Diarylquinolines Are Bactericidal for Dormant Mycobacteria as a Result of Disturbed ATP Homeostasis*

An estimated one-third of the world population is latently infected with Mycobacterium tuberculosis. These nonreplicating, dormant bacilli are tolerant to conventional anti-tuberculosis drugs, such as isoniazid. We recently identified diarylquinoline R207910 (also called TMC207) as an inhibitor of ATP synthase with a remarkable activity against replicating mycobacteria. In the present study, we show that R207910 kills dormant bacilli as effectively as aerobically grown bacilli with the same target specificity. Despite a transcriptional down-regulation of the ATP synthase operon and significantly lower cellular ATP levels, we show that dormant mycobacteria do possess residual ATP synthase enzymatic activity. This activity is blocked by nanomolar concentrations of R207910, thereby further reducing ATP levels and causing a pronounced bactericidal effect. We conclude that this residual ATP synthase activity is indispensable for the survival of dormant mycobacteria, making it a promising drug target to tackle dormant infections. The unique dual bactericidal activity of diarylquinolines on dormant as well as replicating bacterial subpopulations distinguishes them entirely from the current anti-tuberculosis drugs and underlines the potential of R207910 to shorten tuberculosis treatment.

Mycobacterium tuberculosis infection results in more than 2 million deaths per year and is the leading cause of mortality in people infected with HIV 2 (1). The global epidemic of tuberculosis (TB) is fuelled by co-infection of HIV patients with TB and a rise in multidrug-resistant TB strains (2). Despite the fact that TB control programs have been in place for decades, approximately one-third of the world population is latently infected with M. tuberculosis. Reactivation of latent TB is a high risk factor for disease development, particularly in immunocom-promised individuals, such as HIV-infected patients. For global control of the TB epidemic, there is an urgent medical need for new drugs active against dormant or latent bacilli. These socalled sterilizing drugs would be able to shorten the current 6-month treatment duration for drug-susceptible TB and also offer new treatment opportunities for latent TB.
Tubercle bacilli enter lungs of healthy individuals by inhalation, where they are phagocytosed by the alveolar macrophages that eliminate most of the invading mycobacteria (3). However, a small proportion of bacilli survive and exist in a nonreplicating, hypometabolic state, and these bacilli are tolerant to killing by bactericidal anti-TB drugs, such as isoniazid (4). They can linger in these altered physiological environments for an individual's lifetime and maintain the capability of causing active TB after reactivation. The pathophysiological conditions in human lesions, thought to lead to persistence, are reduced oxygen tension, nutrient limitation, and acidic pH (5,6).
Recently, we identified a new chemical class, diarylquinolines (DARQs) that demonstrate potent anti-mycobacterial activity on replicating bacilli both in vitro and in vivo (7), and the lead compound, R207910 (or TMC207), is currently in Phase IIb clinical trials for the treatment of patients with multidrug-resistant TB. R207910 acts by specifically targeting the membrane-bound c-subunit of F 1 F 0 -ATP synthase, the ATP-synthesizing machinery of the cell (8). During synthesis of ATP, the energy stored in the electrochemical proton gradient across the membrane is utilized to drive protons from the periplasmic space into the cytoplasm through the F 0 subunit and supplying a torque to the F 1 unit to convert ADP into ATP (9,10). Under nonrespiratory conditions, in several bacteria, ATP synthases can function in the reverse direction, hydrolyzing ATP to ADP through ATPase activity and as such pumping protons from the cytoplasm into the periplasmic space (11). In this way, ATP synthase generates the membrane potential required for the uptake of nutrients. In dormant mycobacteria, a considerable remodeling of the respiratory chain has been reported, including down-regulation of cytochrome aa 3 type oxidase and upregulation of the cytochrome bd type menaquinole oxidase (12,13). Several studies also suggest that hypometabolic, nonreplicating mycobacteria have decreased requirements for ATP synthase, since the genes encoding the components of the ATP synthase operon are down-regulated, both in vitro and in vivo (12)(13)(14). It is, however, an open question whether ATP syn-thase plays an essential role in dormant bacilli and whether the enzyme is responsible for ATP synthesis and/or maintenance of the membrane potential. Furthermore, it is not known whether, despite ATP synthase down-regulation, this enzyme can still be utilized as a drug target in dormant bacilli and whether DARQs are still able to effectively kill dormant bacilli.
