In Vitro and in Vivo Activity of mTOR Kinase and PI3K Inhibitors Against Leishmania donovani and Trypanosoma brucei.

Kinetoplastid parasites, including Leishmania and Trypanosoma spp., are life threatening pathogens with a worldwide distribution. Next-generation therapeutics for treatment are needed as current treatments have limitations, such as toxicity and drug resistance. In this study, we examined the activities of established mammalian target of rapamycin (mTOR)/phosphoinositide 3-kinase (PI3K) inhibitors against these tropical diseases. High-throughput screening of a library of 1742 bioactive compounds against intracellular L. donovani was performed, and seven mTOR/PI3K inhibitors were identified. Dose-dilution assays revealed that these inhibitors had half maximal effective concentration (EC50) values ranging from 0.14 to 13.44 μM for L. donovani amastigotes and from 0.00005 to 8.16 μM for T. brucei. The results of a visceral leishmaniasis mouse model indicated that treatment with Torin2, dactolisib, or NVP-BGT226 resulted in reductions of 35%, 53%, and 54%, respectively, in the numbers of liver parasites. In an acute T. brucei mouse model using NVP-BGT226 parasite numbers were reduced to under the limits of detection by five consecutive days of treatment. Multiple sequence and structural alignment results indicated high similarities between mTOR and kinetoplastid TORs; the inhibitors are predicted to bind in a similar manner. Taken together, these results indicated that the TOR pathways of parasites have potential for the discovery of novel targets and new potent inhibitors.


Introduction
Neglected tropical diseases including leishmaniasis, human African trypanosomiasis (HAT, or sleeping sickness), and Chagas disease are caused by the pathogenic protozoans Leishmania spp., Trypanosoma brucei subspecies, and Trypanosoma cruzi, respectively. Globally, there are nearly one hundred thousand deaths and over 22 million reported cases associated with these diseases annually [1,2]. There are three types of leishmaniasis: cutaneous leishmaniasis (CL), mucocutaneous leishmaniasis, and visceral leishmaniasis (VL) [3,4]. The disease is distributed throughout 97 countries, mainly in Africa, Asia, and Latin America, but the fatal form of VL mostly occurs in Ethiopia, South Sudan, India, Bangladesh, and Brazil [5,6]. Current treatment options include sodium stibogluconate, which is only available for administration via injection; amphotericin B, which is repurposed from antifungal treatment; and miltefosine, being the only orally administered drug [7]. and showed efficacy against T. brucei in an animal model [37]. In a CL animal model, it was found that treatment with rapamycin and GSK-2126458 resulted in significant decreases in footpad swelling and parasitemia in the draining lymph nodes in infected animals [38]. These compounds are shown to kill parasites directly. On the other hand, Khadem et al. have demonstrated that selective pharmacological inhibition of the host PI3Kδ with CAL-101 results in significant reductions of parasite burdens in VL and CL animal models [40]. These findings suggest that mTOR/PI3K inhibitors have potential applications for treatment of kinetoplastid diseases.
In this study, we report the results of an in vitro screening of bioactive compounds against intracellular L. donovani. We identified new mTOR/PI3K inhibitors for further assessment for use on different forms of kinetoplastid parasites. Potent compounds in vitro were subjected to in vivo models of VL and HAT. Structural modeling of identified inhibitors binding to the TORs of kinetoplastid parasites were predicted to provide insights on the potential mode of inhibition at molecular level.

Intracellular Leishmania Pilot Screening
We used a high-content screening system to assess a library of 1742 compounds with potential bioactivity against intracellular Leishmania. Human acute monocytic leukemia (THP-1) cells infected with L. donovani were treated with compounds; images were analyzed to quantify the numbers of parasites and host cells and infection ratios. The reproducibility and robustness of the assay was evaluated by duplicate runs. The correlation between two independent screenings indicated a high degree of linear relationship, with an R 2 value of 0.922. There were 74 compounds that had activity in set 1 or set 2 (threshold > 60%), and 32 with activity in both sets ( Figure 1A). To evaluate assay quality, the Z factor was calculated using the average values from duplicate testing. The Z factor was given by Z = 1 − 3(σ c + + σ c − )/(µ c + − µ c − ), where σ c + /σ c − were the standard deviation values of the positive/negative controls and µ c + /µ c − were the corresponding mean values. The Z value of the screening using the infection ratio was 0.798; this result indicated excellent assay quality ( Figure 1B). The screening results indicated that 50 compounds were active, based on a >60% threshold of the inhibition of intracellular parasite survival ( Figure 1B). Since this number included the compounds showing activity due to killing of the host cells (low selectivity), we plotted the parasite survival inhibition versus the host cell viability to further filter parasite-selective compounds. With >60% parasite survival inhibition and >60% host cell viability, a final number of 20 compounds was selected out of the 1742 with the overall hit rate of 1.15% ( Figure 1C).

