Preclinical candidate for the treatment of visceral leishmaniasis that acts through proteasome inhibition

Significance Safer and more effective oral drugs are urgently required to treat visceral leishmaniasis (VL), a neglected parasitic disease that kills 20,000–40,000 people each year in parts of Asia, Africa, and Latin America. Here, we describe the development of GSK3494245/DDD01305143/compound 8, a small molecule that demonstrates clinical-level efficacy in a mouse model of VL. Compound 8 exhibits attractive biological and biosafety properties, resulting in its selection as a preclinical candidate. Target deconvolution and cryo-EM studies reveal that compound 8 is a potent and selective inhibitor of the chymotrypsin-like activity of the parasite proteasome binding in a site sandwiched between the β4 and β5 subunits. Compound 8 is progressing toward human clinical trials, raising hopes of improved therapeutics for this disease.


Synthetic methods
Chemicals and solvent were purchased from the Aldrich Chemical Co., Fluka, VWR, Acros, Fisher Chemicals, Fluorochem, Apollo Scientific and Alfa Aesar. All chemical were used as received unless otherwise stated. Air-and moisture-sensitive reactions were carried out under an inert atmosphere of argon in oven-dried glassware. Analytical thin-layer chromatography (TLC) was performed on pre-coated TLC plates (layer 0.20 mm silica gel 60 with fluorescent indicator UV254, from Merck). Developed plates were air-dried and analyzed under a UV lamp (UV254/365 nm). Flash column chromatography was performed using prepacked silica gel cartridges (230-400 mesh, 40-63 µM, from SiliCycle) using a Teledyne ISCO Combiflash Companion or Combifalsh Retrieve. 1 H NMR spectra were recorded on a Bruker Avance DPX 500 spectrometer (at 500.1 MHz). Chemical shifts (δ) are expressed in ppm recorded using the residual solvent as the internal reference in all cases. Signal splitting patterns are described as singlet (s), doublet (d), triplet (t), quartet (q), multiplet (m), broad (b), or a combination thereof. Coupling constants (J) are quoted to the nearest 0.1 Hz. LC-MS analyses were performed with either an Agilent HPLC 1100 series connected to a Bruker Daltonics MicrOTOF or an Agilent Technologies 1200 series HPLC connected to an Agilent Technologies 6130 quadrupole LC/MS, where both instruments were connected to an Agilent diode array detector. All assay compounds had a measure purity of ≥95% as determined using this analytical LC-MS system (TIC and UV).
a. Chemical synthesis of compound 1 Commercially available.
b. Chemical synthesis of compound 2 Followed described procedure 1 .
The combined organics from all three batches were concentrated and the crude material recrystallized from EtOAc to give 18 (330 g, 1.62 mol, 75%) as pale yellow solid. 1 (20).

2.
CAD solubility assay GSK in-house kinetic solubility assay: 5 mL of 10 mM DMSO stock solution was diluted to 100 µL with pH7.4 phosphate buffered saline, equilibrated for 1 h at room temperature and filtered through Millipore Multiscreen HTS-PCF filter plates (MSSL BPC). The filtrate was quantified by suitably calibrated Charged Aerosol Detector 3 . The upper limit of the solubility was 500 µM when working from 10 mM DMSO stock solution.

Solubility of Solid Compounds in Fasted Simulated Intestinal Fluid (FaSSIF).
This experiment determines the solubility of solid compounds in fasted simulated intestinal fluid (FaSSIF) at pH 6.5 after 4 h equilibration at RT. Then 1 mL of FaSSIF buffer (3 mM sodium taurocholate, 0.75 mM lecithin in sodium phosphate buffer at pH 6.5) was added to manually weighed 1 mg of solid compound in a 2 mL HPLC autosampler vial. The resulting suspension is shaken at 900 rpm for 4 h at RT and then transferred to a Multiscreen HTS, 96-well solubility filter plate. The residual solid was removed by filtration. The supernatant solution was quantified by HPLC-UV using single-point calibration of a known concentration of the compound in DMSO. The dynamic range of the assay was 1-1000 µg/mL.

a. Rate of Kill
The rate of kill of compound 8 was determined in axenic amastigotes using the previously described cidal axenic assay 5 , with one minor modification. Instead of a single readout at 72 h, multiple identical plates were used and read at different time points (24,48, 72 h). Rate of kill curves showing the luminescence signal at each time point for each concentration were plotted.

