Hits and Lead Discovery in the Identification of New Drugs against the Trypanosomatidic Infections

The Neglected Tropical Diseases (NTDs) are a group of 17 pathologies, recognized by the World Health Organization (WHO), which are endemic in 149 countries of tropical and sub-tropical areas of the globe, and affect more than one billion people overall (Feasey et al. 2010). The pathologies are all caused by microparasites or macroparasites. Among the diseases provoked by microparasites, three are caused by kinetoplastidae, a group of flagellated protozoa transmitted by insect vectors: Chagas disease, Human African Trypanosomiasis and leishmaniasis. The focus of the present chapter is on the identification of new drugs for the treatment of trypanosomatidic infections such as Chagas disease, caused by Trypanosoma cruzi, and Human African Trypanosomiasis, caused by Trypanosoma brucei (Filardy et al. 2018).


chagas disease
Chagas disease, also referred to as American Trypanosomiasis, represents a serious medical and socioeconomic burden for over 21 regions of Latin America. More than 8 million people worldwide are estimated to be infected by T. cruzi and more than 10,000 deaths per year are reported. Recently, a reduction in incidence and prevalence of the infection has been observed, but on the other hand, a spread of the disease to previously unaffected regions as a consequence of massive migration fluxes, congenital transmission, and blood and organ donations has been reported, making Chagas disease a world health issue (Figure 1) (Pérez-Molina and Molina 2018).
The disease usually evolves from an acute to a chronic phase. The acute phase arises just after the infection and can last from a few weeks to several months and is characterized by non-specific symptoms (i.e., fever, swelling of skin and mucosa). Thereafter, the disease enters an asymptomatic chronic stage and only 20-30% of the infected people manifest lethal complications after several years (Bern et al. 2011). The treatment of both the acute and chronic stages of Chagas disease is based on chemotherapy. For almost 50 years, benznidazol and nifurtimox ( Figure 2) were the two first-line drugs for the treatment of Chagas disease (Rodriques Coura and de Castro 2002). Unfortunately, their efficacy is limited, and more frequently the side effects that the two drugs elicit during the treatment of the chronic stage induce the patients to quit the therapy. Nifurtimox treatment is discontinued in up to 75% of all cases, and benznidazol, which is better tolerated, is discontinued in 9 to 29% of the cases. Besides chemotherapy, serological cure is recommended for acute congenital Chagas during the first year of life, and for chronically infected children under 14 years (Pérez-Molina and Molina 2018).

Human African trypanosomiasis (HAt)
Human African Trypanosomiasis (HAT), also known as sleeping sickness, is caused by the bloodstream form of the two protozoan parasites Trypanosoma brucei gambiense and Trypanosoma brucei rhodesiense (Büscher et al. 2017). The sleeping sickness is mainly spread in the poorest rural area of sub-Saharan Africa,

color version at the end of the book
where the habitat is suitable for the survival of the tsetse fly, the vector responsible for the transmission of the parasite. In particular, T. b. gambiense is widespread in Central and Western Africa and causes over 98% of chronic HAT, whereas T. b. rhodensiense only infects humans occasionally and is mainly spread in Southern and Eastern Africa (Figure 3). Starting from 2009, and for the first time in thirty years, mostly due to WHO launched programs of surveillance and treatment, less than 10.000 new cases per year of HAT were registered in 2009. This represents a drop of 76% with respect to the 300,000 cases observed in 1995. This trend suggests that a complete eradication of HAT by 2020 is possible (WHO 2017). However, the number of cases registered by the WHO covers only the visible part of HAT incidence, whereas around 70 million people living in sub-Saharan Africa are exposed to T. brucei infection (Simarro et al. 2012).
HAT can occur in an acute or chronic form. The acute form of the disease appears within few weeks of the initial injection of the trypomastigote form of the parasite into the host by the tsetse fly. It is characterized by non-specific flu-like symptoms including fever, headache and joint pain. Thereafter, similar to Chagas disease, the infection can remain quiescent for a long time period and re-emerge several years later (chronic form). The second stage starts when the parasite migrates across the blood-brain barrier into the central nervous system, resulting in neurological symptoms such as confusion, depleted coordination, sensory disturbances and disruption of the normal sleeping cycle (from which the name sleeping sickness was derived). Ultimately, progressive mental deterioration leads to coma and death (Kennedy 2013). Beside T. b. gambiense and rhodesiense, pathogens for humans, other Trypanosoma species (e.g., T. vivax, T. congolense and T. brucei evansi) are etiological agents of a variety of wasting diseases affecting domestic livestock, which collectively go by the name of Animal African Trypanosomiasis (AAT or Nagana). Nagana causes significant damage to cattle farming with 3 million heads lost each year, contributing to significant economical damage in countries with already fragile economic structures. Moreover, animals represent a reservoir for these parasites from which humans may be exposed to infection (Njiokou et al. 2010). Therefore, fighting both HAT and AAT in the one health approach (Okello et al. 2014) is the only way to the complete eradication of this pathology (Giordani et al. 2016). The treatment of HAT, diversified depending on the stage of infection, is only based on chemotherapy (Cullen and Mocerino 2017). Nowadays, only four drugs and one drug combination are employed in therapy ( Figure 2). Pentamidine is currently the first-line option for treating the initial stage of HAT (Sands et al. 1985) whereas suramin is employed as second choice or for the primary treatment of T. b. rhodesiense infection, for which it is more effective than pentamidine. However, both drugs suffer from important limitations: they are administered via injection, require a healthcare professional, are ineffective in stage two and, more importantly, have adverse side effects that are often responsible for the suspension of the treatment. Melarsoprol, eflornithine and nifurtimox, instead, are the only drugs actually effective in the second stage of HAT (Eperon et al. 2014). Melarsoprol was the first drug to be developed for HAT; it has been medically used since 1949 and it has the great advantage of being effective against both T. b. gambiense and rhodesiense infections. However, it derives from arsenic which is the reason for its undesirable effects, including fatal encephalopathy (Fairlamb et al. 2018).
Eflornithine is the only recent drug approved for HAT in 2000. It is a less toxic alternative to melarsoprol, even though it is completely ineffective against T. b. rhodesiense. The main drawback of eflornithine is the complex regimen of intravenous infusions 4-times per day for 14 days of treatment (Chappuis et al. 2005). Today eflornithine is rarely used alone and is normally prescribed in combination with nifurtimox. Nifurtimox alone is less effective against T. brucei than T. cruzi, but in combination with eflornithine, the efficacy of the treatment is comparable to that of eflornithine alone, but with considerably fewer side-effects than the monotherapy. In addition, the regimen of eflornithine in combination with nifurtimox is more manageable, furthermore permitting an oral administration of the two drugs (Priotto et al. 2009).

HAt and chagas diseases' Social Impact and Intervention
Although one seventh of the world's population has an interest in or is exposed to the risk of NTDs, these pathologies remain neglected at several levels. At the local level, they are usually not mentioned by color version at the end of the book affected people, who are afraid of bias and marginalisation by the community. These pathologies barely spread outside the tropical area of origin; they do not represent an immediate risk for the 'occidental' countries that tend to be disinterested. However, the price to be paid for these choices is high: the NTDs have an elevated impact on the life of every single person, family and community, drastically decreasing the quality of life, aggravating poverty and reducing productivity -therefore representing a barrier for the economic growth of already depleted countries (Weiss 2008). Notwithstanding the social and economic burden, research and development for these illnesses is limited. For example, among all the 1,393 new drugs approved between 1975 and 1999, only 13 (0.9%) were targeted for NTDs, and only four of those were developed by pharmaceutical companies. The other nine were discovered as a result of veterinary and military research (Trouiller et al. 2002). Therefore, in order to encourage the research and development of new drugs that could be more effective, safe and economically sustainable, and to overcome the lack of interest of the major pharmaceutical companies in this health issue, associations between public and private institutions (PPP, public-private partnership) were recently initiated (Liese et al. 2010 (Eperon et al. 2014). A pivotal Phase II/III trial was completed in 2016, although two additional 'plug-in' trials are currently ongoing to assess the efficacy of the drug in special population groups not examined in the initial cohort. In addition, DNDi is also running a Phase III trial examining the effectiveness of fexinidazole in patients treated on both an out-patient basis and in a clinical hospital setting-depending on status. These trials are going to be completed by March 2020 (Mesu et al. 2018). Besides acoziborole and fexinidazole, which are two successful examples of new chemical entities identified for the treatment of trypanosomatidic infections, new drug candidates continuously enter the drug discovery pipeline.

