*Opinion* **From Colonial Research Spirit to Global Commitment: Bayer and African Sleeping Sickness in the Mirror of History**

#### **Ulrich-Dietmar Madeja \* and Ulrike Schroeder**

Bayer AG Pharmaceuticals, Müllerstrasse 178, 10785 Berlin, Germany; ulrike.schroeder@bayer.com **\*** Correspondence: ulrich-dietmar.madeja@bayer.com

Received: 27 January 2020; Accepted: 4 March 2020; Published: 10 March 2020

**Abstract:** In the early 20th century, a series of epidemics across equatorial Africa brought African sleeping sickness (human African trypanosomiasis, HAT) to the attention of the European colonial administrations. This disease presented an exciting challenge for microbiologists across Europe to study the disease, discover the pathogen and search for an effective treatment. In 1923, the first "remedy for tropical diseases"—Suramin—manufactured by Bayer AG came onto the market under the brand name "Germanin." The development and life cycle of this product—which today is still the medicine of choice for Trypanosoma brucei (T.b), hodesiense infections—reflect medical progress as well as the successes and failures in fighting the disease in the context of historic political changes over the last 100 years.

**Keywords:** African sleeping sickness; development of treatment; suramin; medical history; political history

By the middle of the 19th century, several international health conferences had already set themselves the goal of protecting the colonial countries from tropical diseases such as smallpox, cholera, plague and yellow fever. Then, at the beginning of the 20th century after several epidemics across equatorial Africa, the colonial powers became aware of an additional disease, African sleeping sickness (human African trypanosomiasis, HAT), which posed a threat not only to Europeans travelling to the colonies but also to the local population, as well as to the economic value of the colonies themselves for their ruling countries.

The colonial powers' response was immediate for humanitarian reasons, but humanitarianism reflected the sentiment of the time—"a mixture of benevolent condescension and outright racism" [1] towards the local population—as illustrated by the colonialists' claim that they were "saving hapless Africans from the diseases that plagued them" [1]. However, their intervention was also based on practicalities, such as the need to protect their source of manpower in the thinly populated equatorial zone. Transport at this time relied on human porters or canoes because pack animals were unable to survive in regions infested with tsetse flies. This new epidemic and the resulting deaths among the local population not only caused transport problems, but also aggravated agricultural development, thus threatening to thwart plans to further exploit the colonies [1,2]. However, the turn of the century was also the heyday of microbiology in Europe. Motivated scientists were eager to discover new pathogens, explore life cycles of vectors and develop vaccines and potential new treatments. African sleeping sickness, which posed such a problem for colonial administrations, thus presented a great challenge for microbiologic research [1].

In fact, the challenge was so great as well as worrying that from 1901 to 1913 the colonial administrations sent a total of 15 special research missions to Africa to study the new disease [3,4]. Eight of these missions were from Great Britain, who hosted a major conference on African sleeping

sickness in London in 1907–1908 attended mainly by scientists from Britain, France, Germany and Portugal. Recent research findings as well as potential new drugs for treatment and prophylactic measures were discussed. At the same time, Germany and Great Britain were also active on a political level. They implemented steps to stop Africans infected with the disease from crossing borders and then, in 1911, they signed an agreement outlining their plans to jointly fight African sleeping sickness in West Africa.

Bayer, which had been involved in the production and sale of synthetic dyes since its founding in 1863, expanded its activities at the end of the 19th century to include other business areas. In 1891, the chemist and later CEO of Bayer, Carl Duisberg (1861–1935), established a scientific laboratory and efficient research department in Wuppertal-Elberfeld where dyestuffs and their intermediates were developed, as well as pharmaceuticals (e.g., acetylsalicylic acid, Aspirin™). Unlike the first discoveries of medicines which were based on success in the treatment of disease symptoms, research at this time increasingly focused on combating newly discovered pathogens. It was German physician and Nobel prize winner Paul Ehrlich (1854–1915) who defined the term chemotherapy as the "creation of a chemical that would attack a specific pathogen" [1]. He demonstrated that dyes could also be effective against specific pathogens. The so-called trypan dyes were identified as being effective against animal trypanosomes but turned out to be too toxic to be used for treatment of human African trypanosomiasis.

In 1909, Wilhelm Roehl (1881–1929), Ehrlich's former assistant, asked Duisberg to help by providing dyes and money for animal experiments. Heinrich Hoerlein, head of Bayer's Pharmaceutical Department since 1910, realized the importance of Roehl's work and employed him. In 1916, with the help of a small team of chemists, he developed the first effective drug for the treatment of African sleeping sickness: the compound "Bayer 205," a colorless and odorless urea derivative which was later named Suramin.

In 1921, the compound was successfully tested on animals and passed on to the Hamburg Clinic for Tropical Diseases for further testing. There, the English engineer Christopher G. James had been suffering from sleeping sickness for over eight months. James had contracted the infection in Rhodesia and seemed to have little chance of survival. After only a few injections with "Bayer 205," however, Christopher G. James was well again and was able to travel back to Africa [5].

Encouraged by this success, Bayer sent an expedition to South Africa to carry out the necessary field trials on site, despite the difficult conditions after the First World War. The German microbiologist and pharmacologist Friedrich Karl Kleine (1869–1951), an expert in African sleeping sickness research, took the lead. In November 1921, the expedition, equipped with 30 kg of "Bayer 205," set off from Cape Town to Rhodesia (Figures 1–3).

The scientific experiments began in January 1922 and showed that oral applications only had a temporary effect and that injections were much more effective. The results were so convincing that the Governor General of Belgian Congo even invited the expedition to continue its work in the southern Congo region. In 1923, the "remedy for tropical diseases"—Suramin—came onto the market under the patriotic name "Germanin" (Figure 4).

After the First World War, scientific collaborations across Europe slowly resumed. As part of its war reparations and in an attempt to regain possession of its former colonies, Germany expressed its willingness to reveal the secret formula of Suramin. After consultations, France and Great Britain agreed not to accept the offer [1]. A French pharmacologist Ernest Fourneau (1872–1949) succeeded in reverse-engineering the drug based on patents that Bayer had taken out and renamed it "Fourneau 309" [1,6–8]. The pharmaceutical company Rhône-Poulenc marketed it then under the trade name "Moranyl."

**Figure 1.** (Bayer Archive. 1921. (Picture 0-19682)) Route of the Bayer expedition starting on 2 November 1921 in Cape Town, travelling to the areas most affected by African sleeping sickness in what is today Tanzania, Burundi, Ruanda, DR Congo as well as a small part of Mozambique, Zimbabwe and Zambia, where the expedition ended in Kiambi in late 1922.

**Figure 2.** (Bayer Archive. 1921. (Picture 0-34295)) Screening of local population for symptoms of African sleeping sickness in the Urambi camp at Lake Tanganyika in 1921.

**Figure 3.** (Bayer Archive. Karl Friedrich Kleine with a Patient (Picture 0-3438301)) Friedrich Karl Kleine demonstrating the palpation of lymph nodes during screening of patients.

**Figure 4.** (Bayer Archive. Product "Bayer 205", ca. 1934. (Picture 0-25791)) Pack of ampoules of "Bayer 205" (Germanin), ca. 1934.

Despite being effective in the treatment of the first stage of African sleeping sickness, Suramin was not able to reverse the disease course once the trypanosomes had penetrated the blood-brain barrier. As such, Suramin remained the treatment of choice for acute cases of African sleeping sickness [4].

The availability of Suramin played a significant role in controlling the 1920s epidemic and subsequent outbreaks and in significantly decreasing the number of reported cases until the 1940s. At that time new treatments also became available. Pentamidine, discovered in 1940, started to be used for the treatment of the first stage of T.b. *gambiense* infections. In 1949, Melarsoprol was discovered and used for the treatment of both T.b. *gambiense* and T.b. *rhodesiense* infections. Due to being derived from arsenic, this treatment had many undesirable side effects, some even fatal. Furthermore, increased resistance to Melarsoprol had been observed in certain focal disease areas, particularly in central Africa.

Over the years, the colonial powers introduced extensive screening of populations at risk by mobile teams and implemented early vector control measures [9]. The disease was under control by the mid-1960s with fewer than 5000 cases reported across the African continent (Figure 5) [5].

**Figure 5.** Number of cases of African sleeping sickness reported and population screened [5,10–12]. Graphs show the correlation between population screened (black circles) and number of reported cases (grey columns). Screening and surveillance of endemic disease areas are key to control African sleeping sickness.

By the mid-1960s, most countries affected by African sleeping sickness became independent. Support from the colonial powers ended and most African countries experienced an era of political instability and economic downturn. The effect on health services and on the control and prevention of endemic tropical diseases was disastrous. Consequently, disease control programs were stopped and population screening declined considerably. This situation provoked a new epidemic of African sleeping sickness.

During the 1950s to early 1970s, extensive insecticide spraying was the method of choice for vector control and resulted in a significant reduction in tsetse fly populations, as they are the vector for African sleeping sickness, but concerns about the environmental effect of DDT (Dichlordiphenyltrichlorethan) led to a worldwide ban in the late 1970s.

Since the mid-1970s, decreasing vector control and disease prevention as well as screening and treatment of populations at risk resulted in a steady increase in the number of reported cases of African sleeping sickness. This most recent epidemic lasted until the late 1990s. Around the year 2000, the scale of African sleeping sickness had once again almost reached the levels of the epidemics seen at the beginning of the 20th century (Figure 5) [11,13].

Suramin was added to the "WHO Model List of Essential Medicines" in 1979 and became the medicine of choice for T.b. *rhodesiense* infections, even though the published clinical evidence to support the use of the product remained limited. With no more disease control programs in place in Africa, the demand for medicines to treat African sleeping sickness decreased to a very low level. The only new product, Eflornithine, was registered in 1990. This molecule has shown to be less toxic than Melarsoprol but is only effective against T.b. *gambiense*. Furthermore, the treatment regimen is complex and difficult to apply.

In 1999, pharmaceutical companies started to question whether to continue manufacturing products which were in such low demand. Aventis proposed a substantial increase in price for further supplying Pentamidine, a second-line option to Suramin since 1937. The production of Melarsoprol, used for the treatment of second-stage disease caused by T.b. *rhodesiense* when the central nervous system is involved, became uncertain because of ongoing discussions in Europe over manufacturing the required raw materials containing arsenic. On several occasions, Bayer, which was still producing Suramin on the grounds that no alternative treatments were available, also threatened to halt production [14].

After the colonial-motivated push for research and control of African sleeping sickness in the early 20th century, the increased efforts of WHO, national control programs, bilateral cooperation and nongovernmental organizations (NGOs) have been able to reverse the curve of reported cases after the year 2000, ending the last major epidemic. The availability of novel diagnostic tools such as rapid diagnostic tests (RDTs) and raising awareness in the political will of local countries to fight African sleeping sickness played a crucial role in advancing disease control.

WHO also started public-private partnerships with pharmaceutical companies to secure the supply of the essential medicines to treat African sleeping sickness. In 2001, WHO and the pharmaceutical companies Bayer AG and Aventis (now Sanofi) reached an agreement to provide their medicines to treat African sleeping sickness free of charge for endemic countries. Pharmaceutical companies started to engage beyond donation of medicines and to provide much needed financial contributions for screening of focal disease areas by mobile intervention teals, surveillance and disease mapping, as well as training and public awareness programs. Within the period 2000–2018 the number of reported cases dropped by 95%.