Intrinsic clearance (CLint)
Test compound (0.5 µM) was incubated with female CD1 mouse liver microsomes (Xenotech LLC TM ; 0.5 mg/mL 50 mM potassium phosphate buffer, pH7.4) and the reaction started with addition of excess NADPH (8 mg/mL 50 mM potassium phosphate buffer, pH7.4). Immediately, at time zero, then at 3, 6, 9, 15 and 30 minutes an aliquot (50 µL) of the incubation mixture was removed and mixed with acetonitrile (100 µL) to stop the reaction. Internal standard was added to all samples, the samples centrifuged to sediment precipitated protein and the plates then sealed prior to UPLCMSMS analysis using a Quattro Premier XE (Waters Corporation, USA).
XLfit (idbs, UK) was used to calculate the exponential decay and consequently the rate constant (k) from the ratio of peak area of test compound to internal standard at each time-point. The rate of intrinsic clearance (CLint) of each test compound was then calculated using the following calculation: CLint(mL/min/g liver) = k x V x Microsomal protein yield Where V (mL/mg protein) is the incubation volume/mg protein added and microsomal protein yield is taken as 52.5 mg protein/g liver. Verapamil (0.5 µM) was used as a positive control to confirm acceptable assay performance.
a. Mouse and dog PK Mouse pharmacokinetic studies were conducted in male CD-1 mice and male beagle dogs. 3 animals per group were used for each study performed.
A dose of 3 mg/kg was administered intravenously, in a bolus form to mice and infused over a 30 min period to dogs. For all the intravenous studies compound were dissolved in 5% DMSO/20% kleptose in saline.
A dose of 10 mg/kg was orally administered by gavage to the two species in a suspension of 1% methylcellulose to investigate oral pharmacokinetics.
Following discrete oral gavage dosing from a suspension of 1% methyl cellulose at target doses of 10, 30, 100 and 300 mg/kg, respectively, serial blood samples (10 µl) were taken via lateral tail vein at 0.08, 0.25, 0.5, 0.75, 1, 2, 4, 6, 8 and 24 h post dose and 1/9 diluted with miliQ water, frozen on dry ice, then stored at approximately -20°C or below until analysis. Samples were analyzed for parent compound using a method based on protein precipitation followed by UPLC-MS/MS. Pharmacokinetic parameters, namely clearance (CL), volume of distribution at steady state (Vss) and bioavailability (F%), were estimated using Phoenix, WinNonLin 6.3.
c. Data Analysis Data analysis and calculation of pharmacokinetic parameters was performed using Phoenix, WinNonLin Version 6.3.
Following the intravenous administration, the whole blood clearance was calculated by determining the dose administered to each animal and dividing by the AUC0-∞ . The estimate of the volume of distribution at steady state (VSS) was calculated as MRT*CL, where MRT is the mean residence time, calculated by AUMC0-∞ /AUC0-∞.
For both intravenous and oral administration, the systemic exposures were determined by calculating the area under the blood concentration time curve (AUC) from the start of dosing to the last observed quantifiable concentration (AUC0-t) by using the linear up-log down trapezoidal rule. The slope of the terminal elimination phase was estimated by linear regression of the terminal data points (minimum 3 points) from a natural log concentration versus time plot of the data. The half-life (t1/2) of the terminal elimination phase was calculated as t1/2 = 0.693/λ.

7.
In vivo efficacy in female Balb/C mice To establish infection in mice, sodium stibogluconate-sensitive L. donovani (LV9, WHO designation: MHOM/ET/67/HU3) amastigotes were isolated from the spleen of a heavily infected donor hamster (Janvier, France) 7 . An inoculum containing 1.0 x 10 8 amastigotes/mL in Dulbecco's modified eagles medium (high glucose) was prepared and female BalbC mice (Harlan, UK) then infected intravenously via the tail vein with a 0.2 mL bolus (equivalent to ~2.0 x 10 7 amastigotes). Infection was left to establish for 7 days. At day 7, mice (n=5/dose group) were treated with compound 8 once or twice daily orally as a suspension in 0.5% hydroxypropylmethylcellulose, 0.4% tween 80 and 0.5% benzyl alcohol. With twice daily regimens, the second daily dose was administered approximately 8 h following the first dose. Dose regimens evaluated were: 1. 50 mg/kg twice daily for 5 days 2.
25 mg/kg once daily for 10 days Miltefosine, as the only oral treatment for visceral leishmaniasis, was used as a positive control in each evaluation, dosed once daily at 30 mg/kg as solution in sterile water. 30mg/kg was chosen as it represents the ED90 in mice 8,9 and therefore the required benchmark by which to compare any potential new oral therapies. For reference, a detailed study on the pharmacokinetics of miltefosine in mice has recently been published 10 . The minimal efficacy required by DNDi is 95% reduction in parasitaemia 11 . Vehicle dosed animals formed the negative control group.
Spot blood samples were taken for bioanalysis from each dose group (n=3/5 mice) following the first dose at day 1 and the last day of dosing.
Three days following completion of each dosing regimen, all mice were culled, liver smears prepared from each animal and stained with Rapi-diff II (Biostain Ready Reagents, UK). The number of amastigotes/500 liver cells were then counted microscopically and parasite load expressed in Leishman Donovan units (LDU): the mean number of amastigotes per 500 liver cells x mg liver. Expression of parasite load as LDU is the standard approach for assessing efficacy 12,13 . The individual animal LDU following oral dosing of compound 8 at 50 mg/kg b.i.d. for 5 days, 25 mg/kg b.i.d., 10 mg/kg b.i.d. and 3 mg/kg b.i.d. for 10 days and 25 mg/kg u.i.d. for 10 days were determined.

