Improved Phenoxyalkylbenzimidazoles with Activity against Mycobacterium tuberculosis Appear to Target QcrB

The phenoxy alkyl benzimidazoles (PABs) have good antitubercular activity. We expanded our structure–activity relationship studies to determine the core components of PABs required for activity. The most potent compounds had minimum inhibitory concentrations against Mycobacterium tuberculosis in the low nanomolar range with very little cytotoxicity against eukaryotic cells as well as activity against intracellular bacteria. We isolated resistant mutants against PAB compounds, which had mutations in either Rv1339, of unknown function, or qcrB, a component of the cytochrome bc1 oxidase of the electron transport chain. QcrB mutant strains were resistant to all PAB compounds, whereas Rv1339 mutant strains were only resistant to a subset, suggesting that QcrB is the target. The discovery of the target for PAB compounds will allow for the improved design of novel compounds to target intracellular M. tuberculosis.

M ycobacterium tuberculosis was a leading cause of death by infectious disease in the world in 2015. 1 Contributing to over 1.4 million deaths and infecting an estimated one-third of the population, M. tuberculosis is a leading global health problem with major social and economic burdens. 1 Over the past decade, there has also been an alarming increase in multidrug-resistant (MDR-TB) and extensively drug-resistant (XDR-TB) strains that are refractory to conventional therapeutics. 1 Because of this, there is a dire need for novel drugs and preventatives to combat this disease.
The past few years have seen promising leads for new antituberculosis drugs with several advancing to clinical trials. One series in the early stages of development, the phenoxy alkyl benzimidazoles (PAB), has shown good promise with minimum inhibitory concentrations (MICs) against M. tuberculosis in the nanomolar range, sterilizing activity under starvation conditions and low cytotoxicity against eukaryotic cells. 2 In this study, we further characterized the SAR for the PAB series and synthesized compounds with antitubercular activity in the low nanomolar range, good cytotoxicity profiles, and intracellular activity against M. tuberculosis. We also identified the probable target of PAB compounds as QcrB, a component of the cytochrome bc 1 oxidase of the electron transport chain. This knowledge will enable the elucidation of the mechanism of action for the strategic design of new compounds and combination therapies to treat M. tuberculosis infection.
■ RESULTS AND DISCUSSION PAB Structure−Activity Relationship. The alkyl benzimidazoles were synthesized according to Scheme 1. Condensation of appropriate 1,2-diaminobenzene derivative with propionic acid yielded the benzimidazole intermediate (1) upon heating. Alkylation of the benzimidazole intermediate (1) was accomplished by reacting with dibromoalkane to form N-(bromoalkyl)benzimidazole (2). This in turn was treated with the corresponding anilines, thiophenols, and phenols to yield the corresponding benzimidazole alkylamine, alkyl thioether, and alkyl ethers.
In our previous study, we observed that compounds with heteroatoms (O, S, and N) as part of the linker were tolerated, a linker length of three to four carbons was favored, and ethyl at the 2-position and methyl at the 6-position on the benzimidazole were the most favorable substituents on the core. 2 We first investigated aryl and heteroaryl thioethers for their activity against M. tuberculosis. Since compound 4 from our previous work showed good activity (Table 1), 2 we made analogues (6 − 18) to explore the SAR around the phenyl group of 4. In our previous studies on aryl ethers, we observed that the 4-methyl substitution improved both activity and selectivity. In the case of the aryl thioether analogues, 4-methyl substitution (6) did not show any improvement in activity, while 4-methoxy substitution (7) decreased activity by . Replacement of the phenyl with biaryls or a heteroaryl ring, such as naphthalene (8), quinoline (9), quinazoline (10), benzothiazole (11 and 12) or benzimidazole (13 and 14) decreased activity while replacement with thiazole (15 and 16) or tetrazole (17) resulted in no activity (MIC > 20 μM). The 4-phenylthiazole (18) was the most active thioether analogue with an MIC of 0.73 μM. However, it was fairly cytotoxic to eukaryotic cells (TC 50 = 9.7 μM), despite the selectivity index (SI = 13) being similar to the earlier compounds.
In our previous SAR study, we observed that replacing the phenyl ether in compound 3 with 5-(4-chlorophenyl)oxadiazol-2-yl thioether as in 19 improved activity. 2 We attempted to improve this further by modifying the linker length ( Table 2). Reduction of the linker length from four carbons to three (20) or two (21) was detrimental to the antitubercular activity; in the case of the propyl (20) linker, we observed a 39-fold reduction in activity and further reduction of the linker to an ethyl (21) resulted in complete loss of activity. The addition of a weaker electron-releasing group such as the methyl group on the 6-position of the benzimidazole core (22) retained good potency (MIC = 0.085 μM). The combination of a 6-methyl on the benzimidazole core and a propyl linker (23) was slightly less active in comparison (MIC = 0.34 μM). Interestingly, replacing the 6-methyl with 6-methoxy, a strong electron-donating group, in compound 24, reduced activity by almost 10-fold (MIC = 2.5 μM).
We next explored the influence of nitrogen as the heteroatom in the linker (Table 3). We synthesized compounds (25−35) to investigate the electronic and steric effects of different aromatic substituents on the terminal aniline. The addition of the electron donating groups methyl (25) or methoxy (26) made no improvement in activity, although the methyl analogue was slightly more active. A trifluoromethyl substituent at the meta position (27) had a comparable MIC to compound 5. However, the cytotoxicity was significantly worse than that of 5 and the selectivity was 8-fold lower. The 3,5-dimethoxy (28) lost activity, whereas the 3-halo and 4-methyl substituted anilines (29,30) showed comparable activity to that of 5, The substituents Scheme 1. Synthesis of Amine and Thioether Aryl, Hetero Aryl, and Cycloalkyl Benzimidazoles 3-benzyloxy (31) and 4-pyridyl (32) showed similar activity to that of 5, but the 3-pyridyl (33) was much less active (MIC = 3.7 μM). Moving from the aniline (5) to benzylamine (34) was detrimental to the antitubercular activity. Replacing the terminal aniline with a morpholino group also resulted in loss of activity (35).

