Research Article

A Diarylquinoline Drug Active on the ATP Synthase of Mycobacterium tuberculosis

See allHide authors and affiliations

Science  14 Jan 2005:
Vol. 307, Issue 5707, pp. 223-227
DOI: 10.1126/science.1106753


The incidence of tuberculosis has been increasing substantially on a worldwide basis over the past decade, but no tuberculosis-specific drugs have been discovered in 40 years. We identified a diarylquinoline, R207910, that potently inhibits both drug-sensitive and drug-resistant Mycobacterium tuberculosis in vitro (minimum inhibitory concentration 0.06 μg/ml). In mice, R207910 exceeded the bactericidal activities of isoniazid and rifampin by at least 1 log unit. Substitution of drugs included in the World Health Organization's first-line tuberculosis treatment regimen (rifampin, isoniazid, and pyrazinamide) with R207910 accelerated bactericidal activity, leading to complete culture conversion after 2 months of treatment in some combinations. A single dose of R207910 inhibited mycobacterial growth for 1 week. Plasma levels associated with efficacy in mice were well tolerated in healthy human volunteers. Mutants selected in vitro suggest that the drug targets the proton pump of adenosine triphosphate (ATP) synthase.

After AIDS, tuberculosis (TB) is the leading cause of infectious disease mortality in the world, with 2 million to 3 million deaths per year (1). The TB and HIV epidemics fuel one another in coinfected people, and at least 11 million adults are infected with both pathogens (2, 3). Hence, factors contributing to the TB burden include not only difficulties in implementing TB control programs in many countries but also the recent increase in the number of HIV-infected individuals (4). Although current first-line anti-TB drug regimens can achieve more than 99% efficacy, this is often reduced because of drug resistance (5). As pointed out by the Global Alliance for TB Drug Development, new drugs that could shorten or simplify effective treatment of TB would substantially improve TB control programs (6).

We report on the antimycobacterial properties of the diarylquinolines (DARQs). The lead compound, R207910, not only has several properties, both in vitro and in vivo, that may improve the treatment of TB, but also appears to act at a new target, providing an antimycobacterial spectrum different from those of current drugs. Its clinical potential is being tested in patients.

Chemistry and in vitro antimycobacterial activity. We sought new anti-TB compounds by selecting prototypes of different chemical series and testing them for inhibition of multiple-cycle growth of Mycobacterium smegmatis. A whole-cell assay was preferred because of its ability to concurrently assess multiple targets. Chemical optimization of a lead compound led to a series of DARQs with potent in vitro activity against several mycobacteria, including M. tuberculosis (7). To date, 20 molecules of the DARQ series have a minimum inhibitory concentration (MIC) below 0.5 μg/ml against M. tuberculosis H37Rv. Antimycobacterial activity was confirmed in vivo for three of these compounds (8). The most active compound of the class, R207910, is a pure enantiomer with two chiral centers. A mixture of diastereoisomers is prepared in five steps, and R207910 is isolated from the resulting mixture of four isomers; Fig. 1 shows its structure and absolute configuration. Its chemical name is 1-(6-bromo-2-methoxy-quinolin-3-yl)-4-dimethylamino-2-naphthalen-1-yl-1-phenyl-butan-2-ol; the molecular formula is C32H31BrN2O2 and the molecular weight is 555.51 daltons.

Fig. 1.

Absolute configuration of R207910. From the racemate, two diastereoisomers were purified and separated with a ratio A:B = 40:60 by high-performance liquid chromatography (HPLC). The active diastereoisomer (A) was separated into the corresponding (R,S) (A1) and (S,R) (A2) isomers by chiral HPLC. The active (R,S) isomer is R207910 (A1). The structure of R207910 in the solid state was solved by circular dichroism and x-ray crystallography experiments. The absolute configuration of two asymmetric centers was determined as R,S (the carbon bearing the phenyl substituent is R and the carbon bearing the hydroxyl substituent is S).

Structurally and mechanistically, DARQs are different from both fluoroquinolones (including methoxyquinolones) and other quinoline classes, including mefloquine and its analogs 4-methylquinolines and 4-quinolylhydrazones (913). One of the major structural differences between DARQs and other quinolone or quinoline classes is the specificity of the functionalized lateral (3′) chain borne by the DARQ class.