Here we report that, despite a substantial down-regulation of the ATP synthase operon and reduced cellular ATP levels, DARQs effectively kill nonreplicating M. tuberculosis. We find that dormant mycobacteria have active and functional ATP synthase that is capable of synthesizing ATP, a scarce energy resource in a nonreplicating cell. The depletion of ATP in dormant mycobacteria by the R207910-mediated chemical inhibition of ATP synthase leads to potent bactericidal activity. Thus, targeting processes that generate ATP and concurrently disturb the cellular ATP homeostasis is an effective strategy against dormancy. R207910 exhibits no significant effect on the membrane potential in dormant or replicating mycobacteria, suggesting that ATP synthase is not critical for maintaining the membrane potential but is primarily used for the production of ATP.
We also demonstrate an increased susceptibility of dormant mycobacteria toward R207910 as compared with actively growing bacteria. Using R207910-resistant mycobacterial strains, we show that the drug specifically targets ATP synthase during dormancy. The unique dual bactericidal action of R207910 on dormant as well as actively replicating bacteria probably contributes to its remarkable sterilizing efficacy in mice, where R207910 as a monotherapy was shown to be as effective as the triple combination of rifampin, isoniazid, and pyrazinamide (7,15). This may also explain the ability of this drug to render lungs of infected mice culture-negative faster than first line anti-TB drugs.
In conclusion, despite a number of changes in the energy metabolism during dormancy, the ATP production by ATP synthase remains essential for mycobacterial survival. The current study demonstrates the critical importance of ATP synthase in mycobacterial dormancy and suggests that the ATP synthase is an excellent target to treat latent TB infections.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Culture Medium-M. tuberculosis H 37 Rv, Mycobacterium smegmatis, and Mycobacterium bovis BCG were obtained from the American Type Tissue Culture collection. Mycobacteria were cultured in Middlebrook 7H9 medium (Difco) with 0.05% Tween-80 (Sigma) in log phase for a period of 3-4 days before the start of the experiment. The incubation was done in a 250-ml sterile Duran bottle with stirring conditions 37°C (500 rpm).
Chemical Synthesis of DARQ Analogs-DARQ compounds are synthesized as described in a earlier patent (WO2004/ 011436) and previous reports (7).
MIC Microdilution Assays and Determination of Minimum Bactericidal Concentration (MBC 90 ) and Wayne Cidal Concentration (WCC 90 )-Minimum inhibitory concentration (MIC, g/ml) of M. tuberculosis against several DARQ compounds was determined by using 7H9 liquid broth and 7H10 agar dilution as described in earlier studies (16) with minor modifica-tions using Alamar blue or resazurin for fluorometric measurement (7). MBC 90 (g/ml) was determined as the concentration where 90% of the aerobic grown bacteria were killed after 5 days of treatment by colony-forming unit (CFU) counting using 7H10 agar dilution according to Clinical and Laboratory Standards Institute guidelines. WCC 90 (g/ml) was determined as the concentration where 90% of the dormant bacteria (grown in the Wayne model) were killed after 5 days of treatment by CFU counting using 7H10 agar dilution.
Dormancy Assays-Drastic oxygen depletion (hypoxia model) of mycobacterial cultures was achieved by placing loosely capped tubes containing different drug concentrations inside an anaerobic jar (BBL) along with anaerobic gas generation envelopes (using palladium catalysts), as described previously (17). After 7 days of anaerobiosis, the dormant cultures were harvested by low speed centrifugation, washed twice with 7H9 medium to remove the drugs, and resuspended in drugfree medium. The CFUs of the treated and untreated cultures were determined by plating on 7H10 agar to evaluate the bactericidal activity. In the Wayne model, gradual O 2 depletion induces dormancy in mycobacterial cultures, as previously described (18). Briefly, after induction of dormancy for 18 days by gradual O 2 depletion, cultures without exposure to oxygen were treated further with various drug concentrations for a period ranging from 4 to 21 days, and CFUs were counted on 7H10 plates. In the NO model, dormancy was obtained by incubating cultures with 150 M of a NO donor (diethylenetriamine/nitric oxide (DETA/NO)) for 2 h, followed by treatment with several compounds for 10 h as previously described (19).