The development of new treatment regimens and products has been part of this success story. In 2009, the new Nifurtimox Eflornithine Combination Therapy (NECT) was introduced for the treatment of T.b. *gambiense* infections. It simplified the use of eflornithine by reducing the duration of treatment and the number of intravenous infusions. In the same year, the combination of Nifurtimox (produced by Bayer) and originally registered for the treatment of American trypanosomiasis, with Eflornithine (produced by Sanofi) was included in the "WHO List of Essential Medicines" and is currently recommended as first-line treatment for the T.b. *gambiense* form. Since 2009, both pharmaceutical companies have been providing these products free of charge to WHO for endemic countries. The products are packed in a patient treatment kit containing all the material needed for their administration in remote rural areas [15].

In 2005, the development of Fexinidazole as the first-all oral treatment for T.b. *gambiense* infected patients started. On 10 July 2019, the WHO added the product to the "WHO List of Essential Medicines." Fexinidazole is indicated for first stage and non-severe second stage and it can simplify and facilitate case management of human African trypanosomiasis caused by T.b. *gambiense* [16].

Under the lead of WHO and the implementation of efficient disease control programs, the number of reported cases of African sleeping sickness dropped below 10,000 (9878) in 2009 for the first time in 50 years and continued to decline with only 997 new cases reported in 2018, the lowest level since the start of systematic global data-collection 80 years ago [15].

One hundred years ago, Suramin was discovered as the first effective treatment of African sleeping sickness. The history of this product and the motivation for fighting African sleeping sickness have changed significantly over time and reflect the historic medical and political changes. In 2012, the "WHO Roadmap on Neglected Tropical Diseases (NTDs)" set the goal to achieve the sustainable elimination of African sleeping sickness as a public health problem by 2020. This goal was reached by WHO and partners a concerted high-impact approach.

New clinical developments are essential to overcome the development of cross-resistance of currently used substances and to advance treatment options. Surely, one of these "will finally put suramin to rest and ease this esteemed great-grandfather of chemotherapy into a well-deserved retirement" [17]. Along with other interventions, the next goal of achieving disruption of transmission of African sleeping sickness, as outlined in the "WHO NTD Roadmap 2030," seems now to be very feasible.

**Author Contributions:** Both authors equally contributed. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **The Drugs of Sleeping Sickness: Their Mechanisms of Action and Resistance, and a Brief History**

#### **Harry P. De Koning**

Institute of Infection, Immunity and Inflammation, University of Glasgow, Glasgow G12 8TA, UK; Harry.de-Koning@glasgow.ac.uk; Tel.: +44-141-3303753

Received: 19 December 2019; Accepted: 16 January 2020; Published: 19 January 2020

**Abstract:** With the incidence of sleeping sickness in decline and genuine progress being made towards the WHO goal of eliminating sleeping sickness as a major public health concern, this is a good moment to evaluate the drugs that 'got the job done': their development, their limitations and the resistance that the parasites developed against them. This retrospective looks back on the remarkable story of chemotherapy against trypanosomiasis, a story that goes back to the very origins and conception of chemotherapy in the first years of the 20 century and is still not finished today.

**Keywords:** sleeping sickness; human African trypanosomiasis; trypanosoma brucei; drugs; drug resistance; history

#### **1. Introduction**

The first clue towards understanding drug sensitivity and, conversely, resistance, in human African trypanosomiasis (HAT) is that most drugs are very old and quite simply toxic to any cell—if they can enter it. That places the mechanisms of uptake at the centre of selectivity, toxicity and resistance issues for all the older trypanocides such as diamidines (e.g., pentamidine, pafuramidine, diminazene), suramin and the melaminophenyl arsenicals. Significantly, none of these drug classes, dating from the 1910s to the 1940s, were designed for a specific intracellular target and even today identification of their targets has defied all attempts with advanced postgenomic, proteomic and metabolomic techniques—in short, they are examples of polypharmacology, where the active agent acts on multiple cellular targets. One might say they are non-specifically toxic. As such, resistance is unlikely to occur from mutations that change the binding site of a particular intracellular protein. Rather, the resistance mechanisms have been associated with mechanisms of cellular uptake and/or distribution. Some of the newer drugs have a more defined mode of action, and are selective at target level, but resistance to eflornithine, at least, is still associated with the loss of the *T. brucei* transporter that internalises it, rather than with the target enzyme. In the sections below I will examine these issues for each drug separately and show how resistance and treatment failure have changed clinical treatment of sleeping sickness and stimulated the development of the newer generations of drugs, culminating in the latest additions to the arsenal (fexinidazole, acoziborole) [1–4].

#### **2. Diamidines**

The development of the diamidines arose from the observations that advanced (animal) trypanosomiasis is often associated with hypoglycaemia [5,6] and trypanosomes metabolise glucose at a phenomenal rate. This suggested that the chemical induction of hypoglycaemia might be deleterious to trypanosomes in the bloodstream. Several groups tested insulin and other hypoglycaemia-inducing therapies against trypanosomiasis but with at best mild and variable success [7,8]. However, the synthetic insulin substitute synthalin (**1**; for structures see Figure 1) did have remarkable, curative trypanocidal activity [8,9] and, importantly, was not cross-resistant

with the aromatic arsenicals used at the time, nor with suramin ("Bayer 205") [10]. Although it was not immediately clear to what extent this could be attributed to effects on blood sugar levels, that question was rapidly settled by the trypanocidal effects of synthalin on ex vivo trypanosomes [11]. By 1939, Lourie and Yorke, in collaboration with A. J. Ewins of May & Baker Ltd, reported on a large series of new diamidine compounds, among them 4,4'-diamidinostilbene (stilbamidine, **2**) and 4,4'-diamidino,1,5-diphenoxy pentane (pentamidine, **3**) [12]. Stilbamidine was the most active compound—curative with 25–50 μg per 20 g mouse (1.25–2.5 mg/kg b.w.) and a therapeutic index of 30—closely followed by propamidine (**4**) and pentamidine, which displayed a slightly lower therapeutic index of 15. To appreciate the enormous advance this signified, these numbers need to be compared to the dramatically higher minimum curative doses for the aromatic arsenicals then in use: 1000 mg/kg for tryparsamidine (**5**) or 250 mg/kg for atoxyl (**6**), each with a therapeutic index of just two [12]! As stilbamidine appeared to induce adverse neurological sequelae in early clinical trials [13], it was abandoned and pentamidine became the drug of choice for early stage HAT, especially of the *gambiense* variety. The now exclusively veterinary analogue diminazene aceturate ("Berenil", **7**) has also been used initially (and later sporadically) against HAT [14,15], but this practice has long been discontinued.

**Figure 1.** *Cont*.

**Figure 1.** *Cont*.

**Figure 1.** Structural formulas of trypanocides.

Diamidines are believed to be minor groove binders and as such bind to the DNA double helix, particularly targeting AT-rich sequences [16–19], impeding replication and transcription processes in the kinetoplast and/or nucleus. Usually, they accumulate strongly in the trypanosome's single mitochondrion (and mitochondria of cancer cells [20]), the compartmentalisation of these dications being driven by the mitochondrial membrane potential and binding to the kinetoplast DNA (kDNA) (for a schematic of the trypanosome structure, see Figure 2). Indeed, fluorescent diamidines light up the kinetoplast within 1 minute of administration, a process that is much delayed in resistant parasites [21]. Thus, pentamidine is known to accumulate up to mM levels inside trypanosomes [22] and does not exit the cell when the extracellular drug is removed [23]. Furamidine (**8**) and its analogues reportedly accumulate to > 10 mM, associating strongly with kinetoplast and nuclear DNA [17,24]. Similar processes drive mitochondrial accumulation of other cationic trypanocidal agents including isometamidium [25], symmetrical compounds with choline-like head groups [26], furamidines [21,27,28], shielded bis-phosphonium compounds [29] and inhibitors of Trypanosome Alternative Oxidase (TAO) linked to a lipophilic cation [30,31]. Resistance to minor groove binders cannot occur via mutations in the target and the binding affinity does not need to be very high if the accumulation of the drug is to the high local concentrations reported. Thus, resistance is associated with the inability of the diamidine to reach its target, either by preventing its uptake into the cell altogether, or at least preventing its accumulation in the mitochondrion. The latter mechanism was demonstrated in pentamidine-resistant *Leishmania mexicana* parasites [32].

**Figure 2.** Schematic representation of a *T. brucei* trypomastigote, indicating some of the structures and proteins involved in the uptake or mechanism of action of trypanocides.

While pentamidine is used exclusively for the treatment of stage I (haemolymphatic) HAT, there have been reports of successes with 'early late stage' infections, where the parasite has crossed the blood-brain barrier (BBB) but not yet fully penetrated the brain parenchyma [33]. In particular, a study from 1996 reported a cure rate of 94% of early-late stage HAT patients with pentamidine [34]. Thus, pentamidine must be aided across the BBB by a transporter, and Sekhar et al. identified the Organic Cation Transporter 1 (OCT1) as responsible for the process [35], as previously reported in experiments with cell lines expressing all three human OCT isoforms [36]. The reason that pentamidine is not active against cerebral trypanosomiasis, then, is because it is actively extruded again from the CNS to the blood, by P-glycoproteins (P-gp), Multidrug Resistance-Associated Proteins (MRPs) or other extrusion transporters [37]; Yang et al (2014) later reported that pentamidine and the furamidines are not substrates for P-gps [38]. Other diamidines such as diminazene and furamidine are equally ineffective against cerebral trypanosomiasis [39–41], and the distribution of DB75, although readily detectable in-whole brain extracts, was limited to the cells lining the BBB and blood–cerebrospinal fluid (CSF) barriers as it did not penetrate the brain parenchyma [36].

However, DB829 (2,5-bis(5-amidino-2-pyridyl)furan; (**9**)), a close analogue of furamidine, did display remarkable efficacy against cerebral trypanosomiasis in mice and in vervet monkeys [42,43]. This was taken as evidence that DB829 is either recognized more efficiently than furamidine by a BBB transporter importing it into the CSF, or less efficiently extruded from the CSF by a P-gp-type transporter; both compounds have a similar pKa and are dications at physiological pH, precluding any notion that they could simply diffuse across the barrier. However, the late failure of the clinical trials with pafuramidine (**10**), as a result of delayed nephrotoxicity in a small number of patients [40], also impeded clinical development of the all too similar DB829 and its prodrug DB868 (**11**).

