Antitubercular 2 ‑ Pyrazolylpyrimidinones: Structure − Activity Relationship and Mode-of-Action Studies

: Phenotypic screening of a Medicines for Malaria Venture compound library against Mycobacterium tuberculosis ( Mtb ) identi ﬁ ed a cluster of pan-active 2-pyrazolylpyrimidinones. The biology triage of these actives using various tool strains of Mtb suggested a novel mechanism of action. The compounds were bactericidal against replicating Mtb and retained potency against clinical isolates of Mtb . Although selected MmpL3 mutant strains of Mtb showed resistance to these compounds, there was no shift in the minimum inhibitory concentration (MIC) against a mmpL3 hypomorph, suggesting mutations in MmpL3 as a possible resistance mechanism for the compounds but not necessarily as the target. RNA transcriptional pro ﬁ ling and the checkerboard board 2D-MIC assay in the presence of varying concentrations of ferrous salt indicated perturbation of the Fe-homeostasis by the compounds. Structure − activity relationship studies identi ﬁ ed potent compounds with good physicochemical properties and in vitro microsomal metabolic stability with moderate selectivity over cytotoxicity against mammalian cell lines.


■ INTRODUCTION
Tuberculosis (TB) caused by Mycobacterium tuberculosis (Mtb) continues to be one of the leading causes of death and morbidity globally, claiming 1.2 million lives in 2018. 1 The emergence and spread of multi drug-resistant (MDR) and extensively drugresistant (XDR) strains of Mtb that are resistant to the firstand second-line drugs have further exacerbated the situation. 2 The rise in anti-microbial resistance warrants the search for new drugs with unique modes of action that can bypass existing modes of resistance or can be used as adjunctive therapy to compensate for those that are vulnerable to promoting resistance. While considerable progress has been made toward establishing a TB drug pipeline, the high attrition rate in clinical development reinforces the need to continually replenish the pipeline with high-quality leads that act through the inhibition of novel targets. 3 In light of the above, we have been engaged in early TB drug discovery to identify and progress new chemical series with potentially novel modes of action lacking cross-resistance with existing drugs. 4,5 One approach to identify compounds with a novel mode of action is to phenotypically screen compound libraries against Mtb growing under different media conditions with different carbon sources followed by a biology triage process to weed out actives hitting frequently encountered targets such as the respiratory system (e.g., QcrB) and cell−wall synthesis (e.g., MmpL3 and DprE1), as well as DNA damaging agents. Herein, we describe structure−activity relationship (SAR) and target identification studies of one novel chemical series, 2-pyrazolylpyrimidinonesrepresented in Figure 1. This series was identified through high-throughput screening of a Medicines for Malaria Venture (MMV) compound library.
■ RESULTS AND DISCUSSION Phenotypic Hits with a Novel Mode of Action. A highthroughput screen of a ∼530,000 diverse set of compounds from MMV compound library against Mtb was conducted at the National Institute of Allergy and Infectious Diseases of the National Institutes of Health (NIAID/NIH, U.S.). Reconfirmed hits were followed up for determination of minimum inhibitory concentration (MIC) on multiple media that identified a cluster of pan-active pyrazolylpyrimidinones represented by compounds 1 and 2 ( Table 1). The biology triage of these actives (described in the target identification section), suggestive of a unique mechanism of action (MoA), generated further interest in the exploration of these hits. We recognized that pyrazolylpyrimidinones potentially possess iron-chelating properties by virtue of their two sp2 heteroatoms in a 1,4-relationship on adjacent rings, which might play a crucial role in the SAR and mode of action. 6 During these studies, we came across the related pyrazolopyrimidinone (PZP, Figure 2a) compound reported as an intracellular iron chelator in Mtb. 7 PZP is reported to chelate iron through the pyrimidinone carbonyl and 2-position pyrazole nitrogen. Similarly, bidentate iron chelation by the pyrazolylpyrimidinones from the pyrimidinone nitrogen and pyrazole nitrogen as depicted in Figure 2b may be hypothesized. A single-crystal X-ray structure of the pyrazolylpyridines−Fe complex is reported in the literature which supports this hypothesis. 8,9 This aspect of metal chelation by the compounds was considered while planning further SAR exploration and target identification studies with the series.
Synthesis. In order to explore the SAR profiles of the series, various routes were explored toward the synthesis of diverse range of analogues as represented by compounds 3−54 in Tables 2−5. To facilitate exploration of a broad array of substituents off the pyrazole ring and pyrimidinone core, a concise and scalable synthesis leading to an advanced intermediate was desired. As summarized in Scheme 1, this was achieved by the ring formation reaction of 6-substituted-2thiomethyl-pyrimidin-4-one 55 via condensation of the appropriate β-keto-ester and thiourea, and subsequent methylation of the thiol. 10 Intermediates 55a−g were reacted with hydrazine hydrate in refluxing ethanol to give the versatile 2hydrazinylpyrimidin-4(1H)-ones 56a−g, which gave access to analogues for SAR exploration on the pyrazole ring and the R1 substituent on the pyrimidinone core. Similarly, substitution of a phenyl group at the R2 position was achieved through bromination at C5 of 6-methyl-2-thiomethyl-pyrimidin-4-one (55, R1 = CH 3 ) followed by a Suzuki cross-coupling reaction with phenyl boronic acid to provide 55h, which was then treated with hydrazine to form 56h.