Hit Characteristics and Selection of mTOR/PI3K Inhibitor
Among the selected 20 hits, five compounds were in the category of mTOR/PI3K inhibitors. In the screening library, 58 mTOR/PI3K inhibitors were present, and by proportion, 8.62% (= 5/58) was found active within this category of inhibitor. This rate is relatively higher compared to the overall hit rate of 1.15% ( Figure 1D). The mTOR complex inhibitors, such as rapamycin and its analogs deforolimus, temsirolimus, and everolimus, were included in the library, but did not show potent inhibition at the screening concentration (10 µM). Previously reported antileishmanial compounds, such as dactolisib (NVP-BEZ235) [37] and GSK2126458 [39], were found active, but the latter one showed some toxicity against the host cell. The compounds known to exhibit activity in vivo via modulation of the host immune response (i.e., CAL-101) had no direct effects on the parasite [38].
With two additional mTOR/PI3K inhibitors at the border of threshold, a total of 22 compounds were subjected to dose-dilution assays for activity confirmation. In the result, compounds such as dopamine uptake inhibitors (GBR12909 and 3-CPMT), protein kinase A inhibitor (H89), VEGFR inhibitor (cediranib and ENMD-2076), and EGFR inhibitor (afatinib) had either low activity (half maximal effective concentration, EC 50 ) or high toxicity (CC 50 ). Other compounds, including Y29794, metaphit, Ro 106-9920, SKF 96365, and NNC05-2090 were confirmed to be active, but did not group together into a category of inhibitor. In the case of the mTOR/PI3K inhibitors, all the compounds were well-confirmed with potent activities and sufficient selectivity index (SI) values (Table 1). Based on these results, we sought to focus the subsequent investigations on this class of inhibitors ( Figure 2). well-confirmed with potent activities and sufficient selectivity index (SI) values (Table 1). Based on these results, we sought to focus the subsequent investigations on this class of inhibitors ( Figure 2).

