter BP [band pass], 515 to 560 nm; barrier filter LP [long pass], 580 nm) using an automatic image analysis system (Cometa, release 2.1; Sarin, Florence, Italy). The migration distance between the edge of the comet head and end of the tail (tail length [TL]) provided representative data on genotoxic effects. The samples were coded and evaluated blind (50 cells for each of two replicate slides per data point). All of the tests were performed at least three times. Ethyl methane sulfonate (EMS) (2 mM) was used as a positive control.

Modified comet assay for detection of oxidized bases.

DNA glycosylase and endonuclease III (ENDOIII) are able to recognize and cleave specific DNA base modifications, such as oxidized pyrimidine bases. Strand breaks can be detected by the comet assay (performed at pH 12.1). Oxidatively generated damage can be evaluated easily by comparing the DNA migration in enzyme-treated and buffer-treated slides (30, 31). Briefly, after cell lysis, the slides were washed three times with enzyme buffer (0.1 M KCl, 0.5 mM Na₂EDTA, 40 mM HEPES, 0.2 mg/ml bovine serum albumin, pH 8, with KOH) and incubated with ENDOIII (or in buffer alone as a control, detecting only strand breaks and alkali-labile sites). Hydrogen peroxide (50 μM), an inducer of oxidative stress, was used as a positive control. ENDOIII was isolated from bacteria containing overproducing plasmids (A. Collin, University of Oslo). The enzyme-treated gels reveal alkali-labile sites and strand breaks (ALS/SB) as well as oxidized bases (ALS/SB + OX). Assuming a linear dose response, the subtraction of (ALS/SB) from (ALS/SB + OX) gives a measure of oxidized bases. OX DNA was calculated using the following formula, in which the control sample is represented by dose 0: (TL_+ENDOIII − TL_−ENDOIII)_{treated sample} − (TL_+ENDOIII − TL_−ENDOIII)_{control sample}.

The SPSS 11 (SPSS Inc., Chicago, IL, USA) statistical package was used to analyze statistical differences between samples. Statistical differences between controls and treated samples first were determined with the nonparametric Wilcoxon rank-sum test for each experiment. The mean values from the repeated experiments were used in a one-way analysis of variance. Analysis of variance was followed by single or multiple pairwise comparisons (Dunnett's test). Least-squares linear regression analysis was used to calculate specific activity. Furthermore, to compare the slope of the dose-response curves, assuming a linear dose response, the angular coefficients were calculated; they represent the DNA migration increases per dose unit.

MN assay.

The micronucleus (MN) assay was performed using blood samples from healthy, nonsmoking males, as provided by the AVIS (Italian Association of Voluntary Blood Donors). Lymphocytes were separated by Lymphoprep density gradient (Axis-Shield PoC As, Oslo, Norway) and, after two washes in RPMI 1640 medium, cultured at a concentration of 5 × 10⁵ cells/ml in RPMI 1640 containing 15% fetal calf serum (FCS), 2% (vol/vol) KaryoMAX phytohemagglutinin (Invitrogen Ltd., Collegeville, PA, USA), 2 mM l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin. The cultures were incubated at 37°C in 5% CO₂ for 72 h. Cytochalasin B was added 44 h after the start of incubation at a final concentration of 6 μg/ml; at 48 h, the lymphocytes were treated with WSP934. Sterile DMSO (1.6 μl/ml) was used as a solvent control, and the positive controls were EMS (120 μg/ml), bleomycin (6 μg/ml), and demecolcine (0.5 μg/ml). Each treatment was tested with cells from two donors and performed in duplicate (i.e., four cultures were set up for each treatment). After incubation, the lymphocytes were collected and resuspended in a mild hypotonic solution (0.075 M KCl) and then added to an ice-cold 5:3 acetic acid-methanol solution. After centrifugation (500 × g, 10 min), the pellets were resuspended in cold (−20°C) methanol and maintained at −20°C (at least 24 h). The cells then were washed twice (7:1 methanol-acetic acid, −20°C), plated onto cold degreased slides, air dried, and then stained with 2% (vol/vol) Giemsa (Carlo Erba, Milan, Italy). Scoring was done using an Exacta-Optech (Munich, Germany) light microscope at ×400 magnification. Micronuclei were scored according to the criteria described by Fenech (32, 33). At each dose, at least 1,000 binucleated (BN) lymphocytes for each culture were examined for the presence of one or more MN. The MN frequency in 1,000 BN cells then was calculated for each treatment. Assays with too few BN cells to determine the MN frequency (due to extensive cytotoxic effects) were classified as toxic. Cell cycle parameters were evaluated by classifying 1,000 cells as either mononucleated, binucleated, trinucleated, or tetranucleated, and the percentages of the different cell types were calculated. The nuclear division index (NDI) was calculated by the formula NDI = (M1 + 2M2 + 3M3 + 4M4)/N, where M1 through M4 indicate the number of cells with 1 to 4 nuclei and N indicates the total number of cells scored (34).