Syntheses of compounds 3−7 (Table 2) with changes in the central core ring are summarized in Scheme 2. Compounds masking the pyrimidinone amide were synthesized from compound 1 via a Mitsunobu reaction to form methyl ether 3 and N-methylation of 1 with iodomethane using potassium carbonate in N,N-dimethylformamide (DMF) to afford N-   Journal of Medicinal Chemistry pubs.acs.org/jmc Article methyl pyrimidinone 4. The nucleophilic aromatic substitution reaction of 2,6-dichloro-4-iodopyridine with sodium methoxide gave intermediates 57 and 58 in a 2:1 ratio, which upon separation allowed for the synthesis of compounds 5 and 6, respectively. Briefly, the three-step synthesis from 57 and 58 involved a Suzuki cross-coupling reaction with phenylboronic acid to afford intermediates 59 and 60, respectively, followed by a Buchwald Hartwig amination to introduce the pyrazole (intermediates 61 and 62, respectively), and finally Odemethylation with boron tribromide to afford compounds 5 and 6, respectively. Pyrimidin-2(1H)-one 7 was synthesized by the cyclization of ethyl benzoylacetate with urea to provide uracil 63, which was then reacted with phosphorus oxychloride in the presence of a catalytic amount of DMF to give 2,4-dichloropyrimidine 64 (Scheme 2). 12 The activated 2,4-dichloropyrimidine readily undergoes nucleophilic substitution with the pyrazole at the 4chloro position by heating in the microwave at 80°C (intermediate 65) and subsequent hydrolysis of 2-Cl afforded   13 which was reacted with the corresponding 5-membered heterocycle to displace 2-Cl using cesium carbonate under microwave heating at 80°C (Scheme 3). Compound 18 with an N-methylimidazole was synthesized by condensing 1-methyl-1H-imidazole-2-carboximidamide with ethyl benzoylacetate in low yields.
In order to introduce groups with polar functionalities at the R1 and R2 positions on the pyrimidinone core, 2,4,6trichloropyrimidine was a useful and accessible starting material when the appropriate β-keto-ester was not readily available to pursue synthetic routes, as shown in Schemes 1 and 3. Treatment of 2,4,6-trichloropyrimidine in aqueous 1,4-dioxane, brought to the alkaline reaction with NaOH, led to hydrolysis at room temperature with the formation of 4,6-dichloro-2hydroxypyrimidine 67 and 2,6-dichloropyrimidin-4(3H)-one 68, with the former precipitating out as the sodium salt (Scheme 4). 14 The filtrate was concentrated to approximately half the volume and cooled to afford a precipitate which was filtered off and acidified to pH 2 using 5 N HCl to give the required   15 A strong preference for the C6 position has been observed in Suzuki cross-coupling reactions. 16−20 Coupled with the use of a lower temperature for the Suzuki coupling, this allowed for the sequential introduction of different substituents. Thus, 2,6-dichloropyrimidin-4(3H)-one 68 underwent a Suzuki cross-coupling with the appropriate boronic acid, using Pd(OAc) 2 /dppf or PdCl 2 (dppf) 2 as the catalyst to afford compounds 69a−d. Coupling of intermediates 69a−d with 3,5dimethylpyrazole was achieved by heating to 130°C under microwave irradiation using 15 mol % Pd 2 (dba) 3 , xanphos, and Cs 2 CO 3 in 1,4-dioxane to afford the desired compounds 22−25 (Table 3). A slightly modified procedure, as shown in Scheme 5, was used to prepare amide 26. Methyl 2,6-dichloropyrimidine-4carboxylate was treated with sodium methoxide at 0°C to obtain intermediate 70.
To determine if polarity at the R2 position on the pyrimidinone core was tolerated, compounds 27−29 (Table  3) were synthesized using the route, as shown in Scheme 6. 4-Fluorobenzaldehyde was condensed with thiourea and ethyl cyanoacetate using potassium hydroxide in refluxing ethanol to give 2-mecapto-5-cyanopyrimidinone intermediate 73. 21 The thiol of intermediate 73 was methylated using methyl iodide in the presence of sodium hydroxide followed by displacement with hydrazine in refluxing ethanol to give 2-hydrazinyl-5-  condensed with acetylacetone by heating in acetic acid-ethanol at 85°C to obtain compound 27, which was hydrolyzed to amide 28 by heating in concentrated sulfuric acid at 85°C. 22 The conversion of amide 28 to acid 29 was achieved via nitrosation by heating with tert-butyl nitrite in acetic acid at 85°C followed by hydrolysis. 23 Compounds 30−47 (Table 4) were synthesized to explore the SAR around the R5 position of the pyrazole ring, keeping the C6-position on the pyrimidinone core fixed as trifluoromethyl.