In Vitro Activity Against Kinetoplastid Parasite
The confirmatory intracellular Leishmania assays found that the tested mTOR/PI3K inhibitors had EC 50 values, ranging from 0.14 to 13.44 µM (Table 1 and Figure 3). The most potent compound was NVP-BGT226 with EC 50 value of 0.14 µM, which is approximately 2.5 and 30 times more potent than amphotericin B and miltefosine, respectively. The structurally similar compounds, Torin2 (EC 50 = 0.25 µM) and dactolisib (EC 50 = 0.36 µM), exhibited activities in the submicromolar range. Torkinib and sapanisertib share the same scaffold, and PKI-402 and WYE-125132 are structurally similar. These inhibitors were moderately potent with EC 50 values similar to that of miltefosine (Table 1). We then sought to check any relationship between the intracellular Leishmania survival inhibition values to mTOR/PI3K inhibition values from published data (Table S2) [46][47][48][49][50]. Interestingly, the inhibitors that are active against mammalian PI3K at low nanomolar range, such as Torin2, dactolisib, and NVP-BGT226, were mostly found active against the parasite except PKI-402. and sapanisertib share the same scaffold, and PKI-402 and WYE-125132 are structurally similar. These inhibitors were moderately potent with EC50 values similar to that of miltefosine (Table 1). We then sought to check any relationship between the intracellular Leishmania survival inhibition values to mTOR/PI3K inhibition values from published data (Table S2) [46][47][48][49][50]. Interestingly, the inhibitors that are active against mammalian PI3K at low nanomolar range, such as Torin2, dactolisib, and NVP-BGT226, were mostly found active against the parasite except PKI-402. We further tested the seven compounds against L. donovani promastigotes ( Figure S1), T. brucei bloodstream form ( Figure 4) and intracellular T. cruzi. In the L. donovani promastigote growth inhibition assay, the EC50 values ranged from 0.009 to 7.46 μM; NVP-BGT226 was again the most potent compound ( Table 1). The correlation of activity between the promastigote and amastigote assay was R 2 = 0.64. Two structurally similar compounds, WYE-125132 and PKI-402, were 6.3-and 27.4-fold more potent against the promastigotes compared with the intracellular amastigotes ( Figure  5A). In T. brucei assay, the compounds were generally more active and especially NVP-BGT226 had an EC50 value in the picomolar range (0.000054 μM). Except for sapanisertib and torkinib, all the compounds had submicromolar activities with values approximately similar to pentamidine. The We further tested the seven compounds against L. donovani promastigotes ( Figure S1), T. brucei bloodstream form ( Figure 4) and intracellular T. cruzi. In the L. donovani promastigote growth inhibition assay, the EC 50 values ranged from 0.009 to 7.46 µM; NVP-BGT226 was again the most potent compound ( Table 1). The correlation of activity between the promastigote and amastigote assay was R 2 = 0.64. Two structurally similar compounds, WYE-125132 and PKI-402, were 6.3and 27.4-fold more potent against the promastigotes compared with the intracellular amastigotes ( Figure 5A). In T. brucei assay, the compounds were generally more active and especially NVP-BGT226 had an EC 50 value in the picomolar range (0.000054 µM). Except for sapanisertib and torkinib, all the compounds had submicromolar activities with values approximately similar to pentamidine. The extracellular forms of L. donovani and T. brucei displayed a high correlation of activity with R 2 = 0.80 ( Figure 5C), but the correlation between T. brucei and the amastigotes of L. donovani with R 2 = 0.51 ( Figure 5B). These differences may be due to genus-related differences between the parasites or to the effect of the extra host cell membrane on the penetration of compounds, or both. Furthermore, since dactolisib (NVP-BEZ235) showed varying range of in vitro activity depending on the subspecies of T. brucei in the previous report, the tested compounds in this study may exhibit different inhibitory activity to other subspecies of T. brucei or in different species of Leishmania [51,52]. One interesting point is the differences of antileishmanial and antitrypanosomal potency among Torin2, dactolisib and NVP-BGT225, in which the compounds share a similar structure. NVP-BGT226 and dactolisib share the same scaffold of 3-methyl-1-phenyl-8-(pyridin-3-yl)-1,3-dihydro-2H-imidazo[4,5-c]quinolin-2-one, and the corresponding structure in Torin2 is 1-phenyl-9-(pyridin-3-yl)benzo[h] [1,6]naphthyridin -2(1H)-one. Since all three compounds differ by the substitutions on the phenyl and pyridine-3-yl group, the difference of potency by the sequence of NVP-BGT226 > Torin2 > dactolisib potentially have resulted from such substitutions [53][54][55][56]. The group led by Pollastri has shown that the corresponding substitutions can improve the activity of dactolisib against the in vitro growth of T. brucei as well as SI [37]. The same set of compounds was tested using a T. cruzi-infected human osteosarcoma cell line (U2OS) model. However, most of the compounds had high toxicity against the host, with a low selectivity index (data not shown). Based on the activity correlations, the compounds are expected to directly act on the parasite and share a similar mechanism of action or a target between different parasites. yl)benzo[h] [1,6]naphthyridin-2(1H)-one. Since all three compounds differ by the substitutions on the phenyl and pyridine-3-yl group, the difference of potency by the sequence of NVP-BGT226 > Torin2 > dactolisib potentially have resulted from such substitutions [53][54][55][56]. The group led by Pollastri has shown that the corresponding substitutions can improve the activity of dactolisib against the in vitro growth of T. brucei as well as SI [37].The same set of compounds was tested using a T. cruzi-infected human osteosarcoma cell line (U2OS) model. However, most of the compounds had high toxicity against the host, with a low selectivity index (data not shown). Based on the activity correlations, the compounds are expected to directly act on the parasite and share a similar mechanism of action or a target between different parasites.