RNA Isolation and Real Time Quantitative PCR-Dormant and aerobically grown bacterial pellets were resuspended in Trizol and disrupted with acid-washed glass beads in the Mixer Mill (MM 301). Total RNA was isolated as previously described (20) and cleaned up with the RNeasy Protect Bacteria minikit (QIAgen). Reverse transcription was carried out from 0.5 g of total RNA using random hexamer primers and Superscript II RT as described by the manufacturer (Invitrogen). The sequence of the TaqMan FAM (6-carboxyfluorescein) and TAMRA (6-carboxy-N,N,NЈ,NЈ-tetramethylrhodamine) probes (Eurogentec) and primers for the mycobacterial genes analyzed in this study are available upon request. Real time quantitative PCR was carried out using the ABI PRISM 7900HT sequence detection system (Applied Biosystems) as described by the manufacturer. Amplification of endogenous 16 S ribosomal RNA was performed to standardize the amount of cDNA sample added to each reaction.
Total Cellular ATP Measurement Assay-ATP levels of log phase aerobic and dormant cultures of M. tuberculosis were measured using the ATP bioluminescence assay kit (Roche Applied Science), as previously described (8). ATP levels were followed during dormancy (Wayne model) from day 0 to day 21. Additionally, M. tuberculosis cultivated under Wayne dormancy conditions for 18 days were treated with R207910 (0.006 -6 g/ml), isoniazid (0.3 g/ml), and dicyclohexylcarbodiimide (DCCD) (100 g/ml) for 4 days, and ATP levels were measured.
Measurement of ATP Synthesis Activity-ATP synthesis of dormant M. smegmatis grown under Wayne conditions for 10 days was measured as described previously (8). Briefly, inverted membrane vesicles were prepared using a precooled French pressure cell at 20,000 p.s.i. The membrane vesicles were preincubated with R207910 or DCCD under stirring conditions at room temperature for 10 min. The ATP synthesis activity was determined by energizing the membranes with NADH and quantifying the amount of ATP produced using the luciferin/ luciferase system (ATP Bioluminescence Assay Kit HS II; Roche Applied Science). Data are presented as averages Ϯ S.E.
Measurement of the Membrane Potential-Membrane potential was measured using tetra-[ 3 H]phenylphosphonium bromide (Amersham Biosciences) in aerobic and Wayne dormant (10 days) M. smegmatis, as previously described (21). Briefly, cells were energized with 20 mM glucose, tetra-[ 3 H]phenylphosphonium bromide was added, and cells were centrifuged through silicon oil. Supernatant and cell pellet were dissolved in scintillation fluid and counted with a liquid scintillation analyzer (TRI-CARB 2100TR; Packard). Membrane potential (in mV) was calculated according to the Nernst relationship. Values represent averages Ϯ S.E. of four independent experiments. Comparisons were performed using Student's t test (StatXact). p values of Ͻ0.05 were considered as statistically significant.

Bactericidal Activity of R207910 on Dormant M. tuberculosis-
The effect of DARQ lead compound, R207910, on the growth of nonreplicating M. tuberculosis was tested in three different in vitro dormancy models. As a control for target specificity during dormant conditions, we tested the efficacy of R207910 on the R207910-resistant strain of M. tuberculosis (BK12), carrying a point mutation in the ␣-helix of the c-subunit of the ATP synthase complex (7). Both wild type and mutant strains were subjected to dormancy using either gradual (Wayne) or drastic (hypoxia) O 2 depletion or NO treatment in the presence of 0.1 and 10 g/ml R207910 (Fig. 1).
We treated Wayne dormant M. tuberculosis for 7 days with R207910 to determine its killing efficacy. As can be seen in Fig.  1A, R207910 at 10 g/ml leads to a 1.8-log 10 reduction in CFU counts, whereas no apparent effect was observed at 0.1 g/ml. The viability of dormant bacilli was not affected by isoniazid treatment, an inhibitor of biosynthesis of cell wall mycolic acids. In contrast, metronidazole at 100 g/ml led to a 1.9-log 10 reduction in CFUs as compared with untreated control. Metronidazole is a prodrug that is effective only under anaerobic conditions. However, it has been shown to be ineffective against latent mycobacteria in a Cornell mouse infection model (22). DCCD, a nonspecific ATP synthase inhibitor, showed a drastic killing effect at 100 g/ml (Fig. 1A).