ACS Infectious Diseases
Article substitution on the phenyl ether (47) or aniline (48) had minimal effect on activity, but this substitution resulted in a 10-fold shift in the activity of a corresponding phenyl thioether (49). Both 3-chloroaniline (50) and 4-chlorophenyl thioether (53) were less active, but a 3-trifluoromethyl substitution on aniline (51) retained all activity of parent compound 44. We also tested 3-benzyloxy aniline (52) with good activity. We finally probed the effect of combination of the optimized residues, namely, the three-carbon linker and a 6-methyl substitution on the benzimidazole core (54−58). Four compounds had a selectivity index of >200 (54, 55, 56, and 57) ( Table 5). The aniline analogues 54, 55, and 57 had the best activities with comparable MICs of 0.061 μM, 0.067 μM, and 0.070 μM, respectively. While 54 had no cytotoxicity at 50 μM, the addition of one (55) or two (56) methyl groups on the phenyl ring caused a stepwise increase in cytotoxicity toward eukaryotic cells, thus reducing the selectivity index from >819 (54) to 470 and 200 for 55 and 57, respectively ( Table 5). The methoxy analogue (56) was 5-fold less active (Table 5).
Several phenoxy alkyl imidazoles 59−68 were evaluated to study the importance of the benzimidazole core for the antitubercular activity of PAB ( Table 6). Many of the compounds lost activity (59, 60, 64, and 67), while others maintained only moderate activity compared to benzimidazole variants (61, 62, 63, 65, 66, and 68). These data show the benzene of the benzimidazole core is not strictly required; several imidazole variants maintain moderate activity against M. tuberculosis. However, the benzene ring does appear to confer optimal antitubercular activity to PAB compounds.
PAB Compounds Target QcrB. To identify the target of PAB compounds, we isolated resistant mutants against three different PABs. Of the seven confirmed mutants, five contained mutations in qcrB, the Cytochrome b subunit of the cytochrome bc 1 oxidase (Rv2196), while two of the three mutants raised Table 6. SAR of Phenoxy Alkyl Imidazole Analogues with Varying Linker Length a MIC 99 is the minimum concentration required to completely inhibit growth of M. tuberculosis in liquid culture. MICs of active compounds are the average of two independent experiments ± standard deviation. b TC 50 is the concentration required to inhibit growth of Vero cells by 50%. TC 50 is the average of two independent experiments ± standard deviation. c Selectivity index (SI) is TC 50 /MIC. ND, not determined. NC, not calculated.