Using the same methodology, as shown Scheme 1, the condensation of 56d with the corresponding 1,3-dicarbonyl compound afforded compounds 30−32 exclusively as one regioisomer as confirmed by NOESY NMR (Scheme 7). In addition, 56d provided access to compound 33, which served as a versatile intermediate and allowed for several transformations to form N-substituted pyrazole derivatives. The reductive amination of acetaldehyde with 33 in the presence of sodium cyanoborohydride in methanol gave ethylaminopyrazole 34. Attempts to synthesize the N,N-dialkyl derivative from compound 34 failed as the corresponding derivatives decomposed on standing. Compound 47 was synthesized in the same  Compound 38 was synthesized from hydrazine 56d and N-(2methoxyethyl)-3-oxobutanamide with Lawesson's reagent, which allows for the efficient installation of substituted-5amino groups on the pyrazole core in a single step. 24 However, this methodology failed in the case of the synthesis of compounds 36, 37, and 39, with 2-(5-hydroxy-3-methyl-1Hpyrazol-1-yl)-6-(trifluoromethyl)pyrimidin-4(3H)-one being obtained as the major product. An alternate highly regioselective route to 5-(substituted-amino)pyrazoles 36, 37, and 39 was achieved following a modified literature procedure. 25 As shown in Scheme 8, the displacement of one methylthio group of the α-oxoketene dithioacetal 75 26,27 by the appropriate amine in the presence of acetic acid in ethanol afforded the N,Sacetals 76a-c, which were useful three-carbon 1,3-electrophiles. 28 Cyclocondensation with hydrazine 56d in refluxing ethanol, gave access to 5-(substituted-amino)pyrazoles 36, 37, and 39. The regioisomeric products with 3-(substitutedamino)pyrazoles could be isolated in minor quantities and showed much less potent activity against Mtb (data not shown).
Compound 35, with an O-ethyl group at R5, was synthesized, as shown in Scheme 9. The condensation of ethyl acetoacetate Compounds 48−50 with substitutions on the pyrazole C4 position were synthesized, as shown in Scheme 10. The bromination of compound 12 gave compound 48, which on Suzuki cross-coupling with pyridine-3-boronic acid gave compound 49. Attempts to displace C4-Br of 48 with morpholine to synthesize compound 50 were unsuccessful. In an alternative route, the morpholine ring was preinstalled on acetylacetone via bromination with N-bromosuccinimide (intermediate 80) followed by the reaction with morpholine in the presence of triethylamine to give intermediate 81. This intermediate was then condensed with hydrazine 56d by refluxing in ethanol-acetic acid to give compound 50.
Compound 51 was synthesized in a similar fashion to compound 33, by condensing hydrazine 56d with 3-cyclopropyl-3-oxopropanenitrile (82) (Scheme 11). The synthesis of compounds 52−54 was accomplished from 83, as shown in Scheme 11. The reaction of 83 with 85 under microwave heating gave acid 53. This compound was then coupled with dimethylamine using HATU to give compound 54. Pyrazole ester 83 was converted to the corresponding dimethylaminomethyl derivative 84 via a standard two-step protocol involving the reduction of the ester to the aldehyde followed by reductive amination with dimethylamine. Intermediate 84 was reacted with 2-methylsulphonylpyrimidinone 79 under microwave heating to obtain compound 52.
Formal Hit Assessment to Determine the Pharmacophore Critical for Antitubercular Activity. Although the initial hits, 1 and 2 (Table 1), showed reasonably potent antitubercular activity (MIC < 2 μM), they suffered from unfavorable physicochemical properties including low solubility, low to moderate microsomal metabolic stability (MS), and a low selectivity index (SI) between Mtb MIC and mammalian cell toxicity as determined by measuring 50% inhibitory concentrations (IC 50 ) determined against a CHO cell-line (an epithelial cell-line derived from the Chinese-hamster ovary). A formal hit assessment study was undertaken to determine the minimum pharmacophoric features critical for antitubercular potency and to understand the SAR features needed to improve the SI. Initially, compounds with single point modifications on the pyrimidinone and pyrazole portions of the molecule were synthesized and profiled in a first wave of assays comprising Mtb MIC, CHO IC 50 , and aqueous solubility. The substituents on the pyrimidinone and pyrazole rings were labeled R1-R5, as shown Figure 1 (see Table 2). Masking of the pyrimidinone amide as −OCH 3 (3) or N-CH 3 (4) abolished the Mtb activity, indicating the essentiality of the NH, which might be involved in the critical hydrogen bond (HB) donor−acceptor interactions with the target. Each of the nitrogens of the pyrimidinone ring were found to be essential for maintaining Mtb activity (5 and 6). Interestingly, compound 7 with the carbonyl functionality between the ring nitrogens pyrimidin-2(1H)-one vs pyrimidin-4(3H)-one as in 1 maintained similar Mtb activity. The additional SAR explorations in this work were limited to the pyrimidin-4(3H)-one core as in compound 1, considering the ease of synthesis. Journal of Medicinal Chemistry pubs.acs.org/jmc Article Compound 8 with a methyl group added at the pyrimidinone C5 (R2) position, while keeping a phenyl substituent at C6 (R1), retained a good MIC value (2 μM) while improving aqueous solubility (120 μM) and the SI (12.5) for cytotoxicity. However, isomeric compound 9 with C5-phenyl (R2) and C6methyl (R1) showed a poor MIC value (9 μM) with a lower SI of 1, whereas the fusion of the phenyl ring with the pyrimidinone ring to form the corresponding quinazolinone (compound 10) retained potency but with a low SI of only 2. A phenyl substituent at the R1 (pyrimidinone C6) position was necessary to maintain Mtb activity. Compounds 11 and 12 with CH 3 and CF 3 groups, respectively, at R1 instead of the phenyl substituent showed a 10−15-fold loss in activity. As compound 12 with the CF 3 group at R1 showed higher solubility and SI relative to 1, this modification was included in further explorations for potency and selectivity. Overall, these SAR observations suggested the requirement for hydrophobic substitutions on the pyrimidinone ring for the retention of potent in vitro activity against Mtb.