In the hypoxia model, M. tuberculosis cultures were subjected to drastic O 2 depletion in an anaerobic chamber. R207910 reduced the mycobacterial viability by 2.1-log 10 CFUs at 10 g/ml but had no apparent effect at 0.1 g/ml (Fig. 1B). Metronidazole tends to be slightly more potent than in the Wayne model, leading to a 2.5-log 10 CFU reduction at 100 g/ml. This can be explained by the extreme hypoxic conditions created in the anaerobic chamber, and as such, the effects of metronidazole, which works best in a severely O 2 -depleted environment, tend to be enhanced.
NO has been shown to inhibit aerobic respiration in several bacterial systems, and it induces a dormancy response in mycobacteria similar to oxygen depletion (19). Treatment of midlog  (Fig. 3A). In our study, 2 h after treatment with DETA/NO, mycobacteria were exposed for 10 h to R207910 and control drugs. DETA/NO alone had no effect on mycobacterial survival as compared with untreated control (Fig. 1C). Furthermore, isoniazid (10 g/ml) had no effect on these dormant bacilli, whereas the viability of the bacilli was drastically reduced by DCCD. However, metronidazole (100 g/ml), in contrast to the O 2 depletion dormancy assays, was not bactericidal, because as a prodrug, it is active only in an O 2 -limiting and not in an NO environment. R207910 (10 g/ml) led to a 1-log 10 drop in CFU counts, whereas no apparent effect was observed at 0.1 g/ml. It has been previously shown that NO dissipates in the reaction tubes over a period of time, and by 16 h, the bacteria tend to lose their dormant phenotype (19). However, the effects of R207910 on NO-treated mycobacteria in this model ( (Table 1). Rifampicin and isoniazid, included as controls, were both more potent in aerobically grown and significantly less for the dormant bacilli. In contrast, R207910 was slightly more active on dormant bacilli (WCC 90 of 1 g/ml), as compared with aerobically grown bacilli (MBC 90 of 2 g/ml). This suggests that dormant bacteria are slightly more prone (about 2-fold) to R207910 bactericidal activity as compared with actively growing mycobacteria (WCC 90 /MBC 90 ratio of 0.5).
To further evaluate this result, we measured killing kinetics of R207910 and isoniazid on Wayne dormant and aerobically grown bacilli (Fig. 2). For this purpose, both cultures were diluted to equal A 600 values or ϳ10 7 to 10 8 CFU/ml and treated with R207910 (10 g/ml). For aerobic cultures, we observed an initial 2.2-log 10 drop in CFUs by the end of the first week of treatment with R207910 and an additional 0.8-log 10 reduction in CFUs by day 14. In contrast, treatment of dormant mycobacteria resulted in an initial 1.8-log 10 drop in CFUs by the end of the first week, but by day 14, killing was more pronounced, with an additional 3.0-log 10 drop in CFUs versus untreated control. As expected, isoniazid had no effect on dormant bacterial cultures. However, in aerobic cultures, isoniazid killed about 2.2 log 10 during the first week, with no additional killing during the second week.
R207910-mediated killing of dormant bacilli at a 10 g/ml concentration was highly potent, since no bacteria could be counted by day 14 (Fig. 2). In contrast, the killing kinetics of R207910 at 1 g/ml on Wayne dormant M. tuberculosis suggested a characteristic biphasic killing with a 2.9-log 10 drop in CFUs by day 14 and no further reduction by day 21 (Fig. S1).
In summary, by measuring the bactericidal concentrations as well as killing kinetics, we demonstrated that R207910 kills dormant M. tuberculosis at least as effectively as actively replicating bacilli. This makes R207910 the only TB drug with equal bactericidal activities on different bacterial subpopulations.