R207910 has a unique spectrum of potent and selective antimycobacterial activity in vitro (Table 1). The range of MICs for the international reference strain M. tuberculosis H37Rv and six fully antibiotic susceptible isolates was 0.030 to 0.120 μg/ml (Table 1), versus 0.500 μg/ml for rifampin and 0.120 μg/ml for isoniazid (14). R207910 demonstrated similar in vitro efficacy against M. tuberculosis clinical isolates resistant to the TB agents isoniazid, rifampin, streptomycin, ethambutol, pyrazinamide, and moxifloxacin. R207910 did not inhibit M. tuberculosis purified DNA gyrase, the target for quinolones.

Table 1.

MICs of R207910 that inhibited 99% of the growth of different mycobacterial species.

Mycobacterial species Number of strains Range of MICs for multiple strains (μg/ml) Median MIC (μg/ml)
M. tuberculosis, H37Rv 1 0.030
M. tuberculosis, fully susceptible clinical isolates 6 0.030-0.120 0.060
M. tuberculosis resistant to isoniazid 7 0.003-0.060 0.010
M. tuberculosis resistant to rifampin 1 0.030
M. tuberculosis resistant to isoniazid and rifampin 2 0.030-0.030 0.030
M. tuberculosis resistant to isoniazid and streptomycin 1 0.010
M. tuberculosis resistant to ethambutol 1 0.010
M. tuberculosis resistant to pyrazinamide 1 0.030
M. tuberculosis resistant to fluoroquinolone 2 0.060-0.120 0.090
M. bovis 1 0.003
M. avium/M. intracellulare (MAC) 7 0.007-0.010 0.010
M. kansasii 1 0.003
M. marinum 1 0.003
M. fortuitum 5 0.007-0.010 0.010
M. abscessus 1 0.250
M. smegmatis 7 0.003-0.010 0.007
M. ulcerans 1 0.500

The lack of cross-resistance with currently used anti-TB agents suggests that R207910 retains activity against multidrug-resistant (MDR) TB strains. Indeed, using the BACTEC culture system (15), we observed inhibition of bacterial growth when MDR-TB strains were exposed to fixed concentrations of R207910. All 30 isolates of MDR-TB tested were susceptible to R207910 at 0.100 μg/ml; of these, 17 (57%) were susceptible to R207910 at 0.010 μg/ml (Table 2). Using the same method, we observed similar susceptibility among 10 fully antibiotic-susceptible strains. Low MICs were also found for other mycobacterial species, including M. bovis, M. kansasii, and M. ulcerans, as well as species naturally resistant to many other anti-TB agents and involved in opportunistic infections, such as M. avium complex (MAC), M. abscessus, M. fortuitum, and M. marinum (Table 1).

Table 2.

Susceptibility of drug-resistant M. tuberculosis to two concentrations of R207910, as measured by the BACTEC radiometric system.

Resistance pattern Total number of strains Number of strains inhibited by 0.100 μg/ml Number of strains inhibited by 0.010 μg/ml
M. tuberculosis, fully susceptible isolates 10 10 1 (10%)
M. tuberculosis, resistant isolates 40 40 22 (55%)
Multidrug-resistant M. tuberculosis (resistant to at least rifampin and isoniazid) 30 30 17 (57%)
M. tuberculosis resistant to isoniazid, rifampin, streptomycin, and ethambutol 13 13 8 (62%)
M. tuberculosis resistant to isoniazid 38 38 20 (53%)
M. tuberculosis resistant to rifampin 30 30 17 (57%)
M. tuberculosis resistant to streptomycin 25 25 15 (60%)
M. tuberculosis resistant to ethambutol 20 20 12 (60%)

The activity of R207910 appeared to be specific for mycobacteria. R207910 had much higher MICs for Corynebacterium and Helicobacter pylori (MIC 4.0 μg/ml), and especially for other organisms such as Gram-positive Nocardia, Streptococcus pneumoniae, Staphylococcus aureus, and Enterococcus faecalis, or Gram-negative Escherichia coli and Haemophilus influenzae (table S1).