A few polar modifications on the pyrazole ring were attempted in order to evaluate the scope of reducing lipophilicity and improving drug-like properties. Changing the C5′-methyl (R5) on the pyrazole to oxygen abolished the activity as observed for compound 13, whereas an amino group at the C5′-position as in compound 14 resulted in the retention of activity comparable to compound 1 (3 μM) and significantly improved solubility (180 μM). Unfortunately, this polar modification did not improve the SI. N-acetyl amide derivative 15 retained the activity and SI profile similar to parent amine 14. Compound 16, with an unsubstituted pyrazole, lost potency considerably (MIC 50 μM). A similar loss in potency was observed when the pyrazole ring was replaced with other 5-membered rings like triazole or imidazole, as in compounds 17 and 18, respectively. This suggests the critical nature of an appropriately substituted pyrazole moiety for maintaining antitubercular activity.
The SAR observations regarding the critical pharmacophoric features required for maintaining and/or improving antitubercular activity along with the scope for further SAR studies are summarized in Figure 3. In summary, the pyrimidin-4(3H)-one ring was retained for further SAR studies considering the critical nature of the cyclic amide as well as the ring nitrogens for maintaining the antitubercular activity. The SAR observations suggest the important role of hydrophobic substituents at the R1 and R2 positions for potency. We explored the scope of polar interactions further away from the main scaffold by extending polar groups from the aryl ring at the R1 position. In addition, we also explored a few polar groups at the R2 position instead of a methyl group, which could influence HB-donor acceptor interactions of the pyrimidinone ring amide. The 3,5disubstituted pyrazole ring was critical for maintaining activity, but there was a clear scope to extend substituents from the R5 position to improve potency and physicochemical properties. The scope of substitution at the R4-postion remained to be explored.
SAR of Polar Groups at R1 and R2 Positions of the Pyrimidinone Core. Several polar R1 and R2 substituents on the pyrimidinone ring were incorporated in order to expand the scope relative to activity (Table 3) and increase solubility and SI. Compound 19 with a 4-fluorophenyl group at the R1 position retained activity and cytotoxicity similar to compound 1. The replacement of the phenyl ring with pyridine rings considerably improved solubility, as observed for compounds 20 and 21, although with some loss in Mtb activity. Compound 20, with a 4pyridyl ring, was more potent (MIC 9 μM) than compound 21 with a 3-pyridyl ring (MIC 50 μM) leading to the speculation of the role of HB-interactions through the 4-pyridine nitrogen. Hence, compounds 22 to 25 with various polar substitutions at the para-position of the phenyl ring at the R1 position were prepared. Unfortunately, all these compounds lost activity. Next, we considered the option of inserting a polar linker between the pyrimidinone and phenyl rings. Among the various groups attempted, only compound 26 with a 4-flurophenyl group extended by a carboxamide linker at the R1 position, retained weak activity (25 μM).
Based on the encouraging data for compound 8 with a C5methyl group on the pyrimidinone, we explored adding a few polar groups such as CN, CONH 2 , and CO 2 H at this position, but unfortunately this led to the loss of activity (compounds 27, 28, and 29). Overall, these observations indicated very limited scope for adding polar substitutions on the pyrimidinone ring. Hence, we turned our attention to exploring possible interactions from the pyrazole ring.