The first resistance mechanism identified for the diamidines pentamidine and diminazene (the most-used drugs for early-stage HAT and for AAT, respectively) was loss of the P2 aminopurine transporter [44–46], which is encoded by the gene *TbAT1* [47]. Deletion of this gene did result in modest loss of in vitro pentamidine sensitivity in *T. b. brucei* bloodstream forms (and high-level resistance to diminazene [48]) but not to the extent that it would lead to clinical treatment failure. Indeed, Graf et al. [49] established that field isolates of *T. b. gambiense* from HAT patients that carry the *TbAT1* resistance allele were highly resistant to diminazene but only marginally less sensitive to pentamidine, compared to strains carrying the reference *TbAT1* allele. The *TbAT1R* allele contained several amino acid changes and a 1-codon deletion, compared to the reference. These mutations were systematically evaluated, separately and in groups, by expressing the various mutant forms in a *tbat1*-null strain of *T. b. brucei* [50]. Surprisingly, none of the single amino acid mutations changed pentamidine transport or sensitivity much, but the introduction of two and particularly three such changes at the same time all but disabled the transporter's capacity for pentamidine uptake. The mutational studies were combined with homology modelling of the TbAT1 protein and produced a strong model of substrate binding, in which both aminopurines and pentamidine are bound via dual H-bonds with Asp140 and aromatic interactions with phenylalanines 19 and 316 [50]. This binding mode confirmed earlier binding models based on substrate binding energies with a large set of adenosine and diamidine analogues [51,52].

However, the transport of [3H]-pentamidine by *T. b. brucei* is inhibited only partially by P2/TbAT1 substrates adenine and adenosine, leading to the unambiguous conclusion that (an)other transporter(s) must also be involved [45,53]. Two additional transporters were identified, with Km values of 36 ± 6 nM and 56 ± 8 μM (compare 0.43 ± 0.02 μM for TbAT1) and were accordingly designated the High Affinity Pentamidine Transporter (HAPT1) and the Low Affinity Pentamidine Transporter (LAPT1) [53]. Adaptation of the *tbat1*-null cell line to higher levels of pentamidine in vitro produced a strain, B48, that had lost HAPT1 activity and was 130-fold resistant to pentamidine in vitro (and was also highly resistant to melaminophenyl arsenicals, vide infra) [54]. This clonal strain had lost the HAPT1 transporter but retained normal LAPT function [54]; efforts to induce even higher levels of pentamidine and induce mutations in LAPT1 were unsuccessful [55]. The conclusion from this work was that the HAPT1 transporter was the most important contributor to pentamidine sensitivity in *T. brucei*, with LAPT playing only a minor role at therapeutically relevant drug concentrations.

A genome-wide screen for genes conferring pentamidine sensitivity with an RNAi library identified two loci, encoding for the plasma membrane proton ATPases HA1, HA2 and HA3, and for the aquaglyceroporins TbAQP2 and TbAQP3, respectively [56]. The involvement of the proton pumps indicates that pentamidine uptake is dependent on the proton-motive force or plasma membrane potential (as previously reported for procyclic *T. b.* brucei [53]), which the ATPases help maintain, but the latter RNAi 'hit' was harder to explain, as aquaporins are not known to engage in the uptake of relatively large molecules such as pentamidine, and certainly no cations. However, deletion of TbAQP2, but not of TbAQP3, resulted in a high-level of pentamidine resistance [57], and expression of TbAQP2 in *Leishmania mexicana* promastigotes, which are normally not very sensitive to pentamidine, increased their sensitivity ~40-fold and introduced a high affinity transport function that was indistinguishable from HAPT1 of *T. b. brucei* by Km and inhibitor profile [55].

One possible explanation for the implausible uptake of dicationic pentamidine by an aquaporin is that pentamidine simply binds to the extracellular face of TbAQP2 and is internalised through endocytosis when the protein is internalised for turnover [58]. However, the rate of high affinity pentamidine transport in *T. brucei* procyclic cells is > 10-fold higher than in bloodstream forms [53,59], despite a much lower rate of endocytosis [60]. Moreover, although TbAQP2 is solely localised to the flagellar pocket in bloodstream forms [57], it is located all over the plasma membrane of procyclic forms, and the flagellar pocket is the sole site of endocytosis in trypanosomes [61]. For the same reasons, the high rate of pentamidine uptake in *L. mexicana* promastigotes expressing *TbAQP2* [55] seems incompatible with the endocytosis model. The alternative explanation is that pentamidine does in fact traverse the TbAQP2 channel, and this is explained by the unique changes to the selectivity filter motifs characteristic of aquaporins; these changes have the combined effect of making the pore much wider and removing the cation-excluding arginine residue [57,62]. This allowed docking of the

stretched-out pentamidine in a low-energy conformation inside the pore of a model of TbAQP2 [62]. Much work has subsequently been done to supply definitive proof of either model, and this has definitively come down on the side of uptake through the TbAQP2 pore [63].

In summary, pentamidine resistance is principally associated with changes in *TbAQP2* through mutations, deletions and chimeric rearrangements with the adjacent *TbAQP3* gene. Interestingly, resistance can be bypassed with a nanotechnology formulation of pentamidine in chitosan nanoparticles coated with single domain nanobodies that specifically target trypanosomal surface proteins [64].

#### **3. Melaminophenyl Arsenicals**

Arsenic-based drugs were the very first treatments against sleeping sickness, starting with the use of inorganic sodium arsenite to treat animal trypanosomiasis or 'nagana' by David Livingstone in 1847 or 1848 [65] and David Bruce in 1895 [66]. Although neither of these pioneers achieved a full cure with this treatment (the animals eventually relapsed), it led to the development of the first organo-arsenic compound for sleeping sickness, atoxyl (**6**) as early as 1905, at the time of the major sleeping sickness epidemic in Central and East Africa [67]. However, atoxyl, meaning 'non-toxic' had severe side effects and was only active against early stage HAT, and was followed by tryparsamide (**5)** in 1919 [68]. Tryparsamidine was the first drug to be active against the late stage; however, it was only used against *gambiense* HAT and was considered of little use to either the early or late stage of the Rhodesian form [7,69,70]. A very similar compound, acetyl-*p*-amino-O-oxophenylarsenic acid, known as Fourneau 270 or Orsanine (**15**), was synthesised by Ernest Fourneau at the Institut Pasteur in Paris [71] and had very similar properties as tryparsamide [72,73]; indeed, it was claimed to have twice the selectivity index of tryparsamidine [74] and was in use for approximately 15 years [75].

The melaminophenyl arsenicals (MPA) replaced these earlier arsenicals, particularly tryparsamide, partly because melarsoprol (**12**) had better penetration into the central nervous system ('rendered trypanocidal the cerebrospinal fluid of rabbits' [76]) and tryparsamidine was poorly active against the early stage of the disease [14]. In fact, the MPAs were the first cure for late stage Rhodesian sleeping sickness, which had been considered incurable up to then [72]. More importantly yet, resistance to tryparsamide had developed and had become widespread throughout the early 1940s, to the point that the drug had become ineffective in most cases, particularly in French West Africa and Belgian Congo [77]; in 1947, 80% of new cases in Congo were reportedly resistant to what was then the only treatment for late stage HAT [7,14]. Moreover, resistance to both atoxyl and tryparsamide had been shown to be highly stable over prolonged periods, with Yorke and Murgatroyd concluding that it 'persists indefinitely', even when passed repeatedly through tsetse flies [78]. The MPAs were introduced in the late 1940s [14,69], despite reservations about their toxicity [79], largely driven by concerns about tryparsamide resistance. Extensive experience with MPAs in French West Africa was contained in a 1953 report by the Service Général d'Hygiène Mobile et de Prophylaxie de l'Afrique Occidentale Française [80] and concluded that the proposed "detoxified melarsen oxide", Mel B (melarsoprol, "Arsobal") was at least as toxic as melarsen oxide (**13)** itself but less active, and strongly recommended a return to melarsen oxide. However, the report also acknowledged that the MPAs as a class were a step forward and active against strains resistant to the older arsenicals. Ian Apted, in his comprehensive 1970 review of HAT treatments, states that melarsoprol, developed from melarsen oxide by reaction with Dimercaprol, also called British anti-Lewisite (BAL; antidote for arsenic poisoning) was less toxic than the original melarsen and melarsen oxide [72]. However, even after melarsoprol became the standard treatment for late stage sleeping sickness, the high level of severe adverse effects, with an estimated 10% of patients suffering from reactive encephalitis, half fatal [81,82] remained acceptable only for lack of alternatives. The introduction of an optimised 10-day administration schedule in 2005 [83] reduced the total amount administered and saved money but did not reduce the adverse effects significantly. A further and highly promising development was the proposal of melarsoprol cyclodextrin complexes as the first oral treatment of HAT [84]. In mouse models, this protocol appeared to be much safer, and it

negated the need for intravenous administration of melarsoprol, as a caustic solution in propylene glycol. The melarsoprol cyclodextrin complex was awarded orphan drug status by the European Medicines Agency (EMA) in October 2012 [85] and by the U.S. Food and Drug Administration (FDA) in 2013. A protocol for phase 2 clinical trials of oral complexed melarsoprol in late stage *T. b. rhodesiense* HAT in Uganda was subsequently approved by the EMA (Peter Kennedy, personal communication). However, this has not yet been implemented for lack of funding, presumably because new human trials with arsenicals would be hard to get funding for, and NECT and fexinidazole came along as timely alternatives, at least for *gambiense* HAT (see below).

The discovery of the resistance mechanisms for the MPA melarsoprol (and its veterinary analogue cymelarsan (**14**)) is very similar, and parallel to that of the diamidines. Indeed, the phenomenon of pentamidine-melarsoprol cross-resistance (MPXR) was first reported almost 70 years ago [86,87], and at the time hypothesised to be due to the presence of similar motifs in the melamine-phenyl group and benzamidine moieties and thus 'loss of permeability to, or adsorption affinity of, the melamine grouping in the melarsen-resistant strain may possibly prevent initial uptake of the amidine-type drugs' as well [86]. This insight, based solely on cross-resistance patterns and a faint structural similarity, has now been well validated [88].

The association of MPAs with uptake via P2/TbAT1 was first discovered by Carter and Fairlamb in 1993, who found that out of a large number of biochemicals only adenine, adenosine and the transport inhibitor dipyridamole partly blocked the trypanocidal activity of melarsen oxide, and that an MPA-resistant strain had lost the function of one purine transporter, which they termed P2 [89]. This was later confirmed by other researchers [90], and the joint recognition motif for adenosine, diamidines and MPAs formally established [51,52,91]. As related in the previous section, the P2-encoding gene, *TbAT1*, was identified in 1999 and an allele bearing multiple polymorphisms was found to confer resistance [47]. A similar resistance allele was reported from Uganda in 2001 and was found more frequently in melarsoprol relapse patients than in those cured [92,93]. This was followed by experimental evidence that the deletion of *TbAT1* led to a loss in in vitro MPA sensitivity [48].