8.
Preclinical safety studies a. Genotoxicity i. Bacterial Mutation (Ames) Assay Compound 6 was tested in a bacterial mutation screening assay (Ames test) with Salmonella typhimurium TA1535, TA1537, TA98, TA100 and Escherichia coli WP2uvrA(pKM101) in the presence and absence of an exogenous mammalian oxidative metabolism system (S9-mix). The maximum concentration tested and analysed was 2500 µg per plate, limited by precipitation and in accordance with current guidelines (OECD 471, 1997 14 ). Compound 6 was found not mutagenic in this bacterial mutation screening assay when tested in the presence and absence of S9-mix.
Compound 6a was tested in a bacterial mutation screening assay (Ames test) with Salmonella typhimurium TA1535, TA1537, TA98, TA100 and Escherichia coli WP2uvrA(pKM101) in the presence and absence of an exogenous mammalian oxidative metabolism system (S9-mix). The maximum concentration tested and analysed was 1500 µg per plate, limited by precipitation. Compound 6a was found mutagenic in strain TA1537 when tested in the absence of S9-mix.
Compound 7a was tested in a bacterial mutation screening assay (Ames test) with Salmonella typhimurium TA1535, TA1537, TA98, TA100 and Escherichia coli WP2uvrA(pKM101) in the presence and absence of an exogenous mammalian oxidative metabolism system (S9-mix). The maximum concentration tested and analysed was 1500 µg per plate, limited by precipitation. Compound 7a was found not mutagenic in this bacterial mutation screening assay when tested in the presence and absence of S9-mix.
Compound 8 was tested in a bacterial mutation screening assay (Ames test) with Salmonella typhimurium TA1535, TA1537, TA98, TA100 and Escherichia coli WP2uvrA(pKM101) in the presence and absence of an exogenous mammalian oxidative metabolism system (S9-mix). The maximum concentration tested and analysed was 5000 µg per plate, in accordance with current guidelines (OECD 471, 1997 14 ).
The bacterial mutation screening assay was a plate incorporation test carried out as previously described 15 and according to current methodology.
ii. Mouse Lymphoma Assay Compound 8 was negative in the mouse lymphoma screen when tested for 3 hours in the presence of S9-mix and 24 hours in the absence of S9-mix. The maximum concentrations analysed were 410.45 µg/mL (1 mM, the maximum concentration in accordance with current guidelines and 100 µg/mL (limited by toxicity), for the 3 and 24 hour treatments, respectively (OECD 476, 1997 16 )).
The mouse lymphoma screen was carried out using test methodology based on established procedures for in vitro gene mutation testing 17,18,19,20 . b.
7 day rat toxicology studies A 7 day oral repeat-dose study in male rats was conducted to determine the toxicity and toxicokinetics of compound 8. Groups of male rats (4/group toxicity; 3/group toxicokinetics) were given 0, 100, 300 or 1000 mg/kg/day compound 8 once daily for up to 7 days by oral gavage. In animals given 100, 300 and 1000 mg/kg/day, blood concentrations of compound 8 were generally quantifiable to 24 h after dosing on both day 1 and day 7 (1000 mg/kg/day animals were not sampled on day 7 due to early termination). Systemic exposures (AUC0-t and Cmax) increased in an approximately proportional manner with increasing dose on Day 1. On Day 7 exposures increased in a greater than proportional manner between 100 and 300 mg/kg/day. This was driven by the increase in systemic exposure from Day 1 to Day 7 at the 300 mg/kg/day dose. There is a 37-fold margin at 300 mg/kg/day, where minimal microscopic and clinical pathology changes were observed, to the predicted human blood AUC0-24 of 27 ng/h/mL after a 600 mg dose b.i.d.
The 1000 mg/kg/day dose level was not tolerated with all animals terminated early due to the clinical signs observed. The 100 and 300 mg/kg/day dose levels were well tolerated over the 7 days. Microscopic (mild to moderate) changes were observed at 1000 mg/kg/day in the kidney, liver and pancreas. At 300 mg/kg/day minimal/mild microscopic changes were observed in the liver (increased incidence of PAS reactivity of hepatocytes compared to control [no evidence of hepatocellular rarefaction on H&E stained sections]), and in the pancreas of a single animal at this dose (diffuse acinar cell degranulation). At 1000 mg/kg/day increases in alanine aminotransferase, glutamate dehydrogenase, alkaline phosphatase, glucose, phosphorus, calcium and a reduction in chloride concentrations were seen. A lower urea concentration was seen in a single male. Increases were seen in total bile acids (≥100 mg/kg/day), total bilirubin and cholesterol (≥300 mg/kg/day) and creatinine (300 mg/kg/day only). Individual males given 100 or 300 mg/kg/day had increased reticulocyte counts with correlating increases in medium and high reticulocyte subpopulations.
b. RITseq library screening An RNAi library screen was carried out as described previously 22 . Briefly, the RNAi library was induced on day 0 with tetracycline (1 µg/mL) and maintained under blasticidin (1 µg/mL) and phleomycin (1 µg/mL) selection at a minimum of 2.5 × 10 7 cells in 150 ml of media. Following induction for 24 h, cultures were selected with 6.3 nM compound 7 (~3x EC50) was added. Cultures were split and supplemented with fresh drug as required. DNA was extracted from drug-resistant cells on day 12. RNAi target fragments were then amplified by PCR using the LIB2f and LIB2r primers (Table  S6). For high throughput identification of fragments, the PCR products were fragmented and sequenced using an Illumina HiSeq platform at BGI (Beijing Genomics Institute).
Reads were mapped to the T. brucei 927 reference genome (v9.0, tritrypdb.org) with Bowtie 2 24 using the following parameter: very-sensitive-local-phred33. The generated alignment files were manipulated with SAMtools 25 and a custom script to identify reads with barcodes (GCCTCGCGA) 22 . Total and bar-coded reads were then quantified using the Artemis genome browser 26 . Hit-lists generated from RIT-seq analyses excluded selected large gene families, including variant surface glycoproteins, and genes immediately adjacent to hits.
c. Generation of drug-resistant parasites Compound 7-resistant lines were generated by sub-culturing a clone of wild-type L. donovani in the continuous presence of this compound. Starting at a sub-lethal concentration of 3 nM compound 7, the drug concentrations in 3 independent cultures were increased in a step-wise manner, usually by 2-fold. After a total of 120 days in culture, when promastigotes were able to survive and grow in >100 nM compound 7, the resulting cell lines were cloned by limiting dilution in the absence of compound. Three clones (RES I-III) were selected for further biological study.

