Design and Synthesis of Brain Penetrant Trypanocidal N-Myristoyltransferase Inhibitors

N-Myristoyltransferase (NMT) represents a promising drug target within the parasitic protozoa Trypanosoma brucei (T. brucei), the causative agent for human African trypanosomiasis (HAT) or sleeping sickness. We have previously validated T. brucei NMT as a promising druggable target for the treatment of HAT in both stages 1 and 2 of the disease. We report on the use of the previously reported DDD85646 (1) as a starting point for the design of a class of potent, brain penetrant inhibitors of T. brucei NMT.


■ INTRODUCTION
Human African trypanosomiasis (HAT) or sleeping sickness is prevalent in sub-Saharan Africa 1 with an estimated "at risk" population of 65 million. 2 The causative agents of HAT are the protozoan parasites Trypanosoma brucei gambiense and Trypanosoma brucei rhodesiense 3,4 transmitted through the bite of an infected tsetse fly. HAT progresses through two stages. In the first stage (stage 1), the parasites proliferate solely within the bloodstream. In the second, late stage (stage 2), the parasite infects the central nervous system (CNS) causing the symptoms characteristic of the disease, such as disturbed sleep patterns and often death. 5 Currently, there are a number of treatments available for HAT, though none are without issues, including toxicity and inappropriate routes of administration for a disease of rural Africa. 6 Research has revealed enzymes and pathways that are crucial for the survival of T. brucei, and based on these studies, a number of antiparasitic drug targets have been proposed. 7−10 T. brucei Nmyristoyltransferase (TbNMT) is one of the few T. brucei druggable targets to be genetically and chemically validated in both in vitro and in rodent models of HAT. 7,11,12 NMT is a ubiquitous essential enzyme in all eukaryotic cells. It catalyzes the co-and post-translational transfer of myristic acid from myristoyl-CoA to the N-terminal glycine of a variety of peptides. Protein N-myristoylation facilitates membrane localization and biological activity of many important proteins. 11,13 NMT has been extensively investigated as a potential target for the treatment of other parasitic diseases including malaria, 14 leishmaniasis, 15 and Chagas' disease 16,17 resulting in the identification of multiple chemically distinct small molecule inhibitors. 18 NMT has also been shown to be a potential therapeutic target for human diseases such as autoimmune disorders 19 and cancer. 20,21 Previously we have reported the discovery of compound 1 (Figure 1), 7,22,23 which showed excellent levels of inhibitory potency for TbNMT and T. brucei brucei (T. br. brucei) proliferation in vitro and was used as a model compound to validate TbNMT as a druggable target for stage 1 HAT. 7,22 However, 1 is not blood−brain barrier penetrant, a requirement for stage 2 activity. Two approaches were taken to increase the brain penetration of 1. A classical lead optimization approach is described elsewhere. 24 This article describes a second approach that used a minimum pharmacophore of 1 aiming to derive a structurally distinct series of potent TbNMT inhibitors with brain penetration, as leads for the identification of suitable candidates for the treatment of stage 2 HAT.