drug discovery Approaches
Usually, the drug discovery for NTDs is carried out using classical ligand-based approaches, target-based approaches and phenotypic screening. Ligand-based approaches, which are discussed later in the chapter, are typically focused on the implementation of already known active compounds, natural compounds or drugs in therapy. Chemical structure modifications are then applied on the basis of classical medicinal chemistry strategies (Wermuth et al. 2015). Target-based approaches involve screening a library of compounds against a protein target and then optimizing the compounds for potency against the enzyme, selectivity, cellular activity, and pharmacokinetic properties. However, there are relatively few validated drug targets across the trypanosomiatidic infections (Frearson et al. 2007). Targets of major interest in the field of drug discovery are reported in Gilbert et al. (2013) and, in this chapter, we describe the main achievements in the fields of ergosterol biosynthesis, folate metabolism, phosphodiesterases, cysteine proteases and trypanothione metabolism. On the other hand, phenotypic screening has the advantage of identifying compounds that are active against the whole cell, meaning issues such as cell uptake and cell efflux have already been addressed (Haasen et al. 2017). There has been a major emphasis on phenotypic approaches to drug discovery for neglected diseases and a number of notable successes have been reported, like the above mentioned fexinidazole and acoziborole (Eperon et al. 2014). Nonetheless, a balanced portfolio of carefully selected ligand-and target-based approaches together with phenotypic approaches is probably the best strategy for drug discovery for neglected tropical diseases. This must be complemented by studies in animal models, which are reviewed in the final section of this chapter.

Synthetic compounds active against Trypanosoma cruzi and Trypanosoma brucei
Over the years, numerous papers have been published aiming at new, more effective and safe drugs with fewer side effects. In the context of this chapter, the synthesis and SAR studies of compounds with interesting biological activity are described, covering related literature from 2010 to the present.

Quinoline derivatives
A series of 10 2-alkylaminomethylquinoline derivatives were synthesized (Muscia et al. 2011) and were tested against different developmental stages of T. cruzi (epimastigotes, trypomastigotes and amastigotes). The interesting antiparasitic activity in vitro against epimastigotes of 5 compounds (IC 50 = 3.4-11.8 μM vs 5.8 μM for benznidazole) led to further evaluation against trypomastigotes and amastigotes. Compound 1 was the most active against trypomastigotes with IC 50 = 3.1 μM, with cytotoxic activity against COS-7 line slightly better than the reference drug (CC 50 = 770.9 μM vs 706 μM for benznidazole). Furthermore, compound 1 showed good selectivity for both trypomastigotes and amastigotes with SI values of 248.7 and 60.2, respectively. Reid et al. (2011) described the synthesis of derivatives of N-benzyl-1,2-dihydroquinolin-6-ols ( Figure 5) and their evaluation against T. brucei rhodesiense STIB900. The authors proceeded with an extensive SAR study in order to decipher the effect of the substituents on the biological activity. These studies are summarized in Figure 6. The most active compounds in vitro were 2, 3, 4 and 5 with IC 50 values 0.012 ± 0.001, 0.011 ± 0.001, 0.013 ± 0.004 and 0.014 ± 0.004 μM, respectively. These compounds were selected for in vivo evaluation and were administered to mice infected with T. brucei rhodesiense STIB900. Compound 2 suppressed parasitemia but a relapse occurred after 12.75 days after injection. Conversely, the ester prodrug 3 and the hydrochloride salts 4 and 5 resulted in cure even though at higher dosages (4 days × 50 mg/kg ip) than the reference drugs diminazene (4 days × 10 mg/kg ip) and pentamidine (4 days × 5 mg/kg ip). Hiltensperger et al. (2012) described the synthesis of a library of quinolone-type compounds bearing a benzamide function in position 3 and an amine heterocycle in position 7 ( Figure 7) for SAR purposes. These compounds were tested against T. b. brucei and T. b. rhodesiense.
In the context of the SAR studies, synthetic intermediates were tested as well. It is apparent that the amidation at position 3 is essential for activity but only for benzylamide and not for phenylamide derivatives, indicating the necessity for flexibility in position 3. In addition, the presence of cyclic and acyclic amines at position 7 enhances the activity. The most potent compound proved to be compound 6 which exhibited promising in vitro activity against T. brucei (IC 50 = 47 nM) and T. b. rhodesiense (IC 50 = 9 nM) combined with low cytotoxicity against macrophages J774.1. Hence, compound 6 was chosen for in vivo evaluation in a murine model of HAT. A preliminary formulation for oral administration was developed but was unsuccessful against T. b. rhodesiense (STIB900) infected NMRI mice.     Upadhayaya et al. (2013) described the synthesis of a series of compounds based on quinolines and indenoquinolines with variable side chains. From this library, five compounds ( Figure 8) showed interesting activity. These compounds have shown better activity than benznidazole and nifurtimox (IC 50 = 0.25-0.65 μΜ) against T. cruzi and T. brucei.
A series of new 3-nitrotriazole and 2-nitrotriazole-linked quinolines and quinazolines were synthesized and their antitrypanosomal activity was tested (Papadopoulou et al. 2017). Even though almost all compounds showed activity against T. cruzi and T. b. rhodesiense, only 2 chloroquinoline derivatives (12 and 13, Figure 9) exhibited satisfactory selectivity indices.
Particularly, compound 12 which bears a nitroimidazole moiety showed IC 50 values of 1.29 and 1.11 μM against T. b. rhodesiense and T. cruzi, respectively, while compound 13, which bears a nitrotriazole moiety, showed IC 50 values of 0.038 and 0.574 μM, respectively. Especially compound 13 exhibited an excellent selectivity index against T. b. rhodesiense (SI = 1937). Caminos et al. (2012) described the synthesis of 25 N,N'-disubstituted diamines through reductive amination of free aliphatic amines with various substituted benzaldehydes. This library was screened for antiparasitic activity in vitro against T. brucei and T. cruzi. In terms of SAR studies, the length of the aliphatic chain between the two amino functions was studied as well as the substitution of the benzyl moiety attached to the amines. The most promising compounds were 14 and 15 that bear a 4-OBn substitution and the aliphatic chain comprises 3 or 4 carbons ( Figure 10). These compounds exhibit medium activity against T. cruzi (IC 50 = 0.78 and 0.76 μM, respectively); however, they are significantly more active against T. brucei (IC 50 = 0.062 and 0.097 μM, respectively) and with higher selectivity indices (SI = 97 and 60, respectively).
Prompted by the in vitro results, these compounds were tested in vivo against female BALB/c mice and they proved to be more active than benznidazole in both the acute and chronic phase animal models.

Acridones
A small library of acridones was synthesized and their antiparasitic activity was tested (Montalvo-Quirós et al. 2015). The most active compounds are depicted in Figure 12. Acridones 18-20 showed excellent activity against T. b. rhodesiense (IC 50 = 0.069, 0.007 and 0.062 μM, respectively), accompanied by very good selectivity (SI > 2000). The N-methylation of these compounds to the corresponding acridones 21-23 did not alter the activity of 22 and 23 but increased the activity of 21 10-fold (IC 50 = 0.007 μM).
Compounds 18-20 were chosen for preliminary in vivo evaluation in the T. b. rhodesiense mouse model. Compound 19 confirmed its activity since three out of four mice were cured, with mean day of relapse over fifty days.