The statistical analysis of MN frequency was performed using the chi-square test. NDI data were analyzed by Student's t test.

RESULTS

In vitro activity.

We previously reported (11, 17) that the melamine nitrofuran WSP934 has very potent activity against T. brucei rhodesiense STIB 900, with a 50% effective concentration (EC₅₀) of about 11 nM, which is around the same order as that of melarsoprol (6 nM), a currently licensed drug. For comparison, the EC₅₀ of nifurtimox is 1.5 μM. This compound is a nitrofuran registered for use against Chagas' disease and, in combination with eflornithine, to treat human African trypanosomiasis (2). Nifurtimox shows more than 100-fold less activity against trypanosomes, which might relate to the melamine group enhancing uptake into the parasites. Work has shown that the melamino group can be taken up into T. brucei through other transporters in addition to the P2 transporter (11, 17). To further characterize WSP934, the EC₉₀ was measured and found to be 26 nM, again similar to that of melarsoprol (20 nM). This EC₉₀ value underlines the potent antitrypanosomal activity of WSP934.

In vivo activity.

We previously reported that WSP934 is curative of T. brucei brucei STIB 795 when given i.p. at 20 mg/kg for 4 consecutive days, starting at day 3 postinfection (Table 1) (11, 17). WSP934 then was investigated in the more stringent T. brucei rhodesiense STIB 900 acute mouse model, in which parasites are believed to invade extravascular sites comparatively early in infection. At the dose that was curative in the STIB 795 model, the compound was not curative in the STIB 900 model, although it did cause a significant enhancement in the life expectancy and one mouse survived until day 60 (11, 17). Increasing the dose to 40 mg/kg led to a cure in the T. brucei rhodesiense STIB 900 model. This is a significant result, as very few compounds are able to cure this model. For example, pentamidine, one of the drugs registered for stage 1 HAT, is unable to cure the stringent disease model at 4 doses of 25 mg/kg. The compound was significantly more potent than nifurtimox, consistent with observations from the in vitro activities.

TABLE 1.

In vivo activity of WSP934 in stage 1 models of HAT

Compound	In vivo activity values forb:
	T. brucei brucei STIB 795			T. brucei rhodesiense STIB 900
	Dose (mg/kg, 4 doses each)	No. cured/total no. infected	MSD (days)	Dose (mg/kg, 4 doses each)	No. cured/total no. infected	MSD (days)
WSP934	20 i.p.	4/4	>60	20 i.p.	1/4	>35.3
				40 i.p.	4/4	>60
				100 p.o.	1/4	30
				100 p.o.a	2/4	45.8
Melarsoprol	0.5 i.p.	0/4	42.5	0.5 i.p.	0/4	19.5
	1 i.p.	2/4	>60	1 i.p.	0/4	21
				2 i.p.	2/4	45
				8 i.p.	4/4	>60
Nifurtimox	20 i.p.	0/4	9.8	20 i.p.	0/4	6.25
	100 p.o.	0/4	26.8	100 p.o.	0/4	16
Control	0		6	0		6.5

^aEight doses dosed twice per day, giving a 4-day treatment in total.

^bi.p., intraperitoneal injection; p.o., oral administration; MSD, median survival days.