■ COMPOUND RATIONALE AND DESIGN
To aid compound design, and to significantly lower molecular weight and polar surface area (PSA), the chlorines and the sulfonamide moieties of 1 were removed to define a minimum pharmacophoric scaffold (Figure 2A). This scaffold was chosen because the piperidine makes a key interaction through the formation of a salt bridge with NMT's terminal carboxylate. 10 This interaction is highly conserved across the binding modes of NMT inhibitors covering multiple chemotypes including 1 ( Figure 2B); known antifungal NMT inhibitors such as Roche's (2-benzofurancarboxylic acid, 3-methyl-4-  29 Attempts to crystallize TbNMT had proved to be unsuccessful; therefore, the fungal Aspergillus f umigatus NMT (Af NMT) 24,30 was used as a surrogate model for TbNMT in this study. Af NMT is 42% identical to TbNMT; however, within the peptide binding groove the level of identity is 92%. Previously, a selection of molecules from series 1 were assayed against Af NMT and TbNMT using the SPA biochemical assay and pIC 50 values compared using linear regression analysis. The pIC 50 values were shown to be correlated with an R-squared value of 0.73 suggesting that Af NMT is a suitable surrogate system for study within this chemical series (see Supporting Information).
This minimum pharmacophoric scaffold had low molecular weight (237) and low PSA (12 Å 2 to maximize the potential for CNS penetration) from which we could design varied chemistry ( Figure 2A) to either access the serine pocket (occupied by the pyrazole moiety in 1) or the peptide recognition region, as seen in the peptomimetic compound highlighted in red ( Figure 2C).
Compound Design. The adopted compound design strategy covered both compounds based on 1 (where common sulfonamide bioisosteres 31 and pyrazole mimics were included) and compounds based on the binding pocket structural features, probing these with diverse hydrogen bond acceptor (HBA) and hydrogen bond donor (HBD) groups. We employed high throughput chemistry, using technologies and techniques such as scavengers and solid supported reagents enabling arrays to be made in parallel. Three different but complementary chemistries Figure 1. Compound 1. *Potencies were determined against recombinant TbNMT and HsNMT1, and against bloodstream form T. brucei brucei (T. br. brucei) and MRC-5 proliferation studies in vitro using 10 point curves replicated ≥2. a Calculated using Optibrium STARDROP software. b Ligand efficiency (LE), calculated as 0.6· ln(IC 50 )/(heavy atom count) using T. brucei NMT IC 50 potency. 25 IC 50 values are shown as mean values of two or more determinations. Standard deviation was typically within 2-fold from the IC 50 . c Enzyme selectivity calculated as HsNMT1 IC 50 (μM)/TbNMT IC 50 (μM).

Journal of Medicinal Chemistry
Article of Suzuki couplings, amidations, and Mitsunobu reactions were chosen to explore all positions around ring A ( Figure 3).
Crossing the blood−brain barrier (BBB) was an essential part of our chemistry design and presented its own challenges. Improving the BBB permeation of molecules has been widely studied and in silico prediction methods developed based on known CNS penetrant and nonpenetrant compounds. 32,33 Examination of the physicochemical properties of molecules and their influence on affecting BBB permeability has suggested some guiding principles and a physicochemical property range to increase the probability of improving the BBB permeability. 33 The top 25% CNS penetrant drugs sold in 2004 were found to have mean values of PSA (Å 2 ) 47, HBD 0.8, cLogP 2.8, cLogD (pH 7.4) 2.1, and MW 293. They suggested the following maximum limits when designing compounds as PSA < 90 Å 2 , HBD < 3, cLogP 2−5, cLogD (pH 7.4) 2−5, MW < 500. As this was the first round of compound design, we restricted the compounds to the following parameters: PSA 40−70 Å 2 , HBD < 3, cLogP 2−4.5, MW 250−400.
Virtual libraries of all possible compounds that could be constructed from our in-house chemical inventory were constructed and minimized to ensure that a wide region of chemical space was explored, and structures were not biased to one region. Reaction schemes, intermediates, and examples of compounds made are described in the Supporting Information.
■ RESULTS AND DISCUSSION Scaffold Array Results. No compounds made in the Suzuki chemistry (1, Figure 3) derived series had a potency <10 μM against TbNMT (see Supporting Information for compounds made). Table 1 shows the potency against TbNMT for selected examples from the amide (3−7), homologated amide (8−12), and ether series (17−24). The most potent compound in the amide series was 3 (TbNMT IC 50 1.7 μM). Amides directly linked to the phenyl ring in the 3-position were found to be more potent than the corresponding 4-substituted analogues (5 vs 4 and 7 vs 6). The homologated amide series in comparison to the amide series were on the whole >3-fold more potent (6 vs 12) with the most potent compound achieving a TbNMT IC 50 value of 0.07 μM (10). In the homologated amides series the 4position amides showed greater potency than the corresponding 3-position analogues (the opposite trend to the amide series). Further optimization of both the directly linked and homologated amide series failed to improve the potency or the pharmacokinetic properties.
The ether array produced compounds with good levels of activity against TbNMT, the most active of these achieved an IC 50 value of 0.5 μM (24). The more potent compounds were substituted in the 4-position, on average showing around 10-fold greater potency over their 3-position analogues, e.g., 3-position compound 22 (35 μM) vs 4-position compound 23 (1.2 μM) or 3-position compound 20 (13 μM) vs 4-position compound 21 (1.4 μM). Interestingly, the replacement of the sulfonamide in structure 1 with an ether linkage (17) was completely inactive against TbNMT (IC 50 > 100 μM). This was surprising, as methyl ethers are considered possible sulfonamide bioisosteres. 31 Compound 24 was not selective over human NMT (HsNMT1) but exhibited an EC 50 of 2 μM in the T. br. brucei proliferation assay, with good microsomal stability and moderate levels of selectivity against proliferating human MRC-5 cells (Figure 4).
The crystal structure of 24 bound to Af NMT ( Figure 4A) shows the ligand binds in the peptide binding groove in an overall U-shaped conformation, with the ligand wrapping round the side chain of Phe157. The central aryl rings of 24 lie perpendicular to each other allowing the ligand to sit in the cleft formed by the side