SAR of the Pyrazole Ring. Further exploration on the pyrazole ring was planned, maintaining the CF 3 as the R1 substituent on the pyrimidinone, as it increased the solubility and SI exemplified for compound 12. In addition, the smaller CF 3 group kept the molecular weight and lipophilicity on the low side compared to the phenyl. The addition of bulkier alkyls like ethyl and cyclopropyl (compounds 30 and 31) at the R5 position on the pyrazole were tolerated for Mtb activity relative to 12 but with a deterioration of the SI, whereas a slight loss in activity was observed for compound 32 with the phenyl group at the R5 position. Isomers of these compounds with bulkier groups at the R3 position could not be synthesized because of the selectivity observed in the condensation reaction of hydrazine with requisite diketones favoring the placement of bulkier groups at the R5 position. Compound 33 with a NH 2 at the R5 position maintained a profile similar to compound 12, whereas compound 34 with an aminoethyl at the R5 position showed increased toxicity against CHO cell lines without a significant improvement in activity. Compound 35 with an Oethyl group at the R5 position had a potency and selectivity profile similar to compound 12. The extension of polar groups like −NH 2 or −OH via a short ethylamino linker from the R5 position as in compounds 36 and 37, respectively, was detrimental to activity against Mtb. Compounds 38 and 39, with the neutral ether substituents like methoxyethylamino and pyranamino, respectively, at the R5 position, showed more potent activity (MIC 1.5−2 μM) as compared to 12 while the SI was about the same. Based on these observations, it is speculated that the R5 position lies in a more hydrophobic region of the compound binding site within the target in Mtb. Hence, we decided to examine aryl groups, some with polarity, at R5 toward maintaining activity while improving solubility with polar substituents placed distant from the main scaffold.
Compound 40 with a 4-F-aniline at the R5 position showed Mtb potency and a SI profile similar to 39, with an aminopyran at position R5, but had lower aqueous solubility (15 μM) as expected because of the overall increase in lipophilicity with an additional aryl ring. Compound 41 with a 2,4-difluoroanilino group at the R5 position was the most potent compound (MIC 0.2 μM) obtained so far in these studies but was poorly soluble (<5 μM) in aqueous media. The addition of polar groups like CN and dimethylcarboxamide (42 and 43) at the para-position of the R5-anilino groups decreased activity (MIC 9 and >50 μM, respectively). Similarly, compound 44 with a 4-pyridylamino at the R5 position was inactive against Mtb up to the highest concentration tested (MIC > 50 μM). Interestingly, compound 45 with a 2-pyridylamino group at the same R5 position was quite potent with a MIC of ≤1 μM. The compound had moderate aqueous solubility (25 μM) with a modest SI of 25 for CHO cell-line toxicity, thus making it a valuable compound for MoA studies (see below). A few compounds with polar substituents para to the amino group on the 2-pyridyl ring were synthesized and screened with a view to improving physicochemical properties. For example, compound 46 with a 4-morpholino-2-pyridylamino group at the R5 position retained good activity (MIC = 2 μM) with improved solubility (180 μM) but was significantly more cytotoxic (SI 0.6).
Compound 47, with a 3-pyridyl at the R1 position and an ethylamino group at the R5 position, showed modest activity with an MIC of 6.25 μM and CHO IC 50 of >50 μM.
A limited SAR exploration at the R4 position on the pyrazole was conducted to evaluate the scope for further improvement in potency and selectivity. Compound 48 with 4-bromopyrazole showed only moderate activity (MIC 25 μM, CHO IC 50 > 50 μM) and was used as an intermediate for further synthesis. Replacing the bromine with a 3-pyridyl ring (49) abolished Mtb activity (MIC >50 μM), whereas the morpholine ring at this position (50) was well tolerated with an MIC = 6.25 μM and a cleaner cytotoxicity profile (CHO IC 50 > 50 μM). In comparison, the R3 position on the pyrazole showed a very limited scope for further modifications. In general, a cyclopropyl group as in compound 51 was well tolerated for Mtb potency but larger groups in general gave weakly active compounds. Polar groups like dimethylaminomethyl, carboxylic acid, and amide as in compounds 52, 53, and 54, respectively, led to considerable loss of activity against Mtb (MICs > 50 μM).
In summary, various modification on the pyrimidinone and pyrazole portions of the scaffold showed potential for improvement in potency against Mtb in vitro and MICs as low as 200 nM could be achieved. However, in general there was a narrow scope to improve the SI against mammalian cell-line toxicity. The potential impact of both these activities in vivo remains to be determined.
Physicochemical, DMPK, and Safety Profiles. Presented in Table 6 are physicochemical and in vitro DMPK properties of representative compounds, including solubility, microsomal MS, and human plasma protein binding (PPB). Compounds showed moderate to high aqueous solubility with excellent in vitro microsomal stability, with the exception of compound 8 that was only moderately stable in microsomes. A small number of compounds that were tested for the inhibition of liver cytochrome P450 enzymes did not show significant inhibition up to 20 μM (data not shown), indicating minimal possibility of metabolism-based drug−drug interactions of these compounds, which is a desirable attribute of a TB drug. 29,30 The compounds tested in the series were highly bound to the human plasma most likely due to the albumin binding of acidic pyrimidinone. 31 The safety profiles of compounds 1 and 12 were evaluated across a panel of 21 liability targets (39 functional assays), which included cell-based GPCRs and ion channels in both agonist and antagonist readout, and biochemical functional assays for nuclear hormone receptors and phosphodiesterases. 32 Both compounds showed no significant inhibition or activation of the enzyme/receptors at 10 μM test concentration. Both compounds showed hERG IC 50 > 30 μM. In summary, the in vitro safety profile of the compounds indicated no obvious safety liability even though the reason for mammalian cytotoxicity observed was not clear.
