Establishment of a Structure–Activity Relationship of 1H-Imidazo[4,5-c]quinoline-Based Kinase Inhibitor NVP-BEZ235 as a Lead for African Sleeping Sickness

Compound NVP-BEZ235 (1) is a potent inhibitor of human phospoinositide-3-kinases and mammalian target of rapamycin (mTOR) that also showed high inhibitory potency against Trypanosoma brucei cultures. With an eye toward using 1 as a starting point for anti-trypanosomal drug discovery, we report efforts to reduce host cell toxicity, to improve the physicochemical properties, and to improve the selectivity profile over human kinases. In this work, we have developed structure–activity relationships for analogues of 1 and have prepared analogues of 1 with improved solubility properties and good predicted central nervous system exposure. In this way, we have identified 4e, 9, 16e, and 16g as the most promising leads to date. We also report cell phenotype and phospholipidomic studies that suggest that these compounds exert their anti-trypanosomal effects, at least in part, by inhibition of lipid kinases.


■ INTRODUCTION
Human African trypanosomiasis (HAT), also known as sleeping sickness, is caused by an infection with a subspecies of the eukaryotic protozoan parasite, Trypanosoma brucei. Trypanosoma brucei gambiense occurs in Western and Central Africa and is responsible for over 90% of all reported cases of infection, whereas Trypanosoma brucei rhodesiense is found in Eastern and Southern Africa. Both subspecies are transmitted by the bite of infected tsetse flies, and it is estimated that about 60 million people are at risk in sub-Saharan Africa, with at least 10 000 cases occurring annually. 1 Established therapies were introduced in the mid-to-late 20th century and have severe safety and efficacy limitations, and drug resistance is emerging against some treatments. 2 Thus, there is an urgent need to develop new safe, effective, and affordable therapeutics that can be orally administered and are stable under tropical conditions. 3 However, financial incentives for drug discovery against HAT are quite limited because of the economically disadvantaged regions where this disease is endemic. As a strategy to overcome this disincentive for drug discovery, we have hypothesized that medicinal chemistry knowledge against classes of human drug targets could be repurposed to facilitate rapid and cost-effective drug discovery against parasite drug targets. In this "target repurposing" approach, existing drugs and drug-like compounds serve as early hits or leads from which to optimize parasite-specific therapeutics. 4 Kinase inhibitors represent one promising class of compounds in both humans and parasites. As a pivotal class of enzymes central to cellular signaling, kinases have been identified as key targets for inflammation, 5,6 cancer, 7 and a wide range of other therapeutic indications. Indeed, kinases are estimated to represent 22% of the druggable human genome. 8 The genome of T. brucei encodes 176 kinases, and the kinome of the related parasites Trypanosoma cruzi and Leishmania spp. consists of highly orthologous enzymes, 9,10 some of which are beginning to emerge as druggable targets of potential intervention for such parasitic infections. 11−15 We recently reported that NVP-BEZ235 (1, Figure 1A), currently a phase III clinical candidate for cancer, showed a subnanomolar growth inhibitory phenotype in T. brucei and good-to-modest activities against cultures of T. cruzi and Leishmania major. 12 Recognizing, of course, that 1 is a potent human kinase inhibitor, we started to study the structure− activity relationships (SAR) of this class of compounds in an attempt to reduce the inherent host cellular toxicity and to allow assessment and subsequent improvement of the selectivity profile over human kinases. Furthermore, the physicochemical properties of 1 do not lend it to CNS exposure (a requirement for HAT therapeutics), as suggested by GSK internal models of CNS penetration and by other predictive models recently disclosed, such as the central nervous system multiparameter optimization (CNS MPO) score. 18

■ RESULTS AND DISCUSSION
In order to establish the SAR of this chemotype, we looked toward the docking studies of 1 that were previously reported using a homology model of the human kinase domain of PI3Kγ, showing that the binding of the inhibitor to the hinge region of the kinase is made mainly through three H-bond interactions ( Figure 1A). 16 We anticipated that such interactions would also be important for the parasitic kinase(s) by which 1 effected its potent growth inhibition. We therefore divided the structure of 1 into three regions (R 1 , R 2 , and R 3 ), as shown in Figure 1B, for systematic modulation.