As reviewed elsewhere [94], concerns of melarsoprol treatment failure had been rising throughout the 1990s and early 2000s [95–98]. The confirmation of *TbAT1* resistance alleles, as well as the well-documented toxicity of melarsoprol [81] and the confirmation that the treatment failures could not be attributed to individual patients' differences in melarsoprol pharmacokinetics and distribution [99], shifted first-line treatment in many centres, including Omugo in Uganda, to eflornithine monotherapy. Resampling of clinical isolates 4 years later no longer detected the *TbAT1<sup>R</sup>* allele in Omugo; in contrast, the mutant allele was readily amplified from patients, including five relapse cases, in the Moyo treatment centre, also in Northern Uganda, that had continued to use melarsoprol as the first-line drug [100]. Similarly, no *TbAT1R* alleles and no MPA-resistant isolates were found in a 2007 clinical study in South Sudan—an area that had also switched to eflornithine in 2001 after high melarsoprol relapse rates [101]. However, in the latter study it remained unproven whether the *TbAT1* mutations had disappeared after melarsoprol treatment had been largely discontinued or that, alternatively, the treatment failures had not been due to *TbAT1* mutations in the first place, as no sampling had been done before the switch to eflornithine. Conversely, the Ugandan study had not tested the *TbAT1R*-bearing isolates MPA for sensitivity in vitro or in a controlled animal model, and thus the link of clinical melarsoprol failure and *TbAT1* mutations has remained formally unproven, although highly plausible. Specifically, it could not be ruled out that additional factors (whether patient- or trypanosome-related) also played a role.

Meanwhile, Bridges et al. [54] showed that MPA resistance was much higher in laboratory strains that had functionally lost both the TbAT1/P2 transporter and the HAPT1 transport functions. As related in the previous section, HAPT1 was found to be encoded by *TbAQP2* [57] and the MPA resistance was due to a chimeric rearrangement in the *TbAQP2-TbAQP3* locus [55]. Several studies subsequently found such TbAQP2-3 chimeras, or outright TbAQP2 deletions, in clinical isolates from South Sudan and the DRC, demonstrating a clear link between *AQP2* mutations and MPXR [49,62,101]. The definitive word on this was the demonstration that the introduction of a wild-type *AQP2* gene in the resistant *T. b. gambiense* isolates restored sensitivity, whereas the expression of two different chimeric AQP2/3 genes, from an MPXR *T. b. gambiense* isolate and from the MPXR *T. b. brucei* clone B48, into an *aqp2*/*aqp3* null cell line of *T. b. brucei*, did not [55,102]. *TbAQP2* deletions were also found in two *T. b. rhodesiense* strains adapted in vitro to pentamidine and melarsoprol, respectively [103], and as such there is little remaining doubt, if any, that mutations in the *TbAQP2* gene are the principal determinant of MPXR, and that its unique pore architecture is what made *T. brucei* spp. highly sensitive to them in the first place. Interestingly, the veterinary trypanosome *T. congolense*, which lacks paralogues of both *TbAT1* [104] and *TbAQP2*, is orders of magnitude less sensitive to pentamidine than the *brucei* species.

#### **4. Suramin**

While melarsoprol and pentamidine were developed in the 1930s or 1940s, suramin (**16)** is by some distance the oldest trypanocide still in routine clinical use. It was developed out of a series of trypanocidal dyes tested by Paul Ehrlich, starting with Nagana Red (**17**), which displayed only weak trypanocidal properties, followed by the more water-soluble form Trypan Red (**18**) in 1904 [105], which turned out to be both curative and prophylactic for *T. equinum* infections in mice [105–107]. As Jim Williamson [7] put it: "This was the classic first cure of an experimentally produced disease by administration of a single dose of a synthetic organic substance of known chemical composition", and it has had enormous impact on the pursuit of chemotherapy. The 7-amino derivative of Trypan Red was used in a trial in Africa by Robert Koch, but unsuccessfully [107]. Further experimentation with Nagana Red also led to the concepts of acquired drug resistance ('serum-fast') in infectious agents and the whole concept of specific targets for different drugs ('chemo-receptors') to explain cross-resistance patterns observed [7,108]. However, Nagana Red and its successors, such as Trypan Blue (**19**), all displayed unacceptable side effects at curative doses. Only Trypan Blue was effective in animal models of trypanosomiasis [109,110] and was taken into use as a veterinary drug (against babesiosis), but it stained the meat and skin blue, which did not serve to make it popular and precluded its use on human patients [107]. Further development to an active (and colourless!) trypanocide was undertaken by Maurice Nicolle and Felix Mesnil at the Institut Pasteur [111] in collaboration with by Wilhelm Röhl and Bernhard Heymann at Bayer [7], who via Afridol Violet (**20**), the first of the symmetrical ureas of the series, and after synthesis of >1000 of related structures, found 'Bayer 205' in 1916 [112]. This was introduced clinically under the name Germanin, and the formula was kept secret and supplied only to German clinicians, i.e., in German colonial territories [113,114]. As related by Dietmar Steverding [107], the German authorities offered the formula of Bayer 205 to the British Government in return for their lost colonies after World War I, but this was declined. The formula of Germanin was elucidated by Fourneau in 1924 [115,116], and promptly reissued under the name Fourneau 309 [114]. Bayer confirmed the structure 4 years later [107].

For decades now, suramin has only been used for early stage *T. b. rhodesiense* infections, with pentamidine preferred against the *gambiense* disease. There have been few reports of treatment failure with suramin, and it is generally assumed that most of these could have been related to misdiagnosis of cerebral stage HAT. Apted also proposed that in some cases suramin may simply not attain the curative concentration in blood [117], which would constitute treatment failure rather than resistance. However, Pépin and Milord, in their authoritative 1994 review [118], discuss several reports of significant relapse rates in East Africa, and suramin resistance can also be induced experimentally [119], but it is reportedly less stable than tryparsamide resistance, with experimental *T. b. rhodesiense* strains gradually regaining full sensitivity [78].

Suramin has six negative charges at physiological pH and this ensures of course that it will not cross biological membranes unless aided by an active process. One consequence is that suramin must be administered parenterally (i.v. because i.m. and s.c. causes inflammation and necrosis at the injection site [118]) and has no activity against cerebral trypanosomiasis as it is unable to cross the BBB. Another is that this large (MW = 1296 for the free acid, 1429 for the sodium salt), clumsy, un-drug-like molecule that breaks all the Lipinski rules [120], must enter the trypanosomal

cell body in order to impact its viability, via an active mechanism. The size and charge of the molecule all but precludes uptake via a nutrient transporter, channel or suchlike. It was proposed that suramin, which binds strongly to Low Density Lipoprotein (LDL) was taken up together with this serum protein by *T. brucei*, via receptor-mediated endocytosis [121,122], but later work, while consistent with uptake via endocytosis, found no correlation between LDL internalization rates and suramin sensitivity [123].

As for several other trypanocides, genome-wide screening for loss-of function with an RNAi library in bloodstream form *T. b. brucei* revealed new details of the suramin mode of action and resistance [56]. Most of the hits from this screen concerned its mechanism of uptake, confirming endocytosis, whereas it gave few clues about its mechanism of action, strengthening the view that suramin exhibits polypharmacology. Indeed, suramin has been shown to inhibit many trypanosomal enzymes including dihydrofolate reductase [124], thymidine kinase [125], all the glycolytic enzymes [126] and many others [118,127]. The RNAi screen revealed the suramin receptor to be an Invariant Surface Glycoprotein, ISG75, and also highlighted the involvement of a number of endosomal proteins, lysosomal proteases (Cathepsin L) and a lysosome-based member of the Major Facilitator Superfamily, designated MFST [56]. This has resulted in the current model of suramin uptake, via binding (either free or in complex with a serum protein) to ISG75, delivery to the lysosome by the endosomal system, degradation of the proteinous receptor and finally exit from the lysosome via MFST in to the cytoplasm [127,128].

#### **5. Eflornithine**

Eflornithine (dl-α-difluoromethylornithine; DFMO; **21**) is chemically a close analogue of the amino acid ornithine and pharmacologically a suicide inhibitor of Ornithine Decarboxylase (ODC), i.e., it binds the protein active site irreversibly, via a chemical reaction with a cysteine residue (Cys360 in mouse ODC [129]). ODC is the key enzyme in the cellular production of polyamines (spermine, spermidine, putrescine), which are essential for cell division and as such eflornithine was developed to inhibit cancer cell proliferation [130]. While this was insufficiently effective, due to the very rapid turnover/replacement rate of mammalian ODC (t1/<sup>2</sup> ~20 min), the drug is currently being investigated by cancer researchers as a chemoprevention agent [131,132].

Eflornithine started being tested, with both oral and i.v. formulations, against *gambiense* sleeping sickness from the mid-1980s, with promising results even against late-stage disease [133–136]. Because it was able to cure melarsoprol-refractory cases and patients already too frail to survive the arsenic-based treatment, it became known as the 'resurrection drug' [137]. By the early 2000s, the consensus treatment regimen was established as 100 mg/kg b.w. every 6 h for 14 days, by infusions [138]. However, the treatment was much less successful against *rhodesiense* sleeping sickness than against the *T. b. gambiense* infection [139,140]. The relative refractoriness of *T. b. rhodesiense* was also seen in a test with clinical isolates in mice [141], and Iten et al. [142] concluded that *T. b. rhodesiense* are innately tolerant to eflornithine.

Probably the most important reason that eflornithine worked better against trypanosomiasis than against cancer is that the *T. b. gambiense* ODC is very stable, with a half-life time in excess of 18 h. Thus, the irreversible inhibition of the enzyme by eflornithine ensures that the cell is deprived of polyamines, which it cannot obtain any other way, for a long time (African trypanosomes do not have polyamine transporters, as there are no polyamines in the blood). This seems also to be a main difference with *T. b. rhodesiense* (ODC t1/<sup>2</sup> ~4.3 h), although the total ODC activity in this species is also higher than in *T. b. gambiense* [141,143]. There was no difference in DFMO uptake between *T. b. rhodesiense* and *T. b. gambiense* [143]. However, several early studies showed reduced eflornithine uptake in resistant cells, which were readily produced in the laboratory [144,145].

One debate [146] with respect to the eflornithine mechanism of action was whether its uptake might be transporter-mediated [145], or by simple diffusion [143,144,147]. This debate has been definitively settled in favour of mediated uptake, as should be expected of a highly soluble, zwitterionic compound with an experimental LogP of −2.7. Vincent et al. [148] induced eflornithine resistance in *T. b. brucei* bloodstream forms and found no mutations in *ODC* but again saw a strongly diminished rate of

eflornithine uptake. However, a systematic amplification and sequencing of amino acid transporter genes identified deletions of the gene encoding amino acid transporter 6 (*TbAAT6*) in two independently generated resistant lines. Specific ablation of this transporter by RNAi in a sensitive line resulted in eflornithine resistance, whereas the (re)-introduction of a wild-type *TbAAT6* allele into a resistant strain restored sensitivity [148]. This paper was almost immediately followed by the publication of two other, independent studies using RNAi library screens, which each also identified *TbAAT6* as the main determinant of eflornithine resistance [56,149].