Oral bioavailability is desirable for the development of compounds for treatment of HAT (35). Therefore, we investigated the activity of WSP934 when given orally (Table 1). Oral dosing with 100 mg/kg for 4 days led to a significant increase in life span (30 days) compared to that of the control (6.5 days) in the T. brucei rhodesiense STIB 900 model. One mouse still was disease free after 60 days. Repeating the regimen, dosing twice a day for 4 days, cured 2 out of 4 infected mice and increased life span (46 days), indicating a reasonable degree of oral activity.

We next investigated the activity of WSP934 in a stage 2 model of infection (Table 2). At 10 doses of 50 mg/kg given i.p., the survival time of mice was increased, although cure was not found. In the same model, 5 melarsoprol doses at 10 mg/kg was curative for all mice treated. These results do not exclude that subcurative levels of drug reached the CNS thanks to the prolonged dosing, but it is also possible that prolongation of survival time relates to sustained prophylactic blood levels of drug, which prevented reestablishment of hemolymphatic disease.

TABLE 2.

In vivo activity of WSP934 in stage 2 model of HAT^a

Compound	T. brucei brucei GVR35 in vivo activity
	Dose (mg/kg, i.p. [no. of doses])	No. cured/total no. infected	MSD (days)
WSP934	50 (5)	0/3	43
Diminazene diaceturate	40 (1)	0/5	47
WSP934	50 (10)	0/5	99
Diminazene diaceturate	40 (1)	0/5	68
Melarsoprol	10 (5)	4/4	>180

^aDiminazene clears parasites from the bloodstream. Relapse comes from brain infection. The MSD and the days of relapse may vary in different experiments due to a number of factors; hence, it is included for each experiment.

ADME properties.

WSP934 shows high potency in vitro. However, the inability to kill trypanosomes in the CNS indicates that the compound does not possess ideal pharmacokinetic properties for CNS penetration and, to a lesser extent, absorption by the gastrointestinal tract. In addition to the potency against the molecular target, the drug-like properties of new compounds are key to their in vivo efficacy (36, 37). Therefore, a series of experiments was carried out to investigate the ADME properties of WSP934 and to see if we could use these results to better understand the in vivo efficacy data (Table 3).

TABLE 3.

Physicochemical properties and microsomal stability of WSP934

Solubility (concn, in mg/ml) at pH:		Lipophilicity (logD) at pH:		Protein plasma binding (%)	t½ (min) for liver microsomes (with NADPHa)
1.2	7.4	1.2	7.4		Human	Mouse
0.85 (1.2)	0.033	−1.36	0.05	51	>100	40

^aThe microsomes are stable in the absence of NADPH.

WSP934 had reasonable solubility at pH 1.2 but much lower solubility at pH 7.4. The higher solubility at pH 1.2 probably is due to salt formation. The logD (lipophilicity) value at pH 1.2 was −1.36, and it increased to 0.05 at pH 7.4. The combination of very low lipophilicity, low water solubility, high polar surface area (136 Ų) (38), 5 hydrogen bond donors, and 7 hydrogen bond acceptors indicates that the molecule has low passive cellular permeability. We predict that there is carrier-mediated uptake of this compound into T. brucei. Active transport into the parasite and low passive permeability into mammalian cells may account for the selectivity of the compound for parasite over mammalian cells. However, this also could account for low oral and CNS bioavailability.

There are a number of factors which are important in maintaining the level of compound in the bloodstream in a form that is available to the parasite. WSP934 appeared stable to metabolism by human liver microsomes, while the stability toward mouse (the model used in efficacy studies) liver microsomes was lower but still acceptable for a lead compound. WSP934 also had very low protein plasma binding, indicating relatively high unbound levels in the blood (Table 3).

The compound also was investigated for its ability to inhibit several cytochrome P450 enzymes as an indication of possible liabilities with regard to drug-drug interactions. Results showed that there was no significant inhibition of these enzymes (see Table S1 in supplemental material).

A pharmacokinetic study was carried out in male, Swiss white NMR1 mice at an i.p. dose of 40 mg/kg (Fig. 2 and Table 4). This is the same breed of mouse, same route of administration, and same dose as those used for the in vivo efficacy studies. On administration of WSP934, it immediately appeared in the bloodstream. A rapid initial clearance of the compound from the bloodstream was followed by a steady decrease in levels. The terminal half-life was calculated at 3.2 h. In spite of the rapid initial clearance of compound from the plasma, levels were greater than the EC₉₀ for trypanocidal activity (7 ng/ml) for at least 12 h. This was true at total levels and free (non-protein-bound) levels. The rapid initial loss of compound from the bloodstream may be due to distribution of the compound to the body tissues. Once the equilibrium between the blood and tissue is established, the rate of elimination of WSP394 from the bloodstream decreases.

