Discovery and Structure–Activity-Relationship Study of N-Alkyl-5-hydroxypyrimidinone Carboxamides as Novel Antitubercular Agents Targeting Decaprenylphosphoryl-β-d-ribose 2′-Oxidase

Magnesium plays an important role in infection with Mycobacterium tuberculosis (Mtb) as a signal of the extracellular environment, as a cofactor for many enzymes, and as a structural element in important macromolecules. Raltegravir, an antiretroviral drug that inhibits HIV-1 integrase is known to derive its potency from selective sequestration of active-site magnesium ions in addition to binding to a hydrophobic pocket. In order to determine if essential Mtb-related phosphoryl transfers could be disrupted in a similar manner, a directed screen of known molecules with integrase inhibitor-like pharmacophores (N-alkyl-5-hydroxypyrimidinone carboxamides) was performed. Initial hits afforded compounds with low-micromolar potency against Mtb, acceptable cytotoxicity and PK characteristics, and robust SAR. Elucidation of the target of these compounds revealed that they lacked magnesium dependence and instead disappointingly inhibited a known promiscuous target in Mtb, decaprenylphosphoryl-β-d-ribose 2′-oxidase (DprE1, Rv3790).


■ INTRODUCTION
Tuberculosis (TB) remains a global threat despite more than a century of research on its etiologic agent, Mycobacterium tuberculosis (Mtb). According to the WHO, in 2016, 10.4 million people were newly infected by TB, and 1.3 million people died from this disease. 1 Treatment of drug-sensitive TB requires at least 6 months of chemotherapy using drugs that were introduced into the market several decades ago, whereas treatment of multidrug-and extensively drug-resistant TB incorporate more drugs into an already complex treatment regimen, thereby increasing the duration of chemotherapy up to 24 months. The financial burden of TB chemotherapy also increases dramatically for the diagnosis and treatment of drugresistant TB and was estimated to have incurred a global cost of US$2.0 billion in 2017, severely straining resourceconstrained TB-control programs. 1 New agents that inhibit novel targets are one strategy to strengthen the repertoire of candidates in the drug-development pipeline that may lead to simplified chemotherapy for drug-sensitive TB and better options for drug-resistant-TB therapy. 2, 3 Mg 2+ is important in all bacteria, serving, for example, as an important counterion for ATP in numerous enzyme reactions, enabling the formation of the tertiary structure of the bacterial ribosome, and contributing to the stability of lipid bilayers in cell envelopes. 4−6 In Mtb, two systems have been implicated in Mg 2+ homeostasis, PhoPR and MgtC, which directly or indirectly exert broad control over various aspects of mycobacterial metabolism. The PhoP system, which in Enterobacteriaceae regulates Mg 2+ levels and controls virulence, 7 plays a role in controlling several families of virulenceassociated methyl-branched fatty acid containing acyl trehaloses that are unique to the pathogenic mycobacterial species. 8 PhoP-deficient mutants cannot grow in low concentrations of Mg 2+ , do not replicate in macrophages, and fail to establish or maintain infections in mouse lungs. 9 MgtC is a Mg 2+responsive virulence factor thought to function by interacting with the F 1 F 0 ATP synthase, 10 which was discovered in Mtb by homology to a well-studied analogue in Salmonella. 11 Like mutants lacking PhoP, mgtC mutants are impaired for growth in low concentrations of Mg 2+ in a mildly acidic environment, do not replicate in macrophages, and fail to grow in murine lungs and spleens. 12 The reliance by Mtb on Mg 2+ as both a signal and a cofactor spurred our interest in potential therapeutic strategies that impact these systems.
Target-based approaches to TB drug discovery have not afforded any of the drugs in the preclinical or clinical development pipeline. Whole-cell screening has been most successful in identifying hits of interest with subsequent efforts dedicated to lead development and unraveling mechanisms of action. 13 However, the hit rates of commercially available small-molecule libraries are often quite low, and the hits often possess low structural diversity and poor physicochemical properties. 14 In contrast, small-molecule collections based on privileged scaffolds offer the potential of libraries encompassing attractive physicochemical properties and containing core scaffolds known to target a diversity of cellular targets, thereby increasing the likelihood of acceptable hit rates with compounds that inhibit specific targets. Privileged structures are defined as chemical scaffolds that could afford potent and selective small-molecule regulators for different kinds of biomolecules. 14−16 Therefore, privileged structures typically display favorable druglike properties, resulting in more highquality leads. The dihydroxypyrimidine or pyrimidinedione scaffold is considered a privileged scaffold found in many biologically active molecules. The broad-ranging pharmacological roles generated by the pyrimidinedione moiety include antiviral agents, 17,18 antibacterial and antifungal agents, 19−21 and kinase inhibitors. 22 Among them, Raltegravir, a Mg 2+dependent integrase inhibitor developed by MSD, is a good example of an antiretroviral drug with a pyrimidinedione scaffold. 23 −25 In an effort to select hits that involve Mg 2+dependent processes, we were interested in exploring the activity of the Mg 2+ -chelating pyrimidinediones.