While clinical reports about eflornithine resistance in the field are scarce, the fact remains that it is easily induced in the laboratory and that a single point mutation in a non-essential gene (*TbAAT6*) will cause a high level of resistance. Considering that the dosage regimen of eflornithine monotherapy is already severe as well as expensive, and cannot easily be increased in amount or duration, this placed serious question marks over the longevity of the drug, and was a major driver, along with both the cost, duration and logistics of administration, and the adverse effects of 14-day i.v. eflornithine [150], in the development of combination therapies. This included a trial, reported in 2002, of a short treatment with eflornithine followed by three daily injections with melarsoprol, giving a probability of cure of 93% [151]. In 2006, a trial with 54 patients was reported, testing three combinations: melarsoprol-nifurtimox, melarsoprol-eflornithine and eflornithine-nifurtimox [152]. The trial was halted because of the toxicity of the melarsoprol-plus-nifurtimox combination (which also gave only a 44% cure rate), but the eflornithine-nifurtimox performed significantly better than the eflornithine-melarsoprol combination (94.1 versus 78.4%; *P* < 0.05) and resulted in fewer adverse effects. In 2007, two studies, in Uganda [153] and the Republic of Congo [154], described further trials with eflornithine-nifurtimox combination therapy (NECT), which were subsequently extended to a full multi-centre non-inferiority trial [155]. The overall conclusions were that the combination is non-inferior to eflornithine monotherapy and has considerable advantages such as protection against resistance, lower cost, easier and shorter administration as well as a reduction in adverse effects by 50% [155,156].

NECT was added to the WHO Essential Medicines list in April 2009 and was adopted as the new first-line treatment for late stage *gambiense* sleeping sickness [156]. At this point, the phasing-out of melarsoprol was considered to be one of the main remaining challenges, as it was still used for ~50% of cases in the Democratic Republic of Congo (DRC), for instance [156]. In 2012, a pharmacovigilance evaluation of 1735 patient outcomes from 9 different countries found that although adverse effects were quite common (60.1%) serious adverse effects were rarely observed (1.1%) and the case fatality rate was 0.5% [157]. This of course compared very well with melarsoprol, the use of which by then had dropped to 12% of all second stage *gambiense* HAT cases [150]. Two further clinical reports, describing NECT outcomes in 684 second stage patients in the DRC [158] and 109 patients in Uganda [159] provided a further evidence base for the selection of NECT as the first-line treatment for cerebral stage infection with *T. b. gambiense*. Based on the low efficacy of eflornithine against *T. b. rhodesiense*, no clinical trials with either eflornithine monotherapy or NECT were initiated against late stage *rhodesiense* sleeping sickness, and to date melarsoprol remains the only approved treatment for that condition.

#### **6. Nifurtimox**

Nifurtimox (**22**) has been used since 1969 against *Trypanosoma cruzi* (i.e., Chagas Disease) [160,161] and, given the urgent need to replace melarsoprol for late-stage HAT, has been investigated as a possible treatment for African trypanosomiasis as well [69]. As related by Janssens and De Muynck [162], the first tests with nifurtimox against African trypanosomiasis were conducted by Marc Wéry who found, apparently to everybody's great surprise, 'a definite activity on the chronic infection' in rats, justifying a first trial in humans. That first trial, of just four patients, used 3×120 mg or 3×150 mg daily for 60 or 120 days or however long the drug was tolerated. The trial results were mixed, and it was concluded that nifurtimox was not the sought-after, reliable and low-toxicity replacement of melarsoprol that was hoped for, but that it might serve as a drug of last resort for melarsoprol-refractory patients that were thus untreatable at the time [162]. A subsequent trial in Zaire with 12.5–15.0 mg/kg

b.w. daily (in three doses) for 2 months reported cures of 7-out-of-8 melarsoprol-refractory cases and 5-out-of-7 new late-stage cases [163]. In contrast, using essentially the same treatment schedule, Pépin and co-workers reported a relapse rate of 63% in 1989 and concluded that a higher dose would be necessary for cure in most patients [164]. However, after a trial with 30 mg/kg b.w./day for 30 days they concluded that this regimen was 'significantly toxic' and that the (only slight) improvement in efficacy did not outweigh the increase in toxicity [165]. Their overall conclusion was that nifurtimox is inferior to eflornithine, then becoming available, as a mono-therapy treatment for arseno-tolerant HAT. Van Nieuwenhove, summarizing the emerging evidence on eflornithine and nifurtimox monotherapy in 1992, strongly advocated trials of combinations of the three available treatments of late-stage HAT [166]. While a treatment regimen starting with melarsoprol alone 2 doses) followed by 10 days of nifurtimox-plus-melarsoprol gave superior cure rates to melarsoprol or nifurtimox alone [167], the combination of eflornithine with nifurtimox (NECT) was eventually adopted, as related in the previous section.

With the trialling and implementation of NECT, the question of nifurtimox' mode of action and potential resistance mechanism became one of acute importance. It had long been known that nitro-heterocyclic trypanocides can generate free radicals such as superoxide [168], although direct evidence of this being the principal mechanism of action for nifurtimox and related nitro compounds was lacking. However, the deletion of a gene encoding a glycosomal and cytosolic superoxide dismutase, *TbSODB1*, significantly increased sensitivity to nifurtimox and benznidazole, which was restored to wild-type levels upon re-introduction of the gene; this was specific to *TbSODB1* as deletion of the related glycosomal *TbSODB2* had no effect [169]. Further confirmation of the mechanism of action came with the identification of a mitochondrial NADPH-dependent nitroreductase, NTR, that proved to be essential for the efficacy of a nitro-heterocyclic drug in *T. brucei* and *T. cruzi* [170]. Both species have one copy of this type 1 nitroreductase in their genome. A *T. cruzi* cell line induced for nifurtimox resistance was highly cross-resistant to a variety of nitro-heterocyclic compounds and purified NTR was shown to efficiently reduce all of them. Worryingly, deletion of just one *TbNTR* allele created a heterozygous *NTR*+/- *T. brucei* that displayed no growth phenotype but was three-fold resistance to the nitro-heterocyclic drugs; the double deletion (*ntr*-null) *T. brucei* cell line was even more resistant but also displayed significant growth impairment, leading to the conclusion that the gene is essential [170]. The same gene was subsequently identified in genome-wide RNAi library screens with nifurtimox and benznidazole [171]. As overexpression of *NTR1* resulted in hypersensitivity to the nitro drugs [170] it is clear that *NTR1* is the main determinant of nitro-heterocyclic sensitivity in trypanosomes, and that a single mutation in one allele could eliminate their small therapeutic window, as the main biochemical difference with host cells, as pertains to nifurtimox' mode of action, is cancelled.

This conclusion was put to the test by Alan Fairlamb, who induced two nifurtimox-resistant *T. brucei* lines in vitro, which both showed cross-resistance to other nitro-heterocyclic drugs including fexinidazole [172]. The resistant strain displayed unimpeded virulence in mice and, most worryingly, nifurtimox had little effect on the in vivo progression of the infection with this strain [172].

The accumulation of nifurtimox across the BBB was investigated by Sarah Thomas using both a murine perfusion system and an in vitro model based on immortalised human BBB cells [173,174]. Nifurtimox readily crossed the BBB and blood–CSF barriers and this was not significantly different in healthy mice or infected mice where the barrier integrity had been compromised [173]. This almost certainly means that the uptake of nifurtimox at the BBB is trans-cellular rather than paracellular (i.e., between cells). Nor was there any difference between the standard model and mice deficient in the BBB efflux transporter P-gp. This was consistent with an earlier study in rats, which found that 35S-nifurtimox is rapidly distributed throughout the host, and that both the blood-brain or placental barriers were permeable to the drug [175]. This could be the result of its lipophilicity (octanol-saline partition coefficient=5.46 [173]), which might allow a simple diffusion across membranes. Certainly, no transporter for nifurtimox uptake has, to date, been identified, either in the host or in trypanosomes, including with RNAi library screens [171]. Yet, nifurtimox, bearing a polar nitro

group, is not *very* lipophilic, the trans-cellular uptake at the BBB is consistent with a transporter, and it has been argued that almost all drugs require a transport mechanism [176] and that absence of proof for the involvement of a transporter simply means we haven't looked well enough [177]. Moreover, Jeganathan et al. reported a higher concentration of 3H-nifurtimox in the CNS compared with plasma [173], which would be hard to explain without active import, especially since there is also a component of extrusion across the barrier (see below). Of potential importance for NECT, eflornithine reduced CNS accumulation of 3H-nifurtimox by > 50%, presumably by interfering at the level of a transporter at the barrier [173], although that result could not be reproduced in the later study with cultured human BBB cells [174]. If nifurtimox uptake is indeed transporter-mediated, the failure to identify transporter genes with the genome-wide RNAi library could reflect uptake by multiple transporters (non-dependence on a single gene) or the gene being essential (knockdown being lethal).

Although the existence of a plasma membrane transporter importing nifurtimox has not (yet) been established, Thomas's studies with the BBB did provide evidence for extrusion across the BBB, despite the non-involvement of P-gp [173,174]. For instance, coadministration with pentamidine enhanced CNS accumulation in the mouse perfusion system, likely indicating an interaction with the 3H-nifurtimox efflux transporter at the barrier [173], as also observed with 3H-pentamidine distribution [37]. In the follow-on study using immortalised human BBB cells, it was found that accumulation across the barrier was strongly increased by inhibitors of the breast cancer resistance protein (BCRP) but not of P-gp; BCRP is an ATP-dependent efflux transporter of the ABC super family and depletion of ATP in the cell line had the same effect as inhibitors of BCRF [174].

Nifurtimox appears to be similarly effective against *T. b. rhodesiense* and *T. b. gambiense*, at least in vitro [178,179]. However, if nifurtimox has been tested clinically on *rhodesiense* HAT, I have been unable to find any reference to it. Certainly, no systematic trials have been held and, considering that NECT is unlikely to work on that infection because of the eflornithine component (see above), the effort would seem almost redundant, and the ethics might be debateable. As such, melarsoprol is still the only approved option for late-stage *rhodesiense* sleeping sickness, although the Drugs for Neglected Diseases initiative (DNDi) started a programme in September 2018 to develop fexinidazole, the newly approved oral drug for late stage *gambiense* disease [180] that does not require co-treatment with eflornithine [181], for the equivalent condition with *T. b. rhodesiense* (https://www.dndi.org/diseases-projects/portfolio/fexinidazole-tb-rhodesiense/).

As regards fexinidazole, the EMA issued a positive opinion in November of 2018, based on the DNDiFEX004-6 trials, clearing the way for its implementation [182]. However, fexinidazole monotherapy was somewhat less effective than NECT (91.2% vs. 97.6% cure) [181], within the predetermined acceptability margin. However, as pointed out by François Chappuis, this is compensated for by the ease of administration and approved access to medicine that this oral formulation brings to the HAT treatment options [181,183]. Moreover, fexinidazole, was reportedly > 98% curative for early and early-late stage *gambiense* HAT (trial NCT03025789) and can thus be used without having to perform the invasive lumber punctures still required for determination of HAT stage [183]. The most important limitation of fexinidazole, however, seems to be a relatively low curative rate of 86.9% for the patients with the most severe meningoencephalitic HAT (defined as > 100 white blood cells/ml CSF) [1,182]. Thus, for these patients NECT remains the best option [182,184].