The docking studies mentioned above showed that the nitrogen atom of the quinoline substituent in the R 3 position could play an important role in binding to Asp933. Very recently, a new mTOR/PI3K dual inhibitor, PF-04979064 (2, Figure 1B), was disclosed by Pfizer that contains a 3-pyridine in the same position as the 3-quinolinyl, reinforcing the importance of this substituent for activity against the human targets. 17 To evaluate the importance of this region of the molecule to confer activity against the parasite, we replaced the quinoline with a variety of aromatic and nonaromatic substituents.
Preparation of compounds 4−10 was accomplished via various cross-coupling reactions using common intermediate 3 (Scheme 1); Suzuki conditions were employed in the preparation of compounds 4, direct amination using a  16 (B) General regions of the compound's core and the structure of 2, a recently disclosed mTOR/PI3K inhibitor. 17

Scheme 1 a
Buchwald−Hartwig reaction for compounds 5, or coppercatalyzed conditions in aqueous ammonia for compound 6. Compound 10 was prepared utilizing Sonogashira coupling conditions. Palladium-mediated dehalogenation of 3 provided 9. Compound 8 was easily obtained utilizing Negishi coupling conditions with chloro(methyl)zinc and palladium catalyst. Biological assessments of these compounds are summarized in Table 1 and are discussed below. Synthesis of analogues with variations of both R 1 and R 3 was achieved starting from 11, which was prepared via a three-step sequence starting from 2-amino-5-iodobenzoic acid with an overall yield of 51% (Scheme 2). 19 Preparation of anilines 12 was accomplished via nucleophilic aromatic substitution with the requisite amine, and nitro group reduction using iron provided intermediates 13. Cyclization of the imidazolidinone ring was obtained with diphosgene (compounds 14), and Nmethylation gave intermediates 15. Final compounds (16 and 17) were synthesized from 14 and 15 using Suzuki crosscouplings in a microwave reactor. We also prepared the ringopened analogue 20 from 13 using Suzuki conditions. Structure and biological activity of the compounds prepared in Scheme 2 are summarized in Table 2.
Biological Assessments. As shown in Table 1, replacing the quinoline with a 3-pyridine (4b), with the nitrogen atom in the same spatial region, decreased potency by 8-fold (EC 50 = 16 nM) compared to that of 1 and exhibited high HepG2 cytotoxicity (TC 50 = 575 nM). The same 3-pyridine with an ethyne spacer (10) only loses 4-fold activity (EC 50 = 8 nM), but it displays even worse selectivity. In fact, 10 is a clinical backup molecule for 1 from Novartis, 20 so this host cell activity is not unexpected.
Installation of a 4-pyridine (4a) at the R 3 position also rendered a potent molecule (EC 50 = 136 nM), with 25-fold selectivity over host cells. Overall, the pyridine substituents presented a more favorable profile than an unadorned aromatic phenyl ring (4c), which has lower potency and reduced selectivity.
Following the promising profile exhibited by the pyridinesubstituted analogues, other heteroaromatic substituents were explored. For example, the five-membered ring methylimidazole afforded a highly potent compound, 4e (EC 50 = 51 nM), albeit with only a 12-fold selectivity window. However, more bulky benzothiophene 4f was similarly potent (54 nM), yet with no HepG2 toxicity observed. Table 2. Biological Data for R 1 and R 3 Variations a Data were not obtained due to low solubility. b EC 50 average was obtained from two replicates, with SD lower than 0.1 log units. c SD lower than 0.2 log. d TC 50 average was obtained from two replicates, with SD lower than 0.5 log units. e Selectivity = TC 50 /EC 50 . f Stock concentration: 2.5 mM.
Introducing cyclic or linear amines or amides at the R 3 position led to complete loss of activity against T. brucei and HepG2. Eliminating R 3 substituents from 1 altogether (9) resulted in a ∼300-fold loss of potency, although this compound still remained in submicromolar range (EC 50 = 624 nM). We observe that an aromatic system (preferably heteroaromatic) is needed in the R 3 region to afford potent anti-trypanosomal activity (cf. 8). Nevertheless the choice of this group influences HepG2 cytotoxicity and the overall physicochemical profile of the molecule.