d. DNA sequencing
Genomic DNA was harvested from three independently generated L. donovani resistant clones and an age match control, as previously described 27 . PCR was performed using Platinum PCR Supermix (Invitrogen), as per manufacturer's instructions. To sequence the β4 subunit (LdBPK_360340.1), an 889bp fragment was generated using primers SubunitB4FW and SubunitB4RV, then sequenced using primer SubunitB4FW (Table S6). To sequence the β5 subunit (LdBPK_361730.1), an 1113bp fragment was generated using primers Subunitβ5FW and Subunitβ5RV, then sequenced using primer SubunitB5FW. PCR amplification and DNA sequencing were performed in triplicate for each independently generated cell line. The Leishmania donovani BPK282A1 sequences of the two subunits were identified in TriTrypDB and were a direct match with the age match control (drug-sensitive).

e. Generation of overexpression constructs and LdBOB transgenic cell lines
The genes encoding the mutated versions of the β4 (LdBPK_360340.1) and the β5 (LdBPK_361730.1) subunits of the proteasome, identified in our drug-resistant parasites, were synthesised (GeneArt, Thermo Fischer Scientific). The synthetic genes, flanked by BamHI sites were cloned into the equivalent site of pIR1-SAT. The accuracy of all assembled constructs was verified by sequencing.
Mid-log-phase L. donovani promastigotes (LdBOB) were transfected with overexpression constructs using the Human T-Cell Nucleofector kit and the Amaxa Nucleofector electroporator (program V-033). Following transfection, cells were allowed to grow for 16-24 h in modified M199 medium with 10% fetal calf serum prior to appropriate drug selection (100 µg/mL nourseothricin). Cloned cell lines were generated by limiting dilution, maintained in selective medium, and removed from drug selection for one passage prior to experiments.
f. T. brucei RNAi constructs and cell lines PCR primers (Table S6) were designed using RNAit 28 to generate a 500-bp fragment conferring specific knockdown to Tb927.8.6620 (Table S7) and a 543-bp fragment of Tb927.9.15260 PCR fragments were cloned in the pRPaSLi construct for the generation of stem-loop dsRNA under the control of tetracycline as the trigger for RNAi 28,29 . Constructs were digested with AscI, EtOH-precipitated, and resuspended (1 µg/mL) in sterile water. 2T1 strains, containing a tetracycline repressor, were electroporated using program X-001 of the Nucleofector II electroporator (Amaxa, Cologne, Germany) 30 following the addition of 5 µg of linearized DNA mixed in 100 µL of cytomix 31 . Transformants were cloned by limiting dilution under phleomycin (1 µg/mL) and hygromycin (2.5 µg/mL) selection. Puromycin susceptibility (1 µg/mL) was tested for full integration of the construct and expression of stem-loop RNAi was induced with concentrations of tetracycline ranging from 1 ng -µg/mL.