Exposure of M. tuberculosis in log-phase growth to concentrations of R207910 at 10 times MIC resulted in a reduction in bacterial load of 3 log units after 12 days, indicating that R207910 has bactericidal activity in vitro (table S2). The killing effect was not increased by higher concentrations of the compound, which suggests that the killing was time-dependent rather than concentration-dependent.

Isolation of mutants, cross-resistance, and postulated drug target. Mutants of M. tuberculosis and M. smegmatis resistant to R207910 were selected in vitro to quantify the proportion of resistant mutants arising (by comparison with rifampin), to assess the cross-resistance pattern, and to investigate the mechanism of action.

From selection experiments, the proportion of resistant mutants that emerged was 5 × 10–7 and 2 × 10–8 at 4 times MIC, and 5 × 10–8 and 1 × 10–8 at 8 times MIC, for M. tuberculosis and M. smegmatis, respectively (8). In the case of M. tuberculosis, these proportions were comparable to those of rifampin-resistant mutants that emerged (10–7 to 10–8). In addition, the susceptibility of R207910-resistant M. tuberculosis strains remained unchanged relative to other anti-TB agents, including isoniazid, rifampin, streptomycin, amikacin, ethambutol, and moxifloxacin. Analysis of R207910 mutants of M. tuberculosis and M. smegmatis showed that there were no mutations in the DNA gyrase regions gyrA and gyrB, sequences in which quinolone resistance typically develops.

The genomes of the resistant M. tuberculosis strain BK12 and the two resistant M. smegmatis strains R09 and R10, as well as the parental M. smegmatis, were sequenced to near-completion. Point mutations that conferred R207910 resistance were identified by comparative analysis of the genome sequences of susceptible and resistant strains of M. tuberculosis and M. smegmatis (Fig. 2). The only gene commonly affected in all three independent mutants encodes atpE, a part of the F0 subunit of ATP synthase. This finding indicates that R207910 inhibits the proton pump of M. tuberculosis ATP synthase. The two point mutations identified—Asp32 → Val (D32V) for M. smegmatis and Ala63 → Pro (A63P) for M. tuberculosis—are both in the membrane-spanning domain region of the protein.

Fig. 2.

atpE protein sequence alignments for M. tuberculosis and M. smegmatis mutants. (A) Mtb_S: drug-sensitive strain of M. tuberculosis H37Rv, atpE (residues 1 to 81; Swiss-Prot accession number Q10598). Mtb_R: drug-resistant strain of M. tuberculosis BK12, atpE (1 to 81; EMBL accession number AJ865377). Homo sapiens ATP5G3 (66 to 142; accession number Ensembl ENSP00000284727). (B) Msm_S: drug-sensitive strain of M. smegmatis, atpE (1 to 86; EMBL accession number AJ862722). Msm_R09 and Msm_R10: drug-resistant strains of M. smegmatis atpE (1 to 86; EMBL accession numbers AJ865378 and AJ865379, respectively). Homo sapiens, ATP5G3 (66 to 142; accession number Ensembl ENSP00000284727). Shading indicates amino acid similarity using BLOSUM62 matrix (19) (black, high; purple, medium). Yellow highlight indicates the proton-binding glutamic acid; red highlight indicates the amino acid of the sensitive strain (arrow); blue highlight indicates the amino acid of the resistant strain (arrow). Single-letter abbreviations: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; Y, Tyr.

Complementation studies have verified that the mutant atpE gene is responsible for resistance to R207910, implying that the atpE gene product is the target of R207910 in mycobacteria. Wild-type M. smegmatis was transformed with a construct expressing the F0 subunit from mutant M. smegmatis R09 (D32V). This rendered the cells resistant to R207910, with a MIC identical to that of the R207910-resistant strain M. smegmatis R09 (D32V) (table S3). In addition, when the plasmid was reisolated from these transformants and the ATP F0 operon was sequenced, it was shown to have retained the mutant allele (D32V).