In the R 1 position, we first evaluated the importance of the substituents in the para position of the aromatic ring in R 1 with 16a−c ( Table 2). These derivatives exhibited a 10−15-fold decrease in potency when compared to that of 1. Surprisingly, the most potent compound was the one without a para substituent (16c, EC 50 = 24 nM), and both the nitrile (16b) and trifluoromethyl (16a) analogues showed similar antitrypanosomal potency (EC 50 = 91 and 103 nM, respectively). In addition, these three compounds exhibited an excellent selectivity profile against HepG2. This data shows that when R 3 = 3-quinolinyl the presence of a substituent in the para position of the phenyl ring of the R 1 position is important, but not essential, for anti-trypanosomal activity. Interestingly, when the R 3 group is 4-pyridyl, the presence of a nitrile group (16d and 16f) or amine moiety (16h) in the R 1 para position seems to afford improved potency over p-methyl or methoxy (16e and 16g). This is perhaps consistent with the predicted binding of 1 to the human p110α orthologue, where the nitrile established hydrogen bonds with Ser774, contributing to a strong binding to the human target. 16 We observe a loss in activity when substituents on both the R 1 and R 3 substituents are eliminated simultaneously (23, Table 3).
One of the medicinal chemistry goals of this work was to improve the physicochemical properties of the new compounds. Compound 1 is very lipophilic (cLogP = 5.81, chromLogD = 4.75) 21 and displays very poor solubility (19 μM) (Supporting Information, Table S2). 22 The high number of aromatic rings surely contributes to the establishment of favorable π-stacking interactions, resulting in low solubility. Not unexpectedly, R 3 quinolinyl compounds 16a−c also presented very low solubility. However, with a 4-pyridyl substituent at R 3 , a drastic reduction in cLogP was observed (compare 1 vs 4a and 4c), and one of these compounds (4a) maintained good antiparasitic activity (EC 50 = 136 nM, TC 50 = 3.43 μM).
Using 4a as reference, increase of potency was achieved by removing the two α-methyl groups (16d) or by replacing the −C(CH 3 ) 2 CN group by a dimethyl amine (16h). We hypothesize that the improved positioning of a nitrogen lone pair for hydrogen bonding lends 16h to having improved potency. However, introduction of a methyl (16e) slightly decreases potency relative to that of 4a, and incorporation of a methoxy group (16g) has little impact on activity. Addition of a larger, nonaromatic heterocycle or replacement of the R 1 group with methyl reduces or abrogates activity (16i−k).
The R 2 substituent on the urea ring adjacent to the quinoline core was also explored in terms of methylation versus nonmethylation of the free NH for compounds 16/17b, 16/ 17d, 16/17h, and 18/19 (Table 3). However, no clear trends were evident. Compound 16h is also the only example where methylation afforded toxicity in HepG2 (TC 50 = 2.34 μM). Opening the imidazolone ring the carbonyl group in the "ureaimidazo" ring seems to play an important role. Disruption of the aromatic ring, affording the free amine (20), led to a significant loss in potency (EC 50 = 0.767 μM). From a physicochemical property perspective, we do not observe a trend in solubility and permeability when comparing R 2 = Me versus H.
As mentioned above, we note that 1 does not penetrate the blood−brain barrier (BBB), a crucial requirement for new drugs in order to be effective in the second stage of HAT. In evaluating analogue designs, we employed the multiparameter optimization (MPO) algorithm developed by Pfizer scientists Table 3 a EC 50 average was obtained from two replicates, with SD lower than 0.1 log. b TC 50 average was obtained from two replicates, with SD lower than 0.5 log. c Selectivity = TC 50 /EC 50 . f Stock concentration: 2.5 mM. g Stock concentration: 1.25 mM.
for CNS-active drugs. 18,23 This aforementioned publication showed a correlation between six fundamental physicochemical properties (cLogP, cLogD, MW, TPSA, hydrogen-bond donors, and the pK a of the most basic center) and their impact on CNS-acting agents as well as other critical ADME, toxicity, and binding efficiencies. By summing scores for each of these six properties, one can ascertain the likelihood that a given compound should penetrate the CNS. Importantly, a subsequent report successfully applied this approach prospectively for design of CNS-penetrant PI3K-α inhibitors. 24 As a point of reference, the CNS MPO score of 1 is 3.22, which is lower than the desirable range of CNS-active molecules (≥4). From the new analogues synthesized and tested here, only four shown in Tables 1−3 scored equal to or worse than 1, and these were primarily due to unfavorable LogP, LogD, and molecular weight values. The MPO score of our most potent analogue (4b) improved to a score of 3.9, which is a direct effect of MW and cLogP decrease compared to those of 1. However, although 16d is slightly less potent than 4b, we observe significant improvement in cellular selectivity and predicted CNS activity (MPO score = 4.8).