Thiohydantoins
During a high throughput screen for potential inhibitors of T. brucei in vitro, Buchynskyy et al. (2017a) discovered the substituted 2-thiohydantoin 24 (EC 50 = 346 nM) as a potential hit compound. An extensive  SAR study which comprised changes to each of the phenyl moieties in terms of lipophilicity, site of substitution, electron donating and electron withdrawing groups and the combination of them, led to two new compounds 25 and 26 ( Figure 13). These compounds showed excellent T. brucei activity in vitro (EC 50 = 3 nM for 25 and EC 50 = 2 nM for 26). For this purpose, they were also tested in vivo in the acute phase model of HAT. Both compounds cured all treated mice as no parasitemia was detected for 60 days after infection.
In the last decade, quinoxaline derivatives have been characterised as potential anti-trypanosomatid agents (Ancizu et al. 2009). Numerous compounds have been synthesized and evaluated and thus, some structural requirements have been established against trypanosomatids in vitro. Hence, the main structural requirements comprise the existence of the N-oxide moiety and the insertion of electron withdrawing groups on the quinoxaline ring.
Benitez et al. (2011) synthesized an extensive library of quinoxalines N,N'-dioxides as potential anti-T. cruzi agents. Initially, the evaluation of these compounds was realized in vitro against the epimastigote form of the Talahuen 2 strain of T. cruzi by calculating the percentage of growth inhibition (PGI). The most active compounds were 39-47 ( Figure 18). In addition, the ID 50 concentrations (50% inhibitory dose) were assessed with the most active compounds being 39 (ID 50 = 0.4 μM), 40 (ID 50 = 0.7 μM) and 42 (ID 50 = 0.39 μM). Active compounds were also tested against two other T. cruzi strains, namely, colombiana and Y strains. In particular, the most potent compounds, 40 and 42, were tested in vivo in murine models of acute Chagas' disease against the bloodstream trypomastigote form of T. cruzi, the CL Brener clone showing a very interesting biological profile.
The aforementioned structural prerequisites concerning quinoxalines were confirmed by Torres et al. (2013). 18 new quinoxaline N,N'-dioxides were synthesized and evaluated against the epimastigote form of the Talahuen 2 strain of T. cruzi by calculating PGI. 14 out of 18 compounds showed PGI values of 100% while the most active compounds were 48-53 ( Figure 19) with IC 50 values varying from 0.4-2.6 μM. Compound 52 was the most active compound (IC 50 = 0.4 μM).
In an effort to further optimize 55 (NEU-617) as an antiparasitic lead compound, Devine et al. (2015) proceeded with the synthesis of new derivatives involving several replacements on the quinazoline scaffold. Even though 55 was superior against T. brucei compared to the new derivatives, several hits were discovered against other protozoal species, i.e., T. cruzi, L. major and P. falciparum. Thus, compound 56

Benzamide derivatives
Based on their previous findings (Hwang et al. 2010), Hwang et al. (2013a) reported their optimization efforts for chloronitrobenzamides (CNBs). Compound 58 was found to be active against T. b. brucei (EC 50 = 0.92 μM) and for that reason, a new SAR study was conducted which was aimed towards the discovery of new leads in which the chloronitrobenzamide moiety remained intact. Numerous compounds were synthesized and tested and finally compounds 59-61 were the most potent against T. b. brucei (EC 50 = 0.006, 0.027 and 0.041 μM, respectively) ( Figure 22).
In addition, these compounds were tested against two other species, T. b. rhodesiense and T. b. gambiense. All compounds showed activity against T. b. rhodesiense (EC 50 = 0.013, 0.007 and 0.011 μM, respectively) as well as against T. b. gambiense (EC 50 = 0.036, 0.002 and 0.001 μM, respectively). This fact indicates that there is no species variation concerning the sensitivity against CNBs. However, these compounds exhibited poor to modest solubility, an issue that should be addressed in their further development.
In an effort to reduce complexity and molecular weight and at the same time, improve the pharmacological properties of CNBs, Hwang et al. (2013b) synthesized a new library of compounds by replacing the chloronitrobenzamide moiety. Even though most of the compounds were inactive, one potential lead (compound 62, Figure 23) was found which bears the 4-chloropyridine carboxamide moiety. This compound showed good activity against T. b. brucei (EC 50 = 0.12 μM) and excellent activity against T. b. rhodesiense (EC 50 = 0.018 μM) and T. b. gambiense (EC 50 = 0.038 μM).
Pyridyl benzamides were further explored as potential anti-trypanosomatidic agents (Ferrins et al. 2014). An extensive SAR study was conducted based on hit compound 63 (IC 50 = 3.03 μM). A large library was synthesized and tested against T. b. brucei, T. b. rhodesiense and T. cruzi. Compounds 64-68 ( Figure 24) showed activity with IC 50 values ranging between 0.41-1.1 μM against T. b. brucei and 0.045-0.64 μΜ against T. b. rhodesiense, with the most active compound being 65 (IC 50 = 0.045 μM). However, all compounds were inactive against T. cruzi. These compounds with low molecular weight have favorable properties for further optimization and have a very good prediction for CNS penetration which is essential for treating second stage HAT.
In the context of a high throughput screening campaign of 87000 compounds against T. b. brucei, Rahmani et al. (2015) identified a new class of potential trypanocides, pyrazine carboxamides. This screening delivered a starting hit, compound 69 (EC 50 = 0.49 μM) and subsequent SAR studies were conducted concerning structural changes around the core. Several compounds showed very good activity against T. b. brucei and T. b. rhodesiense but the most active compounds were 70 with EC 50 values of 0.035 and 0.024 μM against T. b. brucei and T. b. rhodesiense, respectively and compound 71 with EC 50 values 0.025 and 0.038 μM, respectively ( Figure 25). Worth mentioning is that this class of compounds showed     very low cytotoxicity and was selective for T. brucei, since no significant activity was observed against T. cruzi. Cleghorn et al. (2015) conducted a phenotypic screening of compounds against T. b. brucei followed by a mammalian cell counterscreen (MRC-5 cells) to exclude nonselective compounds. This work led to indoline-2-carboxamide 72 ( Figure 26) which showed good activity (EC 50 = 27 nM) and selectivity over mammalian cells (> 1600 fold). With compound 72 in hand, SAR studies were realized in terms of core modifications and the effect of stereochemistry on the chiral center. These studies led to the discovery of two new compounds, 73 and 74 (racemic), with lower activity against T. b. brucei (EC 50 = 60 and 80 nM, respectively) but with improved metabolic stability and enhanced in vivo exposure. These compounds exhibited excellent pharmacokinetic properties, and resulted in a full cure in a HAT stage 1 mouse model, but unfortunately only a partial cure in stage 2. Buchynskyy et al. (2017b) identified N-(2-aminoethyl)-N-phenyl benzamides as an interesting class of compounds with potential antiparasitic activity against T. brucei. Compound 75 (Figure 27) was selected as a hit compound for its activity (EC 50 = 1.21 μM) against T. b. brucei, its selectivity over mammalian cells (> 30 fold) and its drug-like features, including low molecular weight, clogP, H-bond donors and acceptors. The outcome of SAR studies was the discovery of some new lead compounds with EC 50 values ranging from 0.001-0.031 μM with compound 76 being the most potent (EC 50 = 0.001 μM against  T. b. brucei and 0.002 μM against T. b. rhodesiense). However, in murine efficacy models of HAT infection, the compounds showed only partial cures or suppression.

Chalcones
Chalcones form another class of compounds with broad spectrum pharmacological activities. Recently, Borsari et al. (2017) reported the synthesis of a library of thirteen 2-hydroxy chalcones bearing methoxy groups. All the synthesized compounds were tested against the bloodstream form of T. brucei and the intracellular T. cruzi at 10 mM. Compounds 77-79 ( Figure 28) were the most potent against T. brucei with EC 50 values 1.3 ± 0.02, 2.1 ± 0 and 2.1 ± 0.4 μΜ, respectively and with selectivity index > 12. Bhambra et al. (2017) synthesized a small library of pyridylchalcones which were tested against T. b. rhodesiense. Even though none of these compounds showed better activity than the reference drugs pentamidine or melarsoprol, three compounds (80-82) can be considered as potential leads with IC 50 values 0.29, 0.40 and 0.41 μM, respectively ( Figure 29). Fytas et al. (2011) discovered acetohydroxamic acids as a new class of compounds with potential antitrypanocidal activity. The synthesis of these compounds was realized through the attachment of an acetohydroxamic acid moiety to the imidic nitrogen of 2,6-diketopiperazines. A SAR study revealed that compounds 83 (S-enantiomer), 84 (R-enantiomer) and 85 (racemic mixture of 83 and 84) ( Figure 30) showed low nanomolar activity against bloodstream-form T. brucei with IC 50 values 6.8 ± 1.4, 9.1 ± 0.2 and 17± 1 nM, respectively, while compound 83 along with compounds 86 and 87 displayed significant activity against T. cruzi with IC 50 values 0.21 ± 0.04, 5.51 ± 0.68 and 3.62 ± 0.31 μM, respectively. Overall, compound 83 proved to be the most active compound against both species. In addition, these compounds showed very good selectivity indices. The fact that replacement of the hydroxamic acid moiety led to compounds with decreased activity makes this group a requisite for trypanocidal activity.