FIG 2.

Logarithm of the concentration of WSP934 (ng/ml) found in the plasma of Swiss white NMR1 mice following i.p. administration of 40 mg/kg of compound. Most data points are the means from 3 measurements. Total (⧫) and unbound (■) concentrations are plotted. The EC₉₀ for activity against T. brucei rhodesiense STIB 900 (▲) is shown for comparison. Individual data points are shown in Fig. S1 in the supplemental material.

TABLE 4.

Plasma pharmacokinetic parameters of WSP934-treated NMR1 mice (one dose of 40 mg/kg, i.p.)

Parametera	Mean value
Cmax (ng/ml)	2,809
Tmax (h)	0.10
AUC0-12 (ng/ml · h)	1,263
AUC∞ (ng/ml · h)	1,524
t½ (h)	3.24

^aC_max, maximum plasma concentration; T_max, time to reach C_max; AUC, area under the plasma concentration-time curve from time 0 to 24 h; AUC_∞, area under the plasma concentration-time curve from time zero to infinity; t_1/2, terminal elimination half-life.

Genotoxicity.

Some nitroheterocycles are known to have problems with genotoxicity (39, 40). For example, the development of megazol as a potential treatment for HAT was halted due to its genotoxicity (25, 41–43). Therefore, we conducted a number of in vitro assays on WSP934 to profile any potential genotoxic liabilities.

Ames Salmonella/microsome test.

The standard first regulatory test for genotoxicity is the Ames test. We describe WSP934 data compared to those of the 5-nitrofuran nifurtimox. Data are reported for a number of different strains (TA98, TA100, TA1535, TA98NR, and TA100NR) of Salmonella with or without S9 (Fig. 3; also see Table S2 in the supplemental material).

FIG 3.

Revertants induced by WSP934 and nifurtimox in different Salmonella strains, with or without exogenous metabolic activation. Data are expressed as means ± standard deviations (SD) (n = 3). Significant increases of revertants are in boldface. −S9, in the absence of S9 mix; +S9, in the presence of S9 mix.

WSP934, as well as nifurtimox, was more active for base pair substitution (TA100) than frameshift (TA98) induction. The difference in activity on TA100 and TA1535 is due mainly to the enhancement of error-prone repair by the plasmid pKM101 in TA100, since both strains are sensitive to base pair substitution mutagens. Microsomal fraction S9 weakly reduced both cytotoxic and genotoxic effects (TA100 strain) of WSP934. However, the genotoxic effects of nifurtimox did not appear to be altered by S9. Furthermore, whereas the mutagenicity of nifurtimox strongly depended on its nitroreduction (both in TA100 and TA98 strains), the genotoxic potential of WSP934 could be considered independent of, or even possibly slightly decreased by, nitroreductase activity, as indicated by the data on NR strains. This suggests mechanisms of genotoxicity other than nitroreduction by the classical nitroreductase activity, which is lost in the NR strains. It also indicates that nifurtimox and WSP934 have very different mechanisms of genotoxiciy.

Cytotoxicity on human cells.

Although the Ames test is considered an important first step in the assessment of genotoxicity, the fact that it employs bacteria while genotoxic liability ultimately relates to mammalian cells led us to use several mammalian cell cytotoxicity/genotoxicity tests as well.

After short-term exposure of leukocytes to WSP934 (1 h of treatment, up to the solubility limit, i.e., 100 μg/ml), cell survival suggested no overt cytotoxicity (data not shown). A longer-term toxicity assay on cultured lymphocytes (24-h treatment) also failed to show significant effects induced by the compound. However, a significant decrease in viability (P < 0.05) was observed when the same 24-h assay was applied to the U937 cell line (Fig. 4A). The U937 cells double in number over 24 h, while lymphocytes do not divide over this period; therefore, eventual toxic effects on their proliferation rate could not be observed using these short incubation times (i.e., 1 and 24 h). To determine whether WSP934 had cell cycle-dependent cytotoxic effects on lymphocytes as well, they were treated for 48 h (a time sufficient to allow these cells to divide). In this case, lymphocytes showed a behavior similar to that of U937 cells treated for 24 h (Fig. 4B). Taking into account both curve slopes and the time necessary for mitosis in the cell types, it is likely that WSP934 is cytostatic (at least to mammalian cells) in a way similar to that of demecolcine (Fig. 4C).