The pharmacokinetic parameters of representative compounds, 12, 47, and 40 were measured from mouse blood at intravenous doses of 2 mg/kg and oral doses of 20 mg/kg in the mouse ( Table 7). The compounds were well tolerated following either route of administration, with no obvious effects noted in the animals over the course of the exposure. Compounds 12 and 40 showed a low volume of distribution and low rate of clearance with moderate oral bioavailability of approximately 40%. Compound 47 showed rapid blood clearance and high volume of distribution with an oral bioavailability of 33%. Further analysis of mouse blood samples from the PK studies of 47 showed the glucuronide as the major metabolite. However, further optimization of physicochemical properties to improve pharmacokinetic properties, along with improved SI, is warranted in order to identify a compound suitable for in vivo efficacy studies in TB animal models.
Bactericidality and Activity against Clinical Isolates. The compounds were found to be bactericidal against replicating Mtb showing 2 log CFU reduction at 1−2× MIC over the time period of 7 days (Fig. S1). Importantly, we also evaluated the activity of 1, 12, 15, and 20 against a panel of clinical isolates (Table S1). All the tested compounds retained MICs against clinical isolates within the 4-fold range of MICs against the drugsensitive Mtb H37Rv strain. In order to explore the MoA, potent compounds were initially screened against various tool strains to rule out the involvement of known mechanisms and/or targets. The tested compounds retained activity against a QcrB mutant (A396T) and did not show hypersensitivity against a cytochrome-bd oxidase knockout mutant strain (cydKO), 33 thereby eliminating them as potential targets. The compounds did not show a positive signal in two standard bioluminescence reporter assays: PiniB-LUX 34 detects modulation in the iniB expression, if a compound targets Mtb cell−wall biosynthesis, and PrecA-LUX 34 detects modulation in the recA expression, an indicator of genotoxic compounds. A strain carrying a mutation in DprE1 (C387S), which confers resistance to other DprE1 inhibitors, was not resistant, suggesting DprE1 is not the target. However, strains carrying mutations in MmpL3 were cross-resistant to compounds (Table 8). 35 The compounds showed an increase of 8−100-fold in MIC values against the strains with either MmpL3 F255L or MmpL3 V681I or MmpL3 G596R , whereas, there was no change in activity against the MmpL3 F644L mutant. Most of these mutations lie within the region required for proton translocation. 36 Variability in the MICs against these mutant strains is suggestive of the fact that the compounds may bind differently to the protein depending on the mutation. To investigate whether or not compounds 1, 2, 12, 15, 47, 40, and 38 retain target selectivity for MmpL3 in Mtb cells, we asked whether or not conditional depletion of MmpL3 in a mmpL3 hypomorph (Grover et al., manuscript under review) would sensitize Mtb to the growth inhibitory effects of the compounds. However, transcriptional silencing of mmpL3 resulted in no significant change in the MICs of the compounds, confirming that MmpL3 is not the direct target of this series (Table S2). We hypothesize that MmpL3 acts as a transporter of these compounds across the cell membrane as the compounds can form heme-like iron-complexes, and MmpL3 is known to act as a heme transporter. 37,38 Hence, MmpL3 is likely to be responsible for building resistance toward compounds through the impairment of transport of compounds across the cell membrane rather than being a direct target. It is to be noted that the compounds were PiniB-LUX negativeindicating a non-cell wall MoA.
Based on the structural similarity of the compounds with the published thymidylate kinase inhibitors, 39,40 we asked whether this compound series can inhibit Mtb thymidylate kinase, which is an essential target encoded by Rv3247c. To investigate this, we tested the activity of the compounds 1, 2, and 12 against the Rv3247c hypomorph. However, we did not observe any MIC modulation of the compounds upon silencing of Rv3247c, suggesting that the tested compounds are not the thymidylate kinase inhibitors.
Treatment with 45 and 12 Affects Genes Involved in Fe Homeostasis. In continuation of our efforts to explore MoA, the effects of 45 (at 1 μM and 10 μM) and 12 (at 20 μM) treatment on the gene expression of Mtb was investigated. Mtb cultures were treated with the compound for 6 h, harvested, and the RNA extracted for transcriptional profiling by microarray analysis. The data obtained from the transcriptional profiling of Mtb exposed to 45 and 12 were similar (Table S3). We performed indepth analysis of the data obtained from samples treated with 45, which was tested at two concentrations. The 50 most upregulated (>4-fold) genes were selected in order to understand their potential contribution in the cellular response based on the MoA of 45 (Table S4). This indicated a transcriptional signature related to iron-sequestration. The upregulated genes included rv2377c-78c-79c-80c-81c-82c-83c-84-85-2386c (mbtA-H)all of which are associated with mycobactin biosynthesis or regulation, 41 which are known to be upregulated in response to iron deprivation.