Biochemical Kinase Selectivity. On the basis of the cellular potency and selectivity profiles, we prioritized 14 compounds, plus 1, to be tested against PI3K-α, -β, -γ, and -δ as well as mTOR, and this data is shown in the Supporting Information (Table S1). We note that although seven of these compounds retain nanomolar levels of mTOR activity, in most cases we do observe decreases in potency against the mammalian kinases tested.
Next, on the basis of an evaluation of anti-trypanosomal activity, selectivity, MPO score, and the biochemical selectivity assays, we selected four compounds (plus compound 1) to initiate mechanism of action studies. These four prioritized compounds are shown in Table 4.
Phospholipidomics Analysis. We originally selected 1 to investigate its potential as an trypanocidal agent on the basis of its activity as a PI3K/mTOR inhibitor in mammalian cells and the fact that PI3K activity is essential for the bloodstream form of the T. brucei parasite. 25 In the interest of elucidating the potential mechanism of action against T. brucei cells, we performed a lipidomic analysis to discern if PI kinases are likely targets of these analogues. Lipid extracts of cells grown in the presence of the best compounds at sublethal doses (200 nM) for 12 h were analyzed by ES-MS. Survey scans in positive and negative ion mode between 600 and 1000 m/z showed a wide range of expected phospholipid species (Supporting Information Figures S1 and S2, respectively). There were no significant differences in the phosphatidylcholine (PC) and sphingomylein (SM) species between the control and cells grown in the presence of the other compounds ( Figure S1). Additionally, only minor differences were observed in the negative ion mode survey scans between the control and the treated cells ( Figure  S2). However, variations were observed in the relative ratios of the PI species 862 m/z (18:0/18:2), 886 m/z (18:0/20:4), and 912 m/z (18:0/22:5), as well as relative to 826 m/z PG (18:0/ 22:4). This could be a reflection in changes in either the formation of PI (hence, PG increases as PI decreases) because they have a common lipid donor (cytidine-diphosphatediacylglycerol) or variation in the usage of PI (i.e., for PIP formation). Alternatively, as is often the case, changes in the lipid profile could be a reflection in the cells succumbing to (or adapting to) a defect in a major cellular process, such as endoor exocytosis or cell cycle arrest.
Because these compounds are thought to target PI3K, negative ion mode scans between 950 and 1300 m/z were conducted to focus upon singly and multiply phosphorylated PIs. As seen in Figure 2A for the control cells, PIP species are observed at ∼992 m/z and ∼1040 m/z, corresponding to 40:5-8 and 44:7-9, respectively, as well as PIP 3 (44:9-12) species at ∼1215 m/z.
Compounds 1 and 16g ( Figure 2B,C) show almost complete absence of all phosphorylated PIs, clearly indicating that these compounds are targeting PI kinases (e.g., PI3K). Compounds 4e and 16e ( Figure 2C,E) also show significant (albeit incomplete) reduction in PIP levels, although this reduction is significant compared to that in the control cells (Figure 2A), indicating that they are affecting metabolism of PIPs, most likely phosphorylation of PI. Cell Phenotype Analysis. BEZ235 is known to inhibit human PI3K/mTOR; thus, we used the human kinase domain of PI3K-γ (including Ser774, Val855, and Asp933, mentioned above) to search in the T. brucei genome geneDB for homologous enzymes. Two high-homology PI3Ks were identified: the previously characterized TbPI3KIII (Tb927.8.6210), which is orthologous to the yeast Vps34, and an uncharacterized, potentially pseudo TbPIK (Tb927.11.15330). In addition, three of the four TORs described in trypanosomes, TbTOR1, TbTOR2, and TbTOR3, were also identified. In the sequences of these five proteins, similar positions for the important residues, Ser774, Val855, and Asp933, were observed. To resolve the possible target(s) of compound 1, we compared the phenotype of cells treated with 1 with the previously described phenotypes caused by RNAi of these various kinases. TbPI3KIII depletion in T. brucei induced a phenotype characteristic of a defective endocytosis, 25 similar to the phenotype observed in cells treated with 1. Furthermore, ES-MS analyses show complete absence of PIPs, suggesting that 1 is indeed targeting one or more of the PI/PIP kinases, including TbPI3KIII ( Figure 2).