FIG 4.

Cytotoxicity induced by different compounds in the U937 cell line and cultured lymphocytes treated for 24 h (A) or 48 h (B). (C) Demecolcine and DMSO treatment of lymphocytes (48 h) also are reported as controls. The dashed line indicates cell seeding. Data are expressed as means ± SD (n = 4).

Comet assay.

The comet assay is a way to measure damage to DNA of mammalian cells. Results obtained by this test are reported in Fig. 5 and Table S3 in the supplemental material. In the comet assay performed at pH >13, WSP934 induced a significant increase of DNA damage for doses of ≥16 μg/ml. A clear dose-response relationship in the range of 8 to 100 μg/ml was observed (DNA migration increase per dose unit, 0.35; R² = 0.97). At pH 12.1, significantly (P < 0.05 by Student's t test) less DNA migration was detectable than at pH >13 (the lowest effective dose was 25 μg/ml; the dose-response relationship was in the range of 16 to 100 μg/ml, with a DNA migration increase per dose unit of 0.17; R² = 0.92). These results show that WSP934 is able to produce both DNA strand breaks and alkali-labile sites, the latter being more easily detected at the highest pH (more alkaline-resistant lesions also are detectable at pH >13) (44).

FIG 5.

Total and specific oxidative DNA damage detected by the comet assay in fresh human leukocytes treated with WSP934 and nifurtimox (37°C, 1 h). Data are expressed as means ± SD (n = 3). Significant differences (P < 0.05) with respect to dose 0 (Dunnett's post hoc test) are in boldface.

Nitro-compounds often are prodrugs which exert their toxic effects only after activation by nitroreductases. Reduction of the nitro-group by eukaryotic type II nitroreductase leads to the production of reactive oxygen species, which results in oxidative stress (45). To assess if the genotoxicity of WSP934 also was linked to oxidative mechanisms, as expected for nitro-compounds, we proceeded to specifically measure oxidatively damaged DNA by using bacterial repair endonuclease III in the modified comet assay protocol (46) performed at pH 12.1 (Fig. 5; also see Table S3 in the supplemental material). ENDOIII treatment significantly (P < 0.05 by Student's t test) increased DNA migration. The induction of oxidation of DNA bases by WSP934 treatment is indicated for doses of ≥25 μg/ml (Fig. 5; also see Table S3). Nifurtimox induced significant effects at lower doses (4 μg/ml) than WSP934, even if the new compound was more active than nifurtimox at the highest doses.

Micronucleus assay.

A micronucleus assay also was performed to determine in situ DNA damage (Fig. 6; also see Table S4 in the supplemental material). This test allows detection of both aneugenic (whole chromosome) and clastogenic (chromosome breakage) damage in interphase cells. This system allows identification of fixed damage, as the cells where this event occurred are examined only after they have already undergone mitosis. The comet assay measures clastogenic effects, but it can be considered to be a more sensitive test, as it also detects primary damage induced through several mechanisms of DNA alteration (i.e., base alkylation and base oxidation) that still can be repaired before giving rise to a fixed mutation. Results showed that the nuclear division index (NDI) was significantly (P < 0.05) decreased by both WSP934 and nifurtimox treatments for doses of ≥16 μg/ml. A weak nonsignificant increase of micronuclei (25 μg/ml), followed by a strong decrease (related to the blockage of nuclear division), was detected in WSP934-treated cells. Nifurtimox induced a significant increase of MN at 100 μg/ml.

FIG 6.

Nuclear division index (NDI) and mean frequency of micronuclei (MN) in binucleated cells and in human lymphocytes treated with WSP934 and nifurtimox. nd, not determinable. An asterisk indicates significant differences with respect to dose 0 (chi-square test).