In this work, a novel series of N-alkyl-5-hydroxypyrimidinone carboxamides with antibacterial activity against Mtb is reported. The initial hit compound, P01, was derived from high-throughput screening of a library provided by St. Jude Children's Research Hospital that contained diverse pyrimidinedione derivatives as a privileged substructure. 26 Preliminary structure−activity-relationship (SAR) studies of the hit against Mtb were explored, as well as preliminary characterization of its biochemical target, decaprenylphosphoryl-β-D-ribose 2′-oxidase (DprE1), identified by resistance selection followed by whole-genome sequencing.

■ RESULTS AND DISCUSSION
High-Throughput Screening of Pyrimidinedione Library. The pyrimidinedione scaffold is a core for a diverse range of compounds inhibiting a multitude of potentially Mg 2+dependent targets on the basis of the cation dependence of Raltegravir binding to its target. As a result, we decided to screen under a growth condition that supported robust in vitro growth with modest Mg 2+ concentrations to exclude compounds that exert toxicity simply by cation chelation. Thus, a pyrimidinedione library containing 6207 compounds was screened for inhibition of growth of Mtb in a single-point assay at 10 μM final concentration. In medium containing glucose/glycerol/Tween 80 as the carbon sources with a final Mg 2+ concentration of 0.2 mM, 38 hits (hit rate of 0.6%) giving >50% growth inhibition over a 3 day incubation period were generated. These hits were further validated in dose titrations by full MIC determination. To observe any Mg 2+dependence of the hits and to diminish the effects of protein binding, these assays were performed in minimal medium (GAST) containing either 0.06, 6, or 60 mM MgCl 2 , representing the lowest concentration that still supports Mtb growth, the standard in vitro Mg 2+ concentration, and the 10fold higher in vitro Mg 2+ concentration, respectively. Three compounds having similar scaffolds exerted submicromolar inhibition of Mtb growth for up to 19 days of incubation with little evidence of rescue in response to increasing extracellular Mg 2+ concentrations. One of these (P01) was chosen for further follow-up after consideration of structural aspects. As shown in Figure 1, P01 is a N-methyl-5-hydroxypyrimidinone derivative conjugated with a substituted benzyl group on the left-hand side and a benzyl-substituted carboxamide moiety on the right-hand side. Preliminary biological profiling of resynthesized P01 was performed to verify that the compound was active independent of carbon source and to determine whether the compound inhibited targets in mycolyl-arabinogalactan biosynthesis or induced DNA damage as previously described. 27 The effect of the compound on the inhibition of respiratory metabolism was also assessed using the cyd::aph strain. 28 The compound was found to be active under all replicating conditions with decreased activity in the slowgrowth assay wherein Mtb is subjected to the suboptimal environment of acidic pH and nitrosative stress ( Table 1). The cell-wall-reporter assay hinted that the compound might affect a target in cell-wall biosynthesis, as evidenced by the upregulation of the iniBAC promoter.
Synthesis of Derivatives and Their Structure−Activity Relationship. To understand the structural aspects of the initial hit that were critical for antitubercular activity, a series of pyrimidinedione derivatives, modified in each moiety of the original structure, were designed and prepared. The general synthesis of N-alkyl-5-hydroxypyrimidinone carboxamides P01−P13 is described in Scheme 1. For the formation of the dihydroxypyrimidine ring, previously reported procedures starting from nitrile derivatives (1) were used. 18,29−31 Briefly, amidoximes 2 were prepared from nitriles 1 after reactions with salt-free hydroxylamine. Next, Michael-type addition reaction of 2 with dimethyl acetylenedicarboxylate followed by intramolecular cyclization generated dihydroxypyrimidine intermediates 4−6. After benzoyl protection of the 5-hydroxyl

Journal of Medicinal Chemistry
Article group, alkylation on the N3 position mediated by Cs 2 CO 3 followed by a thermal amidation reaction produced the desired products, P01−P13.