Journal of Medicinal Chemistry
Article chain of Tyr263, Tyr393, and Leu436. The cleft is formed by the movement of the side chain of Tyr273; a feature observed in the binding mode of benzofuran ligands 26,27 and subsequent derivatives. 34 The pyridyl nitrogen of 24 forms an interaction with Ser378 in a similar orientation as the trimethyl-pyrazole group of 1, and the piperidine moiety interacts directly with the C-terminal carboxyl group of Leu492.
Compound 24 does not interact with His265, an interaction formed by the sulfonamide in 1 (overlaid with 24, Figure 4B), which potentially explained the drop off in potency between 1 (TbNMT 0.002 μM) and 24 (0.5 μM). Despite this loss of activity, 24 had comparable ligand efficiency (LE) 35 of 0.33 to 1, LE = 0.36, and in combination with the observed binding mode, gave us confidence that the design strategy was valid.
Optimization of Compound 24. With the aim of increasing potency against TbNMT, the diphenyl piperidine ring was replaced with the dichlorophenyl-pyridyl-piperidine moiety of 1. This change reduced the logP by ∼1 log unit from 4.3 for 24, with an increase in PSA from 34 Å 2 (19) to 50 Å 2 , which was within the acceptable guidance limits for BBB permeability 32,33 to give 29 (synthesis shown in Scheme 1).
Compound 29 ( Figure 5) exhibited a 4-fold improvement in potency against TbNMT (IC 50 0.1 μM) and improved efficacy in the T. br. brucei proliferation assay (EC 50 0.7 μM), while retaining good microsomal stability (1.4 mL/min/g) and LE (0.33). Encouragingly, 29 showed good levels of brain penetration (brain−blood = 0.4), a significant improvement over 1 (brain− blood < 0.1), 22 indicating that the strategy of reducing MW and PSA was a valid approach (1, PSA 101 Å, MW 495). The crystal structure of 29 bound to Af NMT ( Figure 6) was determined showing the ligand adopted a conformation similar to 1 with the biaryl system sitting in plane with the 2,6-dichlorophenyl ring stacking in plane with the side chain of Tyr273. Key interactions between the piperidine N to Ser378 and the piperazine to the Cterminal carboxyl group are retained from 24.
Replacement of the 2,6-Dichlorophenyl Ring. Optimization of 29 focused on modifications to the central 2,6dichlorophenyl ring to increase enzymatic selectivity relative to HsNMT1 (0.3 μM, 3-fold compared to TbNMT IC 50 ). These modifications were made employing the same chemistry as outlined in Scheme 1, by varying the starting substituted bromophenol used in the Mitsunobu step. These 2,6dichlorophenyl modifications are detailed in Table 2.
None of the core modifications improved potency against TbNMT when compared to 29 (Table 2) nor LE and enzyme selectivity, although some demonstrated increased levels of potency against HsNMT1 (37, HsNMT1, IC 50 0.01 μM). The reason for this increase in HsNMT1 activity was not explained using the available crystal structure data. Certainly inhibitors of human NMT such as 37 are of potential interest in the treatment of cancer, 20 and further elaboration of the core could be explored.
Pyridyl Headgroup SAR. The next phase of optimization focused on modifications to the ether pyridyl ring of 29 shown in Table 3. These compounds were made using the same common phenol intermediate (Scheme 2), applying solid phase reagents such as polystyrene bound triphenylphosphine, and running reactions and purifications in parallel using commercially available alcohols or alcohols derived from commercially available carboxylic acids or esters after reduction with borane or lithium aluminum hydride (see Supporting Information).
Modifications to the pyridyl headgroup showed encouraging results with 47 equipotent to 29 (IC 50 ≈ 0.1 μM) but with ∼65fold selectivity over HsNMT1, equivalent activity in the T. br. brucei proliferation assay, and promising microsomal stability (C int 4.2 mL/min/g). Compound 30, though, had equivalent activity to 29 in the MRC-5 counter screen, indicating that HsNMT1 activity may not have been driving the MRC-5 toxicity.
Homologation of the linker to the pyridyl group did not improve potency, as did groups on the pyridyl ring at the 6-(48) or 4-positions (49), though both 48 and 49 showed equivalent activity in the T. br. brucei proliferation assay to 29. The crystal structure of 29 overlaid with the trimethylpyrazole of 1 suggested that additions of methyl substitution may have been beneficial to potency ( Figure 7A) because the trimethyl substitution of pyrrole in 1 was essential for activity. Subsequent crystal structures of 48 showed that the binding pocket the pyridyl headgroup accesses is small and that these substituents in the case of 49 forced the ether pyridyl ring to twist in the pocket to avoid steric clashes with its dichlorophenyl ring, and for 48, the 4methyl forces the pyridyl ring out of the pocket. In both cases, the direct hydrogen bond from the pyridyl nitrogen to the serine was broken, but 48 still formed an interaction, though this was now water mediated ( Figure 7B).
Alternative Nonpyridyl Head Group SARs. To advance the series, two regions within the structure were modified with the aim to improve potency, first examining pyridyl replacements and modifications to the pyridyl ring and replacement of the piperazino-pyridine moiety. First, the pyridyl ring was replaced with a range of five-membered heterocycles, mainly thiazoles, with various substitutions; see Table 4. The most potent of these showed levels of promising activity against TbNMT (IC 50 ≈ 0.05−0.06 μM; 58 and 57). The SAR around 57 was tight. The removal of either methyl groups (60 and 65) lost activity against TbNMT; in addition, substitution of the 2-methyl group with either ethyl (64) or isopropyl (66) lost all activity in the T. br. brucei proliferation assay. Compound 57 showed good stability to microsomal turnover (C int 2.4 mL/min/g) but also improved selectivity over MRC-5 cytotoxicity. Both 58 and 57 showed equivalent levels of potency against HsNMT1 (IC 50 ≈ 0.03−0.08 μM) and again showed very different MRC-5 activities, indicating that MRC-5 toxicity may not be entirely driven by HsNMT1 activity.
Replacement of the Piperazino-Pyridine Moiety. We had previously validated TbNMT as a druggable target in the