Pharmacokinetic studies in mice. After single or multiple oral administration in male Swiss SPF mice, R207910 was rapidly absorbed. After a single dose of 6.25 mg/kg body weight, maximum plasma concentrations (Cmax) reached 0.40 to 0.54 μg/ml within 1 hour; after a dose of 25 mg/kg, Cmax reached 1.1 to 1.3 μg/ml within 2 to 4 hours. Areas under the curve for concentration versus time (AUCs) were 5.0 to 5.9 μg·hour/ml after a dose of 6.25 mg/kg; AUCs were 18.5 to 19.4 μg·hour/ml after a dose of 25 mg/kg (table S4). R207910 was extensively distributed to tissues, including lungs and spleen (Fig. 3A) (table S4). Half-lives ranged from 43.7 to 64 hours in plasma and 28.1 to 92 hours in tissues (table S4). No accumulation of R207910 was observed after five daily oral doses, indicating that slow redistribution from tissues contributed to the relatively long half-lives in plasma. The long half-lives and consequent prolonged exposure are important factors determining the duration of activity in vivo and provide a rationale for less frequent dosing regimens.

Fig. 3.

(A) Mean plasma and tissue concentrations of R207910 (J) in mice (n = 2) after a single oral dose of J at 25 mg/kg. Black diamonds, plasma; red triangles, liver; blue crosses, kidney; green squares, heart; yellow crosses, spleen; purple circles, lung. (B) Effect of a single oral dose of J in the nonestablished infection murine TB model. Red diamonds, controls (untreated mice); blue squares, rifapentine (10 mg/kg); yellow triangles, J (50 mg/kg); black squares, J (100 mg/kg). Values are means ± SD.

To better understand the dose-response and exposure-response relationships between R207910 and TB infection, we combined data from in vivo efficacy and separate pharmacokinetic studies in mice. A single-dose pharmacokinetic study comparing Cmax values and AUCs after doses of 6.25, 25, and 100 mg/kg confirmed that AUCs showed a better correlation with dose than did Cmax values, reflecting limitations in the rate of absorption at higher dosing regimens. Dose linearity was better for tissues, in terms of both Cmax and AUC. The data suggest that maintaining average plasma levels of 0.3 μg/ml throughout a dosing interval of 24 hours—which is achieved with a dose of 100 mg/kg per week—is necessary to achieve the optimal effect of monotherapy in mice infected with strain H37Rv.

Pharmacodynamic studies in mice. A single dose of R207910 (50 mg/kg) in the nonestablished infection murine model of TB infection resulted in a bacteriostatic effect [decrease of 0.5 log units in bacterial load in the lungs, in terms of colony-forming units (CFU)]; this effect lasted for 8 days and was similar to the effect of rifapentine (10 mg/kg) (Fig. 3B). By contrast, at 100 mg/kg, a single dose of R207910 had a bactericidal effect (decrease of up to 2.5 log units in bacterial load in the lungs) that lasted for 8 days, after which bacterial growth resumed at a rate similar to that seen in the controls. The extended effect of a single dose provides further support for a dosing regimen less frequent than five times per week.

In mouse studies using oral treatment of R207910 for 5 days per week for 4 weeks, starting the day after inoculation, the minimal dose of R207910 that prevented mortality in nonestablished infection was 1.5 mg/kg, and the minimal effective dose (MED) preventing gross lung lesions was 6.5 mg/kg. In mice receiving doses of 12.5 mg/kg, the bacterial load per organ was reduced from 5 to 2 log units (P < 0.0014) (Fig. 4A). Thus, the minimal bactericidal dose was very close to the MED; this finding confirms the time-dependent rather than concentration-dependent activity of R207910 in vitro. At 25 mg/kg, the activity of R207910 was significantly better than at 12.5 mg/kg (P < 0.0014).

Fig. 4.