In Vivo Efficacy of mTOR/IP3K Inhibitors in VL Mouse Model
Commonly used animal models for evaluating compound efficacy in vivo are BALB/c (for acute) and hamster (for chronic) models [57][58][59][60][61][62]. For this proof of concept in vivo study, L. donovani infected BALB/c model was used to evaluate the ability of compound to reduce parasites in liver. Based on their EC50 and SI values from the intracellular Leishmania assay, we selected Torin2, dactolisib, and NVP-BGT226 (Table 1). Prior to the efficacy evaluation, the doses were selected based on the results of a literature search [45,63,64].And in pilot toxicity study in mice all the compounds were well tolerated without displaying any overt signs of toxicity during the five consecutive days of treatment. Using the same dosing scheme, the efficacies of the compounds were evaluated in VL mouse model. Miltefosine-treated mice, showed 87 ± 2.13% inhibition of parasitemia in liver compared to the

In Vivo Efficacy of mTOR/IP3K Inhibitors in VL Mouse Model
Commonly used animal models for evaluating compound efficacy in vivo are BALB/c (for acute) and hamster (for chronic) models [57][58][59][60][61][62]. For this proof of concept in vivo study, L. donovani infected BALB/c model was used to evaluate the ability of compound to reduce parasites in liver. Based on their EC 50 and SI values from the intracellular Leishmania assay, we selected Torin2, dactolisib, and NVP-BGT226 (Table 1). Prior to the efficacy evaluation, the doses were selected based on the results of a literature search [45,63,64]. And in pilot toxicity study in mice all the compounds were well tolerated without displaying any overt signs of toxicity during the five consecutive days of treatment. Using the same dosing scheme, the efficacies of the compounds were evaluated in VL mouse model. Miltefosine-treated mice, showed 87 ± 2.13% inhibition of parasitemia in liver compared to the vehicle-treated control mice. In the test groups, significant decreases in parasitemia by 54 ± 3.18%, 53 ± 9.50%, and 35 ± 5.81% in the NVP-BGT226-, dactolisib-, and Torin2-treated mice, respectively (Table 2 and Figure 6), were observed and the weight of spleens were similar to that of miltefosine-treated control mice ( Figure S2). In the Giemsa-stained liver smear, residual parasites not cleared by the tested compounds were observed inside host cells ( Figure 6B). Even though some activity was present, the tested compounds did not show comparably improved inhibition over miltefosine at a given dose. Another point to mention is the dose used for this study. In the in vitro assay, NVP-BGT226 showed CC 50 value of 2.66 µM, but no sign of toxicity was observed in vivo at 5 mg/kg dose. Since the dose was selected based on the previously reported literature that does not show adverse effect, the toxic effect by the compound was not observed in vivo [64]. For more thorough investigations, the maximum tolerated dose of compounds in infected animals can be determined to test efficacy of compounds with varying range of doses. Parasite burden (LDU, Leishman-Donovan Unit) was evaluated 7 days after compound administration. Data are presented as mean ± SD; NA, not applicable.

In Vivo Efficacy of NVP-BGT226 in T. brucei Mouse Model
The Torin2 and NVP-BGT226 compounds had more potent activities compared with pentamidine and were thus chosen for testing in T. brucei acute infection model. Since the efficacy of

In Vivo Efficacy of NVP-BGT226 in T. brucei Mouse Model
The Torin2 and NVP-BGT226 compounds had more potent activities compared with pentamidine and were thus chosen for testing in T. brucei acute infection model. Since the efficacy of NVP-BEZ235 was evaluated in the T. b. rhodesiense infected mouse model in the previous work by Pollastri et al., the compound was not selected for testing in this study [37,51,65]. T. brucei-infected mice with absence of any treatment all died within 6 days post-infection ( Figure 7A). For the drug testing, mice were treated with pentamidine (control), Torin2, or NVP-BGT226 at 30, 15, or 5 mg/kg for 5 days via the per os route, respectively, and the survival and parasitemia results are presented in Figure 7. In the group treated with NVP-BGT226, 80% survived during 14 days post-infection followed by death afterwards, and 20% of the mice survived more than 25 days without parasitemia observed. In the Torin2-treated group, 80% of the mice survived for at least 13 days after infection, but parasitemia was observed still by day 5. NVP-BGT226 exhibited slightly improved efficacy over the Torin2-treated group with extend 1-2 days delay in survival. In addition, compared with the previous reports of NVP-BEZ235 [38], NVP-BGT226 exhibited somewhat improved efficacy in the T. brucei acute mouse infection model.