To validate the importance of the alkyl group at the N3 position, P14, without the methyl substituent, was synthesized by direct thermal amidation from intermediate 4 (Scheme 2).

Journal of Medicinal Chemistry
Article During N3-methylation in the presence of MeI and Cs 2 CO 3 , the addition of excess amounts of reagents, higher temperature, and a longer reaction time gave rise to the formation of overmethylated compound 14, which was the precursor of P15. In addition, the O-Me regioisomer, 15, was also produced as a byproduct during methylation. Structurally, the O-Me group in the pyrimidine ring was easily substituted by benzylamine during the thermal amidation reaction, allowing the reaction time for the synthesis of P16 to be shortened to avoid oversubstitution by benzylamines. To incorporate another methyl group in the acidic 5-hydroxyl position, 16, generated from 10 by debenzoylation, was methylated by trimethylsilyldiazomethane to afford the methylated compound, P17. 17 The synthesis of P18, which has the ketone functionality at the position of C6 normally occupied by the amide, is shown in Scheme 3. The 5-hydroxyl group in 16 was benzylated, and subsequent basic hydrolysis afforded 18. 32 The carboxylic moiety of 18 was transformed into Weinreb amide 19, which underwent selective monoaddition of Grignard reagent produced by (2-chloroethyl)benzene to generate the corresponding ketone, 20, with P18 obtained after debenzylation. 33 All final compounds (P01−P18) were fully characterized by LC-MS and 1 H and 13 C NMR. In addition, HRMS and melting-point analyses were performed for some active compounds. Antitubercular activities of synthetic compounds in two different media and their cytotoxicity against HepG2 cells are summarized in Table 2.
Initial SAR studies around the hit compound, P01, focused on identifying key moieties or functionalities that play important roles in antitubercular activity. This focused library of 18 compounds (P01−P18), constructed for preliminary SAR, guided the next steps for hit-to-lead modification, despite its small size. Compounds P02−P03 and P15 were synthesized to evaluate the left-hand-side modification of R 1 . The benzene ring was necessary for antitubercular activity, as seen by complete loss of potency in the absence of this moiety (P03), with the o-CF 3 substitution (P01) further improving potency. A methyl substitution in the benzylic position was detrimental for activity, as seen by the 5-fold increase in the MIC of P15 compared with that of the initial hit, P01.
To assess the importance of the methyl group in the N3 position, two compounds, including P04 with an ethyl group and P14 without an alkyl substituent, were synthesized. Replacement of the methyl (P01) by the ethyl group (P04) showed good potency with improved MICs (2.3 μM in 7H9/ ADC and 0.4 μM in GAST), but deletion of the methyl at this position (P14) was not tolerated. Not surprisingly, the O-Me regioisomer (P16) was inactive against Mtb. Methyl protection of the acidic 5-hydroxyl group also resulted in loss of antitubercular activity. These results demonstrate that the Nalkyl-5-hydroxypyrimidinone core was critical for retaining potency against this pathogen.
The compounds P05−P13 and P18 were synthesized to determine SAR of the right-hand side R 3 position. As shown in Table 2, the benzyl group at the R 3 position was important for antitubercular activity because its replacement by a phenyl group (P05) reduced potency around 4-fold over P01, and the saturated cyclohexyl methyl group (P06) was also less potent with likely increased protein binding, as evidenced by its lack of activity in BSA-containing Middlebrook 7H9-based medium. Replacement of the phenyl with certain heteroaryl groups, such as the 2-picolyl moiety (P07), was tolerated at the R 3 position, but its regioisomer, 3-picolyl (P08), showed a decrease in activity. Initial investigations into substituent effects on the benzene ring revealed that o-F (P09) was tolerated with a potency similar to that of P01, and it was >4fold more potent than its regioisomer, P10, with p-F. Interestingly, the compound with the benzodioxole moiety replacing the benzene ring (P11) was completely inactive against Mtb, indicating that substituent effects in the benzene ring were important for the antitubercular activity of this scaffold. Conformational restriction of amide linkage by Nmethylation (P12) as well as methylation of the benzylic position (P13) resulted in loss of activity. In addition, the amide functionality was important because P18, with a ketone replacing the amide at R 3 , was inactive. None of the derivatives were cytotoxic against HepG2 cells and additionally did not exert overt mitochondrial toxicity as seen by lack of cytotoxicity during growth of these cells on galactose as a carbon source ( Table 2).