Hydroxamic acid derivatives
A small library of hydroxamic acid derivatives which inhibit human histone deacetylases was synthesized and tested against cultured bloodstream form T. brucei (Kelly et al. 2012) . Most of these compounds exhibited very good activity in the submicromolar range. Specifically, the most promising compounds were 88-90 ( Figure 31) with IC 50 values 0.034 ± 0.002, 0.064 ± 0.005 and 0.086 ± 0.009 μM, respectively. Zelisko et al. (2012)       T. b. gambiense with IC 50 values 0.26 and 0.42 μM, respectively; thus, it can be considered as a potential lead for further optimization. Cardoso et al. (2014) reported the synthesis and the anti-T. cruzi activity of 24 2-(pyridine-2-yl)-1,3thiazoles by ultrasound-assisted synthesis. The majority of this class of compounds demonstrated better activity than the reference drug benznidazole, with compounds 92 and 93 ( Figure 33) being the most potent with IC 50 values 1.2 μM each against T. cruzi trypomastigotes. In addition, this class of compounds showed no cytotoxicity against HepG2 cells. Russell et al. (2016) published their work on hit to lead optimization of compound 94 ( Figure 34). The latter emerged as a hit compound through a high-throughput screening because it exhibited interesting activity (IC 50 = 0.80, 1.5 and 2.3 μM for T. b. brucei, T. b. rhodesiense and T. cruzi, respectively) and very good physicochemical properties. An extensive SAR study was conducted towards the discovery of new leads. Indeed, 2-(fluorophenyl)thiazole 95 was found to exhibit excellent activity against T. b. brucei (IC 50 = 0.024 ± 0.003 μM, SI = 433) and T. cruzi (IC 50 = 0.020 ± 0.01 μM, SI > 3162). Compound 95 was tested in vivo in Balb/c mice infected with the trypomastigote form of T. cruzi Brazil strain with encouraging results but unfortunately it was rapidly metabolized.

Thiazole derivatives
In a phenotypic high-throughput screening of 700,000 compounds, compound 96 ( Figure 35) was identified, among others, as a potential inhibitor of T. brucei (IC 50 = 0.632 μM, SI = 162 against T. b. rhodesiense). Patrick et al. (2016Patrick et al. ( , 2017 with two back-to-back papers reported the efforts towards the optimization of a hit compound. A SAR study was conducted by focusing on changes to both phenyl rings of compound 96. Therefore, 72 new derivatives were synthesized and tested against T. b. rhodesiense STIB900 and L6 rat myoblast cells for cytotoxicity in vitro. 44 compounds were more active than 96 and eight of them showed IC 50 values below 100 nM. The most potent compound was 97 (IC 50 = 9 nM, SI > 18,000) but unfortunately, the in vivo test showed that it was unable to cure infected mice. Even though the administration of 97 caused reduction of parasitemia, relapses occurred. This was attributed to poor metabolic stability. Prompted by these results, a second paper described a new SAR study of 65   additional compounds with modifications in other sites on hit 96 with the intention of maintaining the same activity and improving the metabolic stability. This led to compound 98 which was less active than 97 (IC 50 = 35 nΜ) but with enhanced metabolic stability and brain penetration. Oral dosing of 98 cured five out of five mice in both acute and chronic murine models of HAT.
Many thiazole derivatives were synthesized and evaluated for their antiparasitic activity during the last year by several research groups. Compound 99 (Filho et al. 2017) showed activity against the promastigote form of T. cruzi (IC 50 = 0.37 μM), compound 100 (Thompson et al. 2017) with IC 50 = 0.055 μM against T. cruzi, and compounds 101 and 102  with IC 50 values 1.2 and 1.6 μM, respectively against the trypomastigote form of T. cruzi (Figure 36). Bhambra et al. (2016) reported the synthesis of a small library of 2,6-disubstituted-4,5,7trifluorobenzothiophenes. Compounds 103 and 104 ( Figure 37) demonstrated interesting antitrypanosomal activity against T. b. rhodesiense with IC 50 values 0.60 and 0.53 μM, respectively and no toxicity to mammalian cells. It is apparent that the activity of these compounds can be attributed to the existence of the benzimidazole moiety since its replacement resulted in loss of activity. Vega et al. (2012) prepared a series of nitro-indazolin-3-ones which were tested in vitro against epimastigote forms of T. cruzi. Compounds 105-107 ( Figure 38) showed interesting trypanocidal activity (IC 50 = 0.93, 2.39, 1.17 μM, respectively) and low unspecified toxicity. The authors found structural similarities with other known antiprotozoan drugs during their search, and therefore they introduced a new scaffold for further research and optimization. Tapia et al. (2014) synthesized a series of indole-4,9-dione and their activity was evaluated against the epimastigote form of T. cruzi, Y strain. All new compounds showed better activity and selectivity compared to the reference drug nifurtimox. Interestingly, compound 108 ( Figure 38) demonstrated excellent nanomolar inhibitory activity (IC 50 = 20 nM), and high selectivity index (SI = 625).

Imidazole derivatives
A high-throughput screening of 303,286 compounds revealed a class of imidazole-based inhibitors (109-111, Figure 39) that can inhibit infection of mammalian host cells by T. cruzi trypomastigotes (Bettiol et al. 2009). In an effort to further optimize the activity, cytotoxicity and bioavailability, new  inhibitors of T. cruzi CYP51 were synthesized (Andriani et al. 2013) with ring-constrained structures. The most promising inhibitors were 112 (EC 50 = 72 nM) and 113 (EC 50 = 80 nM).
The SAR study revealed that the substitution of the phenyl ring with lipophilic groups is essential. The absence of these groups was accompanied by loss of potency. In addition, replacement of the imidazole moiety by other nitrogen heterocycles, such as pyrazole or triazole, resulted in loss of activity. Hence, the presence of an imidazole ring is required for T. cruzi CYP51 binding.
More compounds bearing an imidazole ring have been synthesized and tested against the drugtargetable enzyme CYP51 due to structural similarities with its inhibitors. In particular, compounds 114-116 (Figure 40), even in their racemic form, showed very interesting biological activity against T. cruzi Talahuen C2C4 amastigote stage (Friggeri et al. 2013) with IC 50 values 14, 5 and 36 nM, respectively. The authors wanted to further explore the activity of the single enantiomers which led to the conclusion that S-enantiomers were more active than the corresponding R-enantiomers. Compound (S)-114

Computational and experimental structure-based drug design strategies against Trypanosoma cruzi and Trypanosoma brucei
In the above section, we presented a large number of synthetic compounds that have been optimized for their anti-trypanosomal properties; however, this was often done without taking specific molecular targets into account. Target-oriented methods provide an alternative strategy or can present an interesting complement to optimization based solely on anti-parasitic properties. In this section, we review some of the recent literature on structure-based drug design (SBDD) efforts to tackle trypanosomatidic infections by targeting specific metabolic pathways or enzymes that have been extensively explored in target-guided drug discovery approaches during the past years, such as CYP51, folate pathway proteins, phosphodiesterases, cysteine proteases or enzymes of the trypanothione metabolism. In the discovery of new hits and lead optimization, experimental methods and computational techniques are widely used together to facilitate the process of compound library design and iterative improvement in the context of known (or predicted) biomolecular targets. We describe approaches to SBDD for key targets, as well as techniques for tackling typical challenges in SBDD, such as selectivity, off-target binding and deciphering the mechanism-ofaction (MoA) of a drug. We conclude the section with a brief summary of the methods typically employed in anti-trypanosomatidic drug discovery, their potential applications and limitations.