DISCUSSION

There is an urgent need for new therapies to treat HAT, and there is little in the drug discovery compound pipeline. After the development of the diamidine prodrug pafuramidine maleate was stopped for toxicity issues, two compounds currently remain in clinical trials for stage 2 HAT treatment: fexinazole and the oxaborole SCYX-7158, both being progressed by DNDi (2). A third compound, the diamidine derivative CPD-0802, is in advanced preclinical development (47).

Nitroheterocycles have consistently demonstrated potent trypanocidal activity. Benznidazole and nifurtimox are licensed for the treatment of Chagas' disease caused by Trypanosoma cruzi, and nifurtimox (a 5-nitrofuran) also is currently used, as a combination therapy with eflornithine, to treat stage 2 HAT due to T. brucei gambiense. As mentioned above, fexinazole (a nitroimidazole) is in clinical trials for this disease. Other nitroimidazoles (48) and other nitroheterocycles (49) have been shown to be potentially trypanocidal.

We reasoned that by attaching nitroheterocycles to moieties that enable selective uptake into trypanosomes (e.g., the melamine moiety), we could selectively target such compounds to trypanosomes but not mammalian cells, increasing their potency and selectivity. The lead compound WSP934 was synthesized by following this logic as a potential agent for the treatment of HAT. The compound shows very potent activity in cellular models of infection, comparable to melarsoprol. WSP934 also shows good activity in rodent models of stage 1 infection, notably being able to cure the stringent T. brucei rhodesiense STIB 900 model. It is much more potent than nifurtimox, which has undergone clinical trials for HAT (15, 16, 50) and now is registered for treatment as NECT, a combination therapy with eflornithine (2).

Ideally, any new treatment for HAT would be an oral treatment for stage 1 infection and an oral and/or injectable treatment for stage 2 infection. Therefore, we undertook further efficacy studies with WSP934 to see if it would be suitable for further development. The compound shows good in vivo efficacy when given intraperitoneally but has reduced activity when given orally and very weak activity in the CNS model. The lower oral activity of the compound may be a feature of the physicochemical properties of the molecule, reducing passive diffusion across membranes. It is possible that larger or longer treatment regimens could achieve cure of stage 1 disease by oral delivery, but this would need to be investigated further. The lack of CNS penetration also is probably due to the physicochemical properties of the molecule (38), preventing passive diffusion across the blood-brain barrier.

In terms of other parameters, WSP934 looks promising. While it shows a moderate metabolic turnover in mouse microsomal assays, the instability is not unacceptable. Furthermore, in human microsome assays the compound appears stable. WSP934 does not appear to bind significantly to the five key cytochrome P450 enzymes, and there is low protein plasma binding. Pharmacokinetics data show that when the compound is dosed intraperitoneally at 40 mg/kg, it maintains levels above the free IC₉₀ for at least 12 h in mice, accounting for its efficacy in the mouse model of infection.

Given the very potent activity of this compound against T. brucei and the good efficacy in a mouse model of infection when dosed intraperitoneally, WSP934 is clearly getting into the parasite at a high level. This may well be due to active transport through the P2 and related transporters. However, in order to obtain selective uptake through this route, the compound needs to have low passive permeability into mammalian cells. Low passive permeability is not consistent with high levels of passive oral or CNS uptake. The challenge in using this approach is to design compounds which are selectively taken into the parasite but are still orally bioavailable and penetrate the CNS.

We selected the nitroheterocycle moiety as a trypanotoxic unit as a result of a substantial body of evidence that indicates that trypanosomes can be exceptionally vulnerable to members of this class of compounds. Our rationale has been that by attaching this unit to moieties, such as the melamine moiety, that allow selective uptake into trypanosomes (51), we might increase potency against trypanosomes and selectivity against host cells. A consequence of this approach could be the pharmacokinetic difficulties that we have encountered, in that the compounds are designed to preferentially enter trypanosomes and not host cells (thus, permeation of the intestine and blood-brain barrier is restricted). Even here, however, recent evidence indicates it is possible to design polar molecules that can be carried over intestinal epithelia and the blood-brain barrier, probably by specific transporters, and still retain the capacity to enter trypanosomes to very high concentrations (52).