The pharmacokinetic (PK) parameters of C57BL/6 mice administered a single 10 mg/kg oral suspension of P01 indicated absorption was fairly low (C max 0.25 μg/mL, AUC of 1.06 μg·h/mL), but the half-life of 4 h and V d of 33.7 L/kg were acceptable (Table 3). Even though PK parameters are encouraging as a starting point of in-depth medicinal chemistry, efficacy studies in mice were not pursued with this hit compound because PK modeling suggested a much higher dose would be needed to achieve an average plasma concentration above the MIC of P01 (∼1.9 μg/mL in 7H9/ ADC) because of the high murine metabolic rate.
Activity Validation in Vitro and ex Vivo and Target-Protein Identification using Whole-Genome Sequencing. The hit compound, P01, was bactericidal to actively replicating Mtb in vitro, with a 1−2 logarithmic reduction in colony-forming units observed over 7 days with weak dosedependence across a wide range of concentrations ( Figure 2A), but it lacked bactericidal activity against anaerobic nonreplicating Mtb cells generated in the Wayne model of hypoxic adaptation ( Figure 2B). 34 The parental hit compound (P01) and three of its in vitro active analogues (P04, P07, and P09) were also active against Mtb replication in J774 murine macrophages, indicating that the compound retained bactericidal activity under these conditions and was also able to reach the pathogen in its phagosomal niche ( Figure 2C).

Article
To identify the putative target of P01, resistant mutants were generated by plating of Mtb on solid-medium plates containing 1×, 2×, 5×, and 10× MIC concentrations of the compound, with a mutation frequency of 7 × 10 −9 obtained at 10× MIC. Twenty-four P01-resistant mutants were tested for their level of resistance by MIC determination against P01, and all were found to be more than 10-fold resistant compared with the parental strain. Four of these resistant strains were submitted for whole-genome sequencing. All four were found to contain the same Y314H mutation in decaprenylphosphoryl-β-D-ribose 2′-oxidase (DprE1, Rv3790). DprE1 together with DprE2 catalyzes the epimerization of decaprenylphosphoryl ribose (DPR) to decaprenylphosphoryl arabinose (DPA) in the synthesis of cell-wall-linked arabinan. 35 DprE1 has already been established as a highly vulnerable target in mycobacterial cell-wall biosynthesis as seen by the high number of chemically diverse scaffolds that inhibit this essential target, leading to cellular-growth inhibition both in vitro and in vivo. 36,37 The high frequency of hits against this target is likely also driven by its extra-cytoplasmic localization in the periplasmic face of the plasma membrane. The validity of inhibiting this target as a viable strategy in TB chemotherapy will be tested in the ongoing clinical trial of a benzothiazinone compound (PBTZ169). 38 To further confirm that the N-alkyl-5-hydroxypyrimidinone carboxamides inhibited DprE1, we tested the vulnerability of a panel of diverse dprE1 mutants to P01. Strains with the Y314C and P116S substitutions but not the N346S mutation in DprE1 were found to be cross-resistant to P01 (Table 4). Y314C has previously been observed to confer resistance to the DprE1 inhibitor TCA1 39 but not to the benzothiazinone scaffold. 40 Target Validation using a Regulated Expression of the dprE1−dprE2 Locus. The on-target activity of the Nalkyl-5-hydroxypyrimidinone carboxamides was further con-

Journal of Medicinal Chemistry
Article firmed using a strain in which the level of the DprE1−DprE2 epimerase complex was transcriptionally regulated by the tetracycline repressor, TetR. In this strain, dprE1−dprE2 transcription is dependent on the presence of tetracycline. Removal of tetracycline results in repression of dprE1−dprE2 and a concomitant increase in susceptibility to known DprE1 inhibitors, such as TCA1 39 and the benzothiazinone BTZ043 41 ( Figure 3A,B), whereas ethambutol, which inhibits the downstream arabinosyltransferases, showed equipotent efficacy during dprE1−dprE2 repression ( Figure 3C). As expected, active compounds P01 and P04 exerted more growth inhibition of the mycobacterial cells during dprE1−dprE2transcriptional repression ( Figure 3D,E), whereas the inactive homologue P13 did not result in growth inhibition of this regulated mutant ( Figure 3F). The higher resistance of the tetracycline-regulated cells during treatment with TCA1, P01, and P04 is likely driven by higher basal levels of protein of one or both of the DprE subunits. The clear shift in susceptibility during regulated expression of the subunits of this epimerase complex establishes the on-target activity of the N-alkyl-5hydroxypyrimidinone carboxamides in whole cells.