Prediction of Inhibitor Binding to Kinetoplastid TORs
We then sought to further investigate the structural aspects of inhibitor binding. First, multiple sequence alignment analysis was performed between mTOR and kinetoplastid parasite TORs ( Figure  8). In the case of L. donovani, LdTOR1 showed sequence identity of 39.77% to mTOR, followed by LdTOR2 at 36.77% and LdTOR3 at 32.08%. L. major, T. brucei, and T. cruzi TOR1, 2, and 3 had similar sequence identity to mTOR within the range of 30%-40% (Table 3). Then we build homology models of kinetoplastid TORs and compared the structural similarity with human mTOR using root mean square deviation (RMSD) values. LdTOR1, 2, and 3 had RMSD values of 0.374, 0.428, and 0.291, respectively, with the mTOR structure implying high degrees of structural similarity to the human mTOR structure (Table 3 and Figure 9A-C). Since LmTOR3 was found not essential for the survival of the parasites in the promastigote stage, further analysis of Torin2 and torkinib binding to LdTOR1 and 2 was performed by overlaying Leishmania TOR models onto inhibitor-mTOR structures [43]. Seventeen amino acid residues were located within 4Å of Torin2, and among them, 15 residues were found identical to that of human mTOR, except for the difference of Ala 2248 to Glu 2128 and Ser 2342 to Asn 2227 in LdTOR2. In terms of molecular interaction, a hydrogen bond of N in the tricyclic benzonaphthyridinone ring of Torin2 to the O of W 2119 backbone (2.9 Å) was identified ( Figure 9D). For torkinib, the -OH and -NH2 moieties were predicted to interact with D 2074 and the backbone of G 2118 , respectively ( Figure 9E). These interactions were also found in the mTOR structures, which suggests a high possibility of LdTOR1/2 inhibition by the inhibitors with similar binding modes.

Prediction of Inhibitor Binding to Kinetoplastid TORs
We then sought to further investigate the structural aspects of inhibitor binding. First, multiple sequence alignment analysis was performed between mTOR and kinetoplastid parasite TORs (Figure 8). In the case of L. donovani, LdTOR1 showed sequence identity of 39.77% to mTOR, followed by LdTOR2 at 36.77% and LdTOR3 at 32.08%. L. major, T. brucei, and T. cruzi TOR1, 2, and 3 had similar sequence identity to mTOR within the range of 30%-40% (Table 3). Then we build homology models of kinetoplastid TORs and compared the structural similarity with human mTOR using root mean square deviation (RMSD) values. LdTOR1, 2, and 3 had RMSD values of 0.374, 0.428, and 0.291, respectively, with the mTOR structure implying high degrees of structural similarity to the human mTOR structure (Table 3 and Figure 9A-C). Since LmTOR3 was found not essential for the survival of the parasites in the promastigote stage, further analysis of Torin2 and torkinib binding to LdTOR1 and 2 was performed by overlaying Leishmania TOR models onto inhibitor-mTOR structures [43]. Seventeen amino acid residues were located within 4Å of Torin2, and among them, 15 residues were found identical to that of human mTOR, except for the difference of Ala 2248 to Glu 2128 and Ser 2342 to Asn 2227 in LdTOR2. In terms of molecular interaction, a hydrogen bond of N in the tricyclic benzonaphthyridinone ring of Torin2 to the O of W 2119 backbone (2.9 Å) was identified ( Figure 9D). For torkinib, the -OH and -NH 2 moieties were predicted to interact with D 2074 and the backbone of G 2118 , respectively ( Figure 9E). These interactions were also found in the mTOR structures, which suggests a high possibility of LdTOR1/2 inhibition by the inhibitors with similar binding modes.  Sequence identity (%) and root mean square deviation (RMSD) were calculated by using SWISS MODEL and PyMOL, respectively.  Sequence identity (%) and root mean square deviation (RMSD) were calculated by using SWISS MODEL and PyMOL, respectively.
In summary, using high-throughput screening, we identified seven mTOR/PI3K inhibitors with potent activity against the trypanosomatid parasites, L. donovani and T. brucei. Among these compounds, NVP-BGT226 was the most potent in vitro and was efficacious in VL and T. brucei animal models. The compound binding predictions based on the structural analysis suggests possible LdTOR1/2 inhibition by mTOR/PI3K inhibitors. This group of inhibitors was extensively developed for anti-cancer treatments, but currently hurdled by toxicity in human clinical trials. Due to this reason, a direct repurposing to kinetoplastids infections may not be readily feasible, but further development of inhibitors that are more selective to the kinetoplastid TORs would be possible. Another approach is utilizing the host immune response modulating properties of this class of compounds, as demonstrated by Khadem and colleagues, and further search for mTOR/PI3K inhibitors that well balance the direct and indirect killing effects. With limited number of highly potent inhibitors against Leishmania, these inhibitors are excellent chemical tools to decipher TOR pathways in kinetoplastid parasites which may lead to the discovery of new drug targets. In summary, using high-throughput screening, we identified seven mTOR/PI3K inhibitors with potent activity against the trypanosomatid parasites, L. donovani and T. brucei. Among these compounds, NVP-BGT226 was the most potent in vitro and was efficacious in VL and T. brucei animal models. The compound binding predictions based on the structural analysis suggests possible LdTOR1/2 inhibition by mTOR/PI3K inhibitors. This group of inhibitors was extensively developed for anti-cancer treatments, but currently hurdled by toxicity in human clinical trials. Due to this reason, a direct repurposing to kinetoplastids infections may not be readily feasible, but further development of inhibitors that are more selective to the kinetoplastid TORs would be possible. Another approach is utilizing the host immune response modulating properties of this class of compounds, as demonstrated by Khadem and colleagues, and further search for mTOR/PI3K inhibitors that well balance the direct and indirect killing effects. With limited number of highly potent inhibitors against Leishmania, these inhibitors are excellent chemical tools to decipher TOR pathways in kinetoplastid parasites which may lead to the discovery of new drug targets.