(A) Minimal effective oral dose of R207910 (J) and effect of dosing frequency in the nonestablished infection murine TB model. Data shown were obtained from mice killed on day 28. Treatment began on day 1 and was administered 5 times per week (5 ×/W) for 4 weeks, except for one group treated with J (12.5 mg/kg) once per week (1 /W; hatched bars). Blue bar, control lung; yellow bar, control spleen (both ×at day 1); red bars, treatment lung; black bars, treatment spleen; H, isoniazid (25 mg/kg). Values are means ± SD. Multiple comparisons among pairs of groups were performed by Bonferroni's method (20). There were significant differences (P < 0.0014) between the control and J 12.5, J 25, or J 50 mg/kg (5 /W); between H and J 12.5, J 25, or J 50 mg/kg (5 /W); between J 12.5 × mg/kg and J 25 or J 50 mg/kg (5 /W); 12.5 mg/kg (5 ×/W). (B) Efficacy of R207910 (J) in the established infection murine TB model. Red bar, mice killed on day 12 just before treatment started; yellow bars, bacterial load in lungs of mice killed on day 42 (after 1 month of therapy); blue bars, bacterial load in lungs of mice killed on day 70 (after 2 months of therapy). Drugs were administered 5 times per week: R, rifampin (10 mg/kg); J, R207910 (25 mg/kg); H, isoniazid (25 mg/kg); Z, pyrazinamide (150 mg/kg). Values are means ± SD. Multiple comparisons among pairs of groups were performed by Bonferroni's method (20). There were significant differences (P < 0.0018) between the control and all treatment groups after 1 and 2 months, between the RHZ combination and any combination containing J after 1 month, and between the RHZ combination and the RJZ and RHZJ combination after 2 months.

At 12.5 and 25 mg/kg, R207910 was significantly more active (P < 0.0014) than isoniazid (25 mg/kg), a drug known for its strong, early bactericidal activity (Fig. 4A). Moreover, at 12.5 mg/kg, a once-weekly dose of R207910 was almost as efficacious as a dose of 6.5 mg/kg given five times per week (Fig. 4A). This is likely a consequence of the long half-life of R207910.

In established infection (treatment beginning 12 to 14 days after inoculation when bacterial load was 5.94 log units), R207910 (25 mg/kg) as monotherapy was at least as active as the triple combination therapy RHZ [rifampin (R) + isoniazid (H) pyrazinamide (Z)] and more active than + rifampin alone (Fig. 4B). However, concern about resistance development would preclude clinical use of R207910 as monotherapy. When added to the first-line triple TB therapy combination RHZ, R207910 (25 mg/kg) yielded a greater decrease in bacterial load in the lungs (relative to the standard RHZ treatment regimen) by 2 log units after 1 month of therapy and by a further 1 log unit after 2 months of therapy (P < 0.0018 in both cases). When substituting each first-line drug of the RHZ combination with R207910 (25 mg/kg), the activity of each combination containing R207910 (J) was significantly improved relative to that of RHZ, particularly after 1 month of treatment (P < 0.0018). In addition, after 2 months of treatment with JHZ and RJZ, the lungs of all animals were culture-negative. The differences among the bactericidal activities of the combinations RHJ, JHZ, and RJZ were not significant. The bactericidal activity obtained by the RHZ combination after 2 months of therapy was matched by the combinations JHZ and RJZ after just 1 month of therapy.

Pharmacokinetic studies in humans. Preclinical safety assessment (including 28-day toxicology in rats and dogs, genetic toxicology, and safety pharmacology) supported the administration of R207910 to humans. In the first clinical study, we explored the pharmacokinetics, safety, and tolerability of R207910 in healthy male subjects in a double-blind, randomized, placebo-controlled design (8). The mean plasma concentration–time profiles after single-dose oral administration of R207910 (escalating doses, 10 to 700 mg R207910) are shown in Fig. 5.

Fig. 5.

Mean profiles of plasma concentration versus time for R207910 (J) after single oral administration in healthy male subjects immediately after a meal. Dose of J: black diamonds, 10 mg; red circles, 30 mg; orange triangles, 100 mg; blue circles, 300 mg; yellow triangles, 450 mg; purple squares, 700 mg (n = 6 per dose studied). Values are means ± SD.

Pharmacokinetic results from the single-ascending-dose study suggest that R207910 was well absorbed after a single oral administration and that the peak concentration was reached at 5 hours (median value) after the dose. After Cmax was reached, the drug concentration declined triexponentially with time. The pharmacokinetics of R207910 were linear up to the highest dose tested (700 mg); both Cmax and AUC increased proportionally with the administered dose. There was no dose-dependent change in the terminal half-life.

