Correspondence: Dirk Bald, Department of Molecular Cell Biology, Faculty of Earth and Life Sciences, VU University Amsterdam, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands. Tel.: +31 205 986 991; fax: +31 205 987 136; e-mail: email@example.com
ATP synthase is a validated drug target for the treatment of tuberculosis, and ATP synthase inhibitors are promising candidate drugs for the treatment of infections caused by other slow-growing mycobacteria, such as Mycobacterium leprae and Mycobacterium ulcerans. ATP synthase is an essential enzyme in the energy metabolism of Mycobacterium tuberculosis; however, no biochemical data are available to characterize the role of ATP synthase in slow-growing mycobacterial strains. