Anti-Trypanosoma cruzi Activity of Metabolism Modifier Compounds

Chagas disease is caused by the protozoan parasite Trypanosoma cruzi and affects over 6 million people worldwide. Development of new drugs to treat this disease remains a priority since those currently available have variable efficacy and frequent adverse effects, especially during the long regimens required for treating the chronic stage of the disease. T. cruzi modulates the host cell-metabolism to accommodate the cell cytosol into a favorable growth environment and acquire nutrients for its multiplication. In this study we evaluated the specific anti-T. cruzi activity of nine bio-energetic modulator compounds. Notably, we identified that 17-DMAG, which targets the ATP-binding site of heat shock protein 90 (Hsp90), has a very high (sub-micromolar range) selective inhibition of the parasite growth. This inhibitory effect was also highly potent (IC50 = 0.27 μmol L−1) against the amastigote intracellular replicative stage of the parasite. Moreover, molecular docking results suggest that 17-DMAG may bind T. cruzi Hsp90 homologue Hsp83 with good affinity. Evaluation in a mouse model of chronic T. cruzi infection did not show parasite growth inhibition, highlighting the difficulties encountered when going from in vitro assays onto preclinical drug developmental stages.


Introduction
Chagas disease (or American Trypanosomiasis) is caused by the protozoan parasite Trypanosoma cruzi (T. cruzi). The disease was discovered more than 110 years ago, but it remains an important public health issue in Latin America. Moreover, the number of detected cases in nonendemic regions of Northern America and Europe has increased due to population movements during the last decades. Nowadays, it is estimated that over 6 million people are affected by the disease which causes more than 10,000 deaths per year [1].
T. cruzi infection clinically progresses in two well differentiated stages. First there is a short acute phase (4-8 weeks long) that is mostly asymptomatic and thus frequently undiagnosed and untreated. After these initial stages, the parasite enters into a persistent and long-lasting chronic phase that may remain asymptomatic or cause cardiac and/or digestive tissue damage in approximately 30% of patients, leading to years of disability and eventually death if left untreated [2,3].

Anti-T. cruzi Activity
Compound activity against T. cruzi mammalian stages was determined using a phenotypic assay based on African green monkey kidney epithelial cells (Vero, ATCC ® CCL-81™) infected with an engineered T. cruzi-Tulahuen strain (DTU VI) expressing beta-galactosidase [21,22]. In every run of the assay we included the reference drug BNZ as control. This had an average IC50 value of 1.63 (0.11) μmol L −1 (Figure 2) which is similar to previous reports [22,23].

Anti-T. cruzi Activity
Compound activity against T. cruzi mammalian stages was determined using a phenotypic assay based on African green monkey kidney epithelial cells (Vero, ATCC ® CCL-81™) infected with an engineered T. cruzi-Tulahuen strain (DTU VI) expressing betagalactosidase [21,22]. In every run of the assay we included the reference drug BNZ as control. This had an average IC 50 value of 1.63 (0.11) µmol L −1 (Figure 2) which is similar to previous reports [22,23].

Figure 2.
Anti-Trypanosoma cruzi Assay Dose-Response Curves. Graphs represent mean results and standard deviation (SD) of at least three independent biological replicas.

Identification of Compounds with Specific Activity Against the Parasite
A secondary cell toxicity assay with monkey Vero cells was performed to test for parasite specificity of dorsomorphin and 17-DMAG compared to host cells. Oligomycin was discarded from further evaluation because it is highly toxic to human subjects [24].
Since compounds under study were metabolic inhibitors, we evaluated their impact on host cells viability with two different methodologies: AlamarBlue and crystal violet assays (see Section 4.5 for details). BNZ, included as reference in the Vero toxicity assays, yielded an average TC50 value of 243.8 (17.04) μmol L −1 in the AlamarBlue-based assay and 140.2 (13.18) μmol L −1 in the crystal violet assay as reported (Table 1, Figure S3) [22,23]. In order to consider any of the two compounds suitable for further progression

Identification of Compounds with Specific Activity against the Parasite
A secondary cell toxicity assay with monkey Vero cells was performed to test for parasite specificity of dorsomorphin and 17-DMAG compared to host cells. Oligomycin was discarded from further evaluation because it is highly toxic to human subjects [24].
Since compounds under study were metabolic inhibitors, we evaluated their impact on host cells viability with two different methodologies: AlamarBlue and crystal violet assays (see Section 4.5 for details). BNZ, included as reference in the Vero toxicity assays, yielded an average TC 50 value of 243.8 (17.04) µmol L −1 in the AlamarBlue-based assay and 140.2 (13.18) µmol L −1 in the crystal violet assay as reported (Table 1, Figure S3) [22,23]. In order to consider any of the two compounds suitable for further progression we determined a selectivity index (SI, i.e., the ratio between TC 50 and IC 50 ) > 10 in both readouts. Such SI ratio is commonly used to consider compounds with specific antiparasitic activity [25].
we determined a selectivity index (SI, i.e., the ratio between TC50 and IC50) > 10 in both readouts. Such SI ratio is commonly used to consider compounds with specific antiparasitic activity [25].

Anti-Amastigote Specific Activity of 17-DMAG
Amastigotes are considered the main target for any prospective drug to treat chronic T. cruzi infections. Since 17-DMAG activity was determined to be anti-parasite specific, we assessed its anti-amastigote activity by means of a biological assay specifically targeting the mammalian replicative form of the parasite. In this anti-amastigote assay, pre-infected cells from which trypomastigotes had been washed away were exposed to the compounds after allowing the intracellular amastigotes 18 h to settle. The standard drug BNZ was included for comparison. Overall, we found that the anti-amastigote activity of 17-DMAG was ≈15× less potent than what we had retrieved in the antiparasitic assay previously described: IC50 = 0.27 (0.017) μmol L −1 versus IC50 = 0.017 (0.001) μmol L −1 (Table 1, Figure 4). Anti-amastigote IC50 of BNZ was 2.02 (0.08) μmol L −1 versus 1.63 (0.11) μmol L −1 . Despite this reduction in the activity levels of 17-DMAG against the amastigote forms, its SI window over Vero cells remained > 10 ( Table 1). 17-DMAG had an average TC 50 value of 6.23 (0.86) µmol L −1 against Vero cells in the AlamarBlue assay (SI = 366.5; Table 1, Figure 3). Then, although crystal violet readout resulted again in higher toxicity [TC 50 = 2.97 (0.37) µmol L −1 ] (Table 1, Figure 3), its SI prevailed > 10 (SI = 174.7). In fact, 17-DMAG showed wider selectivity indexes than those obtained with the reference drug BNZ (Table 1).

Anti-Amastigote Specific Activity of 17-DMAG
Amastigotes are considered the main target for any prospective drug to treat chronic T. cruzi infections. Since 17-DMAG activity was determined to be anti-parasite specific, we assessed its anti-amastigote activity by means of a biological assay specifically targeting the mammalian replicative form of the parasite. In this anti-amastigote assay, pre-infected cells from which trypomastigotes had been washed away were exposed to the compounds after allowing the intracellular amastigotes 18 h to settle. The standard drug BNZ was included for comparison. Overall, we found that the anti-amastigote activity of 17-DMAG was ≈15× less potent than what we had retrieved in the antiparasitic assay previously described: Figure 4). Antiamastigote IC 50 of BNZ was 2.02 (0.08) µmol L −1 versus 1.63 (0.11) µmol L −1 . Despite this reduction in the activity levels of 17-DMAG against the amastigote forms, its SI window over Vero cells remained >10 (Table 1).

Hsp90 is the Target of 17-DMAG
We further inquired if the observed antiparasitic effect of 17-DMAG could be related to the inhibition of the mammalian Hsp90 or if the compound could also be targeting T.

Hsp90 Is the Target of 17-DMAG
We further inquired if the observed antiparasitic effect of 17-DMAG could be related to the inhibition of the mammalian Hsp90 or if the compound could also be targeting T. cruzi Hsp90 homologue Hsp83 [26]. Previous studies have described that 17-DMAG binds with higher affinity to the N-terminal domain of Hsp83 proteins from T. cruzi close relatives Trypanosoma brucei and Leishmania amazonensis, as well as to that of Plasmodium falciparum in comparison to human Hsp90 [27][28][29][30]. Multiple sequence alignment showed a high percentage of identity (>60%) between human Hsp90 and its parasite homologues [31] ( Figure S4). Particularly, identity of N-term domains, where the ATP-binding pocket is located, was 73% between human Hsp90 and T. cruzi Hsp83 ( Figure S4).
Since the 3D structure of human Hsp90 bound to 17-DMAG is available [32], we performed a morphological analysis of the interaction between this compound and T. cruzi Hsp83 protein with PyMol [33]. It revealed a different surface conformation of the ATP-binding pocket between both chaperones ( Figure 5). Alignment-driven in silico mutagenesis entailed substituting human Hsp90 lysine (K) 112 for an arginine (R) 97 residue in T. cruzi Hsp83. In Hsp90 protein, the amine group of L112 forms a hydrogen bond of 3 Å with the C-18 ketone of 17-DMAG [32]. In contrast, the two amine groups of R97 in T. cruzi Hsp83 would interact with the C-18 ketone upon the formation of two bonds of 2.8 and 2.9 Å ( Figure 5). Substitution of isoleucine (I) 91 onto a valine (V) did not seem to affect the water-mediated interaction.
In addition to the visualization of 17-DMAG interaction with the ATP-binding pockets of Hsp90 and Hsp83, we evaluated the binding affinity of 17-DMAG to both chaperones with Vina Wizard software [34]. This analysis predicted a free energy value of −9.7 Kcal/mol for the binding of 17-DMAG to Hsp90, whereas it was −6.2 Kcal/mol for T. cruzi Hsp83.

In Vivo Anti-T. cruzi Activity of 17-DMAG in a Mouse Model of Chronic Infection
Ultimately, antiparasitic activity of 17-DMAG was assessed in an in vivo model of chronic T. cruzi infection with a transgenic parasite expressing luciferase (Brazil strain, DTU I) [35,36]. In this assay, the signal provided by luciferase after injection of its substrate in the animals is proportional to the T. cruzi load, and it is considered a surrogate for parasitemia. Infected mice reached the peak of parasitemia at day 23 and treatment started at day 132 once chronic infection had been established ( Figure 6A). One experimental group was treated intraperitoneally with 30 mg/kg/day of 17-DMAG for five days ( Figure 6A,B), while other was treated five times every three days (over a 15 days period) with same dose ( Figure 6A,C). We used these dosage and regimes following previous works by Meyer and coworkers and Santos and coworkers, respectively [37,38]. Vehicle-treated and BNZ-treated groups (same regimes and dose: 30 mg/kg/day) were included as controls. None of the groups treated with 17-DMAG showed statistically significant differences versus either vehicle or BNZ groups. Similarly, no statistically significant differences in the registered luciferase signal were observed between daily or intermittently treated mice, either with 17-DMAG or BNZ (data not shown). After treatment, mice treated daily and those treated intermittently with 17-DMAG, respectively had mean luminescence values (AU) of 1.95 × 10 6 (±5.60 × 10 5 ) and 8.69 × 10 6 (±1.27 × 10 7 ) photons s −1 . Such values were higher than the 5 × 10 5 photons s −1 threshold indicative of parasite clearance that had been determined by imaging a noninfected mouse injected with luciferase substrate the same as infected [39].  [32]. 17-DMAG structure is represented in green and protein structure in grey. In panel A, arrows indicate morphological differences between both surfaces. In panel C, arrows show the position of expected H-bonds between 17-DMAG and K112 or R97 residues, respectively. Hsp90/83 from panel B is represented as cartoons; in panel C they are represented as sticks [33].
In addition to the visualization of 17-DMAG interaction with the ATP-binding pockets of Hsp90 and Hsp83, we evaluated the binding affinity of 17-DMAG to both chaperones with Vina Wizard software [34]. This analysis predicted a free energy value of −9.7 Kcal/mol for the binding of 17-DMAG to Hsp90, whereas it was −6.2 Kcal/mol for T. cruzi Hsp83.

In Vivo Anti-T. cruzi Activity of 17-DMAG in a Mouse Model of Chronic Infection
Ultimately, antiparasitic activity of 17-DMAG was assessed in an in vivo model of chronic T. cruzi infection with a transgenic parasite expressing luciferase (Brazil strain, DTU I) [35,36]. In this assay, the signal provided by luciferase after injection of its sub- At the dose used (30 mg/kg/day), BNZ did not manage to completely clear the infection. Nonetheless, in the group of animals that received the drug intermittently over 15 days, there was a 6-fold significant reduction of parasitemia after treatment (p < 0.01, comparing levels at days 132 and 147) ( Figure 6C). Finally, it must be noted that mice under the 17-DMAG daily regime showed signs of treatment associated toxicity. These were not observed in the animals that were intermittently treated with the compound. None of the animals treated with BNZ showed signs of drug-associated toxicity. nificant differences in the registered luciferase signal were observed between daily or intermittently treated mice, either with 17-DMAG or BNZ (data not shown). After treatment, mice treated daily and those treated intermittently with 17-DMAG, respectively had mean luminescence values (AU) of 1.95 × 10 6 (±5.60 × 10 5 ) and 8.69×10 6 (±1.27 × 10 7 ) photons s −1 . Such values were higher than the 5 × 10 5 photons s −1 threshold indicative of parasite clearance that had been determined by imaging a noninfected mouse injected with luciferase substrate the same as infected [39].

Discussion
Out of the nine metabolic inhibitors evaluated, three showed a highly potent antiparasitic activity: dorsomorphin, 17-DMAG, and oligomycin. The latter was included to test that a functional host cell mitochondria was important for the parasite survival, but it was discarded from further progression for its reported acute toxicity.
Cell toxicity assessment of the other two active compounds revealed that only 17-DMAG was specific against the parasite. This was concluded upon performing two complementary cell toxicity assays with AlamarBlue and crystal violet substrates [40,41], which measure cell metabolic activity and membrane integrity, respectively. No significant differences in 17-DMAG TC 50 values were recorded, and the fact that toxicity was slightly higher in the crystal violet assay might have been due to its distinct readout addressing membranes integrity.
Dorsomorphin (also known as compound C) was originally identified as an AMPK inhibitor and widely used in cancer studies [11]. In this study it showed a high antiparasitic activity in the anti-T. cruzi assay, but it was also shown to be highly toxic against Vero cells [TC 50 = 0.26 (0.032) µmol L −1 ] in the crystal violet based cell toxicity assay. Flagged as nonspecific against the parasite, it was discarded from subsequent analysis.
Regarding 17-DMAG, to the best of our knowledge, this is the first time that its anti-T. cruzi growth inhibition activity has been described. It was highly potent, in the submicromolar range (IC 50 = 17 nmol L −1 ), in the primary antiparasitic assay. Moreover, its activity was determined to be parasite specific also in the anti-amastigote assay. That 17-DMAG potency was lower in the latter might be due to amastigotes being less accessible to it as intracellular forms. Nonetheless, still a strong submicromolar antiparasitic activity was registered (IC 50 = 0.17 µmol L −1 ).
17-DMAG is a derivative of geldanamycin, which has been reported to inhibit the activity of the Hsp90 protein family [16], and to antagonize HIF1α transcription factor [17] ( Figure 1). Amongst different Hps90 inhibitors like geldanamycin, 17-AAG or 17-DMAG, the latter has been chosen in multiple assays due to its potency, selectivity, and better water solubility [37]. Hsp90 proteins are chaperones modulating availability of client proteins (kinases and nuclear receptors) thus enabling important cellular processes such as cell growth, signal transduction, and development [29]. Eukaryotic cells have three types of Hsp90s: cytosolic Hsp90 with two isoforms (Hsp90α and Hsp90β), Grp94 (glucoseregulated protein 94) at the endoplasmic reticulum (ER), and the mitochondrial Trap1 (tumor necrosis receptor-associated protein 1) [16]. The sequence of Hsp90 proteins is rather conserved and its activity has been documented also in intracellular protozoan parasites such as those within genus Trypanosoma, Leishmania, Toxoplasma, and Plasmodium [37,42]. Three functional domains are found within Hsp90s: a N-term domain with an ATP-binding site and ATPase activity; a middle region where the binding site for client proteins is located; and a homo-dimerization domain with another binding site for ATP at their C-term [16] ( Figure 1; Figure S4). Remarkably, structural studies have shown that compound 17-DMAG binds to the ATP-binding site of the N-term domain [32]. Others have investigated the use of 17-DMAG against various human pathogens [28,30,37,43]. Palma and coworkers reported a higher selectivity of 17-DMAG against L. amazonensis cells (IC 50 = 86.1 nmol L −1 ) than to THP−1 host cells (TC 50 = 10.6 µmol L −1 ) [28]. Similarly, Meyer and coworkers demonstrated its higher selectivity against T. brucei (IC 50 = 3.1 nmol L −1 ) in comparison to L1210 mouse cells (TC 50 = 1000 nmol L −1 ) [37], and P. falciparum growth was also suppressed by 17-DMAG at nontoxic concentrations to the host cells [30]. Notably, in vivo studies have already shown that 17-DMAG, administered either intraperitoneally or orally at a supposedly nontoxic dose of 30 mg kg −1 for five days cured mice of a lethal T. brucei infection decreasing the parasitemia to the limit of detection by day three [37].
Besides, 17-DMAG has successfully reached phase I clinical studies in patients with advanced solid cancers [44][45][46] and lymphocytic and myeloid leukemia [47]. The highest dose reached in these phase I studies without experimenting dose-limiting toxicity was 80 mg/m 2 , even though lower doses were also defined as effective [45,46]. Most common adverse events were fatigue, nausea, vomiting, anorexia, gastrointestinal, liver function changes, and ocular alterations [44,45]. The maximum blood tolerated concentrations were 600-2700 nmol L −1 [37], which would exceed the 17-270 nmol L −1 IC 50 range we determined against T. cruzi in this work.
T. cruzi Hsp90 homologue Hsp83 participates in the parasite cell cycle by controlling the differentiation of insect to mammalian infective forms [26]. It has also been described to have high ATPase activity, particularly stimulated in the presence of peptides [26]. Since a 73% sequence conservation was observed between N-term of T. cruzi Hsp83 and human Hsp90 [48], we decided to follow an approach to visualize a hypothetical Hsp83-17-DMAG interaction. Somewhat, in silico results unveiled a more embracing binding pocket surface to 17-DMAG in the modeled T. cruzi Hsp83 in comparison to human Hsp90 ( Figure 5). The model also showed that the unique interaction of 17-DMAG C-18 ketone with human Hsp90 K112 was substituted by two bonds with T. cruzi Hsp83 R97, which would have an expected shorter binding distance from 3Å to 2.8 and 2.9 Å, respectively. Such differences may allow a swifter binding of 17-DMAG to Hsp83 N-term domain. In other related parasites, Palma and coworkers reported a higher affinity of 17-DMAG to L. amazonensis Hsp83 than to human Hsp90 based on in vitro assays and docking-structural analysis [28]. Additionally, thermal shift assays showed an affinity up to two-fold higher for T. brucei Hsp83 than to human Hsp90 [28]. However, the predicted energy required to disrupt 17-DMAG binding to T. cruzi Hsp83 was lower than that predicted for the human Hsp90 (∆G = −6.2 Kcal/mol (−25.9 KJ/mol), versus ∆G = −9.7 Kcal/mol (−40.6 KJ/mol)). This would indicate a lower binding affinity between 17-DMAG and Hsp83. The predicted binding affinity value of 17-DMAG to the human Hsp90 correlates with that previously reported (∆G = −32.3 KJ/mol) [49]. Our Vina Hsp83 model did not take into consideration the presence of solvent in the interface and the conformational changes that occur during the complex formation. It has been described that water-mediated hydrogen bonds between Hsp90 inhibitor compounds and amino acid residues within N-term domain of T. brucei Hsp83 were important as they simulated the positioning of one of the phosphate groups of ATP [27]. Furthermore, Nilapwar and coworkers reported the importance of watermediated hydrogen bonds by direct bridging or by forming a network of water molecules in human Hsp90 complex formation with geldanamycin, 17-AAG and 17-DMAG [49], and it is also known that conformational changes coupled to ligand binding can contribute to modify the binding affinity [49]. Whether 17-DMAG in vitro antiparasitic effect is due to an inhibition of the host cells Hsp90 or T. cruzi Hsp83 remains to be elucidated. Moreover, since 17-DMAG has also been described to inhibit host cell glycolysis route through HIF1α, a change in the metabolic state of the cells might as well be having an impact in the parasite intracellular multiplication.
The potent and specific activity of 17-DMAG against T. cruzi in vitro, and its formerly described in vivo activity against closely related parasites [37,38], prompted its evaluation in a chronic model of T. cruzi infection. In this, 17-DMAG did not inhibit parasite presence in the infected mice. We did not observe a statistically significant decrease in parasitemia in mice treated intraperitoneally with 17-DMAG at 30 mg/kg daily for five days or those intermittently treated every three days over 15 days (five times in total) in comparison with vehicle inoculated animals ( Figure 6). Neither was that presence eliminated by similar treatment (dosage and regime) with Chagas disease standard drug BNZ. Nonetheless, decay in parasitemia was indeed observed in those animals intermittently treated with BNZ, likely due to longer availability of the drug to act and contribute to reduce parasite presence. Clearance of T. cruzi parasitemia with BNZ in chronically infected mice has been described upon administration of ≥100 mg/kg/day over at least 20 days [50][51][52]. Reason to include a control group of animals treated with 30 mg/kg/day was to parallel test groups in regime and dose.
Administering 17-DMAG at a higher dose than the one evaluated here may not be an option in future experiments, at least in its daily administration regime and in this mouse strain. Among the observed signs of toxicity in mice within daily treated group, there were lethargy and ruffled fur. Formerly, Meyer and coworkers did not report signs of toxicity in T. brucei infected mice treated with the same dose [37]. This could be explained by the use of different mouse strains [53], as well as by the immunological compromise caused by each parasite when treatment started. Herein reported failure of 17-DMAG in a chronic in vivo model reminds that of posaconazole, which also presented a very potent in vitro anti-T. cruzi activity [54]. BNZ and NFX remain as the best possible Chagas disease treatments at the moment. It will be fundamental to determine whether alternative regimes currently under clinical evaluation manage to reduce their frequent adverse effects while maintaining (or increasing) efficacy [9]. Until then, efforts to try to identify a safe and highly effective therapy should continue. AICAR, Akti-1/2, dorsomorphin, etomoxir, and SC79 stock dilutions were prepared in DMSO at 20 mM whereas oligomycin and 17-DMAG were prepared at 2 and 1 mmol L −1 , respectively. The drug sodium salicylate was prepared before use in the appropriate culture media at a starting concentration of 200 µmol L −1 .

Culture of T. cruzi Parasites
T. cruzi parasites from the Tulahuen strain (Discrete Typing Unit VI) expressing βgalactosidase were kindly provided by Dr. Fred Buckner (University of Washington, Seattle, WA, USA) and maintained using LLC-MK2 cells as hosts in DMEM supplemented with 2% FBS and 1% P-S-G [21]. Free-swimming trypomastigotes were purified as described to keep the parasite cycle in LLC-MK2 cells and for the performance of the antiparasitic assays [22].

Assay to Detect T. cruzi Growth Inhibition in 96-Well Plates
Our assay is based on Vero cells as hosts and infective trypomastigotes from Tulahuen strain that express the bacterial β-galactosidase enzyme as reporter activity [21]. Firstly, compounds were added in the first column of a 96-well tissue culture treated plates at an initial concentration of 200, 20, or 10 µmol L −1 and diluted in assay medium following either a 1:2 or 1:3 fold pattern to conform dose-response plate-maps. Vero cells and purified trypomastigotes were harvested, counted, and diluted to a concentration of 1 × 10 6 cells/mL each. Then these were mixed and 100 µL added per well (50,000 Vero cells and 50,000 trypomastigote cells per well; multiplicity of infection or MOI = 1) [56]. The reference drug BNZ was used as control of drug growth inhibition in each run launched, whereas each plate contained its own negative and positive control as previously described [21,56]. After 4 days at 37 • C, 50 µL of a PBS solution containing 0.25% NP40 and 500 µM chlorophenol red-β-D-galactoside (CPRG) substrate were added per well [22]. Plates were further incubated at 37 • C for another 4 h and the absorbance read out at 590 nm using an Epoch Gene5 spectrophotometer. All experiments were performed at least in triplicate.
Additionally, based on Vero cells as hosts and the recombinant T. cruzi strain expressing β-galactosidase, we further adapted the antiparasitic assay described above to determine whether the activity of the compound 17-DMAG was specific against the intracellular amastigote forms. For this, Vero cells were seeded in T-175 flasks (5 × 10 6 cells/flask) in DMEM supplemented with 1% penicillin-streptomycin and 10% FBS and cultured for 24 h. Then, cells were washed once with PBS and free swimming trypomastigotes (1 × 10 7 trypomastigotes per flask; MOI = 1) in assay medium were added and allowed 18 h to infect. Infected cell monolayers were washed with PBS and detached from the flasks [23]. Cells were counted and diluted to a concentration of 5 × 10 5 cells per mL, before adding 100 µL per well to test plates already containing the drugs as previously described. Each plate contained the same controls as previously described. Test plates were incubated for 72 h and their read out was performed as described above.

Cell Toxicity Assays
In this case compounds were added to tissue culture treated 96-well plates following a dose-response dilution pattern 1:2 or 1:3 with starting concentrations of 800 µmol L −1 for dorsomorphin and 64.8 µmol L −1 for 17-DMAG. Cell viability was checked upon cell counting with Trypan blue staining and we only proceeded if it was >95%. Vero cells were diluted at a concentration of 5 × 10 5 cells per mL, before adding 100 µL per well. Each run contained its own negative and positive controls as described. Plates were incubated at 37 • C for 4 days and we used two different readout methodologies. One was based on the traditional AlamarBlue substrate (resazurin-based), which provides a quantitative measure of cells viability by their capacity to reduce resazurin into resorufin [40]. However, given that inhibitors may impact cells redox state, a distorted toxicity profile could have been obtained, and a parallel readout relying on crystal violet was also used. Vero cells grow as monolayers, and adherent cells detach from the tissue culture plate upon cell death. Thus crystal violet dye would only bind to structures in living cells [41]. In the case of AlamarBlue assays, we added 50 µL per well of a PBS solution containing 10% AlamarBlue reagent (ThermoFischer, Waltham, MA, USA) and incubated the plates for 6 h at 37 • C before reading the fluorescence intensity in a Tecan Infinite M Nano+ reader (excitation: 530 nm, emission: 590 nm). In the case of crystal violet assays, cells were washed with PBS and fixed with 100 µL of 10% formol for 15 min. Then, after washing again we added 50 µL of a crystal violet dilution (1:5 v/v) prepared from the stock and left it for 10 min. Crystal violet stock solution contained 250 mg of crystal violet powder (Sigma-Aldrich), 1 mL of EtOH, and 49 mL of H 2 Od. After that, crystal violet excess was eliminated washing with H 2 Od two times and plates were left to dry in the flow cabinet. Finally, we added 50 µL of SDS 2% for 30 min and read out absorbance at 590 nm in an Epoch Gene5 spectrophotometer. All experiments were performed at least in triplicate.