Ethics Statement
All animal studies were performed in strict accordance with the guidelines and principles established by the Korean Animal Protection Law (http://animalrightskorea.org). The use of animals was approved by the Institutional Animal Care and Use Committee (IACUC) of the Institut Pasteur

Ethics Statement
All animal studies were performed in strict accordance with the guidelines and principles established by the Korean Animal Protection Law (http://animalrightskorea.org). The use of animals was approved by the Institutional Animal Care and Use Committee (IACUC) of the Institut Pasteur Korea (IACUC approval number IPK-16003-3 for VL in vivo model and IPK-19002 for acute T. brucei in vivo model).

Inhibitors
The inhibitors used in this study were purchased from MedChem Express (Monmouth Junction, NJ, USA). CO 2 incubator at 37 • C. The T. b. brucei Lister strain 427 (bloodstream form: BSF) was cultivated in HMI-9 medium supplemented with 10% FBS, 100 µg/mL penicillin, and 100 µg/mL streptomycin at 37 • C and a 5% CO 2 atmosphere. The parasites were sub-cultured every 3 or 4 days and were maintained for 10 passages.

Screening of Bioactive Compounds Against Intracellular Leishmania
PMA-treated THP-1 human monocytic cells were seeded at 0.8 × 10 4 cells per well in a 384-well culture plate (Greiner Bio-One, Kremsmünster, Austria) in RPMI-1640 complete medium supplemented with 10% FBS. After 48 h of incubation at 37 • C in the presence of 5% CO 2 , the promastigotes of L. donovani that were incubated with lectin for 30 min at 28 • C were added to the cells at a parasite to cell ratio of 20:1. Infected THP-1 cells were treated with amphotericin B (at 4 µM, positive control), miltefosine (at 10 µM, positive control), and screening compounds (at 10 µM) [66][67][68]. The negative control consisted of THP-1 infected with the parasite with only 0.5% DMSO. After 72 h, the cells that were infected and treated with the drug were washed with serum-free RPMI-1640 medium. The cells and parasites were stained using 5 µM DRAQ5 and 4% PFA. The images were acquired based on reading using an Operetta ® automated microscope (PerkinElmer, Inc., Waltham, MA 02451 USA). They were further analyzed using Columbus TM (PerkinElmer, Inc. Waltham, MA, USA) software to quantify parasite numbers, host cell numbers, and infection ratios. In brief, large-sized nucleus of host cells was first detected using Draq-5 (Thermo Fisher, Rockford, IL, USA) signal and the host cell boundary masking was performed using the low-intensity signals from cytosols (additional feature of Draq-5). Then the small-sized nucleus signal by Draq-5 was used to identify parasites within the area of masked host cell. Infection ratio (IR) was determined with the value of the number of infected cells divided by total number of cells and the average number of parasites per macrophage (P/ϕ) was defined by the value of the number of parasites divided by the number of infected cells in the acquired image. The average IR value of the negative control wells was calculated as 0.53. Compounds selected based on the screening results were further assessed in a dose-dilution manner (two-fold serial dilution for 20 points starting from 100 µM) using the same method.