Journal of Medicinal Chemistry
Article stage 2 model for HAT in mice using 68 as a model compound ( Figure 8A). 24 Compound 68 showed good potency in the T. br. brucei proliferation assay at EC 50 0.001 μM and improved levels of selectivity over MRC-5 cells when compared to 1. We examined hybridizing the 4-C chain derivative of 68, 69, which showed equally good efficacy and potency, and 29 to increase efficacy in the T. br. brucei proliferation assay. Compound 71 (synthesis in Scheme 3) showed increased selectivity over MRC-5 cells and HsNMT1 but showed a significant drop off in efficacy in the T. br. brucei proliferation assay. This was potentially caused by the significant increase in lipophilicity of 71 (logP 5.5) compared to 29 (logP 3.4), resulting in an increased level of nonspecific protein/membrane binding. Given the more favorable logP of 29, further optimization focused on derivatives of 29 rather than 71. Compound 70 (Scheme 3), the NH piperidine of 71, showed no in vitro activity against TbNMT.
Pyridyl Headgroup Optimization. The crystal structure of 29 ( Figure 6) indicated that the methyl substituent on the pyridyl ring was pointing into a small pocket. Chemistry was developed to explore this pocket with various hydrophobic and polar groups as detailed in Scheme 4. Using a common intermediate (ethyl 2chloronicotinate, 72), Suzuki and Negishi reactions were used to install aromatics rings (73) and alkyl groups (74a−c), respectively. Amines were installed through displacement of the chlorine of 72. After reduction of the ethyl esters (73−75) to the corresponding alcohols (76−78), they were reacted using standard Mitsunobu conditions (Scheme 2) to give final products detailed in Table 5.
Good levels of inhibition of TbNMT were observed for all compounds prepared (except 79), some with improved potency over 29. The loss of activity of 79 was most probably caused by the alkoxy-group reducing the basicity of the pyridine ring, making the ring nitrogen a poorer HBA. Compounds 81 and 82 showed promising potency against the parasite (EC 50 = 0.1 μM), with good selectivity compared to MRC-5 cells (81), and had good microsomal stability (81, 1.2 mL/min/mg; 82, 1.6 mL/ min/mg). Compound 81 ( Figure 9) showed significant levels of brain penetration (brain−blood ratio = 1.9), a significant improvement on 29 (brain−blood ratio 0.4) and 1 (brain− blood ratio < 0.1). Compound 81 represents a good lead for further optimization to identify development candidates for stage 2 HAT.