Highly efficient and rapid endocytosis is one of the most important cellular processes for this infective parasite to survive and multiply within the bloodstream of the host, and the activity of TbPI3KIII is essential for this process. 25 To gain insight on the possible target(s) of our compounds, we carried out transferrin uptake experiments to measure receptormediated endocytosis after drug treatment. The analogues of 1 showed varied trypanocidal potency, so we decided to use concentrations twice that of the EC 50 of each of the four compounds in Table 4 over a short treatment period (18 h). These transferrin uptake assays showed that 1 inhibited endocytosis dramatically, as did compounds 16e and 16g ( Figure 3). In addition, the fact that PIP3 levels were dramatically reduced after treatment with compounds 1 and 16g further supports the hypothesis they are affecting PIK activity, including TbPI3KIII ( Figure 2). Compound 16e induced a phenotype defective in endocytosis ( Figure 3); however, the PIP 3 molecular species identified (42:7-9) was not the same PIP 3 molecular species identified in WT cells (44:9-12). These data suggest that 16e reduces PIP 3 44:9-12 formation by inhibiting similar target(s) to that of compounds 1 and 16g, but it may have an additional mode of action because it causes PIP 3 42:7-9 to be observed, which may be present in a different subcellular compartment or organelle.
Conversely, compound 4e showed a different phenotype, as it did not significantly reduce transferrin endocytosis compared to that of the control (Figure 3). In addition, compound 4e affected cell cycle progression by reducing the proportion of cells in G1 and increasing multinucleated cells, whereas compounds 1, 16g, and 16e did not significantly alter cell cycle progression over this short treatment time (18 h) (Supporting Information Figure S3).
Collectively, these data suggest that compound 4e has a different mode of action than 1, 16e, and 16g because endocytosis was not affected, although the induction of changes in the PIP3 species ( Figure 2) suggests PIP metabolism was affected.  Previously published functional analysis showed that TbTOR2 depletion affected endocytosis and cytokinesis, 26 whereas rapamycin (a specific inhibitor of TbTOR2 in trypanosomes) treated cells induce different changes in the lipid profile than those from compound 1 (data not shown). The RNAi of additional TbTOR1 and TOR3 protein kinases 26,27 does not resemble the defect in endocytosis observed by the treatment with 1, suggesting that the TbTOR family of PIKK is not the main target of 1.
These studies cannot determine unequivocally which member of the trypanosome PI3K family may be the actual molecular target; however, our data strongly suggests that TbPI3KIII is the main target of BEZ235 as well as of derivative compounds 16e and 16g. Regardless, it is possible that additional molecular target(s) in the parasite may be affected In summary, we have developed initial SAR for analogues of 1 as anti-trypanosomal agents. We have identified regions of the molecule that allow improvement of selectivity over selected kinases and HepG2 cells, and we have measured and computed physicochemical properties that place analogues in a CNSpenetrant region of properties space, as predicted by the CNS MPO scoring regime. With this in mind, we highlight the compounds in Table 4 as our most promising leads in this series, as they possess significant overall improvements over 1.
We now have preliminary lipidomics and cell phenotype data to suggest primary targets of action within the lipid kinase pathways, and additional target identification work is continuing using orthogonal methods. Work is also ongoing to establish whether the predicted favorable properties hold true in pharmacokinetic and animal efficacy studies, which will be reported in due course.

■ EXPERIMENTAL SECTION
Chemistry. Starting materials were obtained from commercial suppliers and used without further purification unless otherwise stated. Flash column chromatography was carried out using prepacked Isolute Flash or Merck Si60 (15−40 μm) silica gel columns as the stationary phase and analytical grade solvents as the eluent unless otherwise stated. Proton magnetic resonance ( 1 H NMR) spectra were recorded at 400 MHz on a Bruker AMX400 spectrometer and are reported as follows: chemical shifts δ (ppm) (multiplicity, coupling constant in J (Hz), number of protons). Multiplicities are labeled s, singlet; d, doublet; t, triplet; m, multiplet; br, broad; or a combination of these. Total ion current traces were obtained for electrospray positive and negative ionization (ES+/ES−) on a Waters SQ detector. Analytical chromatographic conditions used for the LC/MS analysis were as follows. The column was an Acquity UPLC BEH C18 1.7 μm, 3 × 50 mm. Solvent A was an aqueous solvent consisting of 25 mM ammonium acetate. Solvent B was 90% acetonitrile + 10% water (pH 6.6). Additional chromatographic parameters were as follows: flow rate, 0.8 mL/min; injection volume, 2 μL; column temperature, 40°C; and UV wavelength range, 200−330 nm. The purity of all tested compounds was ≥95% using the analytical method described above unless otherwise stated. Compounds 10 and 11 20 were prepared as previously described, and their characterization is included in the Supporting Information.