Molecular-Docking Study of N-Alkyl-5-hydroxypyrimidinone Carboxamides in DprE1. Molecular modeling was used to understand the potential interactions of synthetic analogues with DprE1. Figure 4A shows the model of P01 in the binding site of DprE1. An extensive hydrogen-bond (HB) network was formed between P01 and the polar residues lining the interior of the binding site. The carbonyl oxygen atom of the pyrimidinone core formed an HB with the side-chain amino group of residue N385. The 5-hydroxyl group, with a pK a of 4.5 as calculated by ACD Laboratories pK a method, 42 was deprotonated, and the negatively charged oxygen atom formed HBs with both the H132 and Q336 side chains. Methylation of this oxygen atom as in P17 disrupted the HBs, hence rendering the compound inactive. The exocyclic amide oxygen on the right-hand side formed an HB with the K418 side chain. The amide −NH group, on the other hand, formed an intramolecular HB with one nitrogen atom of the pyrimidinone core, stabilizing the ligand-bound conformation. Methylation of this group, as in P12, abolished this intramolecular HB, and the size of the methyl group forced the amide to rotate out of the plane of the pyrimidinone core, leading to the inactivity of P12. Replacing the amide − NH group with a methylene, as in P18, had a similar impact that reduced its binding affinity. Furthermore, the methyl group in the N3 position made hydrophobic contact with the side-chain carbon atoms of K367; demethylation here, as in P16, was likely the reason for its reduced activity.
Hydrophobic interactions with the DprE1 active site contributed substantially to the ligand-binding affinity. The trifluoromethyl group on the phenyl ring of the left-hand side made hydrophobic contacts with the side chain of Y314, whereas the phenyl ring itself engaged in hydrophobic interactions with the side chain of V365 at the bottom of the binding site. Deleting the o-CF 3 group lowered the binding affinity several-fold, as shown for P02, whereas removing the whole o-CF 3 Ph moiety, as in P03, made the compound completely inactive. The phenyl ring on the right-hand side made a π−cation interaction with the R325 side chain.

Article
Replacing this group with a cyclohexyl moiety reduced this favorable interaction, and led to the low potency of P06. In addition, this phenyl group made hydrophobic contacts with the side chains of residues L317, L363, and V365. The size of the binding pocket around this phenyl group was not large enough to accommodate P11, hence its much lower potency. The model of P01 was compared with the crystal structure of a known DprE1 inhibitor, TCA1 39 ( Figure 4B). The pyrimidinedione core of P01 binds at the same subpocket as the thiophene core, both forming strong interactions with the H132 side chain. The amide carbonyl groups in both compounds formed hydrogen bonds with the side chain of K418. The distal benzyl group off the amide binds in the same hydrophobic region as the benzothiazole group of TCA1. Additionally, the other side of both molecules engages the Y314 side chain. This docking model and the similar interaction pattern with the DprE1 binding site in comparison with TCA1 provide further evidence that these compounds target Mtb DprE1.

■ CONCLUSIONS
In order to discover a novel candidate for the treatment of TB, we screened a pyrimidinedione library that contains this embedded privileged structure associated with magnesiumchelating properties and generated a new hit, N-alkyl-5hydroxypyrimidinone carboxamide, P01. The SAR of P01 was examined through the construction of 17 analogues, in which we comprehensively modified each moiety in the original scaffold. Even though the MICs of the active compounds were not dependent on Mg 2+ concentration, they showed significant growth inhibition in each media without notable cytotoxicity.