#### **7. A Perspective on Drug Resistance in African Trypanosomiasis**

It is a dogma in the pharmacology of infectious diseases that resistance will (eventually) occur for any drug. Certainly, the infectious agents have many tricks up their proverbial sleeves, often not anticipated [185]. Yet, there is no proof of suramin resistance in African trypanosomes, despite approximately 100 years of usage in East Africa. Additionally, there is no clear proof of pentamidine resistance either, despite intensive use against *gambiense* HAT since the 1940s, including highly successful mass treatments, particularly by the French and Belgian colonial authorities [33]. This even though resistance to either drug can be induced in the laboratory without

apparent loss of viability. Thus, there has been relatively little incentive to develop new treatments for the early stage disease and the target product profile of the DNDi has, since its inception, been for late-stage disease. While there was a significant element in this of the need for new treatments with reduced toxicity and cost, leading to optimisation of use of melarsoprol with a shorter 10-day protocol, for instance [83,186], most of the new drug development was driven by resistance to the then-standard treatment. It is questionable whether melarsoprol would have ever been taken into clinical use if not for the catastrophically high levels of resistance with tryparsamide. Similarly, a major factor in the introduction of eflornithine, in addition to melarsoprol's dangerous toxicity, was the rapid rise in cases refractory to melarsoprol in Central Africa. The well-founded fear of resistance to eflornithine monotherapy (and the anecdotal reports of increasing failures with the drug) helped drive urgent trials with eflornithine and nifurtimox combinations. Every time, we were on our last or only drug against late stage sleeping sickness and the introduction of replacements was sheer necessity.

So, where are we now, in 2020? The number of patients, particularly with *gambiense* HAT, is in steep decline, human-to-human transmission is low, possibly at an all-time low, and we have in hand first-stage treatments that have stood the no-resistance test of time (pentamidine, suramin), and a combination therapy for late stage that is safer than we ever had. The introduction of fexinidazole [181–184] and potentially acoziborole [2,3], safe oral drugs that are active against both stages of the disease, is finally eliminating the need for risky lumbar punctures for determining the disease stage. The fact that there are finally multiple treatment options would allow clinicians to rotate treatments, should the need arise, but as long as continued vigilance keeps transmission very low, resistance is much less likely to develop. Does this mean sleeping sickness is 'done'? We have thought this once before, when cases were few, in the 1960s, and have hopefully learned that we must continue the vigilance, awareness and training of medical personnel. More work is also still needed on early detection/diagnosis, detection of asymptomatic cases, animal reservoirs. As far as resistance is concerned, the cross-resistance between fexinidazole and nifurtimox is of potential concern as any strains surviving fexinidazole monotherapy would likely also cause failure with NECT. Thus, the completion of the clinical development and registration of acoziborole and/or melarsoprol-cyclodextrin complex is still of great importance, as is the trial of fexinidazole for late stage *rhodesiense* HAT. Currently Rhodesian HAT is the most neglected disease, still treated with suramin from ~1920 and the awful i.v. melarsoprol in propylene glycol for the late stage. Hopefully one of these new options will finally put suramin to rest and ease this esteemed great-grandfather of chemotherapy into a well-deserved retirement—and i.v. melarsoprol as well.

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References**


report/comp-meeting-report-review-applications-orphan-designation-october-2012\_en.pdf (accessed on 17 January 2020).


© 2020 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **New Drugs for Human African Trypanosomiasis: A Twenty First Century Success Story**

#### **Emily A. Dickie 1, Federica Giordani 1, Matthew K. Gould 1, Pascal Mäser 2, Christian Burri 2,3, Jeremy C. Mottram 4, Srinivasa P. S. Rao <sup>5</sup> and Michael P. Barrett 1,\***


Received: 16 January 2020; Accepted: 14 February 2020; Published: 19 February 2020

**Abstract:** The twentieth century ended with human African trypanosomiasis (HAT) epidemics raging across many parts of Africa. Resistance to existing drugs was emerging, and many programs aiming to contain the disease had ground to a halt, given previous success against HAT and the competing priorities associated with other medical crises ravaging the continent. A series of dedicated interventions and the introduction of innovative routes to develop drugs, involving Product Development Partnerships, has led to a dramatic turnaround in the fight against HAT caused by *Trypanosoma brucei gambiense*. The World Health Organization have been able to optimize the use of existing tools to monitor and intervene in the disease. A promising new oral medication for stage 1 HAT, pafuramidine maleate, ultimately failed due to unforeseen toxicity issues. However, the clinical trials for this compound demonstrated the possibility of conducting such trials in the resource-poor settings of rural Africa. The Drugs for Neglected Disease initiative (DNDi), founded in 2003, has developed the first all oral therapy for both stage 1 and stage 2 HAT in fexinidazole. DNDi has also brought forward another oral therapy, acoziborole, potentially capable of curing both stage 1 and stage 2 disease in a single dosing. In this review article, we describe the remarkable successes in combating HAT through the twenty first century, bringing the prospect of the elimination of this disease into sight.

**Keywords:** human African trypanosomiasis; sleeping sickness; elimination; chemotherapy; fexinidazole; pafuramidine; acoziborole

#### **1. Introduction**

The current drugs used for human African trypanosomiasis (HAT) (Figure 1) have served their purpose for many years. The incidence of HAT is now at a historic low (fewer than 1000 cases reported in 2018 [1]). Two forms of the disease occur. The one found in West and Central Africa is caused by *Trypanosoma brucei gambiense*, and the other, found in East and Southern Africa, is caused by *Trypanosoma brucei rhodesiense*. The former causes a chronic disease, taking years between infection and death, while the latter may kill within weeks to months. Parasites injected into the bloodstream cause a stage 1 infection, where replication is primarily associated with blood and lymph. However, the parasites then invade other organs, including the central nervous system (CNS). Once replicating in the CNS, the disease progresses to stage 2, where many of the symptoms of sleeping sickness become manifest. The current drugs suffer many drawbacks [2]. For stage 1 disease, either suramin or pentamidine is used for HAT caused by *T. b. rhodesiense* and *T. b. gambiense*, respectively. Both drugs must be given by injection for a prolonged period, and both carry a risk of adverse events. For stage 2 disease, the highly toxic melarsoprol is still the treatment of choice for rhodesiense HAT. Melarsoprol causes an encephalopathic syndrome that is fatal in up to one in twenty people taking the drug [3]. For gambiense HAT, the past decade has seen the introduction of a combination therapy. Intravenous eflornithine is given for 10 days alongside oral nifurtimox for 14 days [4].

**Figure 1.** Different drugs have been used to treat HAT depending on the trypanosome subspecies causing the disease, and whether progression is at stage 1 or 2. For the past decade nifurtimox and eflornithine combination therapy has been the treatment of choice for stage 2 *T. b. gambiense* disease, and pentamidine for stage 1. For rhodesiense HAT, stage 2 is treated with melarsoprol and stage 1 with suramin.

The combination is relatively safe and efficacious. However, the delivery of kilogram quantities of eflornithine and many liters of sterile saline brings substantial logistical difficulty. For melarsoprol, resistance was a growing problem in the early 2000s [5], and for eflornithine [6] and nifurtimox [7] independently, resistance can be readily selected in the laboratory. There is no doubt that better drugs for use against HAT are required [8].

The first two decades of the twenty first century can be seen as a major success story in regards to intervention against this neglected tropical disease. HAT was running out of control in the late twentieth century, with an estimated 300,000 people infected [9]. In response, the international community launched several key initiatives, which can be seen as having converged to turn the tide. Consequently, the twenty first century has witnessed a dramatic change in the trajectory of HAT.

In 1999, Médecins Sans Frontières (MSF) initiated their "Access" campaign to encourage a re-engagement of the pharmaceutical industry with neglected diseases, including HAT [10]. A key study [11] revisited the administration regimen of melarsoprol based on pharmacokinetic data. An effective 10-day administration protocol, which diminished hospitalization time for patients [12], albeit without increasing safety, was introduced to great effect [13]. This IMPAMEL program ("improved application of melarsoprol") was crucial in implementing the first modern clinical trials on HAT, and also in demonstrating the feasibility of conducting trials in extremely resource-limited conditions according to Good Clinical Practice, laying the foundation for future developments.

Eflornithine had been shown to be far safer than melarsoprol as treatment for stage 2 disease [14]. Yet, by the late 1990s, no pharmaceutical company was prepared to make this compound for HAT treatment. However, when it was discovered that the same compound could prevent the growth of unwanted facial hair in women, a number of drug companies saw an opportunity to market the compound for this purpose [15]. MSF, already campaigning for access to essential medicines, were able to make a compelling case that society needed to rethink drug discovery paradigms for neglected diseases [16]. Aventis (now Sanofi) were persuaded to develop the drug and donate it at no cost to the World Health Organization (WHO) for distribution in Africa. Millions of dollars were also provided by Aventis/Sanofi to WHO, who could now develop new screening and intervention programs. The Bill and Melinda Gates Foundation selected HAT to be one of the first diseases they targeted through the Consortium of Parasitic Drug Development (CPDD) [17], and the Drugs for Neglected Diseases initiative (DNDi) was founded through MSF [18] to seek new drugs for diseases including HAT. In diagnostics, the Foundation for Innovative New Diagnostics (FIND) sought novel ways of improving our ability to detect HAT patients [19], and new ways of combating the tsetse fly were rolled out too [20]. A number of pharmaceutical companies also regained an interest in HAT, including GlaxoSmithKline (GSK) through their Tres Cantos Open Lab foundation [21] and the Novartis Institute for Tropical Diseases (NITD) [22]. Small companies too, such as Immtech Pharmaceuticals Inc., Scynexis and Anacor Pharmaceuticals Inc. in the USA, also raised investment to develop new drugs against HAT. The first twenty years of the twenty first century have now seen the clinical trials and ultimate failure of a new orally available diamidine prodrug [23], the registration of the first all-orally available therapy against stage 2 disease [24], and the entry into clinical trials of a compound that may cure stage 2 HAT with a single oral administration [25]. This article outlines the successes seen in the development of new drugs for HAT in the twenty first century.

#### **2. Pafuramidine—A New Paradigm in Anti-Trypanosomal Drug Development**

Among the currently used drugs for HAT is pentamidine, a diamidine that was introduced in the 1940s, and has been the mainstay in the treatment of stage 1 gambiense HAT for nearly 80 years [26]. Another diamidine, diminazene, is used in treating veterinary trypanosomiasis [27]. The diamidines are di-cations, with positive charges at either end, which renders them highly polar, precluding bioavailability if taken orally [28]. Pentamidine is typically given by intramuscular injection for seven days. Das and Boykin showed in the 1970s that methoxy-derivatives of other diamidines acted as orally available prodrugs, and the methoxy group metabolized back to the amidine systemically [29].

The Bill and Melinda Gates Foundation was founded in 2000, a time when the resurgence of HAT was at its pinnacle. Among the Foundation's earliest supported projects was the development of the diamidine methoxy-prodrug approach towards new drugs for HAT through the CPDD.

DB289 (Figure 2) (pafuramidine maleate; 2,5-bis(4-amidinophenyl)-furan-bis-*O*-methylamidoxime), the methoxy product of furamidine (2,5-bis(4-amidinophenyl)-furan), in which the benzamidine moieties are separated by a furan ring, emerged as the lead compound for progression.

Furamidine is highly polar and unable to traverse lipid bilayers without the assistance of transporters (see later). Pafuramidine, however, has much greater capacity to diffuse across membranes [30], including the intestinal epithelium, giving it considerable oral bioavailability. Once systemic, it is metabolized via various cytochrome P450 enzymes and cytochrome b5 reductase [31–36] to furamidine. Preclinical safety and efficacy results [37,38] were sufficient to enable the passage to clinical trials.