General Procedure A: Aryl Halide Displacement. To a suspension of 4-chloro-6-iodo-3-nitroquinoline (11) (8 g, 24 mmol) in AcOH (100 mL) was added a solution of amine (1.1 equiv) in 70 mL of AcOH. The suspension was stirred for 3 h at rt under N 2 and quenched with H 2 O (150 mL). The resulting yellow precipitate was recovered by filtration, dissolved in CH 2 Cl 2 , washed with a saturated aqueous solution of NaHCO 3 (100 mL) and then H 2 O (100 mL), dried over MgSO 4 , filtered, and concentrated.
General Procedure B: Reduction of Nitroarenes. A suspension of nitroarene 12 (2.6 mmol) in ethanol (37 mL) was heated at reflux. To this mixture was added iron (10 equiv) followed by a solution of NH 4 Cl (1.42 g, 26 mmol) in H 2 O (11 mL). The resulting suspension was heated at reflux for 2 h. The hot mixture was then filtered through a Celite pad, and the filtrate was evaporated under vacuum. The residue was dissolved in EtOAc (30 mL) and washed with H 2 O (30 mL), and the aqueous phase was further extracted with ethyl acetate (2 × 20 mL). The organic extracts were combined, dried over MgSO 4 , filtered, and evaporated under vacuum.
General Procedure C: Formation of the C ring. A solution of diamine 13 (10.5 mmol) and triethylamine (1.2 equiv) in CH 2 Cl 2 (175 mL) was added dropwise at 0°C to a solution of trichloromethyl chloroformate (1.1 equiv) in CH 2 Cl 2 (50 mL), and the resultant solution was stirred for 2 h while being allowed to warm slowly to rt. The reaction was quenched with a saturated aqueous solution of NaHCO 3 (150 mL) and stirred for 2 h. Phases were separated, and the aqueous layer was further extracted with CH 2 Cl 2 (2 × 60 mL). Organic phases were combined, dried over MgSO 4 , filtered, and concentrated General Procedure D: N 1 Methylation. An aqueous solution of NaOH (88 mL, 0.15 M) was added to a mixture of intermediate 14 (8.8 mmol), methyl iodide (1.5 equiv), and tetrabutylammonium bromide (0.1 equiv) in CH 2 Cl 2 (180 mL). The resultant mixture was stirred for 24 h at rt in a sealed tube. The reaction was quenched with H 2 O (100 mL), phases were separated, and the aqueous layer was further extracted with CH 2 Cl 2 (3 × 180 mL). Volatiles were combined and evaporated under reduced pressure.
General Procedure E: Suzuki Coupling. A microwave vessel was loaded with aryl iodide 3 (0.149 mmol), K 2 CO 3 (1.5 equiv), Pd(PPh 3 ) 4 5 mol %), and boronate (1.1 equiv) in 1,2-dimethoxyethane (2 mL), EtOH (1 mL), and H 2 O (0.5 mL). The vessel was evacuated, backfilled with nitrogen, and sealed. The mixture was heated in a microwave reactor at 175°C for 15 min (including the ramp time). After cooling, the solids were removed by filtration, H 2 O (10 mL) and CH 2 Cl 2 (10 mL) were added, phases were separated, and the aqueous layer was further extracted with CH 2 Cl 2 (10 mL). Volatiles were combined and evaporated under reduced pressure. Purification was carried out by flash column chromatography over silica gel.