Compounds targeting the lanosterol 14α-demethylase (CYP51)
A particularly prominent example of a widely considered anti-trypanosomatid drug target is an enzyme involved in the ergosterol biosynthesis route, namely the lanosterol 14α-demethylase (or CYP51). CYP51s, which belong to the family of cytochrome P450s, are highly conserved heme-containing proteins involved in the production of plasma membrane building blocks and regulatory molecules Waterman 2004, 2007). Azole-based CYP51 inhibitors are known anti-fungal agents (Lass-Flörl 2011, Denning andBromley 2015) and the trypanosomal ergosterol biosynthesis resembles that of fungi, making the repurposing of anti-fungal compounds a well-explored design strategy (Buckner and Urbina 2012). Furthermore, crystallographic complexes of TbCYP51 and TcCYP51 with anti-fungal compounds, such as fluconazole 123 and posaconazole 124 (Figure 42, Chen et al. 2010), paved the way for further optimization of azole-based and chemically distinct CYP51 inhibitors for trypanosomatidic infections. As another example, Hoekstra et al. (2015) reported the crystallographic complex of a 1-tetrazole-based anti-fungal clinical drug candidate with TcCYP51, providing an additional target-based starting point for future optimization of the compound as an anti-Chagas agent. Hargrove et al. (2013) used X-ray crystallography to determine the binding modes of the promising pyridine-based drug candidates 125 and 126 ( Figure 43) for Chagas disease. These compounds were found to be the first non-azole compounds showing potent inhibition of TcCYP51 as posaconazole 124. Importantly, the structural data show that the coordination bond between the pyridine heterocycles of 125 and 126 and ferric heme iron of TcCYP51 are longer than in the case of azoles. The authors highlight that this may allow for achieving higher selectivity between the trypanosomal enzyme and the human variant since, due to the weaker coordinating interaction, the other non-bonded interactions between the protein and the ligand, which differ between the targets, become more important determinants for compound binding affinities. However, as can be seen in Figure 43, the interaction of compounds 125 and 126 with TcCYP51 is mainly mediated by hydrophobic contacts.
The development of the aforementioned imidazole-based T. cruzi CYP51 inhibitors 112 and 113 (see Figure 39) by Andriani et al. (2013) as well as compounds 114-116 ( Figure 40) by Friggeri et al. (2013) was similarly supported by crystallographic data (directly by structure determination or indirectly by docking studies to existing crystal structures). In a follow-up study, Friggeri et al. (2014) further optimized imidazole-based inhibitors and demonstrated their ability to bind to CYP51 in two different regions: either in the substrate-binding site close to the heme group or in the substrate access channel (compare binding mode of 127 in Figure 44).
Not only crystallography but also docking studies were employed in the design of CYP51 inhibitors. An interesting complement to the efforts guided solely by static crystal structures are molecular dynamics (MD) simulations, performed by Yu et al. (2015Yu et al. ( , 2016, used to study ligand egress routes in T. brucei CYP51 in comparison with human CYP51. These studies highlight differences in the dynamics and composition of the ligand tunnel between the trypanosomal and human CYP51, which could be exploited in the future anti-parasitic drug design efforts to enhance selectivity. Gunatilleke et al. (2012) combined target-based high-throughput screening (HTS) with a screen against T. cruzi-infected skeletal myoblast cells to yield compounds with a known molecular target and activity against the parasite. Further, they used bioinformatics approaches to demonstrate similarities between compounds preferentially bound by CYP51 and compounds bound by several other cytochrome P450 enzymes and an unrelated glutaminyl-peptide cyclotransferase. Inhibitors of these enzymes can hence serve as a potential pool of lead compounds.
It is not uncommon that large libraries of potential hit compounds are screened in silico against large crystallographic databases of enzymatic targets. A drawback of such screens is that many studies lack experimental validation of their results. Nevertheless, in silico studies exploring, for example, potential  (Lepesheva et al. 2010, respectively. Pocket-lining residues and the pocket shape are shown. Black dashed lines indicate coordination bonds between ferric heme iron and a nitrogen atom of the ligand, gray dotted lines with filled circles indicate π-π interactions, gray dotted lines with white circles cation-π interactions and black arrows show hydrogen-bonding interactions. targets of phytochemicals may provide some first insight when a target is not known or may be used to propose possible hits for a desired target. For example, compounds isolated from Nigerian medicinal plants have been studied by Setzer and Ogungbe (2012). They found in silico potential inhibitors of, amongst others, TbCYP51 and two enzymes of the parasitic folate pathway: pteridine reductase 1 (PTR1) and dihydrofolate reductase (DHFR).
Further details on the current structural knowledge about CYP51 in trypanosomatids and the design of inhibitory compounds can be found in the recent review by Lepesheva et al. (2018).