In Silico Study of 17-DMAG Binding to Hsp90/83 N-Terminal Domains
Hsp90 N-term domain was visualized with PyMol software (version 4.4.0) [33]. We used PyMol as well to visualize T. cruzi Hsp83 N-term domain by means of an alignmentdirected mutagenesis upon the N-term domain of the cocrystalized human Hsp90 and 17-DMAG complex (PDB: 1OSF) [32]. We used BLAST [31] to identify the residues that differ between human Hsp90α isoform 2 (NCBI ref.: NP_005339.3) and Hsp83 of different T. cruzi strains. These were obtained from NCBI and TriTrypDB [57,58]. We chose T. cruzi Hsp83 from Dm28c strain as representative (NCBI ref.: PBJ77530.1) for its visualization in PyMol, but the same identity was shown between Hsp83 from other T. cruzi strains and human Hsp90 ( Figure S4).
We used Vina Wizard software (on PyRx-virtual screening tool) [34] to calculate the binding affinity of 17-DMAG to Hsp90/83. The structure of 17-DMAG bound to the N-term domain of human Hsp90 was obtained from the cocrystalized complex. On the other hand, the N-term domain of T. cruzi Hsp83 was modeled using the SWISS-MODEL homology-modeling tool [59] by inputting the T. cruzi Dm28c amino acid sequence. This model was constructed based on T. brucei Hsp83 3D-structure available from the RCSB Protein Data Bank (PDB: 3O6O). These Hsp83 proteins shared a 90% identity, which was 94% between their N-term domains.

T. cruzi Chronic In Vivo Assay
Trypomastigotes from transgenic T. cruzi Brazil strain (DTU I) expressing firefly luciferase were purified, diluted in PBS, and injected in Balb/c mice (10 3 trypomastigotes per mouse) [39]. Parasites from Brazil strain have been described to be highly virulent and susceptible to BNZ [60]. At defined time points, mice were anesthetized by inhalation of isoflurane and injected with 150 mg/kg of D-Luciferin Potassium-salt (Goldbio, St. Louis, MO, USA) dissolved in PBS [35]. Mice were imaged 5-10 min after injection of luciferin with an IVIS 100 (Xenogen, Alameda, CA, USA), and the data acquisition and analysis were performed with the software LivingImage (Xenogen). On day 132 post-infection, treatment was started by intraperitoneal administration of 17-DMAG at 30 mg/kg. Two schedules of treatment were followed: one experimental group was treated daily for five days [37]; the other was treated five times over 15 days [38]. Each group had its positive control group which received a similar dose of 30 mg/kg of BNZ on corresponding days. Negative control group received only vehicle solution [35]. Each group was formed by five mice. On the scheduled days 7, 14, 23 132, 141, and 147 mice were imaged after anesthesia and injection of luciferin as described above.

Data Analysis
Absorbance and fluorescence values respectively derived from the anti-T. cruzi and cell toxicity assays were normalized to the controls [56]. IC 50 and TC 50 values were determined with GraphPad Prism 7 software (version 7.00, 2016) using a nonlinear regression analysis model defined by Equation (1): IC 50 and TC 50 values are, respectively, defined as the compound concentration that inhibits parasite growth by 50% or is toxic to 50% of cells. Values provided are means and standard deviation [mean (SD)] of at least three independent experiments. Similarly, total flux values recorded in the in vivo experiments are shown as mean and SD. We used ANOVA with Bonferroni's correction for multiple comparisons to determine if the differences between groups were statistically significant.
Comparison of luciferase activity levels in T. cruzi-infected mice after treatment was performed between days 132 (day of treatment) and days 141 (for 5 days treatment) or 147 (for 15 days intermittent treatment) by Mann-Whitney test.

Conclusions
Upon evaluation of nine metabolic modifier compounds we identified one with selective anti-T. cruzi activity in the sub-micromolar range: 17-DMAG. It had potent IC 50 (0.017 µmol s −1 ) and good SI rates (SI = 366.5 and 174.7, for the AlamarBlue and crystal violet Vero assays, respectively). In addition, the in silico directed mutagenesis predicted a hypothetical better binding to T. cruzi Hsp83. Regrettably, 17-DMAG did not present parasite inhibitory activity in an in vivo chronic model of T. cruzi infection.

Institutional Review Board Statement:
This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health, USA. The protocol was approved by the Institutional Animal Care and Use Committee of New York University School of Medicine, which is fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC).

Data Availability Statement:
The data presented in this study are available in the article or supplementary material.

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