General Procedure G: Negishi Coupling. A microwave vessel was loaded with aryl iodide 3 (80 mg, 171 μmol), methylzinc chloride solution 2 M in THF (4 equiv), and palladium tetrakis(triphenylphosphine) (0.1 equiv) in THF (4 mL). The vessel was evacuated, backfilled with nitrogen, and sealed. The mixture was heated in a microwave reactor at 75°C for 1 h (including the ramp time). After cooling, the solids were removed by filtration. Saturated solutions of ammonium chloride (5 mL) and EtOAc (10 mL) were added, phases were separated, and the aqueous layer was further extracted with EtOAc (10 mL). Volatiles were combined and evaporated under reduced pressure. Purification was carried out by washing the resultant solid with MeOH. Prepared from 3-pyridinyl boronic acid using General Procedure E. Purification was carried out by flash column chromatography over silica gel (eluent: CH 2 Cl 2 /MeOH 100:0 to 95:5) to give 4b as an offwhite solid (yield: 40%). 1
Dose−Response Assay. For dose−response experiments, serial dilutions of compounds were plotted against compound concentration. Dose−response starting at 10 mM, unless indicated, with 3fold dilutions for 11 points were made in masterplates. Two hundred nanoliters per well from those masterplates was stamped in final assay plates. Controls of 0% response (control 1, 0.2 μL of 100% DMSO) and 100% response (control 2, 0.2 μL of 1 mM Pentamidine) were included in each assay plate in columns 6 and 18, respectively. Plates containing 0.2 μL of 100% DMSO were included in the assay to assess quality through the entire process. To detect growth inhibition, parasites in log phase growth were diluted to a working concentration of 2500 cells/mL in prewarmed HMI-9 medium and gently stirred until dispensation. Fifty microliters of culture was dispensed in compound-stamped black, clear-bottom, 384-well Greiner microplates using a Multidrop Combi Reagent Dispenser (Thermo Scientific) to give a final solvent concentration of 0.4% DMSO. Plates were covered with a lid, and cells were incubated for 70 h at 37°C and 5% CO 2 . After this period, 10 μL of 200 μM resazurin solution in prewarmed HMI-9 was added to each well, and plates were allowed to incubate 2 h more prior to fluorescence reading in a Wallac EnVision multilabel plate reader (PerkinElmer). Raw fluorescent data from the Envision plate reader were uploaded into GSK HTS database (ActivityBase). Activity of each well was normalized as a percentage of inhibition on a per-plate basis using the following equation 100 test well median CTRL1 median CTRL1 median CTRL2 where control 1 represents wells from the same plate containing 0.4% DMSO (0% inhibition, 100% grown control, n = 16) and control 2 represents wells from the same plate treated with pentamidine (100% inhibition, 0% grown, n = 16). A Z′ value greater than 0.4 was required for plate validation during the quality control process. A four-parameter equation describing a sigmoidal dose−response curve was then fitted with an adjustable baseline using ActivityBase XE Runner software. Fitting of dose−response curves and EC 50 determination were normalized as percentage of inhibition based on controls. The curve-fit model was based on a four-parameter logistic equation HepG2 Cell Culture and Growth Assays. The HepG2 cell line (human liver hepatocellular carcinoma cell line, ATCC) was cultured in Eagle's MEM supplemented with L-glutamine, Earle's salts, 10% heat-inactivated FBS, and 1% non-essential amino acids (NEAA) at 37°C and 5% CO 2 in T-175 vented flasks (Corning). The human biological samples were sourced ethically, and their research use was in accord with the terms of informed consent.
Cytotoxicity Assay. This assay was used as a selectivity assay, and compounds were tested at dose−response concentrations against HepG2 in order to identify the level of cytotoxicity. Dose− response starting at 10 mM, unless indicated, with 3-fold dilutions for 11 points was made in master plates (1536). Fifty nanoliters per well was stamped in final assay plates.
Log-phase HepG2 cells were removed from a T-175 TC flask using cell dispersion medium and dispersed by repeated pipetting. Cell density was adjusted to 60 000 cells/mL as the working concentration in prewarmed Eagle's MEM. The seeding density was checked to ensure that new monolayers were not more than ∼50% confluent at the time of seeding (typically 3000 cells per well), before completing preparation of the plates. Five microliters of culture was dispensed in compound-stamped TC treated, Greiner white 1536-well plates using a Multidrop Combi Reagent Dispenser (Thermo Scientific) to a final concentration of 1% DMSO. Cells were incubated for 48 h at 37°C and 5% CO 2 . Viability was determined by CellTiter-Glo kit (Promega) according to the manufacturer's instructions. Briefly, reconstituted CellTiter Buffer was equilibrated to room temperature prior use, and 5 μL per well was dispensed with a Multidrop Combi Reagent Dispenser. Contents were mixed on a plate orbital shaker and incubated for 10 min at room temperature to allow the signal to stabilize before luminescence reading on ViewLux Plate Reader (PerkinElmer).