Another upregulated gene-set rv1342-43-44-45-46-47-48-49 is also associated with the mycobactin biosynthesis or regulation or with iron transport where rv1348 and rv1349 were annotated as iron-regulated transporters, both being essential for growth in vitro. 42−44 The esx-3 gene cluster is composed of 11 genes stretching from rv0282 to rv0292. All of these were upregulated 2−5-fold in response to the stress caused by the compound. rv0287 and rv0288 are predicted to be essential for growth in vitro. Interestingly, ESX-3 is known to be required for the mycobactin-mediated siderophore iron uptake pathway, 45−48 and it is essential in Mtb but not in Mycobacterium smegmatis, a species which, in addition to mycobactins, produces exochelin that functions independently of ESX-3. However, ESX-3 is still required for mycobactin-mediated iron uptake in the latter species. 45,46,48 The nonessential nature of ESX-3 in M. smegmatis has previously proven to be of a great advantage when resolving mechanisms related to ESX-3 and mycobactin iron metabolism. 48 Some of the top upregulated genes are involved in lipid  The observed upregulation of the rv3269-70 operon encoding a stress response protein and a metal cation transporter, respectively, further supports the notion that the cellular response to 45 is one of the disruption of iron homeostasis. Rv0285PE5 which was upregulated in our study, is downregulated in the presence of iron. 49 Importantly, most of the upregulated genes of our data set were found to be downregulated when Mtb was exposed to an excess of iron. 49 Our data set also complements the recently published work by Kurthkoti et al., 50 where authors performed transcriptomics in iron-starved cultures and identified similar upregulated clusters of genes. This strongly suggests that the disruption of Fe homeostasis is an important component of the MoA of 45.
In summary, our data suggest that upon exposure to 2pyrazolylpyrimidinones, Mtb goes into an iron-starvation state. As expected, it upregulates genes involved in the mycobactin (siderophore) synthesis pathway and its regulation. Siderophores are small, high-affinity iron-chelating compounds secreted by microorganisms such as bacteria and fungi. In the process of cellular uptake, the Fe 3+ -siderophore complex is subsequently reduced to Fe 2+ to release the iron.
Iron Plays a Significant Role in the MoA. To confirm the role of iron in the MoA of this series, an Fe rescue experiment was performed by supplementing standard Middlebrook 7H9/Glu/ CAS/Tx media with 200 μM ferric ammonium citrate. The supplementation of Fe 3+ resulted in a 4-fold increase in the MIC (Table S5). Under physiological conditions, iron can exist in either the reduced ferrous (Fe 2+ ) form or the oxidized ferric (Fe 3+ ) form. The redox potential of Fe 2+ /Fe 3+ makes iron extremely versatile when it is incorporated into proteins as a catalytic center or as an electron carrier. Thus, iron is important for numerous biological processes, which include the tricarboxylic acid cycle, gene regulation, DNA biosynthesis, and so forth. Although iron is abundant in nature, it does not normally occur in its biologically relevant but aerobically unstable ferrous form. To evaluate this, we supplemented the Mtb culture with 200 μM ferrous sulfate, resulting in an 8-fold increase in the MIC for 45 (Table S5). Next, we confirmed the ability of Fe 2+ to rescue the Mtb from 45 toxicity in a checkerboard (2D) assay using GAST (glycerol-alanine-salts-Tween 80) media which is devoid of Fe unlike Middlebrook 7H9/Glu/CAS/Tx ( Figure 4A). The inhibition of Mtb growth by 45 in GAST medium supplemented with 12.5 μM Fe +2 (MIC 0.12 μM) decreased in a dosedependent manner with increasing concentration of Fe 2+ and resulted in a ≥32-fold increase in MIC at 100/200 μM Fe 2+ . To determine that 45 forms a complex with Fe 2+ , we performed UV−vis scanning of 45 with increasing amounts of Fe 2+ ( Figure  4B). The band intensity at 270, which was observed for 45 alone, decreased after the addition of Fe 2+ , whereas the absorbance around 285 is shifted to 300 with the increase in intensity, indicating the formation of the 45-Fe 2+ complex.
In line with the observation of better rescue with ferrous than ferric iron and a preference of heme for the former, there is a strong possibility that the compound binds iron to form a hemelike complex that enters Mtb via MmpL3. This also supports the reason for resistance with some of the MmpL3 mutants as described previouslyheme prefers ferrous over ferric. Involvement of a heme-like structure could also explain the associated cytotoxicity of the series, as described earlier. This is because of the toxic nature of heme to eukaryotic cells, which require specific heme-binding proteins to maintain a nontoxic homeostasis.