g. Quantitative RT-PCR
T. brucei RNA was isolated using an RNeasy purification kit (Qiagen) and cDNAsynthesized using a high capacity RNA-to-cDNA kit (Applied Biosystems). PCR primers (Table S6) were designed using the Premierbiosoft's Beacon Designer 6. qRT-PCRs consisted of 1 µL (40 ng) of cDNA, 10 µL of Brilliant III Ultra-Fast QPCR Master Mix (Agilent Technologies), 1 µL (500 nM) each of the forward and reverse primers, and 0.3 µl (30 nM) of reference dye and nuclease-free PCR grade-treated water. PCR was performed using an Agilent Mx3005P machine with the following cycling conditions: 95°C for 3 min; 40 cycles of 95°C for 20 s; then 60 °C for 20 s. The reference gene TERT (Tb927.11.10190, telomerase reverse transcriptase) was used to provide a baseline of transcription levels for normalization of the data. Relative quantification in the tetracycline-induced cell lines were normalized to the un-induced cell line using the ∆∆Ct method.

h. Morphological analysis
For transmission electron microscopy, L. donovani promastigotes were prepared as described previously 32 , post-fixed for 1 h at 4°C with 1% (v/v) osmium tetroxide in 100 mM phosphate buffer (pH 6.5) and rinsed briefly in water prior to fixing or staining en bloc with 3% (v/v) aqueous uranyl acetate. Cells were rinsed further in distilled water and subsequently dehydrated through a graded EtOH series with a final wash in propylene oxide, prior to embedding in Durcupan resin. Sections were stained with 3% (v/v) aqueous uranyl acetate and Reynold's lead citrate prior to examination using a JEOL-1200 EX TEM.

i. FACS analysis
Cultures of bloodstream T. brucei and L. donovani promastigotes at starting cell densities of 5 x 10 5 were incubated with concentrations of compound 7 equivalent to 5x their respective EC50 values. At defined intervals samples of culture were taken, cells were collected by centrifugation (900 g, 10 min, 4°C) and washed twice in PBS before preparation for FACS analysis 33 . After washing, cell densities were adjusted to 5 × 10 5 cells mL, re-suspended in 500 µL of PBS containing 50 µg/mL propidium iodide, 50 µg/mL RNase and 0.1% Triton X-100, incubated at room temperature for 20 min in the dark and analysed using FACSort analytical flow cytometer using Cellquest software (BD Biosciences).

j. Proteasome enrichment by ultracentrifugation
The protocol described by Kisselev et al. 34 was adapted to enrich the proteasomes present in L. donovani and THP1 crude extracts. L. donovani or THP1 cells were resuspended in sucrose buffer (50 mM Tris buffer pH 7.5, 5 mM Magnesium chloride, 1 mM EDTA, 50 mM NaCl, 250 mM sucrose, 2 mM ATP, 1 mM DTT) and lysed by nitrogen cavitation for 15 min (110 bars of 1500 psi). Lysates were centrifuged at 20,000g for 30 min at 4ºC and supernatants containing soluble protein were subjected to an ultracentrifugation step at 300,000g for 2h at 4ºC. The resulting proteasomecontaining pellets were resuspended in sucrose buffer and left on ice for 30 min to complete solubilization. Samples were centrifuged for 10 min at 20,000 g to remove insoluble material. Proteins contained in supernatant (proteasome enriched fraction) were measured by Bradford.

k. Proteasome activity assays
Compounds were dispensed into black 384-well assay plates (Greiner) by acoustic dispensing (LabCyte ECHO). For potency determinations, eleven-point, one in three dilution curves were generated, with a top concentration of 100 µM. Three proteasome activities were measured using a commercially available indirect enzyme-based luminescent assay kit (Promega) with Suc-LLVY-aminoluciferin, Z-LRR-aminoluciferin and Z-nLPnLD-aminoluciferin used as substrates for chymotrypsin-like, trypsin-like and caspase-like activities, respectively. In addition to proteasome samples: L. donovani and THP1-enriched proteasomes, commercially available human purified 26S proteasome (Boston Biochem) was also assessed. Proteasome final assay concentration were 0.025 mg/ml for L. donovani and THP1-enriched proteasomes and 1 nM for human purified 26S proteasome. Proteasome solution was added to the plates containing the compounds one hour before adding the proteasome substrate solution. Substrate solution was then added and after 30 min of incubation luminescent signal was measured in an Envision reader (PerkinElmer).