Raw luminescence data from the ViewLux reader were uploaded into GSK HTS database (ActivityBase). Activity of each well was normalized as a percentage of inhibition on a per-plate basis using the following equation where control 1 represents wells from the same plate containing 1% DMSO (0% inhibition, 100% grown control, n = 128) and control 2 represents wells from the same plate treated with digitoxin (100% inhibition, 0% grown, n = 128). A Z′ value greater than 0.4 was required for plate validation during the quality control process.
As previously described for the primary assay, a four-parameter equation describing a sigmoidal dose−response curve was then fitted with an adjustable baseline using ActvityBase XE Runner software.
Phospholipidomics Analyses. Lipid Extraction. Total lipids from mid log phase cells were extracted by the method of Bligh and Dyer. 30 Briefly, mid log phase cells were collected by centrifugation (800g, 10 min), washed with PBS, resuspended in 100 μL of TDBglucose, transferred to a glass tube containing 375 μL of 1:2 (v/v) CHCl 3 /MeOH, and vortexed. The sample was agitated vigorously for a further 10−15 min. The sample was made biphasic by the addition of 125 μL of CHCl 3 and vortexed, and then 125 μL of H 2 O was added. The sample was vortexed again and centrifuged at 3000g at rt for 10 min. The lower (organic) phase was transferred into a new glass vial, and the aqueous phase was re-extracted with the fresh lower phase. The resultant lower phase lipid extract was dried under a stream of nitrogen and stored at 4°C.
Electrospray Mass Spectrometry Analysis. Lipid extracts were dissolved in 15 μL of CHCl 3 /MeOH (1:2) and 15 μL of acetonitrile/ isopropanol/water (6:7:2) and analyzed with a triple quadrupole mass spectrometer (Absceix 4000 QTrap) equipped with a nanoelectrospray source. Samples were delivered into the spectrometer using either thin-walled nanoflow capillary tips or a Nanomate interface in direct infusion mode (∼125 nL/min). The lipid extracts were analyzed in both positive and negative ion modes using a capillary voltage of 1.25 kV. MS/MS scanning (daughter, precursor, and neutral loss scans) were performed using nitrogen as the collision gas, with collision energies between 35 and 90 V. Each spectrum encompasses at least 50 repetitive scans. Tandem  Assignment of phospholipid species was based upon a combination of survey, daughter, precursor, and neutral loss scans as well previous assignments. 31 The identity of phospholipid peaks was verified using the LIPID MAPS: Nature Lipidomics Gateway (www.lipidmaps.org).
Cellular Phenotype Analyses. Transferrin Uptake. Two million trypanosomes were harvested and washed with TDB-glucose plus 1% BSA, resuspended in 250 μL in the same buffer, and incubated for 10 min at 37°C prior to the addition of 5 μg of AlexaFluor 488conjugated human holo-Tf (Invitrogen). Incubations were carried out for 0 (no Tf) or 3 min at 37°C, after which cells were immediately fixed in 4°C in 1% PFA diluted in cold PBS for at least 1 h. Parasites were finally washed twice with PBS and analyzed with a Becton Dickinson FACSCalibur flow cytometer (BD Biosciences) using BD CellQuest Pro version 4.0.2 software.
FACS Analysis. For cell cycle analysis, samples (1.5 × 10 6 cells) were collected, centrifuged (1400g at 4°C for 5 min), and washed in trypanosome dilution buffer (TDB). The cell pellets were gently suspended in 50 μL of TDB and permeabilized by adding 1 μL of saponin (25 mg/mL) for 3 min. It was then mixed with another 450 μL of TDB. RNase and propidium iodide (PI) were added to the suspension at final concentrations of 10 and 20 μg/mL, respectively, and the samples were incubated at room temperature for 30 min and then stored at 4°C. The DNA content of PI-stained cells and the percentage of cells in each phase of the cell cycle (10 000 cells per sample) were analyzed with a Becton Dickinson FACSCalibur flow cytometer (BD Biosciences) using BD CellQuest Pro version 4.0.2 software.
■ ASSOCIATED CONTENT * S Supporting Information Additional mass spectrometry data; data tables from this article annotated with NEU registry numbers; and biochemical and physicochemical assay details. This material is available free of charge via the Internet at http://pubs.acs.org. All of the data included in this work has also been made available as a publically available data set on www.collaborativedrug.com.