Journal of Medicinal Chemistry
Article Through whole-genome sequencing of resistant mutants, we identified that this series targets DprE1, which is an essential enzyme for pathogen-specific cell-wall biogenesis. Several molecules that target DprE1 are in preclinical and clinical development, 43 but no drug targeting this enzyme has been approved for clinical use to date. The N-alkyl-5-hydroxypyrimidinone carboxamides represent a potentially viable alternative scaffold should current candidates fail to complete clinical development. These compounds have favorable solubility, toxicity, and PK properties that should provide adequate opportunity for lead optimization. Although it was disappointing to not identify scaffolds that potentiated Mg 2+dependent processes in the cell, alternative screening procedures may still identify candidate molecules within this privileged core structure that do.

■ EXPERIMENTAL SECTION
Chemistry. 1 H and 13 C NMR spectra were recorded on a Varian Mercury-300 NMR Spectrometer, and chemical shifts were measured in parts per million relative to a specific solvent signal. Routine mass and purity analyses (LRMS) were performed on an HP Agilent LC-MS series 1100 system equipped with a reverse-phase column (Agilent Poroshell 120 EC-C18, 2.7 μm, 50 × 2.1 mm) and a photodiode-array detector coupled to an Agilent 1946 DSL quadrupole mass selective detector coupled to an Agilent 1946 DSL quadrupole mass selective detector using electrospray ionization (ESI). A gradient mobile phase consisting of acetonitrile/water with 0.1% formic acid and UV detection at 254 and 210 nm were used to confirm all final products (P01−P18) as being ≥95%. Melting points were measured on an Electrothermal 9100 apparatus. Accurate masses (HRMS) were obtained using a Waters LCT Premiere time-of-flight mass spectrometer. The instrument was operated in W-mode at a resolution of 10 000 in positive-ion mode. Ions were generated with Z-Spray electrospray ionization (ESI) with a capillary voltage of 3.4 kV. Accurate masses were determined using the internal-standard method. Most reagents used in the synthetic procedure were purchased from Sigma-Aldrich, Alfa Aesar, and TCI. The progress of the reaction was monitored using thin-layer chromatography (TLC; silica gel 60 F254 0.25 mm), and the products were visualized by UV light (254 and 365 nm). SiliaFlash P60 (40−60 μm) used in flash column chromatography was purchased from Silicycle Inc. Other solvents were purchased from commercial vendors and used without further purification unless otherwise mentioned.
Methyl 5-(Benzoyloxy)-6-oxo-2-(2-(trifluoromethyl)benzyl)-1,6dihydropyrimidine-4-carboxylate (7). General Procedure of Alkylation Mediated by Cs 2 CO 3 (10−13). To a solution of starting material (1.0 equiv) in anhydrous THF (0.2 M) were added Cs 2 CO 3 (2.0 equiv) and alkyl iodide (2.5 equiv). The reaction mixture was stirred at rt for about 5 h in the case of methylation or stirred overnight at 40°C in the case of ethylation. The reaction mixture was diluted with EtOAc and washed with 1 N HCl and brine. The organic layer was dried over Na 2 SO 4 and evaporated in vacuo. The crude mixture was purified by flash column chromatography (EtOAc/hexane) to obtain the desired product. Synthesis of Methyl 5-(Benzoyloxy)-6-methoxy-2-(2-(trifluoromethyl)benzyl)pyrimidine-4-carboxylate (15). To a solution of 7 (1.0 equiv) in anhydrous DMF (0.2 M) was added K 2 CO 3 (2.0 equiv) and MeI (3.0 equiv). The reaction mixture was stirred at rt for 4 h. The reaction mixture was diluted with EtOAc, and washed with 1 N HCl and brine. The organic layer was dried over Na 2 SO 4 and evaporated in vacuo. The crude mixture was purified by flash column chromatography (EtOAc/hexane = 1/9 → 1/4 → 2/3) to obtain two different isomers, the desired product ( Synthesis of Methyl 5-Methoxy-1-methyl-6-oxo-2-(2-(trifluoromethyl)benzyl)-1,6-dihydropyrimidine-4-carboxylate (17). To a solution of 16 (1.0 equiv) in anhydrous CHCl 3 /MeOH (2/1, 0.2 M) was added trimethylsilyldiazomethane (5.5 equiv, 1.1 equiv, 5 times) portionwise for about 1 h. The reaction mixture was stirred overnight at rt. The reaction mixture was evaporated in vacuo and suspended in EtOH. After being stirred for 2 h, the solid was filtered and washed with EtOH and Et 2 O to obtain the desired product as an General Procedure of Thermal Addition and Elimination Reactions (P01−P17). To a solution of starting material (1.0 equiv) in anhydrous MeOH (0.2 M) was added each amine (3.0 equiv or 2.2 equiv in the case of P16). The reaction mixture was refluxed until completion of reaction (TLC or LC-MS). In most cases, overnight reflux was enough to complete the reaction, except with P05 (3 days), P12 (5 days), and P16 (30 min). The reaction mixture was diluted with EtOAc and washed with 1 N HCl and brine. The organic layer was dried over MgSO 4 and evaporated in vacuo. The crude mixture was suspended in EtOH or Et 2 O for crystallization of each desired product.