Folate metabolism
Targeting the kinetoplastid folate pathway, unlike the corresponding malarial (Hawser et al. 2006) and bacterial (Yuthavong et al. 2005) pathways, requires not only the inhibition of dihydrofolate reductase (DHFR), but also of pteridine reductase 1 (PTR1) (Bello et al. 1994, Sienkiewicz et al. 2010. The latter enzyme is mainly responsible for the reduction of pterins, but can be upregulated when DHFR is inhibited and serve as a bypass for folate reduction to provide necessary educts for DNA synthesis and thus ensure parasite survival (Dawson et al. 2006, Vickers andBeverley 2011). In T. brucei, PTR1 was validated as a potential drug target by the gene knockout and RNA interference experiments , Sienkiewicz et al. 2010. Herein, we mostly focus on inhibitor design for PTR1 from T. cruzi and T. brucei. Multiple drug design approaches have been used to target PTR1, starting from optimization of the substrate scaffold, through virtual screening (VS), to fragment-based drug design (FBDD).
An example of a classical substrate-like compound is methotrexate 129 (MTX), which is a subnanomolar inhibitor of DHFR, and an approved anti-cancer drug (Zuccotto et al. 1999, Shuvalov et al. 2017). However, it has also been shown to inhibit PTR1 at the sub-micromolar level (Dawson et al. 2006, Cavazzuti et al. 2008. Notably, the substrate, MTX and most other inhibitors of PTR1 occupy the main binding site in a π-sandwich between the nicotinamide of the NADPH/NADP + cofactor and a phenylalanine residue (Figure 46). MTX adopts a similar binding mode to folate 130, but with a flipped orientation of the pteridine ring ( Figure 46).
The aformentioned binding modes of folate and MTX were starting points in the study by Tulloch et al. (2010), who selected 3 scaffolds-pteridine, pyrrolopyrimidine and pyrimidine-for inhibitor design and optimization against T. brucei PTR1 (TbPTR1). Five pyrrolopyrimidine-based inhibitors were identified and crystallized in TbPTR1. Among those, 131 and 132 ( Figure 47) with K i values against TbPTR1 of 0.36 μM and 0.40 μM, respectively, were found to attain modest ED 50 values against the T. brucei bloodstream form (BSF): 274 ± 7.5 μM and 123 ± 3.3 μM, respectively. 132 was also found to show a synergistic effect when combined with MTX. Later, Khalaf et al. (2014) used a similar approach for the development of 61 additional pyrrolopyrimidines, of which 23 were crystallized in TbPTR1, greatly expanding the structural data available for this enzyme. Two of the crystallized compounds, 133 and 134 (Figure 48), had TbPTR1 K i app values of 0.23 μM and 0.14 μM, respectively, and were found to have T. brucei BSF IC 50 values of 3.20 μM and 0.25 μM, respectively, in Creek's minimal medium. However, these derivatives showed substantial toxicity in in vivo mouse models and thus were not suitable for further development.
VS approaches were used to identify novel non-pteridine scaffolds, including the 2-aminothiadiazole core and flavonoids as potential binders of the PTR1 substrate pocket (Ferrari et al. 2011, Borsari et al. 2016). Recently, a library of 2-aminothiadiazole-based TbPTR1 inhibitors has been developed (Linciano et al. 2017). X-ray crystallography and docking simulations confirmed the classical binding mode for these series (cf. Figure 46), with the amino moiety of the scaffold core interacting with a serine residue close to the active site, cofactor ribose and phosphates (Figure 49). Two thiadiazole-2,5-diamines 136 and 137 ( Figure 49) reached TbPTR1 IC 50 values of 16.0 μM and 25.0 μM, respectively. Despite being unable to inhibit the T. brucei BSF as single agents, both PTR1 inhibitors potentiated the EC 50 of MTX by 4.1 and 2-fold, respectively.
Another class of hits against PTR1 were flavonoids, identified by computational and experimental screening of a natural products library (Borsari et al. 2016), with flavonols showing the most promising inhibitory effect on TbPTR1. A combination of crystallographic experiments and computational docking was then employed to explore the multiple possible binding modes and deduce a structure-activity relationship (SAR) for a set of synthetic flavonols. In a follow-up project, the same methods allowed in-depth characterization of the determinants of the activities of several flavonols and the corresponding flavanones .
In another VS study, Dube et al. (2014) used a combination of docking and pharmacophore modeling to screen the ZINC database for putative inhibitors of TbPTR1 and predict their activities. The VS of   (Dawson et al. 2006), with folate 130-on PDB-ID 3bmc (Tulloch et al. 2010) and with 138, illustrating the non-classical binding mode outside the π-sandwich-on PDB-ID 3gn2 (Mpamhanga et al. 2009). color version at the end of the book fragments against TbPTR1 performed by Mpamhanga et al. (2009) identified 2-aminobenzothiazole and 2-aminobenzimidazole, binding in the substrate pocket, as promising cores that were further used for inhibitor development. Notably, the crystallographic studies revealed that some of the synthesized compounds with a 2-aminobenzimidazole scaffold adopted a non-classical binding mode in the TbPTR1specific subsite adjacent to the primary substrate binding site (Mpamhanga et al. 2009; Figure 46, compound 138). Compounds 138 and 139 ( Figure 50) reached K i app of 0.4 μM and 0.007 μM, respectively, against TbPTR1. Despite its nanomolar on-target potency, 138 displayed an EC 50 against T. brucei of only 10 μM. Spinks et al. (2011) further attempted to optimize the series of 2-aminobenzimidazoles to target the alternative subsite, but anti-parasite activities did not improve, with the lowest EC 50 value being 6.7 μM.
The non-classical binding mode of the amino-benzimidazole series discussed above inspired computational FBDD studies to combine a scaffold with a classical binding mode with an alternative subpocket-targeting scaffold. In principle, computational methods exist that may help in finding the best molecular fragment binding the subsite. However, in the case of the TbPTR1 alternative subsite exploited by Mpamhanga et al. (2009) and Spinks et al. (2011), this was challenging due to the presence of halogen substituents in the most potent compounds. Many state-of-the-art commonly used force fields used in Figure 49. Interaction diagrams of a 2,5-diamino-1,3,4-thiadiazole inhibitor of TbPTR1, 135 and structures of its derivatives 136 and 137. Compounds 136 and 137 were predicted to show a similar binding mode in TbPTR1 by docking studies (Linciano et al. 2017). The interaction diagram is based on the crystallographic complex with PDB-ID 5izc. Pocket-lining residues and the pocket shape are shown. Gray dotted lines indicate π-π interactions and black arrows show hydrogen-bonding interactions.
NAP denotes NADP(H). The majority of compounds have been developed to target TbPTR1 (or Leishmania major PTR1). Although a crystal structure of a close homolog of TcPTR1, TcPTR2, was solved (Schormann et al. 2005), most drug design efforts did not consider the T. cruzi enzyme variants. Mendoza-Martıńez et al. (2015) designed some quinazoline derivatives and used docking studies to support TcPTR2 and TcDHFR as reasonable potential targets. The compounds were evaluated against bloodstream trypomastigotes of two strains of T. cruzi (NINOA and INC-5 strain) and compounds 140-142 ( Figure 51) were found to be particularly interesting, since they showed a better activity profile than the reference drugs, nifurtimox and benznidazole.
Otherwise, for T. cruzi, mainly DHFR was considered in drug design efforts: Schormann et al. (2008) used 3D-QSAR models to guide the design of inhibitors selective for TcDHFR over human DHFR (hDHFR). In the subsequent study, guided by crystallography and docking experiments, they exploited amino acid  Figure 46). The diagrams are based on the crystallographic complexes with PDB-ID 3gn2 and 2wd8 (Mpamhanga et al. 2009), respectively. Pocket-lining residues and the pocket shape are shown. Gray dotted lines indicate π-π interactions, black arrows show hydrogen-bonding interactions and gray arrows indicate halogen-bonds. differences between the pocket entrance regions of hDHFR and TcDHFR, including the substitution of Phe31 to Met49. A series of 2,4-diaminoquinazoline derivatives, similar to trimetrexate, with varying flexible hydrophobic groups in the tail, some of which bound near Met49 of TcDHFR, had about 7-8 times better K i against TcDHFR than hDHFR (Schormann et al. 2010).
The above discussed study of Mendoza-Martıńez et al. (2015) is an example of a dual inhibition approach, which is particularly important for targeting the trypanosomatid folate pathway. Ideally, this requires targeting both PTR1 and parasitic DHFR, while avoiding inhibitor binding to hDHFR, which is quite similar to the parasitic homologue. To address this issue, Panecka-Hofman et al. (2017) applied a variety of techniques (such as computational analysis of crystallographic structures, homology modeling and binding site mapping) for sequence and structural comparison of the folate pathway on-and offtargets. Ligand-target interactions, binding site properties and target conformational variability were computationally compared to yield guidelines for the design and optimization of selective inhibitors of the trypanosomatidic folate pathway. In an extension of this study, the conformational variability of the TcDHFR and hDHFR binding sites was analyzed with the TRAPP (TRAnsient Pockets in Proteins) web server based on the available crystallographic data (Stank et al. 2017). Complementary to the aforementioned work of Schormann et al. (2010), a transiently appearing subpocket was identified in the vicinity of Met49, which corresponds to Phe31 in hDHFR. The ProSAT+ tool integrated in TRAPP also identified position 31 as a known site of an hDHFR mutation that alters binding of the MTX inhibitor.
The majority of drug design efforts for the kinetoplastid folate pathway have so far focused on PTR1/ PTR2 and DHFR. A few other folate pathway enzymes, such as the N 5 ,N 10 -methylenetetrahydrofolate dehydrogenase/cyclohydrolase (DHCH) of T. brucei, were considered in compound development. Eadsforth et al. (2015) designed inhibitors of TbDHCH using a combination of crystallographic and docking experiments starting from an inhibitor of bacterial and human DHCH.
Despite the extensive design and development efforts to obtain potent and selective inhibitors of PTR1 and DHFR, there is often limited transferability of on-target-based activity to an in vitro or even in vivo activity against T. brucei or T. cruzi. For a more detailed overview on the efforts made, possible reasons for this limitation and other potential targets to be considered in the future, the reader is referred to the recent review by Cullia et al. (2018).

Phosphodiesterases
More recently, a phosphodiesterase (PDE) of T. brucei, the PDEB1 isoform, has gained attention as a potential drug target for Human African Trypanosomiasis (HAT). Phosphodiesterases cleave phosphodiester bonds in cyclic nucleotides like cyclic AMP (cAMP), which is an important second messenger involved in regulation of signal transduction. An overview about cAMP signalling in trypanosomatids and differences between parasitic and human pathways that can be exploited for drug discovery can be found in the review by Tagoe et al. (2015). Herein, we focus on the recent efforts made to develop inhibitors of TbPDEB1.
Since inhibitors of human PDEs, like cilomilast, piclamilast, sildenafil and tadalafil were known, design efforts were often focused on repurposing these drugs and developing analogs thereof (Amata et al. 2014, Woodring et al. 2013), but with rather limited success. A virtual screening of the ZINC database was carried out by Jansen et al. (2013) to broaden the chemical space of the potential TbPDEB1 inhibitors. Molecular docking in conjunction with a score based on receptor-compound interaction fingerprints was used to filter suitable compounds. This way, six novel inhibitors of TbPDEB1 with IC 50 values between 10 and 80 μM were identified.
In another study, Orrling et al. (2012) used homology modeling and docking studies to exploit a parasite-specific subpocket, the so-called P-pocket of TbPDEB1 (Figure 53), for the design of the improved catechol pyrazolinone-based inhibitors. The most potent inhibitor 145 (Figure 54) achieved an IC 50 of 49 nM against TbPDEB1 and was able to inhibit parasite trypomastigote proliferation in in vitro studies: T. brucei rhodesiense with IC 50 of 60 nM, T. brucei brucei with IC 50 of 520 nM and T. cruzi with IC 50 of Figure 53. Compound 147 binding to TbPDEB1 (see also Figure 52, based on PDB-ID 5l8c) and targeting the parasite PDE-specific P-pocket. On the left, the protein is shown as a semi-transparent surface with the ligand in black sticks and the location of the P-pocket marked. On the right side, the protein is shown in the same orientation as cartoon and only the P-pocket residues are shown as a surface with residues labeled according to Blaazer et al. (2018).  7.6 μM. However, compound 145 was an even more potent inhibitor of human PDE isoforms. Thus, this study highlights the potential of PDE inhibitors against trypanosomal infections, but also indicates that further optimization of the selectivity for the TbPDEB1 target is required. The P-pocket of TbPDEB1 has been recently further exploited by Blaazer et al. (2018) in their development of selective 4a,5,8,8a-tetrahydrophthalazinone-based inhibitors. Guided by the crystallographic analysis of non-specific TbPDEB1 inhibitors and the analysis of flexibility patterns of a TbPDEB1-inhibitor complex observed by MD simulations, several compounds specifically targeting the P-pocket were developed. 146 and 147 ( Figure 55) were overall the best, both showing K i of 100 nM against TbPDEB1, 10-and 19-fold selectivity for the parasitic over the human enzyme, and had IC 50 s of 5.5 μM and 6.7 μM, respectively, when tested against T. brucei.