l. Western blot analysis
Cultures of L. donovani axenic amastigotes were diluted to either 8 x 10 6 or 8 x 10 7 cells/ml prior to treatment with compound 8 at 30 nM (0.1X whole cell IC50), 300 nM (1X IC50) or 3100 nM concentrations equivalent to 0.1, 1 and 10X the compounds EC50 value. Culture were incubated with compound 8 for 1h at 37ºC. Miltefosine at 80 µM (equivalent to 10X EC50 value) and bortezomib at 200 nM (equivalent to 10X EC50 value) were used as negative and positive controls, respectively. Following incubation, drug-treated parasites were pelleted by centrifugation at 640 g for 15 min and washed twice with cold PBS. Untreated cells were also harvested alongside treated cells to assess the basal levels of ubiquitylated proteins. Pellets were frozen at -80ºC and three cycles of freeze/thawing were conducted to biologically inactivate parasites. Lysis buffer (50 mM Tris-HCl pH 8.5, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 1 mM PMSF and complete mini EDTA protease inhibitor cocktail) was added to each pellet prior to one fast cycle of freeze/thawing. Lysates were cleared by centrifugation (10 min, 20,000g, 4°C) and supernatants were collected. Protein concentrations were determined by standard Bradford assays and Laemmli sample buffer was added to samples. For each condition, 80 µg of each protein sample was separated by SDS-PAGE. Proteins were then transferred onto polyvinylidene difluoride (PVDF) membranes (Roche), blocked with milk at 5% in TBS-tween 0.1% and incubated with anti-ubiquitin P4D1 antibody (Santa Cruz) diluted 1:1,000 in 2.5% TBS-tween 0.1%. DMSO or test compounds, were incubated with the probe for 2h at RT. Samples were then denatured by addition of Laemmli sample buffer, heated to 95°C for 10 min and separated by electrophoresis on a 15% SDS-PAGE acrylamide gel. Following separation, gels were scanned in the fluorescein channel of a ChemiDoc reader (λ excitation/emission = 480/530 nm). Fluorescent proteins were excised and MALDI-TOF analysis carried out by the FingerPrints Proteomics service at the University of Dundee (see Table S8).