Efficacy and Validating Inhibition of N-Alkyl-5-hydroxypyrimidinone Carboxamides against Mtb in Vitro and ex Vivo. In vitro efficacy was performed in aerobic and anaerobic conditions. For aerobic conditions, logarithmically growing Mtb (OD 650 0.2) was diluted 1000-fold in 1 mL of 7H9/ADC media and exposed to 1×, 5×, 10×, 20×, and 50× MIC of P01 for up to 7 days in duplicates. After 1, 2, 4, and 7 days of treatment, appropriate cell dilutions were plated on 7H11/OADC plates for CFU enumeration. For anaerobic conditions, Mtb was cultured in the self-generated oxygen-depletion model as previously described. 46 One milliliter of anaerobic Mtb culture was exposed to 1×, 5×, 10×, 20×, and 50× MIC of P01 in an anaerobic chamber for 7 days. Diluted cells were plated on 7H11/ OADC plates for CFU enumeration.
For the ex vivo efficacy test, J774 cells (5 × 10 5 cells/well) were seeded in flat-bottom 24-well plates (Corning Inc.) in DMEM GlutaMAX (Gibco/ThermoFisher Scientific) supplemented with 10% fetal-bovine serum, 20 mM HEPES, and 0.5 mM sodium pyruvate (hereafter abbreviated as DMEM/FBS). Cells were infected with Mtb (5 × 10 5 cells/well, MOI 1) for 24 h; this was followed by medium removal and washing (2×) with an equal volume of Dulbecco's PBS. Infected cells were fed DMEM/FBS and exposed to test compounds at 1×, 10×, 20×, and 50× MIC in medium (500 μL/well). Cells were incubated at 37°C with 95% humidity and 5% CO 2 in an incubator for 7 days. The medium was changed every 4 days. After 4 and 7 days of incubation, 0.1% SDS was added to each well to lyse the J774 cells. After 5 min, the lysate was thoroughly mixed and diluted in 7H9/ ADC, and appropriate dilutions were plated in duplicate on 7H11/ OADC plates.
Generation and Characterization of N-Alkyl-5-hydroxypyrimidinone Carboxamide Resistant Mutants. To generate mutants against the N-alkyl-5-hydroxypyrimidinone carboxamide scaffold, 10 7 , 10 8 , and 10 9 cells of Mtb H37Rv were plated on 7H11/OADC plates containing with 2×, 5×, and 10× MIC of P01 with drug-free plates used to enumerate bacterial load. The plates were incubated at 37°C for 4 weeks. After 4 weeks, 24 colonies on 10× MIC of P01 were picked and inoculated in 7H9/ADC medium and MIC-tested to confirm their resistance against P01. Genomic DNA of four mutants was isolated by the CTAB method. 47 Whole-genome sequencing was performed and analyzed as described. 48 Generation and Testing of a Regulated Mutant of the dprE1− dprE2 Operon. DprE12−TetON1 was constructed using experimental strategies originally developed for the construction of dual-control mutants. 49 Briefly, Mtb H37Rv was transformed with a plasmid that integrated into the attachment site of phage L5 (attL5) and constitutively expressed dprE1, dprE2, aftA, and embC. This allowed deletion of the native copies of dprE1, dprE2, aftA, and embC by homologous recombination. The genotype of the resulting strain was verified by Southern blotting. We then replaced the constitutive-dprE1−dprE2−aftA−embC-expression plasmid with one in which transcription of dprE1 and dprE2 was susceptible to repression by the tetracycline repressor (TetR). In the resulting strain, dprE12− TetON1, expression of aftA and embC remained constitutive, but transcription of dprE1 and dprE2 could be increased with anhydrotetracycline. The dprE12−TetON1 mutant was maintained in 7H9-based growth medium supplemented with 500 ng/mL anhydrotetracycline/25 μg/mL kanamycin/50 μg/mL hygromycin. To assess the on-target effects of the compounds, cells were harvested, washed, and diluted in drug-free 7H9-based medium to an OD 580nm of 0.01. The cell suspension was split in two, and anhydrotetracycline was added (500 ng/mL) to one of the cultures. Cell suspensions (49.5 μL/well) were transferred to 384-well, black, clear-bottom plates containing a 14-point 2-fold drug dilution in DMSO, giving a final DMSO concentration of 1%. Positive-control drugs included the benzothiazinone BTZ043 and TCA1, and the negative-control drugs, which show no shift in susceptibility upon transcriptional repression, included ciprofloxacin (data not shown), ethambutol, linezolid (data

Journal of Medicinal Chemistry
Article not shown), and isoniazid (data not shown). Plates were wrapped in aluminum foil in stacks of three and incubated at 37°C with 5% CO 2 , and the OD 580nm was recorded at day 10.