Specificity of DARQ Activity in Dormant Mycobacteria-Target specificity of DARQs was assessed by introducing chemical changes in the DARQ scaffold. During dormancy, bacteria might have alternative means of energy or ATP production (e.g. substrate level phosphorylation), and as such, ATP synthase might be redundant. Hence, it is postulated that R207910 might affect other cellular targets as well (23). At first, we studied the correlation between various structural and chemical changes in R207910 and its analogs versus their effect on biological activity. This leads to the determination of a structure-activity relationship. Over 400 DARQ derivatives were prepared and tested on actively replicating and aerobically grown M. tuberculosis cultures in in vitro MIC determination assays (data not shown). Five compounds (Fig. S2A) with MICs between 0.09 and 20 g/ml were selected for evaluating their bactericidal effect on Wayne dormant M. tuberculosis. R207910 at 1 and 10 g/ml led to 1.3-and 2.1-log 10 CFU reduction, respectively, as compared with untreated dormant control cultures ( Table 2). As can be seen in Table 2, an analogue with a methyl substituent on the first aryl moiety of R207910 (named 18040971-AAA) retained  (Wayne) and aerobically grown cultures of M. tuberculosis. The dormant and aerobic cultures, with equal start A 600 values or comparable starting CFUs (10 7 to 10 8 CFU/ml), were treated with 10 g/ml R207910 or 10 g/ml isoniazid. Killing was monitored by measuring CFU/ml after 1, 4, 7, and 14 days on 7H10 plates and plotted as log 10 CFU reduction, calculated against nontreated controls. Note that for dormant mycobacteria treated with R207910 (10 g/ml), the CFU counts were below the detection limit at 14 days (10 2 CFU/ml). its bactericidal activity on dormant bacteria with 1.55-log 10 killing at 1 g/ml. However, replacement of the dimethylamino group by azole groups (25756770-AAA) led to a decrease in the activity on both aerobic (MIC 8.19 g/ml) and dormant bacilli (0.29-log 10 CFU reduction). Interestingly, changing the chain length in R207910 between the hydroxyl and the terminal dimethylamino group (18086601-AAA) also decreased activity on both aerobic (MIC 20.21 g/ml) as well as dormant bacilli (0.11-log 10 CFU reduction). These data suggest a strong correlation between structural variations of the DARQ scaffold and the subsequent effect on their activities against replicating and dormant bacilli (Fig. S2B). The sharp overlap between the bactericidal activities of DARQ analogs on dormant and replicating bacilli implies that their effects are highly target-specific, irrespective of mycobacterial growth conditions and physiological states. In a second approach to characterize the target specificity of R207910 during dormancy, we tested the efficacy of R207910 on the resistant strain of M. tuberculosis (BK12) bearing a mutation in c-subunit of ATP synthase (7). At the highest concentration of 10 g/ml, R207910 had no effect on this mutant strain in Wayne (Fig. 1A), hypoxia (Fig. 1B), and NO (Fig. 1C) conditions. Isoniazid had no effect under these conditions, suggesting that bacilli were indeed in a nonreplicating state. The ATP synthase inhibitor DCCD had a drastic killing effect and led to a 5.3-, 2.8-, and 4.3-log 10 drop in CFUs in Wayne (Fig.  1A), hypoxia (Fig. 1B), and NO (Fig. 1C), respectively. The lack of cross-resistance between these two compounds suggests that although DCCD binds to the same target as R207910, its binding pocket is different. In summary, our results demonstrate that killing of mycobacteria by DARQs critically depends on the structure of the DARQ compound as well as critical amino acid compositions in the target protein.
ATP Synthesis in Nonreplicating Mycobacteria-The dormancy physiological phase involves changes in several metabolic pathways that lead to the transcriptional reprogramming of different bacterial genes. In a first step, we confirmed transcriptional up-regulation of the dormancy regulon in the three dormancy models, as previously reported (12)(13)(14). We explored the effect of dormancy on the ATP synthase operon under O 2 depletion and NO stress conditions by real time quantitative PCR for atpA to atpH. In M. tuberculosis, the ATP synthase consists of a transmembrane F 0 and a cytosolic F 1 subunit. F 0 is composed of Rv1304 (atpB), Rv1305 (atpE), and Rv1306 (atpF), whereas the F 1 subunit is composed of Rv1308 (atpA), Rv1311 (atpC), Rv1310 (atpD), Rv1309 (atpG), and Rv1307 (atpH). The transcriptional response of atpA to -H was significantly down-regulated in both the hypoxic and the Wayne dormancy models (Fig. 3A). However, in NO-treated samples, there was no apparent change in levels of ATP synthase mRNA transcripts. This difference might be explained by the brief NO exposure time (2 h, compared with several days for Wayne and hypoxia). However, both O 2 -depleted and NO-treated cultures showed up-regulation of five characteristic dormancy marker genes (Rv2623, Rv2626, Rv2007, Rv2031, and Rv3130), suggesting that indeed a dormant physiological state was triggered upon exposure to NO or O 2 stress (Fig. 3A) (19,20).