10.
Measuring Proteolytic activity of Leishmania tarentolae Proteasome The chymotrypsin-like catalytic activity of purified Leishmania tarentolae proteasome was monitored by measuring the cleavage of a Rhodamine110-labelled fluorogenic substrate (Suc-LLVY-Rh110-dPro), which alleviates a quench of the Rhodamine110 fluorophore leading to an increase in fluorescence at 535 nm 35 . Test compounds were pre-prepared as a 10 mM stocks in 100 % DMSO and then diluted 1 in 3, over 11 points, prior to dispense of 100 nl of each dilution series into black, low volume, 384-well microplates. Proteasome was prepared as a 0.5 µg/mL working solution in assay buffer (50 mM HEPES (pH 7.5), 150 mM NaCl, 5 mM MgCl2, 5% glycerol, 1 mM CHAPS) and 5 µL was added using a Multidrop Micro (Thermo Fisher Scientific) to all wells containing compound dilutions on the test microplates. The microplates were then centrifuged for 1 min at 1000 rpm and incubated for 15 mins at 25 °C, prior to an assay initiating addition of 5 µl of fluorogenic substrate, which had been prepared as a 1 µM (~ 2x Km) solution in assay buffer, using a Mutidrop Micro. The assay microplates were then centrifuged for 10 secs before using a Tecan M1000 Pro to monitor the increase of fluorescent intensity at 25 °C by excitation of the liberated Rhodamine110 at 485 nm and emission at 535 nm every 125 secs for 20 cycles. These data were then used to calculate initial rates of reactions for all assay wells on the microplate by fitting to a linear regression model. The initial rates of reaction for all compound concentrations tested were normalised to the initial rates determined for 16 replicate control wells per microplate of both uninhibited reaction and no-proteasome background, and subsequently, IC50 values were determined by fitting the normalised data to the following four-parameter logistic model; y = a + (( b -a) / ( 1 + ( 10 ^ x / 10 ^ c ) ^ d ), where 'a' is the minimum, 'b' is the Hill slope, 'c' is the IC50 and 'd' is the maximum. Table S8 shows the Leishmania tarentolae activity for compound 8.
Leishmania tarentolae strain P10 was purchased from Jena Biosciences and cultivated in LEXSY BHI medium, according to the manufacturers' guidelines. Cells were harvested in early stationary phase by centrifugation at 3000g for 30 mins and the pellets were frozen. For the purification, the frozen cell pellets were defrosted on ice, resuspended in twice the pellet volume of 100 mM mM Tris-HCl pH 7.5, 500 mM sucrose, 2 mM EDTA, 100 mM NaCl, 10 mM MgCl2, 4 mM ATP and 2 mM DTT. Cells were lysed by sonication at 10% Amplitude for 30sec on/off for 6 cycles using Soniprep 150 sonicator. The lysates were clarified by gentle centrifugation at 20,000 g for 30 min. The supernatant was further ultra-centrifuged at 300,000 × g for 2 h at 4 °C. Pelleted proteasomes were solubilised in 100mM mM Tris-HCl pH 7.5, 100 mM sucrose, 1 mM EDTA, 150 mM NaCl, 5 mM MgCl2, 4 mM ATP and 1 mM DTT (GF buffer) and incubated for 1 hour at 4 ºC. The suspension was then centrifuged at 20,000g for 30 mins at 4 ºC. The supernatant was filtered using a 0.2 µm filter and passed through GE HiPrep 16/60 Sephacryl S-400 HR column equilibrated with the GF buffer using a GE Akta Pure chromatography system. Fractions of 1.5 ml were collected and assayed for chymotrypsin-like activity using suc-LLVY-[Rh110]-[D-Pro] peptide. Active fractions were pooled and further purified by two rounds of anion exchange chromatography using GE HiScreen CaptoQ and then GE MonoQ 5/50 GL columns, equillibrated with GF buffer. Bound proteins were eluted by linear 0-1M NaCl gradient. Fractions (0.25 mL) were collected and analysed by silver stain and tested for chymotrypsin activity. The purest fractions were pooled together and loaded on GE Superose 6 Increase 5/150 GL column equilibrated with 100 mM mM Tris-HCl pH 7.5, 150 mM NaCl, 5 mM MgCl2, 4mM ATP and 1 mM DTT. Peak fractions were analysed as described above. The final concentration of 0.1 mg/mL was estimated by Bradford assay using Pierce™ Coomassie Plus (Bradford) Assay Reagent.

12.
Cryo-grid preparation and data collection a.
Purified proteasome was used to prepare Cryo-grids with and without compound 8.
For Cryo-grids with proteasome alone, a concentration of 0.1 mg/mL was used. Quanitfoil grids (300 mesh, 1.3/1.2) coated with 2 nm carbon (Agar Scientific) were freshly glow discharged using a Pelco easiGlow at 25 mA for 45 s. Initially, 2.5 µl of purified proteasome was applied. After 30s incubation the grid was double-sided blotted for 0.5 s using a Vitrobot Mark IV (FEI ThermoFisher). Immediately, another 2.5 µL of sample was applied followed by 30 s incubation and 2.5s blotting and then the grid was plunged into liquid ethane. The blot force was kept constant at 2. For the liganded complex, the purified proteasome was buffer exchanged into 100 mM Tris-HCl pH 7.5, 5 mM MgCl2 and 1 mM DTT and concentrated to 0.4 mg/mL using a Vivaspin 500 centrifugal concentrator MWCO 5 kDa. 20 µM of compound 8 was incubated with 0.5 µM of purified proteasome at 37 °C for 30 min. In this case, the sample was applied only once on the grid. All grids were made at 4°C and 100% humidity.
High-resolution data were collected on a Titan Krios electron microscope equipped with a Falcon III direct electron detector at ThermoFischer-Cambridge-Pharma consortium at Cambridge, Nanosciences.
i. Image processing Beam induced motion was corrected in MotionCor2 36 retaining all the frames of the raw movies, collected at pixel size of 1.07 Å, using a patch alignment of 5x5. All subsequent particle picking, particle extraction, 2D and 3D classifications as well as 3D refinements were performed in Relion2.0 37 . Initially, particles were manually picked and 2D class averages obtained using a mask diameter of 240Å then served as references for later autopicking. CTF correction was performed using CTFFIND4 38 and the best quality integrated movies were examined and selected. Particles were extracted from dose-weighted micrographs by setting a box size of 300 pixels for all data sets. After 2-3 rounds of 2D classification, 3D classification, and 3D refinement were performed without applying any mask and using the Thermophilic acidiphilum proteasome structure (EMD-3231) as the initial model (filtered to 30 Å).
Fourier filtering or masking, were created using UCSF Chimera 39 . The mask was created from the final 3D auto-refinement of the particles ( Figure S11 and S15) adjusted to a threshold of 0.0062 for apo and 0.0115 for liganded, low-pass filtered to 15 Å and a soft edge of three pixels was applied. The local resolution of the maps was estimated using the Relion-2.0 local resolution function, setting the same B-factor as the post processing output. No symmetry was applied at any stage of the image processing.