■ CONCLUSIONS
By using 1 as a starting point to identify alternative TbNMT inhibitor scaffolds with physicochemical properties suitable for penetration into the brain to treat stage 2 HAT, we identified an ether linker as a replacement of the sulfonamide of 1. This modification reduced molecular weight and polar surface area, producing a viable alternative series with excellent levels of brain penetration. This work highlights the importance of decreasing the PSA as a way of increasing the probability of brain penetration. Further optimization identified compounds with good levels of TbNMT and T. br. brucei antiproliferative activity and microsomal stability. Though in comparison with the original structure 1, further potency gains against the enzyme and in the parasite proliferation assay are required. This series

Journal of Medicinal Chemistry
Article presents good leads to identify potential development candidates for stage 2 HAT.

■ EXPERIMENTAL SECTION
Synthetic Materials and Methods. Chemicals and solvents were purchased from the Aldrich Chemical Co., Fluka, ABCR, VWR, Acros, Fisher Chemicals, and Alfa Aesar and 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 precoated 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 Presearch ISCO Combiflash Companion 4X or Combiflash Retrieve. 1 H NMR and 13 C NMR spectra were recorded on a Bruker Avance II 500 spectrometer ( 1 H at 500.1 MHz, 13 C at 125.8 MHz) or a Bruker DPX300 spectrometer ( 1 H at 300.1 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), pentet (p), multiplet (m), broad (br), 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. LC−MS chromatographic separations were conducted with a Waters Xbridge C18 column, 50 mm × 2.1 mm, 3.5 μm particle size; mobile phase, water/acetonitrile + 0.1% HCOOH, or water/acetonitrile + 0.1% NH 3 ; linear gradient 80:20 to 5:95 over 3.5 min, and then held for 1.5 min; flow rate 0.5 mL min −1 . All assay compounds had a measured purity of ≥95% (by TIC and UV) as determined using this analytical LC−MS system; a lower purity level is indicated. High-resolution electrospray measurements were performed on a Bruker Daltonics MicrOTOF mass spectrometer. Microwaveassisted chemistry was performed using a Biotage Initiator Microwave Synthesizer.