ACS Infectious Diseases
Article against compound 4 contained mutations in Rv1339, a gene of unknown function (Table 7).
We confirmed targeting of QcrB by PABs by testing the MIC of 30 active compounds against LP-0351551-RM1 (QcrB M342T ), one of the isolated resistant mutants. All compounds tested showed a minimum 5-fold shift in MIC compared to the parental strain with 78% of the compounds having at least a 10-fold higher MIC (Table 8). Most compounds had similar MICs against LP-0341914-RM5 (Rv1339 V90M ) as against the wild-type strain ( Table 9). Six compounds had 5-fold shifts in MIC against the Rv1339 V90M mutant, but in all cases the observed resistance shift was significantly smaller than those observed when the same compounds were tested against the QcrB M342T mutant (Tables 8  and 9). Together, these data indicate the Rv1339 gene product is unlikely to play a major role in resistance to the PAB series and implicates QcrB as the target of PAB compounds.
Structurally, the benzene of the benzimidazole core of the PAB series is not required for antituberculosis activity, as some compounds with just an imidazole ring (PAB-II) still have moderate activity against M. tuberculosis (Table 6). However, the PAB-II compounds appear to have a different target or mode of action, since the QcrB M342T mutant strain was not resistant to this set (Table 10). This argues that while the benzene ring of the core may not be strictly required for whole-cell activity, this ring may be responsible for targeting of QcrB specifically. It would be interesting to elucidate the target of PAB-II compounds to determine their mechanism of action.
As a key component of the mycobacterial electron-transport chain, QcrB receives electrons from menaquinol and serves as a pump shuttling protons into the periplasmic space, thereby establishing the proton gradient required for ATP production. 3 Over the past few years, the cytochrome bc 1 oxidase has been discovered to be the target of novel drugs against numerous infectious diseases including malaria and Leishmania parasites, as well as multiple classes of compounds active against M. tuberculosis, including the drug Q203, an imidazopyridine amide currently in clinical trials. 4−8 Two of the resistance mutations found in our study cause amino acid changes at position 312 of QcrB, which sits in the same pocket as the resistant mutant at position 313 found for Q203 8 . This suggests the PAB compounds may be binding in the same location and have a similar mechanism of action as Q203. To test this, we examined crossresistance of PAB compounds to a strain of M. tuberculosis containing the same mutation found in Q203-resistant isolates (QcrB T313I ). PAB compounds were highly resistant to the QcrB T313I variant; the MIC 90 for PAB compounds ranged from 13-fold up to 200-fold higher compared to the wild-type strain (Table 11). The fold-change in MIC against QcrB T313I was larger than that for QcrB M342T for every compound tested except IDR-0351551, the compound against which QcrB M342T was raised. When Q203 was tested against QcrB W312C , the MIC was greater than 100 nM compared to just 2.8 nM for H37Rv-LP. Together, these data suggest an overlapping binding site between Q203 and PAB compounds.
To provide further evidence supporting QcrB as the target of PAB compounds, we performed initial mechanistic studies