**Figure 2.** Clinical trials with pafuramidine (DB289), which is the prodrug of furamidine (DB75), failed due to the appearance of renal toxicity during extended phase I safety profiling. The aza-derivatives including DB868—a prodrug of DB829, and DB844—a prodrug of DB820, also showed activity against stage 2 disease, but development was halted following identification of the toxicity associated with pafuramidine. Fexinidazole has been approved for use by the European Medicines Agency in 2018. The compound is converted to sulfoxide then sulfone derivatives after administration. Acoziborole is a benzoxaborole in clinical trials, where the efficacy of a single dosing of the drug as an oral medication against stage 2 disease is being assessed.

#### *2.1. Clinical Trials of Pafuramidine*

Ultimately pafuramidine failed in clinical trials [23,39]. However, those trials were of great importance, not only in testing the efficacy of this promising lead compound, but also in informing on any criteria required to conduct clinical trials for regulatory purposes in a cohort of patients in difficult places to work. These trials also helped to establish protocols for patient seeking, screening, drug administration, record keeping and patient follow-up post-treatment. The very fact that over 350,000 individuals were screened in the quest for patients to include in the trials had a major impact on the incidence of the disease. Patients testing positive but failing inclusion criteria (i.e., not clearly in stage 1 disease), were treated with other drugs appropriate to their stage.

An original phase I safety study (unpublished), carried out in healthy Caucasian volunteers in Germany in 2001 showed the drug to be well tolerated in single and multi-dose testing (100 mg twice per day ( b.i.d.)) and up to 600 mg in a single dose.

A phase IIa study [39] in Viana, Angola and the CDTC Maluku, Democratic Republic of the Congo (DRC) with 32 patients showed that 100 mg of pafuramidine orally twice a day for 5 days yielded loss of all visible trypanosomes in blood 24 h after treatment cessation in 93% of cases, compared to 100% efficacy with 7-day pentamidine treatment. However, prolonged follow up began to pick up enhanced relapse rates from this five-day dosing (by 24 months, only 67% of pafuramidine-treated patients were considered cured). 81 patients had entered a phase IIb trial, at the same 5-day dosing (40 for pafuramidine and 41 for pentamidine) [39] by this time. However, the high relapse rate after prolonged follow up in the phase IIa-1 trial prompted a decision to use a 10-day dosing of 100 mg pafuramidine b.i.d. instead, and 30 more patients were enrolled for this phase IIa-2 trial. The dose was well tolerated over 10 days, and was less toxic than pentamidine [39]. The 10-day dosing of pafuramidine gave 93% cure at 3 months follow up, and was selected for a Phase III trial.

The Phase III trial [23], conducted in several centers in the Democratic Republic of the Congo, Angola, and South Sudan recorded a cure rate of 89% at 12 months follow up. The major safety concern noted in these treatment trials was a 7% incidence of increased liver enzymes; however, the incidence of this adverse event was substantially less than in the comparator group treated with pentamidine (77%). A significantly lower percentage (2% vs. 9%) of patients treated with pafuramidine experienced treatment emergent adverse events detected by renal and urinary tract investigations and urinalyses compared to those under pentamidine. Overall, three subjects in the pafuramidine treatment groups were reported to have serious adverse events that could be considered acute renal failure or insufficiency classified as possibly associated with study drugs [23].

As the phase II/III trial was progressing, a further phase I safety trial was conducted in South Africa, necessitated by the increase in the duration of drug administration and the limited number of patients exposed to pafuramidine (unpublished). One hundred healthy sub-Saharan African adult volunteers were treated with 100 mg pafuramidine b.i.d. for 14 days. Generally, a mild and reversible increase of liver enzymes was observed in multiple subjects, which had been anticipated [23]. However, 6 of 100 subjects then experienced delayed renal insufficiency with unclear etiology.

The delayed action might indicate a possible immunological role; HAT is itself immunosuppressive, which could explain why the incidence of kidney injury in treated patients was much lower. Another hypothesis was a genetic origin across a diverse group of individuals, which was supported by a retrospective study on a panel of 34 genetically distinct mouse strains [40]. Urinary secretion of the kidney injury molecule-1 (KIM-1) protein was used as a marker of proximal tubule injury in the kidney. Only a subset of the mice revealed elevated KIM-1. Genetic association studies then showed several genes with alleles associated with the KIM-1 levels. This included PCSK5, a serine peptidase associated with lipid and cholesterol metabolism, and another cholesterol-associated enzyme, sterol O-acyltransferase, among others. Definitive explanations for the kidney injury mechanism, however, remain obscure. DNA from the subjects involved in the clinical trial was not available to determine whether the same genes were associated to kidney damage in humans.

The appearance of unanticipated toxicity with pafuramidine led to the end of clinical development and also a cessation of other activities around the diamidine project. This was unfortunate, since the project had also identified several aza-analogs, including DB820 and DB829 (alongside their respective methoxyamidine prodrugs DB844 and DB868,), which were effective in mouse [41] and vervet monkey (*Chlorocebus pygerythrus*) models of stage 2 infection [42]. Significantly, DB829 accumulated far less (>10 fold) inside mammalian cells than furamidine [43], pointing to a potentially improved safety profile. For HAT, however, the successes that have occurred with other compounds (see below) indicate that the diamidines are unlikely to be resurrected for future work. However, because of their diverse

and intriguing biological activities [44], it is likely to see them re-emerging for other conditions in the future [45].

#### *2.2. Pafuramidine: Mode of Action and Resistance Risk*

Furamidine, like other diamidines, binds to the minor groove of the DNA double helix [28,45]. The ability to bind is structure-dependent, and intriguing work has aimed to tailor this binding to specific DNA sequences, including regulatory elements [44], with a view to controlling gene expression in, for example, human cancer cells. Furamidine is fluorescent, and its accumulation in trypanosomes shows binding to both the kinetoplast (mitochondrial DNA) and nuclear DNA, as well as its accumulation in vesicles assumed to be acidocalcisomes [46]. The compound accumulates to millimolar concentration, even in cells exposed to low micromolar concentrations over 24 hours [46]. Even a brief exposure (<5 min) to 32 micromolar of the compound, followed by wash out, provoked death 48 hours later [47]. At 3.2 micromolar, exposure time rose to 1 hour to achieve this slow commitment to death. Lanteri also showed a profound decrease in the mitochondrial membrane potential associated with the addition of furamidine to *T. brucei* [48].

The F1Fo ATP synthase complex is an essential mitochondrial complex in trypanosomes. In procyclic forms it acts in a classical fashion, producing ATP as part of the electron transport chain. In bloodstream forms, however, it acts in reverse: consuming ATP in order to maintain the essential mitochondrial membrane potential. All of the F1Fo ATPase subunits are encoded in the nucleus, apart from subunit A6, which is encoded in the kinetoplast. Mutations to the nuclearly-encoded gamma subunit (e.g., L262P substitution or an alanine deletion at amino acid position 281) allow the mitochondrial membrane potential to be generated in bloodstream form trypanosomes without subunit A6, rendering the kinetoplast redundant in this life cycle stage. The kinetoplast can be removed from these cells without apparent impact on viability [49]. The ATPase gamma mutants (now kinetoplast independent) are hundreds of fold resistant to phenanthridine compounds; intriguingly however, only very minor resistance (~ 3-fold) was shown for furamidine [50]. This would indicate that, in contrast to phenanthridines, it is not the kinetoplast per se that is the target of furamidine.

Diamidines also act against yeast mitochondria [51] with *Saccharomyces cerevisiae*that are fermenting glucose, rather than respiring using mitochondrial substrates (e.g., glycerol), being 200-fold less sensitive to pentamidine [52].

Mammalian cells too are vulnerable to diamidines. However, the hypersensitivity of trypanosomes to this class is due to their ability to accumulate diamidines to very high concentrations across their plasma membrane through specific transporters. The P2 aminopurine transporter, *Tb*AT1, which normally carries adenosine and adenine, was the first transporter characterized to carry pentamidine [53]. Subsequently, a high affinity pentamidine transporter [54], later revealed as the aquaglyceroporin *Tb*AQP2 [55], was shown to play a dominant role in uptake [56]. A low affinity pentamidine transporter, whose physiological role remains elusive, was also shown to transport pentamidine [54]. Loss of *Tb*AT1 yields only low-level resistance to pentamidine, while loss of *Tb*AQP2 yields high level resistance [56]. Pentamidine, however, is an exception among the trypanocidal diamidines, having a highly flexible central linker chain. The shorter derivatives (e.g., diminazene, and furamidine plus its aza-analogs), primarily use the *Tb*AT1 aminopurine transporter for uptake [57], and its loss gives high level resistance. Discovering a specific motif found on diamidines, aminopurines and also melaminophenyl arsenicals, all of which can enter via the *Tb*AT1 P2 transporter [58], led to several efforts to create new selective trypanocides to enter via the same carrier protein [59–61].

Diamidines can enter mammalian cells and also cross the blood–brain barrier (BBB) [62]. The human facilitative organic cation transporter 1 (OCT1) is one transporter capable of transporting pentamidine and other diamidines in mammalian cells [63,64]. ATP-dependent pumps (e.g., the P-glycoprotein, P-gp) appear to also be able to efflux pentamidine [62], but not furamidine, nor DB829 [65].

#### **3. Fexinidazole: the First Oral Treatment for HAT**

In November 2018, The European Medicines Agency's Committee for Medicinal Products for Human Use (CHMP) offered a positive opinion on the use of oral fexinidazole for the treatment of both stage 1 stage 1 and 2 gambiense HAT [66]. In December 2018 the DRC, the epicenter of the disease, issued marketing authorization allowing use of the drug. Fexinidazole is given once per day for ten days, involving a four-day loading dose of 1.8 g per day followed by six days at 1.2 g per day with 600 mg tablets. The clinical development of fexinidazole is described in an accompanying article in this volume [67].

#### *3.1. Pre-Clinical Development of Fexinidazole*

The pathway for fexinidazole to the clinic was both long and disrupted. The drug, a 2-substituted 5-nitroimidazole, was originally synthesized as part of a program seeking anti-infectives by Hoechst in the 1970s [68]. Frank Jennings and George Urquhart at the University of Glasgow extended trypanocidal testing in the 1980s [69]. They could not demonstrate a full cure against the stage 2 disease model in mice which they had developed using the *T. brucei brucei* GVR35 strain which develops a slowly progressing disease in which the parasite enters the brain and establishes a CNS infection, prior to having killed the mice in an acute infection, as most laboratory strains of *T. brucei brucei* do. Administering fexinidazole as a monotherapy (given once a day for four days at 250 mg/kg), it cured 11/14 mice [69], although sequential treatment with a single dose of suramin (20 mg/kg) or diminazene (40 mg/kg), followed by four consecutive daily doses of fexinidazole, was curative. In the case of the suramin followed by a fexinidazole regime, a single dose of suramin, followed by four daily doses of just 30 mg/kg fexinidazole, was curative [69]. It was later shown [70] that five days of fexinidazole monotherapy at 200 mg/kg was curative (7/8 mice) in the stage 2 model, showing the necessity for prolonged exposure.