Journal of Medicinal Chemistry
Article through a phase separation cartridge. The organic layer was then absorbed onto silica and purified by flash column chromatography running a gradient from 0% ethyl acetate/hexane to 50% ethyl acetate/ hexane to give 15 as a colorless oil (1.76 g, 85% yield). 1

Journal of Medicinal Chemistry
Article mg, 0.42 mmol, 1.5 equiv), diisopropyl azodicarboxylate (DIAD, 66 μL, 0.34 mmol, 1.2 equiv) in anhydrous dioxane (5 mL) in a capped test tube was heated at 60°C for 16 h. The reaction was absorbed onto silica and purified by flash column chromatography running a gradient from 0% ethyl acetate/hexane to 100% ethyl acetate. The resulting product was evaporated in vacuo before dissolving in dichloromethane (10 mL), addition of trifluoroacetic acid (10 equiv) and stirring at RT for 3 h. The reaction was then evaporated in vacuo before dissolving in dichloromethane and loading onto a prewashed SCX cartridge. The SCX cartridge was washed with dichloromethane (3 × 10 mL) and MeOH (3 × 10 mL) before eluting the product with 7 N ammonia in methanol. This was evaporated to give 17 (64 mg, 61% yield). 1 (Table 2).
Test compound (0.4 μL in DMSO) was transferred to all assay plates using a Cartesian Hummingbird (Genomics Solution) before 20 μL of enzyme was added to assay plates. The reaction was initiated with 20 μL of a substrate mix and stopped after 15 min (HsNMT1 or HsNMT2) or 50 min (TbNMT) with 40 μL of a stop solution containing 0.2 M phosphoric acid, pH 4.0, 1.5 M MgCl 2 , and 1 mg mL −1 PVT SPA beads (GE Healthcare). All reaction mix additions were carried out using a Thermo Scientific WellMate (Matrix). Plates were sealed and read on a TopCount NXT Microplate Scintillation and Luminescence Counter (PerkinElmer).
ActivityBase from IDBS was used for data processing and analysis. All IC 50 curve fitting was undertaken using XLFit version 4.2 from IDBS. A four-parameter logistic dose−response curve was used using XLFit 4.2 Model 205. All test compound curves had floating top and bottom, and prefit was used for all four parameters.
Compound Efficacy and Trypanocidal Activity in Cultured T. brucei Parasites. Bloodstream T. b. brucei s427 was cultured at 37°C in modified HMI9 medium (56 μM 1-thioglycerol was substituted for 200 μM 2-mercaptoethanol) and quantified using a hemocytometer. For the live/dead assay, cells were analyzed using a two-color cell viability assay (Invitrogen) as described previously. 22 Cell culture plates were stamped with 1 μL of an appropriate concentration of test compound in DMSO followed by the addition of 200 μL of trypanosome culture (10 4 cells mL −1 ) to each well, except for one column, which received media only. MRC-5 cells were cultured in DMEM, seeded at 2000 cells per well, and allowed to adhere overnight. One microliter of test compound (10 point dilutions from 50 μM to 2 nM) was added to each well at the start of the assay. Culture plates of T. brucei and MRC-5 cells were incubated at 37°C in an atmosphere of 5% CO 2 for 69 h, before the addition of 20 μL of resazurin (final concentration, 50 μM). After a further 4 h incubation, fluorescence was measured (excitation 528 nm; emission 590 nm) using a BioTek flx800 plate reader.
■ ASSOCIATED CONTENT * S Supporting Information fThe Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jmedchem.7b01255.
Experimental details for compounds 3−12 and 17−24; Xray data collection and refinement statistics; correlation of enzyme activity data for inhibitors against Af NMT and TbNMT ; molecular structures of known NMT inhibitors (PDF) Molecular formulas strings (CSV)