Data from a multiple-ascending-dose study (once-daily doses of R207910 at 50, 150, and 400 mg/day) have shown an increase by a factor of ∼2 in the AUC from the dose time to 24 hours later (AUC0–24h) between day 1 and day 14. There was no substantial variation between subjects. This suggests an “effective half-life” of 24 hours. The mean AUC0–24h values were 7.91, 24, and 52 μg·hour/ml at steady state (corresponding to average concentrations of 0.33, 1.0, and 2.2 μg/mL across a dosing interval) with 50, 150, and 400 mg/day, respectively. These average concentrations were greater than the concentration that achieved optimal activity in established infection in mice.

Safety and tolerability in humans. In the first single-dose clinical study, the only adverse events were mild or moderate in severity and were experienced by subjects receiving R207910 and placebo (8). The majority of adverse events reported (8) were considered possibly related to the study medication: placebo (seven of 18 subjects) or R207910 (20 of 36 subjects). In the second study, the good tolerability was maintained; only one subject withdrew from the study, because of a urinary tract infection (unrelated to R207910) after seven doses at 150 mg/day. There were no consistent trends in the vital signs, electrocardiograms, or laboratory safety tests in any of the cohorts.

Discussion. The DARQ R207910 is a member of a new chemical class of antimycobacterial agents and has a MIC equal to or lower than that of reference compounds. Its spectrum is unique in its specificity to mycobacteria, including atypical species important in humans such as MAC, M. kansasii, and the fast growers M. fortuitum and M. abscessus. This antimycobacterial-specific spectrum differs from that of isoniazid, which has very poor activity against MAC. The clinical use of R207910 will be highly targeted to the treatment of TB and mycobacterial infections.

The target and mechanism of action of R207910 are different from those of other anti-TB agents. Inhibition of ATP synthase function may lead to ATP depletion and imbalance in pH homeostasis, both contributing to decreased survival (16, 17). A comparison of the sequences of ATP synthases of different bacteria and of eukaryotic ATP synthase provides a rationale for the specificity of the antibacterial spectrum and, to a lesser extent, the safety profile of R207910. Furthermore, the distinct target of R207910 means that there is no cross-resistance with existing anti-TB drugs, and our studies verified that R207910 is as effective against MDR strains as it is against fully antibiotic-susceptible strains.

R207910 has potent early bactericidal activity in the nonestablished infection murine TB model, matching or exceeding that of isoniazid (Fig. 4A), which is consistent with in vitro observations pointing to a bactericidal activity. Interestingly, the bactericidal potency seen upon administration of a single dose of R207910 (100 mg/kg) in vivo was similar to the effect seen upon continuous exposure of cultures to 0.6 μg/ml (or 10 times the median MIC of the H37Rv strain) in vitro, giving a first indication of the pharmacodynamics of R207910.

R207910 has potent late bactericidal properties in the established infection murine TB model. When given as monotherapy, the bactericidal effect of R207910 exceeds that of the reference compound rifampin, especially during the second month of therapy (Fig. 4B). Substitution of each of the three first-line drugs with R207910 resulted in a significant (P < 0.0018) increase in potency, leading to complete culture conversion of the lungs in some animals after only 2 months of treatment. Mouse studies to assess culture-positive relapse after follow-up without therapy are needed to clarify the sterilizing properties of R207910 in combination with other first-line agents and its potential to shorten duration of therapy. The extent of bacterial load, the most effective drug combinations, and the treatment duration needed for sterilization are all similar in mice and humans (18).

The extended effect of a single dose of R207910 seems to derive from a combination of the observed long plasma half-life, high tissue penetration (especially in the target organs for TB), and long tissue half-life. These are all attributes that are valuable for treatment of chronic infections, and they may also be important for the development of simpler dosing regimens.

Human studies have shown good tolerability, at least during a limited exposure period, with plasma levels ∼8 times those associated with potent in vivo activity in mice. Human pharmacokinetics of R207910 seem to reflect the good oral absorption and sustained plasma levels seen in mice. The combination of low MIC values, a distinct mechanism of action, early and late bactericidal activity, and pharmacokinetic profile makes R207910 a promising TB drug candidate. Conducting clinical development in patients with active pulmonary TB is highly warranted.

Supporting Online Material

Materials and Methods

Supporting Online Text

Tables S1 to S8

Figs. S1 and S2

References and Notes

Stay Connected to Science

Navigate This Article