Cysteine proteases
Two cysteine proteases structurally related to human cathepsin, L: cruzain of T. cruzi and rhodesain of T. brucei rhodesiense, have been subject to many drug development efforts, including many computational and mechanistic studies in the past years, which we review below. More details on inhibitor design against cysteine proteases in Trypanosoma can be found in the recent review by Ferreira and Andricopulo (2017).
Cruzain: Since cruzain is a protease, many inhibitors designed were based on peptides and typically bind covalently (see Figure 56). Database screening strategies led to non-peptide inhibitors of cruzain. Ferreira et al. (2010) illustrated for cruzain inhibitors how the combination of VS, HTS and prioritization of molecular scaffolds can largely avoid false-positives. Wiggers et al. (2013) identified non-covalent cruzain inhibitors from the ZINC database. Further, Palos et al. (2017) performed VS of FDA-approved drugs against cruzain and confirmed the trypanocidal potential of four putative inhibitors in vitro and in vivo, which may thus represent starting scaffolds for further drug development.  Figure 53). The diagrams are based on the crystallographic complexes with PDB-ID 5g2b and 5l8c (Blaazer et al. 2018), respectively. Pocket-lining residues and the pocket shape are shown. Gray dotted lines indicate π-π interactions and black arrows show hydrogen-bonding interactions. Crystallography was also used for exploring compound binding modes, MoA, and for obtaining SAR. In the aforementioned VS study, Wiggers et al. (2013) used a molecular simplification approach to obtain a SAR for one of the compounds and found that the 2-acetamidothiophene-3-carboxamide scaffold was the critical component for the interactions with cruzain. Then, they confirmed the binding mode of this scaffold by crystallographic data, thus validating a potential new non-covalent, non-peptidic cruzainbinding scaffold. Avelar et al. (2015) characterized dipeptidyl nitriles as inhibitors of cruzain by extensive SAR studies and determining a crystallographic complex with compound 148 (Figures 56 and 57).  synthesized and crystallized oxyguanidine analogs of existing cysteine protease inhibitors  . The diagrams are based on the crystallographic complexes with PDB-ID 4qh6, 4pi3 and 4xui, respectively. Pocket-lining residues and the pocket shape are shown. Gray dotted lines indicate π-π interactions, black arrows show hydrogen-bonding interactions, and a black line ended with circles marks a covalent bond between ligand and protein.
and were able to show that one of the compounds displays covalent (149, Figure 57) and the other noncovalent inhibition (150, Figure 57). A variety of computational chemistry methods has been used to study cruzain interactions with inhibitors, to better understand their MoA, and to design new compounds. Hologram quantitative structureactivity relationship, comparative molecular field analysis and comparative molecular similarity index analysis methods combined with docking simulations have been applied to a series of benzimidazolebased cruzain inhibitors by Pauli et al. (2017) to predict binding modes and activities.  used docking studies and QSAR methods to study nitrile-containing cruzain inhibitors and predict their activities, leading to the proposal of new potential inhibitors. de  used 2D-and 3D-QSAR methods to explore the interactions of oxadiazole-based compounds with different cruzain subsites. Elizondo-Jimenez et al. (2017) synthesized benzenesulfonyl and N-propionyl benzenesulfonyl hydrazone derivatives, evaluated their anti-T. cruzi activity and used docking studies to propose covalent binding to cruzain as the compounds' MoA. Silva-Júnior et al. (2016) synthesized several trypanocidal thiophen-2-iminothiazolidines and used molecular docking to demonstrate that the most active compound likely interacts simultaneously with two subsites of cruzain.
To gain more insight into cruzain dynamics and its interactions with inhibitors, MD simulations have been applied together with other molecular modelling techniques. Durrant et al. (2010) used sequence analysis and MD simulations to explore additional binding sites in cruzain. Hoelz et al. (2016) studied free and liganded cruzain dynamics at acidic pH, which corresponds to its environment in the cell. Very recently, Cianni et al. (2018) also used MD simulations to study the binding mode of reversible covalent inhibitors to the less frequently studied specific subsite S3 of cruzain. Finally, Martins et al. (2018) demonstrated a comprehensive computational approach including docking, MD, ab initio and MM/PBSA calculations for binding mode prediction and estimation of the contributions of specific amino acids to binding. Finally, more mechanistic approaches have been carried out, for example by Arafet et al. (2015Arafet et al. ( , 2017 who studied the MoA of peptidyl-epoxyketone-or peptidyl-halomethyl-ketone-based cruzain inhibitors with a QM/MM method. Rhodesain: For rhodesain, several SBDD efforts were inspired by its structural similarity to human cathepsin. Triazine nitrile-based compounds were studied against rhodesain and extensive SAR analyses were performed to explore the binding preferences of the different subsites, supported by a crystallographic complex of the synthesized compounds with the structurally related human cathepsin L enzyme (Ehmke et al. 2013). Schirmeister et al. (2017) aimed at repurposing human cathepsin-targeting dipeptide nitriles for rhodesain and used covalent docking to support the development of a SAR. Later, Giroud et al. (2018) optimized a set of macrocyclic lactams, developed as human cathepsin L inhibitors, to target T. brucei rhodesain and elucidated their binding modes by crystallographic studies. The initial compound 151 ( Figure 58) had a K i of 11 nM against rhodesain, but also a K i of 10 nM against human cathepsin L. One of the optimized, designed pyrazole derivatives 152 (Figure 58) was found to be about 11 times more effective as an inhibitor of rhodesain than human cathepsin L (K i rhodesain 5.2 nM; K i human cathepsin L 55.7 nM) and to be a potent inhibitor of T. b. rhodesiense with an IC 50 of 0.6 nM.