ACS Infectious Diseases
Article surrounding QcrB. Inhibition of QcrB leads to reduced proton pumping across the bacterial membrane. An inability to pump protons disrupts ATP production and increases the concentration of protons within the cell, thus lowering intrabacterial pH (pHIB). As would be expected following QcrB inhibition, PAB compounds caused ATP depletion over a 24 h time-period ( Figure 1A) and significantly decreased pHIB; all PAB compounds caused a reduction in pHIB to at least 6.7 with 8/16 compounds dropping the pHIB below 6.5, similar to that seen with the ionophore monensin ( Figure 1B). The one PAB-II compound tested showed no activity in this assay, consistent with the conclusion QcrB is not its target. Together with the resistant mutant isolation, these data strongly support QcrB as the target for PAB compounds and suggest potential mechanisms for compound activity.
Previous studies identified PAB compounds as bacteriostatic against actively growing M. tuberculosis, while bactericidal against nutrient-starved bacteria. 2 We tested this for two compounds, including one of our most potent compounds, IDR-0578347 (54). Both compounds had similar activity; when tested at 10× their respective MICs, the compounds inhibited growth of M. tuberculosis grown aerobically but displayed bactericidal activity only in the context of nutrient starvation (Figure 2A,B). One possible explanation for the differential activity stems from the elucidation of QcrB as the target. Under replicating conditions individual bacteria contain large amounts of ATP necessary to support growth and division of bacterial cells. Under starvation conditions ATP is still required to maintain basal cellular processes, but the amount of ATP/cell is much reduced. 9 It is possible ATP depletion mediated by PAB compounds during replicative growth is only enough to inhibit growth and force the bacteria into a nonreplicative state, while under starvation conditions any reduction from the basal amount of ATP being produced is enough to cause bacterial cell death. This hypothesis still needs to be tested experimentally.
The number of series of compounds that target QcrB has led to an intriguing debate within the field. On one hand, the number of times this target has been identified highlights the importance of the cytochrome bc 1 complex and ETC to mycobacterial survival. 3 However, there has been concern as to the promiscuity of the target, the similarity to the human electron transport chain, and the ability of M. tuberculosis to adapt by using the bd oxidase as an alternative electron acceptor under oxygen-limited conditions. 10,11 Recent work used the adaptability of M. tuberculosis against itself by using combination therapy targeting the ETC. 12 On the basis of our results, it is possible the PAB series could similarly be used in combination with other electron-transport inhibitors. These studies are currently underway.
As for the potential consequence of targeting a component of the electron transport chain, our previously published data suggests the inhibitors described herein are specific to M. tuberculosis, as PAB compounds do not inhibit growth of several other bacteria, including M. smegmatis. 2 In fact, when tested against M. tuberculosis in a high-content analysis of M. tuberculosis macrophage infection, PAB compounds showed exquisite sensitivity toward bacterial cells with MICs against M. tuberculosis inside macrophages similar to those found in liquid broth and selectivity indexes ranging from 19 to nearly 2000 (Table 12). Several compounds even appeared to be slightly more active inside macrophages than in liquid culture. Together, these data show the penetration of these compounds into eukaryotic cells, while highlighting the safety of the PAB compounds, offering a promising therapeutic window that would allow for treatment of infection while minimizing host cell toxicity.

■ CONCLUSION
This work investigated a range of possible amine and thioether linkages as a strategy to identify improved alkyl benzimidazole

ACS Infectious Diseases
Article molecules. Several compounds showed good potency and selectivity as well as activity against intracellular bacteria. The key determinants for optimal antimycobacterial activity of the compounds are methyl groups in both the C-6 position of the benzimidazole and the para position of the terminal phenyl ring, a three or four carbon atom linker, a nitrogen as the heteroatom on the alkyl chain, and a benzimidazole as the core moiety. The PAB compounds appear to target QcrB, a component of the M. tuberculosis electron transport chain, suggesting they could be useful in combination therapy with other electron-transport chain inhibitors.

Determination of Minimum Inhibitory Concentration (MIC).
MICs were performed as previously described; 14 briefly MIC 99 s were determined against M. tuberculosis H37Rv-LP constitutively expressing codon-optimized mCherry grown in 7H9-Tw-OADC under aerobic conditions. Bacterial growth was measured by OD 590 after 5 days of incubation at 37°C. MIC 99 was defined as the minimum concentration required for >99% growth inhibition using the Gompertz model. MIC 90 values were determined against M. tuberculosis H37Rv-LP or indicated mutants under the same growth conditions. MIC 90 is defined as the concentration of compound required to inhibit growth of M. tuberculosis by 90% and was determined from the Levenberg− Marquardt least-squares plot. 15 Cytotoxicity Assay. Cytotoxicity was determined against the Vero African green monkey kidney cell line (ATCC CCL-81). Cells were cultured in Dulbecco's Modified Eagle Medium (DMEM), High Glucose, GlutaMAX (Invitrogen), 10% v/v FBS (Fetal Bovine Serum), and 1× of Penicillin-Streptomycin