■ CONCLUSIONS
Whole-cell phenotypic screening of MMV compound library led to the identification of 2-pyrazolylpyrimidinones with potent antitubercular activity and a novel mode of action. Detailed SAR studies identified several compounds with potent activity against Mtb with moderate to high aqueous solubility and excellent in vitro microsomal stability. The compounds were bactericidal against replicating Mtb and retained potency against clinical isolates. Transcriptional profiling suggested that upon exposure to 2-pyrazolylpyrimidinones, Mtb goes into an iron-starvation state that may account for lethality. Iron supplementation using a high concentration of ferrous salts showed shifts in MICs of the compounds, further confirming that 2-pyrazolylpyrimidinones perturbs the Fe-homeostasis in Mtb. Further optimization of pharmacokinetic properties, along with improved SI between MIC and mammalian cytotoxicity, is needed to identify a compound suitable for in vivo efficacy studies in mouse TB

■ EXPERIMENTAL SECTION
MIC Testing and Triage Assays. Unless otherwise indicated, an Alamar Blue fluorescence-based broth microdilution assay was used to assess MIC of compounds against Mtb strains, as described previously. 51,52 Rifampicin was included as a control. Biology triage assays were carried out as described by Naran et al. 34 MICs against Mutant Strains. MICs were determined after 5 days of growth in Middlebrook 7H9 medium plus OADC supplement and 0.05% w/v Tween 80 as described previously. 53 Growth was monitored by OD. Data were fit using the Levenberg−Marquardt least-squares plot. MIC was defined as the concentration required to inhibit growth by 90%.
RNA Extraction and Transcriptional Profiling. Total RNA was extracted from Mtb H37Rv that had been treated for 6 h with 1× and 10× MIC of the compound or vehicle control as previously described. 54 RNA quality was confirmed by UV spectral analysis and Agilent 2100 Bioanalyzer returning a RNA integrity number of 8 or higher. Fluorescent-tagged cDNA was prepared via a direct random-primed labeling method as follows. Microarrays were hybridized robotically using a TECAN HS Pro 4800 hybridization station, Agilent 2× gene expression hybridization HI-RPM buffer, and 10× blocking reagent at 65°C for 17 h and washed with Agilent Gene Expression Wash Buffer 1 at room temperature and Gene Expression Wash Buffer 2 at 37°C. Then, slides were dried under a nitrogen gas for 3 min at 30°C. The slides were imaged using an Agilent high-resolution DNA microarray scanner (model G2505C) at 5 μm resolution and 100/10% PMT dual scanning for XDR extended dynamic range. Agilent Feature Extraction software was used for image analysis.
DMPK. All protocols for in vitro DMPK studies and mouse PK studies are available in the supplementary document. Animal studies were conducted following guidelines and policies as stipulated in the UCT Research Ethics Code for Use of Animals in Research and Teaching, after review and approval of the experimental protocol by the UCT Senate Animal Ethics Committee (protocol FHS-AEC 013/032).
Chemistry. All commercial reagents were purchased from Sigma-Aldrich, Combi-Blocks, Enamine, or Fluorochem and were used without further purification. Solvents were used as received unless otherwise stated. Analytical thin-layer chromatography was performed on SiO2 plates on aluminum backing. Visualization was accomplished by UV irradiation at 254 and 220 nm. Flash column chromatography was performed using a Teledyne ISCO flash purification system with SiO 2 60 (particle size 0.040−0.055 mm, 230−400 mesh). Purity of all final derivatives for biological testing was confirmed to be >95% as determined using an Agilent 1260 Infinity binary pump, Agilent 1260 Infinity diode array detector (DAD), Agilent 1290 Infinity column compartment, Agilent 1260 Infinity standard autosampler, and Agilent 6120 quadrupole (single) mass spectrometer, equipped with APCI and ESI multimode ionization sources. Using a Kinetex Core C18 2.6 μm column (50 mm × 3 mm); mobile phase B of 0.4% acetic acid, 10 mM ammonium acetate in a 9:1 ratio of HPLC grade methanol and type 1 water, mobile phase A of 0.4% acetic acid in 10 mM ammonium acetate in HPLC grade (type 1) water, with a flow rate of 0.9 mL/min, DAD; or an Agilent UPLC−MS was used: Agilent Technologies 6150 quadrupole, ES ionization, coupled with an Agilent Technologies 1290 Infinity II series UPLC system Agilent 1290 series HPLC at two wavelengths 254 and 290 nm using the following conditions: Kinetex 1.7 μm Evo C18 100A, LC column 50 mm × 2.1 mm, solvent A of 0.1% (formic acid) water, and solvent B of 0.1% (formic acid) acetonitrile. The structures of the intermediates and end products were confirmed by 1 H NMR and mass spectrometry. Proton magnetic resonance spectra were determined in an appropriate deuterated solvent on a Varian Mercury spectrometer at 300 MHz or a Varian Unity spectrometer at 400 MHz.
Synthesis and analytical data of intermediates, microbiology assays, MIC data against clinical isolates, screening data against MmpL3 under-expressing strain of Mtb, in vitro DMPK assays, and mouse pharmacokinetic studies (PDF) LCMS HPLC and NMRs of lead compounds (PDF) Data from RNA microarray experiment (XLSX) Data from RNA microarray experiment (XLSX) Molecular formula strings and some data (CSV)