Trypanothione metabolism
The redox metabolism of parasites like T. brucei relies, in contrast to humans, on the dithiol trypanothione 153 (N1,N8-bis(glutathionyl)spermidine; see Figure 59) being absent in humans and making enzymes involved in this pathway interesting as drug targets for trypanosomatidic diseases Cerami 1992, Leroux andKrauth-Siegel 2016). One of the best studied and genetically validated targets is trypanothione reductase (TrypR; see Figure 59), which is responsible for reducing trypanothione disulfide Cerami 1992, Krieger et al. 2000). In humans, a similar reaction is performed by glutathione reductase, which, however, has an opposite net charge in the active site, facilitating the development of specific inhibitors (Faerman et al. 1996).
Structure-based modelling provided insights into the TrypR pocket regions that could be targeted in drug design approaches. For example, Patterson et al. (2011) designed 3,4-dihydroquinazoline-based inhibitors and solved several crystal structures of T. brucei TrypR. They observed conformational changes upon ligand binding, initiating the formation of a subpocket, which was exploited in further design and yielded compounds with improved TbTrypR and anti-trypanosomal activity: The initial hit compound 154 ( Figure 60) had an IC 50 of 6.8 µM against TbTrypR and an EC 50 against T. brucei of 40 µM, while the best designed compound 155 ( Figure 60) yielded a TbTrypR IC 50 of 0.23 µM and a T. brucei EC 50 of 0.73 µM.
In a later study, based on a series of peptide-based inhibitors, da Rocha Pita et al. (2012) developed receptor-dependent four-dimensional quantitative structure-activity relationship (RD-4D-QSAR) models (Pan et al. 2003) that allowed for identifying TrypR subsites that may be exploited in future inhibitor development.    Figure 61) displayed K i values of 4 ± 0.5 μM against TcTrypR and 12 ± 2 μM against TbTrypR, when assuming a competitive inhibition mechanism. The IC 50 against T. cruzi was 19.0 μM and against T. brucei rhodesiense-3.5 μM, with SI of 8 for L6 cells. However, there was some evidence that TrypR might not be the only target of the tested compounds. In a follow-up project (De Gasparo et al. 2018), again with the help of crystallographic studies, the cyclohexylpyrrolidine inhibitors were further optimized by retaining fragments that bind to the hydrophobic substrate binding site and by increasing the polarity of solvent-exposed moieties to improve aqueous solubility.
On the other hand, structure-based techniques helped in identifying TrypR as a potential target for some compounds active at the parasite level. This was the case for propyl/isopropyl quinoxaline-7-carboxylate 1,4-di-N-oxide-based compounds, active against T. cruzi, which were found to target TrypR by using docking simulations and enzymatic validation (Chaćon-Vargas et al. 2017). Arias et al. (2017) synthesized a set of nitrofuran derivatives, which demonstrated uncompetitive inhibition of TrypR in subsequent experiments. Docking studies indicated that the inhibitors may in fact be capable of binding to the enzymesubstrate complex, thus explaining the observed behavior. Similar observations were previously made for mesoionic 1,3,4-thiadiazolium-2-aminide , which was also found to dock to TrypR in the presence of the substrate molecule. Finally, natural products or natural product-derived compounds, such as alkaloids or neolignan derivatives were proposed as inhibitors of TcTrypR on the basis of in silico screening, docking studies, and 2D-QSAR analysis (Argüelles et al. 2016, Hartmann et al. 2017. Other enzymes of the trypanothione pathway were also investigated as potential drug targets. A mathematical model of the T. cruzi trypanothione pathway was developed by Olin-Sandoval et al. (2012). The results demonstrated that a polypharmacology approach targeting γ-glutamylcysteine synthetase (γECS) and trypanothione synthetase (TrypS) as the enzymes that exert the highest control on the pathway fluxes may be more promising than targeting TrypR. TrypR was found to have less metabolic control and thus requires very high ligand binding affinities to shut the pathway down. Further, by the same approach, tryparedoxin was identified as a suitable potential drug target, since it shows low catalytic efficiency and exerts large metabolic control (González-Chávez et al. 2015).
In line with the above findings, Benıtez et al. (2016) focused on identifying TrypS inhibitors, using target-based screening of a compound library. Furthermore, Vázquez et al. (2017) discovered buthionine sulfoximine as a dual inhibitor of γECS and TrypS, which was further confirmed by the docking and over-expression experiments.
For a more detailed review on the system and currently known inhibitors of both TrypR and TrypS, the reader is referred to Leroux and Krauth-Siegel (2016).

Perspectives and limitations of target-based drug discovery approaches
Many of the examples presented herein demonstrate that it is not always trivial to translate good activities against a specific biomolecular target into anti-parasitic activity. However, target-guided approaches, like crystallography and computational docking studies, provide important guidelines for scaffold optimization and aid the development of SARs. Further, off-target effects can be minimized by the same techniques and comparative studies of, for instance, sequences, structures and molecular interaction fields. Computational studies of target dynamics provide further hints for desirable flexibility profiles of designed molecules. Moreover, mechanistic insights are gained by studying drug-target interactions. Target-based methods are under continuous development and offer great potential for finding suitable starting points for medicinal chemistry programs and the design of specific, selective compounds with a well-defined MoA. The choice of the target and its impact on the pathway to be inhibited is often a critical bottleneck and mathematical models of trypanosomal metabolic pathways may shed light on which enzymes have most control over pathway fluxes and thus may present the most promising targets. In summary, while many interesting compounds have been developed by target-based design strategies, there are limitations to the approach which are best overcome by combining target-guided approaches with screening of the compounds' antiparasitic properties early on in the drug discovery process.

In vivo murine models in trypanosomatidic infections
Murine models, due to their small size, availability of tools for genetic manipulation and immunological studies, are at the forefront of animal use in science (Vandamme 2015). Notwithstanding, the use of animals in scientific research has always been a socially polarizing subject even with the implementation of the three Rs (Replacement, Reduction and Refinement) proposed more than 60 years ago (Russell and Burch 1959). For an animal model to be considered, there must be similarities related to disease etiology, pathophysiology, symptomatology and also response to therapeutic or prophylactic agents. In this sense, animal models have contributed decisively to the understanding of pathophysiological processes associated with infection and disease and have also been instrumental in pre-clinical vaccine and drug development.
For HAT, the animal models are well established and have predictive value for drug development (Field, Horn et al. 2017). These models were further solidified using highly susceptible BALB/c mice and bioluminescent parasites that enable longitudinal evaluation of drug performance (McLatchie, Burrell-Saward et al. 2013, Burrell-Saward, Rodgers et al. 2015. Most models that exist are for stage 1 disease using the non-pathogenic T. b. brucei S427, although T. b. rhodesiense and to a lesser extent T. b. gambiense models also exist (Giroud, Ottones et al. 2009, Muchiri, Ndung'u et al. 2015, Field, Horn et al. 2017). Established models for stage 2 are also available using T. b. brucei GVR35 that induces a slow progressing infection in invasion of the central nervous system (Jennings, Urquhart et al. 1983, Burrell-Saward, Rodgers et al. 2015. For Chagas disease, the situation is quite different, as the translatability of the existing mice models to human disease is not linear as demonstrated by the posaconazole failure (Francisco, Lewis et al. 2015, Molina, Salvador et al. 2015. Many of the standardization issues that affect Leishmania mice models are also present in Chagas disease with completely different infection outcomes depending on the parasite strain, stage, inoculum and inoculation route affecting treatment susceptibility (Chatelain and Konar 2015). Like in HAT, there are distinct animal models proposed for acute and chronic stages of the disease. The models for the acute stage are more common in drug development, which normally use high doses of parasites and are expected to be fatal after one month (Romanha, Castro et al. 2010). These models are typically used to demonstrate lack of efficacy of compounds and are considered not to be appropriate for lead development (Chatelain and Konar 2015). Chronic models also exist, requiring several months to be established with disease outcome being highly dependent on the strains used (Marinho, Bucci et al. 2004, Marinho, Nunez-Apaza et al. 2009). Therefore, these chronic models are considered time consuming and are expensive for drug development, requiring in-depth studies of the tissue tropism of the used strains (Chatelain and Konar 2015). A known technical limitation of the Chagas models was, for a long time, the detection of infection. This was problematic as these parasites are often absent from blood, requiring extensive examination of distinct tissues (Field, Horn et al. 2017). Nonetheless, this difficulty has been somewhat minimized with the use of bioluminescent parasites that, in conjunction with immunosuppressive treatments with cyclophosphamide, enable a better assessment of cure (Lewis, Francisco et al. 2015). Consequently, there is the notion that the existing mice models for Chagas disease have a limited predictive value for drug development, raising the need to improve the available mice models for Chagas (Chatelain andKonar 2015, Field, Horn et al. 2017).
Hamsters can be infected with several Trypanosoma sp and this species has been considered the most satisfactory and economic animal for maintenance of T. b. gambiense infection for experimental studies in laboratories (Lee and Pan 1980). However, as a rule, the use of hamsters as experimental models for HAT has been considerably less relevant compared to mice models (Pink et al. 2005). In the case of T. cruzi infections, hamster is considered a valuable model for Chagas disease studies in both acute and chronic phases of the infection (Ramirez et al. 1994), since they reproduce a range of different outcomes of the disease in humans (Bilate et al. 2008).
Ultimately, no perfect animal models exist but they have been essential to advance our knowledge of pathogenic mechanisms involved in these diseases and also to develop successful control approaches through vaccine and drug development (Pasquali 2018).

conclusion and Perspectives
In the present chapter, we have highlighted key medicinal chemistry approaches to identifying and assessing new hits and optimizing leads for tackling Chagas disease and HAT. The drug discovery approaches to the identification of new drugs for Chagas disease and HAT take advantage of the scope for collaborative work by different stakeholders, including academic research scientists, national and international organizations and governmental initiatives as well as private research centers and pharmaceutical companies. More financial support for the early stages of research and development is needed. Unfortunately, many research programs start and develop early hits and leads but then fall prematurely into the "valley of death" of unused early and late candidate compounds. A major effort devoted to further developing the most valuable compounds identified should be a core part of future research programs.