ACS Infectious Diseases
Article Solution (100 units/mL of penicillin, 100 μg/mL of streptomycin). Cells were plated into 96-well plates and allowed to adhere for 24 h before exposure to compounds for 48 h. CellTiter-Glo Reagent (Promega) was added to 96-well plates and relative luminescent units (RLU) were measured. Inhibition curves were fitted using the Levenberg−Marquardt algorithm. Toxic concentration (TC 50 ) was defined as the concentration of compound that gave 50% inhibition of growth. Resistant Mutant Isolation. M. tuberculosis H37Rv-LP was plated on 7H10-OADC plates containing 5× or 10× MIC of three PAB compounds 3, 4, and 19 (Compounds 6, 53, and 68 from 2 ). Resistant clones raised against compounds 4 and 19 were isolated, confirmed, and whole genome-sequenced as previously described. 16 The resistant clone raised against compound 3 was isolated and confirmed by sequencing.
ATP Depletion Assay. M. tuberculosis H37Rv-LP was grown in duplicate at 37°C for 24 h in the presence of 3-fold dilutions of PAB compounds starting at 200 μM. ATP levels were measured using the BacTiter-Glo assay kit (Promega) according to the manufacturer's instructions. OD 590 of the duplicate plate was measured using a Synergy H4 plate reader (BioTek). ATP/OD was calculated by dividing relative fluorescent units from the BacTiter-Glo plate by OD 590 readings.
pH Homestasis Assay. M. tuberculosis H37Rv-LP containing a pH-sensitive ratiometric GFP reporter 17 was added to compounds at 100 μM final concentration in phosphocitrate buffer at pH 4.5. After 48 h fluorescence was read and intrabacterial pH was determined based on the ratio between Ex396/Em510 and Ex475/Em510.
Determination of Compound Kill Kinetics. M. tuberculosis H37Rv-LP was inoculated into 7H9-Tw-OADC at 1 × 10 5 CFU/mL in the presence of 10× the respective MIC of either compound 46 or 54 or a DMSO control (final DMSO concentration of 2%). For starvation experiments, H37Rv-LP was washed and resuspended in phosphate-buffered saline (PBS) + 0.05% Tyloxapol at 1 × 10 5 CFU/mL for 2 weeks prior to compound addition. Serial dilutions were performed at the indicated times and bacteria were plated on 7H10-OADC agar plates for CFU enumeration.
Intracellular Activity. Activity against intracellular bacteria was tested as described. 18 In brief, RAW 264.7 cells were batchinfected overnight in RPMI-1640, 10% v/v FBS, 2 mM Glutagro, 1 mM sodium pyruvate medium with M. tuberculosis DREAM8 (an H37Rv variant constitutively expressing dsRED) 19 at a multiplicity of infection of 1. Cells were harvested, washed twice, and seeded into 384-well assay plates at 3.3 × 10 3 cells per well. Infected cells were exposed to compounds for 72 h, and Macrophage nuclei were stained with SYBR Green I dye and imaged on an ImageXpress Micro with a 4× objective using a FITC filter set. M. tuberculosis were imaged using a Texas Red filter set. Images were analyzed by setting a lower bound for pixel intensity for each channel to define nuclei and bacteria above the background. The integrated intensity of each channel above this threshold was determined. Raw data were normalized to the average integrated control wells. Growth inhibition curves were fitted using the Levenberg−Marquardt algorithm. The IC 50 and IC 90 were defined as the compound concentration that produced 50% or 90% of the growth inhibitory response, respectively, for each readout.
General Methods. 1 H NMR (300 MHz) spectra were recorded on a Bruker Biospin NMR spectrometer. Reactions were monitored using thin-layer chromatography (TLC) using Whatman silica gel 60 Å plates with a fluorescent indicator and visualized using a UV lamp (254 nm). Flash chromatography was performed on Grace with GraceResolv Normal Phase disposable silica columns. High-performance liquid chromatography (HPLC) was performed on a Gilson 322 HPLC pump with a Gilson UV/ vis-155 detector and a Phenomenex Gemini C18 column (10 μm, 250 mm × 10 mm). The purity of all final products was >95% as determined by HPLC analysis conducted on an Agilent 1100 HPLC-MS system (Phenomenex Gemini C18 column, 5 μm, 3 mm × 50 mm, 0.45 mL/min, UV 254 nm, room temperature) with gradient elution (5−95% acetonitrile in water over 8 min with all solvents containing 0.05% formic acid). Liquid chromatography− electrospray ionization mass spectrometry (LC−MS/ESI-MS) were acquired on an Agilent LC/MSD-SL with a 1100 HPLC and G1956B mass spectrometer with a Phenomenex Gemini 5 μm C18 110 Å 50 mm × 3 mm column. High-resolution mass spectra (HRMS-ESI) were acquired by the Mass Spectrometry Laboratory at the University of Michigan on an Agilent Q-TOF HPLC−MS.
General Procedure for Preparation of Compounds 20−24. To a solution of benzimidazole in dimethylformamide were added 5 mol equiv of potassium carbonate and 2 mol equiv of 5-(4-chlorophenyl)-1,3,4-oxadiazole-2-thiol. The reaction was stirred overnight at room temperature or until all of the starting material disappeared. The reaction mixture was washed with water and extracted with ethyl acetate. The organics were dried with anhydrous sodium sulfate and concentrated in vacuo. The crude mixture was purified by column chromatography.