In spite of the fact that nitroheterocycles are frequently dismissed because of genotoxicity issues related to several members of the class, it should be noted that there are a number of them in current use as therapeutic agents. Among these is metronidazole, one of the most commonly prescribed drugs for anaerobic bacterial infections. PA824 currently is in trial for use against tuberculosis and is also nongenotoxic in classical assays (53). Nitazoxanide (54) recently has been licensed for treatment of cryptosporidiosis, and this compound also has passed stringent tests on genotoxicity. The 5-nitroimidazole fexinidazole, currently in advanced clinical trials for HAT, is another example of how nitrocompound properties can be exploited to specifically target microorganisms with negligible side effects for humans. Fexinidazole and its metabolites are genotoxic in the standard Ames test but do not show any toxicity in mammalian cell assays (55, 56). The Ames test genotoxicity of this compound is due to activation by bacterial type I nitroreductases (the same type of enzymes found in trypanosomes) following a reaction that cannot be carried out by type II nitroreductases present in mammalian cells. Therefore, while a compound containing aromatic nitro functionality must be looked at in detail, it is possible to develop nitroheterocycles that have acceptable genotoxicity.

WSP934 shows genotoxic potential. In the Ames assay, the predominant form of mutagenic event is base pair substitution rather than frameshift induction. Some nitrocompounds are readily activated by the bacterial nitroreductase found in Salmonella. However, this type I nitroreductase is not present in mammalian cells (57); hence, it can give false positives in the Ames assay. Therefore, WSP934 also was assayed in nitroreductase-deficient strains. Interestingly, genotoxicity also is apparent in these strains, indicating that mechanisms other than the involvement of type I nitroreductase of Salmonella play roles in the genotoxic activity of WSP934. In mammalian cells, the comet assay, but not the micronucleus test, also points to genotoxic potential in these systems, although it is notable that at low doses activity is lower than that of nifurtimox.

WSP934 also was investigated for its cytotoxicity on human leukocytes. There is no overt cytotoxicity in short-term (1-h) exposures. Longer exposures do not show cytotoxic events but do seem to affect cellular division, suggesting that WSP934 has a cytostatic effect, a hypothesis also supported by the decrease of NDI in the micronucleus test.

Analysis of the genotoxic effects of WSP934 reveals that the compound gives rise to several potential mechanisms of gene damage. Probably the most significant is base pair substitution (as shown in Salmonella), which may be caused by oxidative damage to the DNA (as found in human cells). We have not observed any significant effects at the chromosome level for WSP934 (nifurtimox, instead, is capable of inducing a significant increase of micronuclei).

The genotoxic potential of WSP934 certainly will make it difficult to pursue this particular compound in further clinical trials to treat HAT. However, it is noteworthy that in both the Salmonella-based tests and the mammal-based assays, genotoxic potential is witnessed only at concentrations in large excess of those required to kill trypanosomes. Moreover, the pharmacokinetic data indicate that at 40 mg/kg dosing, a maximum concentration of drug in serum (C_max) lower than 3 μg/ml is found, which itself falls well beneath the dose of 16 μg/ml required to induce DNA damage in mammalian cell assays.

In conclusion, it is evident that the trypanocidal activity of WSP934 is exceptional. Its failure to affect a radical cure on oral dosing could be related to its physicochemical properties. Chemical modifications might engender increased oral availability and even blood-brain barrier penetration. It also may be possible to approach the problems of genotoxicity. For example, one study showed that the genotoxic and antitrichomonal activity of a series of nitroheterocycles could be separated (58).

Summary.

WSP934 is a very potent antitrypanosomal compound which has efficacy in vivo. This is one of the more potent leads to have emerged in recent years. In this paper, we described our experiments to determine if WSP934 is suitable for further development or could be exploited as a lead compound for further modification. We showed that WSP934 has sufficient potency for stage 1 disease but probably insufficient uptake across the blood-brain barrier for use in stage 2 disease. Positivity in several experiments for genotoxicity also was detected following treatment with the compound. The combination of these factors precludes further development of this molecule. Additional studies are required to see if WSP934 can be modified to improve its pharmacokinetic properties and to design out the genotoxicity.