In order to evaluate the effect of down-regulation of ATP synthase in dormant bacteria, we measured ATP levels in M. tuberculosis in Wayne conditions over a period of 21 days (Fig.  3B). As shown upon O 2 depletion, a significant decrease in the amount of ATP was observed at day 1 with additional decline by day 14. Similarly, ATP levels in Wayne dormant M. smegmatis were significantly reduced compared with actively growing bacilli (data not shown). At the corresponding time points, these dormant cultures were also analyzed for atpE mRNA expression. As can be seen in Fig. 3B, decreased atpE mRNA transcription at each time point correlates highly with lower ATP levels.
As compared with actively replicating bacilli, there is more than 90% decrease in ATP levels in dormant M. tuberculosis (Fig. 3B, inset). Thus, total cellular ATP levels in dormant mycobacteria are significantly lower compared with replicating bacteria. We subsequently analyzed whether R207910 further decreases the cellular ATP levels in Wayne dormant M. tuberculosis. As can be seen in Fig. 3C, treatment of dormant mycobacteria with R207910 at 0.06 g/ml (1ϫ MIC) results in a significant decrease in ATP levels as compared with untreated controls. Isoniazid (0.3 g/ml) had no effect on ATP levels, whereas DCCD (100 g/ml) drastically reduced ATP levels. This suggests that R207910, even at a low concentration (1ϫ MIC), is highly effective in reducing ATP levels in dormant bacteria. This demonstrates high susceptibility of dormant mycobacteria to lower concentrations of R207910.
This increased susceptibility is observed despite a transcriptional down-regulation of the ATP synthase operon. In order to show that ATP synthase still has the residual functional activity during dormancy, we measured synthesis of ATP by the ATP synthase enzyme in membrane vesicles isolated from dormant M. smegmatis. This ATP synthase activity was completely blocked by R207910 in a dose-dependent manner with an IC 50 of around 1.5 ng/ml (Fig. 4A). These results indicate that in dormant mycobacteria, ATP synthase still shows a residual function, despite overall transcriptional down-regulation and other changes in energy metabolic pathways.
The ATP synthase enzyme complex can have a bimodal function: synthesis of ATP and generation of a membrane potential.

TABLE 2 Bactericidal action of DARQs in killing aerobic and dormant M. tuberculosis
Different DARQ analogs were tested on M. tuberculosis cultures grown in aerobic conditions or in the Wayne oxygen depletion model. MIC 90 values on aerobically grown bacilli were compared with the bactericidal activity (log 10 reduction as measured against untreated control) on dormant bacteria in 7H10 agar. The bactericidal activity was evaluated at concentrations of 1 and 10 g/ml. Each value is the mean of a minimum of two different experimental values, and each experiment was repeated at least two times. A strong correlation between structural variations of DARQs analogs and their biological effects against replicating and dormant bacilli is observed. Despite the low ATP levels and a reduced expression of ATP synthase, dormant bacteria still need to maintain a membrane potential for cellular viability. We tested the effect of R207910 on membrane potential of Wayne dormant M. smegmatis. It was seen that dormant bacilli are able to maintain an energized membrane or membrane potential (Ϫ135 mV), albeit at slightly lower levels compared with aerobically grown bacilli (Ϫ180 mV) (p Ͻ 0.001) (Fig. 4B). The presence of R207910 had no significant effect on this membrane potential both in aerobic and in dormant M. smegmatis (data not shown). This suggests that ATP synthase is not required for the generation of the membrane potential by pumping protons in the periplasm. Taken together, dormant mycobacteria have functional ATP synthase: low but critically needed ATP levels and an energized membrane. Hence, dormant bacteria are not metabolically or physiologically inert but still operate some key cellular processes.