The finding that many nitroheterocycles were genotoxic and potentially carcinogenic led to a general aversion to this class of molecule for its use as pharmaceuticals for several decades. However, nifurtimox, another nitroheterocycle, entered the anti-trypanosomal armamentarium as a combination partner with eflornithine [4]. The nitroheterocycle known as megazol also showed interesting trypanocidal activity [71], whilst other nitroheterocycles such as PA-824 (now registered as pretomanid for tuberculosis [72]) came to the fore.

Thus, interest in trypanocidal nitroheterocycles was rekindled. A series of 830 nitroimidazoles and related compounds were tested at the Swiss Tropical and Public Health Institute against *T. brucei*. *Trypanosoma cruzi*, *Leishmania donovani*, and mammalian cells *in vitro*. The most active and selective molecules were evaluated in secondary *in vitro* assays and in the mouse models of acute or chronic trypanosomiasis. Based on these data, fexinidazole was singled out as the most promising candidate for further development against HAT [70,73], and a dossier of information was compiled to support the safety and efficacy of the drug [67,74].

#### *3.2. Fexinidazole: Mode of Action and Resistance Risk*

Fexinidazole is a prodrug. Its activity depends upon two consecutive electron reductions of the NO2 group by an NADH-specific nitroreductase (*Tb*NTR1) [75]. Diminished NTR activity leads to resistance to fexinidazole [76] and cross-resistance to other nitroheterocycles, including nifurtimox. The fate of the nitro-reduced product is unknown. An orthologous nitroreductase enzyme is also found in *T. cruzi* [75], where it is responsible for the activation of clinically used nitroheterocycles (e.g., benznidazole and nifurtimox). In the case of nifurtimox, *T. cruzi* was shown to reduce the compound to a highly reactive species [77]. Fexinidazole is also under consideration as a new treatment for Chagas disease [78]. For benznidazole, *in vitro* activation also revealed reductive activation and ultimately disintegration of the compound to glyoxal [79]. Another study, however, failed to demonstrate the production of glyoxal in *T. cruzi*, although it revealed a plethora of metabolized

products and adducts of these benznidazole breakdown-products [80]. This led to suggestions that, following its activation by *Tc*NTR1, benznidazole ultimately kills through the modification of numerous metabolite and protein targets in *T. cruzi* [80]. The metabolic fate of fexinidazole, following NTR1 mediated reduction, has not been determined, and the mechanism by which it kills trypanosomes is not known. However, a similar "cluster bomb" effect, with the activated fexinidazole product of the nitroreduction hitting multi-targets, seems likely. NTR1 has been proposed to act physiologically as an NADH dehydrogenase involved in reducing ubiquinone to ubiquinol [81], hence its essentiality even in bloodstream form *T. brucei* that requires ubiquinone-based electron transport in its mitochondrial alternative oxidase system [82]. In nifurtimox-treated *T. brucei*, loss of mitochondrial membrane potential and other mitochondrial morphology defects were noted [83]. Other nitroheterocycles, including fexinidazole, may also exert their effects on the mitochondrion, where the drug's activation occurs. Knockout of a single copy of the *TbNTR1* gene (trypanosomes are diploid, hence have two copies of the gene, one on each copy of chromosome 7), led to diminished activity of the drug, albeit by just 1.6−1.9 fold, whilst double knockout was not possible [76]. Parasites resistant to fexinidazole could be selected *in vitro* under the conditions of increasing drug pressure (up to 20-fold resistance), and these parasites lost the 3' flanking region of one allele of NTR, leading to a 50% reduction in the expression of the gene. The same study selected nifurtimox-resistant parasites, which lost one copy of the gene in developing 6-fold resistance. In both cases, resistance was significantly higher compared to the single NTR1 KO cells, indicating that other, as yet unknown, events beyond partial NTR1 loss, contribute to resistance. A question arises as to whether there is a risk that the current use of nifurtimox could lead to the selection of resistance to fexinidazole and vice versa. However, although diminished nitroreductase activity can yield cross resistance, it is not clear whether parasites, even with diminished nitroreductase, would be viable in the field, if the enzyme were critical in the insect transmitted forms of the parasites. Moreover, with very few cases of gambiense HAT currently reported (fewer than 1000 in 2018), the probability of selecting parasites resistant to either drug under field-settings is currently very low.

NTR1 is a typical enzyme of prokaryote origin. *Salmonella typhimurium*, the bacterium used in the Ames test, possesses nitroreductase genes as well, explaining the positive mutagenicity signal of fexinidazole and its metabolites in the Ames test. In contrast to *T. brucei*, however, nitroreductase activity is dispensable in *S. typhimurium*, and in a nitroreductase null mutant strain fexinidazole was no longer mutagenic [74]. Mammals do not have NTR1 orthologs, and thus, fexinidazole and fexinidazole sulfone were inactive in micronucleus tests with human or mouse cells [74].

#### **4. Acoziborole: A Single Dose Oral Cure for Stage 2 HAT**

As the enhanced effort to counter the surging HAT epidemic of the late twentieth century began, the prospect of an orally available drug that was able to cure stage 2, CNS-involved disease with a single administration seemed remote. Yet, as fexinidazole and pafuramidine were entering their first clinical trials, a novel, promising class of antimicrobial agents appeared on the scene: the benzoxaboroles [84]. These compounds are characterized by a core scaffold based around an oxaborole heterocycle fused to a benzene ring. Pursued by the small Californian company Anacor Pharmaceuticals Inc. (incorporated into Pfizer Inc. in 2016), the benzoxaboroles quickly attracted attention due to their numerous bioactivities. Activity against trypanosomes was first reported in 2010 [85], and DNDi selected this class for further work. The US-based pharmaceutical company Scynexis was contracted to initiate an intensive program of the structure-activity-relationship work to seek benzoxaboroles with good activity against trypanosomes. Crucially, these benzoxaboroles were also to demonstrate pharmacokinetic properties suitable to cure stage 2 HAT; i.e., being capable of reaching the CNS and retaining trypanocidal activity levels for long enough to obtain cure [86].

Cyrus Bacchi, the same investigator who had brought eflornithine forward for HAT in the 1980s [87], showed that a 6-carboxamido-based series cured the *T. brucei* strain GVR35 murine model of stage 2 disease. Structure-activity work eventually brought forward SCYX-7158, a gem-dimethyl 4-fluoro-2-trimethylfluoro benzamide derivative, which cured when given orally [88]. Its modest *in vitro* potency against *T. brucei* (IC50 around 0.6 μM) [88] was offset by good pharmacokinetic properties, giving a 100% cure in a mouse model of the stage 2 disease following a dosing of 25 mg/kg once a day for 7 days oral dosing [88]. Preclinical testing showed no overt toxicity in mice or dogs with a concentration of no observed adverse event limit (NOAEL) of 15 mg/kg. No binding to key proteins (CYP450), serine/cysteine peptidase or hERG channels emerged [89]. SCYX-7158 was also non-mutagenic in the Ames tests or standard mammalian cell genotoxicity assays [89]. Hence, phase I clinical safety trials were approved, and in 2012, SCYX-7158, or acoziborole, became the first new chemical entity resulting from DNDi's program to enter clinical trials for HAT.

A phase I study that included 128 healthy male subjects of sub-Saharan African origin was conducted in 2015 in France to assess safety, tolerability, pharmacokinetics and pharmacodynamics after single oral, ascending doses [90]. Some adverse effects were noted, including headaches, dizziness and GI tract reactions, as abdominal pain, nausea, vomiting, constipation and diarrhea. However, more serious issues were not identified. This trial led to the selection of a 960 mg dose, given as a single administration in three tablets (each with 320 mg of the active compound). Acoziborole's long half-life (17 days), associated with high protein binding (97.8%) also necessitated safety monitoring in the volunteers for 210 days. Outputs from the phase I trials were good enough to enable progression to phase II/III trials in patients in Africa, which started late in 2016 [91]. The recruitment and dosing of patients is now complete, and follow-up is underway. Final results are eagerly awaited, but early indications (unpublished) suggest that the drug has demonstrated remarkable efficacy and safety profiles.

#### *Acoziborole: Mode of Action and Resistance Risk*

The mode of action and resistance mechanisms to benzoxaboroles are emerging. Acoziborole resistance was associated with multiple genetic changes in trypanosomes [92], although it was not possible to assign a specific gene to resistance in that study. However, among the changes in resistant lines was an amplification in the gene copy number of the RNA cleavage and polyadenylation specificity factor subunit 3 (CPSF3) [93]. Subsequently, CPSF3 was identified as a target for benzoxaboroles in apicomplexan parasites *Plasmodium* [93] and *Toxoplasma* [94]. Based on this observation, and the fact that metabolomics experiments had revealed a profound change in the methionine metabolism [95] that might have been related to RNA processing defects, particularly given the multi-methylation of the spliced leader sequence used in trans-splicing in trypanosomes, the effect of over-expression of CPSF3 on sensitivity to the related benzoxaborole, AN7973, was tested [96].

Over-expression of the gene yielded a notable loss of activity [96]. Elegant experiments using a gene over-expression library, followed by a selection of clones over-expressing genes yielding reduced sensitivity to acoziborole itself, also identified CPSF3 [97], and over-expression confirmed loss of sensitivity, pointing to CPSF3 as a target, if not the exclusive target, of these compounds in trypanosomes.

Another trypanocidal benzoxaborole (of the amino-methyl subclass) was shown to be subject to a two-step metabolic processing, involving a primary conversion using amine oxidase in host serum to an aldehyde that is further metabolized to a carboxylate via parasite aldehyde dehydrogenase [98]. Another valinate–amide derivative series shows considerable promise for animal African trypanosomiasis too [99].

#### **5. Conclusions**

The twentieth century ended with human African trypanosomiasis epidemics raging across Africa. A disease that had been brought under control in the 1960s was wreaking havoc. This led to a concerted effort across a range of international organizations, including the WHO and newly formed agencies including DNDi and the CPDD. The arrival of the Bill and Melinda Gates Foundation and a re-alignment of many international aid efforts provided key investment to allow the development of new routes to combat HAT. The development of new drugs was central to the process. CPDD produced an orally available cure for stage 1 HAT that ultimately failed. In the meantime, DNDi brought forward an old nitroheterocycle, fexinidazole, and through the combined forces of persistence and diligence worked their way around numerous perceptual hurdles related to the nitroheterocycle class, so as to bring to the clinic the first wholly oral treatment for stage 1 and 2 HAT. As those trials were proceeding, DNDi created a consortium to bring forward a class of compounds, the benzoxaboroles, from Anacor Pharmaceuticals Inc., through to Scynexis and an extended collaborative team. Encouraging clinical results for the leading candidate compound, acoziborole, now promise to deliver a drug that may cure stage 2 HAT with a single oral dose. The ability of non-profit Product Development Partnerships to bring new medicines to market is now proven. The turnaround in the epidemiology of HAT in the twenty first century is a truly remarkable story and a classic model of how to create medical success.

**Author Contributions:** Writing—review and editing: E.A.D., F.G., M.K.G., P.M., C.B., J.C.M., S.P.S.R., M.P.B. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by The Wellcome Trust WT-103024MA.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