Pharmacokinetic Study of P01. Animal studies were carried out in accordance with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health with approval from the Committee on the Ethics of Animal Experiments of the National Institute of Allergy and Infectious Diseases (LCID-4). Female mice (5 weeks of age) were purchased from Taconic Biosciences and allowed to acclimatize for at least 1 week prior to handling. The P01 suspension was mixed with 1% carboxy methyl cellulose in a mortar and pestle, transferred to a vial, probe-sonicated until homogeneous in appearance, and then stirred overnight. Fifteen mice were administered 10 mg/kg P01 in 0.2 mL by gavage, and 0.1 mL of blood was collected twice from each animal over the experiment and centrifuged to prepare serum.
Serum samples (five per time point: 0.5, 1, 2, 4, 6, and 24 h) for P01 were prepared by mixing 50 μL aliquots of sample, 20 μL of aqueous 1% formic acid, and 50 μL of a cocktail of internal standards (various Mtb active compounds) followed by 400 μL of an acetonitrile/MeOH mixture (3:1) to precipitate proteins. The sample was centrifuged at 9450g for 5 min, and the supernatant was removed for tandem-mass-spectrometry analysis. The calibration curve was based on the internal standard with the best correlation coefficient and fit. Pharmacokinetic parameters were calculated using PK Solutions 2.0 (Summit Research Services).
Drug quantitation was performed by LC-MS/MS using an Agilent 1290 Infinity HPLC coupled to an Agilent 6460C triple-quadrupole tandem-mass-selective detector (QqQ). Water was supplied from a Barnstead nanopure Diamond system with a minimum resistivity of 18.1 MΩ·cm, and HPLC-grade acetonitrile and 98% formic acid were from EMD Millipore Corporation. Mobile phase A consisted of nanopure water, and mobile phase B consisted of acetonitrile, and each was spiked with 0.1% (v/v) formic acid. Each sample (5 μL) was injected into a 2.1 × 50 mm Agilent EclipsePlus C 18 1.7 μm 100 Å column. The HPLC was supplied mobile phase at 0.8 mL/min with a starting gradient of 8% acetonitrile, which was held for 15 s; this was followed by a linear gradient to 95% acetonitrile in 5.25 min, which was then held for 2 min. The QqQ detection was performed in MRM mode with positive-mode electrospray ionization. The tandem MRM signal was produced by converting the M + H + precursor ion for P01 into the product ion m/z 91.1 with a collision energy (CE) of 28 V.
Molecular Modeling. The Mtb DprE1 crystal structure in the public domain was retrieved from PDB (PDB code: 4P8N). 50 This crystal structure had a noncovalently bound ligand 51 and two chains in an asymmetric unit with a high resolution of 1.79 Å. Chain A of this structure was selected for the docking study. The Schrodinger modeling package 52 was used to prepare the protein structure and to generate the Grid file for docking studies. The protein-preparation process involved the following steps: assigning bond orders to protein and ligand molecules, adding hydrogen atoms, creating disulfide bonds, sampling water orientations, assigning protein and ligand charge states at pH 7.0, and minimizing all hydrogen atoms while fixing the positions of all protein heavy atoms. The docking grid was generated using a box of dimensions 20 × 20 × 20 Å around the crystal ligand. The ligand docking was carried with the Glide method. 53 P01 was docked into the DprE1 binding site using both the SP and XP sampling methods in Glide. The best model was selected via visual inspection with the SAR information in mind.