DISCUSSION
Treatment of active TB in humans by a regimen combining four drugs (isoniazid, rifampicin, pyrazinamide, and ethambutol) leads to rapid killing of the bacteria in the sputum during the first 2 months (24). However, relapses occur if the therapy is not continued further for 4 months to kill the remaining bacilli. These bacilli are likely to be located in caseous regions with limited supply of O 2 and nutrients and exhibit a dormant phenotype. Penetration of drugs into these microenvironments tends to be restricted, and these bacteria seem to be tolerant to antimycobacterial drugs, such as isoniazid (25). Therefore, there is an urgent need for the development of new drugs that specifically target dormant bacilli.
Our study demonstrates that DARQ lead compound R207910 effectively killed dormant mycobacteria in all three in vitro dormancy model systems used that mimic the microenvironment encountered by mycobacteria within granulomas in humans. Surprisingly, we observed a higher susceptibility of dormant bacilli to DARQs as compared with aerobically grown cultures, a unique property that distinguishes R207910 from presently used first and second line anti-TB drugs, such as rifampicin (WCC 90 /MBC 90 of 17). This dual action of DARQs might explain why the combination of R207910 with the standard of care regimen (rifampin, isoniazid, and pyrazinamide) led to a 50% faster reduction of CFU counts than the control regimen (7,15). The sterilizing efficacy of R207910 in mice, involving measurement of relapse rates 3 months after cessation of treatment, demonstrated that R207910 reduces the number of relapses compared with the standard of care combination therapy (26). Moreover, treatment of Guinea pigs with R207910 resulted in an almost complete eradication of M. tuberculosis throughout both primary and secondary lesions in the lung granulomas after 6 weeks of treatment (27).
In correlation with transcriptional down-regulation of ATP synthase, the ATP levels are also significantly decreased in dormant mycobacteria, suggesting an energy adaptation for this new physiological state. It has already been suggested that mycobacterial isolates from human lesions use alternative energy pathways, since several genes responsible for the lipid metabolism are up-regulated (28). Considering the bactericidal action of R207910 on dormant M. tuberculosis, it is likely that nonreplicating mycobacteria still need minimal amounts of ATP to survive, provided by a residual activity of ATP synthase. This suggests that dormant mycobacteria can be more prone to alterations in cellular ATP pools due to inhibition of ATP synthase as compared with actively growing bacilli. Our data thus provide a nice example of a down-regulated enzyme acting as a highly efficient target, whereas generally enzymes that are upregulated during dormancy were considered to be promising targets for latent TB.  For dormant mycobacteria, a role for ATP synthase in synthesizing ATP and/or maintaining the membrane potential (by hydrolyzing ATP and pumping protons across the membrane) has been proposed (29). DARQs potently inhibited ATP synthesis activity but even at high concentrations had no significant effect on the membrane potential. This is further supported by the previous finding that the purified ATP synthase of Mycobacterium phlei appears to have only very low or latent ATP hydrolysis activity (30). It has been suggested that for restoring the redox balance and proton motive force during dormant conditions, the nitrate reductase complex, an alternative respiratory enzyme, can generate a proton motive force during dormant growth conditions (31). Furthermore, under low O 2 conditions, there are several changes in the respiratory chain complexes in mycobacteria, including up-regulation of cytochrome bd-type menaquinol oxidase that has a higher oxygen affinity and a non-proton-translocating NADH dehydrogenase (type II) that is required for oxidation of NADH within the cell (28).
In conclusion, our study shows that despite its down-regulation during dormancy, ATP synthase plays an essential role for mycobacterial survival and thus can serve as an ideal drug target for killing dormant as well as actively replicating bacteria. Unraveling the principles of energy metabolism during dormancy and understanding the mechanisms of antibiotic action can reveal critical weaknesses of dormant and persistent bacterial species. In streptococci, ATP synthase was also found to be differentially regulated in biofilm formation (32). The downregulation of F 1 F 0 -ATPase might make these bacteria more susceptible to changes in cellular ATP pools. Thus, validating ATP synthase as a target might also be useful to treat other persistent bacterial infections.