a. Apo and ligand structures
Structure factors of the apo dataset were generated from the sharpened EM map using Refmac 40 within the CCP4 suite of programs 41 . Molecular Replacement was carried out using Phaser 42 with the coordinates of the human proteasome crystal structure 4R3O.
Model building was carried out using Coot 43 , guided by sequence alignments generated for all the individual αand β-subunits of human and Leishmania tarentolae proteasomes. When the higher resolution sharpened EM liganded proteasome map became available model building was switched from the apo structure to the liganded structure. The quality of this map was sufficient to allow backbone tracing for most of the protein with only a few gaps. The amino acid side chain conformations were built by selecting the best fitting rotamer followed by real space refinement in Coot. One half of the 20S proteasome subunit was manually built (subunits A-N) with the two-fold symmetry used to generate the second half (subunits O-b). A small number of water molecules were added based on standard X-ray structure criteria (significant peaks in the map, consistent with favourable locations for hydrogen bonding interactions with well-ordered protein residues). Ligand coordinates and dictionary files were generated using Grade 44 . Multiple cycles of manual model building followed by structure refinement with Refmac were carried out utilising two-fold NCS restraints throughout, initially using jelly body refinement, with restrained refinement used in the final refinement step. The refined liganded proteasome coordinates were then used as a starting point for building the apo structure (omitting the ligands and waters). The individual subunits were adjusted using rigid body refinement in Coot and manually adjusted where intradomain movements were apparent or where side chain conformations were clearly different in the two structures. The refinement protocol was similar to that used for the liganded structure. The EM maps and structures have been deposited in the Protein Data Bank: liganded structure EMD-4590 (PDB 6QM7), apo structure EMD-4591 (PDB 6QM8).

Molecular modelling
Using Prime 45 , the homology model of L. donovani β4-β5 subunits was generated utilising as reference the Cryo-EM structure of L. tarentolae bound to compound 8. The sequence identity between L. tarentolae and L. donovani proteasome β4-β5 subunits was 95% and 98%, respectively. The model was refined by performing a 100 ns MD simulation of the protein-ligand complex using AMBER16 suite of programs 46 . FF14SB forcefield was used for the protein. The geometry of compound 8 was refined using Gaussian 03 47 at the HF/6-31G* level. The optimized geometries were used to calculate electrostatic potential-derived charges (ESP). The forcefield parameters for the ligand were generated with antechamber module, using the general AMBER forcefield (GAFF2.0). The site recognition software SiteMap 45 was used to describe the physicochemical properties of the ligand binding site. Water networks and relative energies were calculated using WaterFLAP (Molecular Discovery). Figure S1. Screening cascade for primary T. cruzi screen that identified compound 1.                 In yellow, the hydrophobic area matching the pyrrolidine ring, calculated with SiteMap (Schrödinger 2018). In human, due to differences residues in the β4 subunit, the hydrophobic pyrrolidine pocket is lost and the area is solvent exposed.      Table S7. Genome-wide RNAi library screening of compound 7. Table summarising the genes within the ubiquitin-proteasome degradation pathway (green) identified as playing a role in compound resistance following screening of the RITseq library in T. brucei. labelled proteins were excised from the gel (see Figure 4C) and MALDI-TOF analysis carried out by the FingerPrints Proteomics service at the University of Dundee. Identity of the recovered peptides detailed below. Catalytically active subunits of the Leishmania proteasome are highlighted in red.

Ethical Statements
• Rat pharmacokinetics and in vivo efficacy: All regulated procedures, at the University of Dundee, on living animals was carried out under the authority of a project licence issued by the Home Office under the Animals (Scientific Procedures) Act 1986, as amended in 2012 (and in compliance with EU Directive EU/2010/63). Licence applications will have been approved by the University's Ethical Review Committee (ERC) before submission to the Home Office. The ERC has a general remit to develop and oversee policy on all aspects of the use of animals on University premises and is a sub-committee of the University Court, its highest governing body.
• Rat toxicology studies and mouse and dog pharmacokinetics: All animal studies were reviewed by GSK's internal ethical review committee and performed in accordance with Animals (Scientific Procedures) Act 1986 and the GSK Policy on the Care, Welfare, and Treatment of Laboratory Animals (UK 1986).
• The human biological samples were sourced ethically and their research use was in accord with the terms of the informed consents. Usage of humansourced macrophages was approved by the "Scottish National Blood Transfusion Service committee for the governance of blood and tissue samples for non-therapeutic use, and donor research".