Parasite Growth Inhibition
L. donovani promastigote and T. b. brucei Lister 427 BSF growth inhibition were assayed by measuring the conversion of resazurin to resorufin. The assays were performed in 384-well plates that were seeded with L. donovani promastigotes or T. brucei 427 BSF (5 × 10 4 cells per well). After seeding, the parasites were exposed to the compounds for 3 days. Resazurin sodium salt (200 µM; R7017; Sigma-Aldrich, St. Louis, MO, USA) was then added, and the samples were incubated for 5 h. After incubation, the parasites were fixed using 4% paraformaldehyde, and the plates were analyzed using a Victor3 TM plate reader (PerkinElmer, Inc., Waltham, MA, USA) at 590 nm (emission) and 530 nm (excitation) [69]. Amphotericin B and miltefosine were used as the reference drugs for the L. donovani promastigote growth inhibition [67]. Pentamidine was used for the T. b. brucei Lister 427 BSF growth inhibition assay [68].

VL Mouse Model
Groups of five-week-old female BALB/c mice (five per group) were injected with 2 × 10 7 hamster spleen-derived L. donovani amastigotes via the retro-orbital venous sinus route. From day 7 post-infection, groups of mice were treated using the drug vehicle only, miltefosine (30 mg/kg), Torin2 (15 mg/kg), dactolisib (50 mg/kg), or NVP-BGT226 (5 mg/kg). The highest dose for each compound without showing adverse effects was selected based on literature search [44,45,49]. The vehicle for the compounds was 5% N-methylpyrrolidone, 15% polyvinylpyrrolidone, and 80% deionized water. Drug dosing solutions were freshly prepared each day. All drugs were administered once daily for 5 days via the per os route. On day 16 post-infection, all animals were humanely euthanized and assessed microscopically using Giemsa-stained liver imprints. Parasite burdens were measured by counting (blinded to treatment) the number of amastigotes per 1000 cell nuclei and multiplying this number by the liver weight (mg) (Leishman-Donovan Unit: LDU) [70]. The LDU values for the drug-treated samples were compared to those of the untreated samples, and the percent inhibition values were calculated.

HAT Mouse Model
BALB/c mice were infected with T. b. brucei Lister 427 (4 × 10 4 cells) by intraperitoneal (i.p.) injection. The mice were divided into groups (n = 5), and drug treatment was performed for five consecutive days by starting from day 1 post-infection administering 30 mg/kg, 15 mg/kg, and 5 mg/kg of pentamidine, Torin2, and NVP-BGT226, respectively. Parasitemia was evaluated daily for 2 weeks by blood collection from the mouse tail vein, and survival was monitored for 1 month. Mice showing impaired health status and/or with a parasite load > 10 8 cells per mL of blood were euthanized.

Statistical Analyses
All of the half maximal effective concentrations (EC 50 ) and the half maximal cytotoxic concentrations (CC 50 ) values were calculated using two independent experiments. The dose-response curves were fitted using GraphPad Prism 6 software (GraphPad Software, San Diego, CA, USA) by using a sigmoidal dose-response equation with a variable hill slope option.
Supplementary Materials: The following are available online, Figure S1: Dose-response curves of miltefosine, amphotericin B, and mTOR/PI3K compounds against L. donovani promastigote growth, Figure S2: Changes in spleen weight in L. donovani-infected BALB/c mice and effects of treatment with compound, Table S1: The list of 1742 compounds, Table S2: The relationships between mTOR/PI3K inhibition activity and intracellular L. donovani inhibition activity.

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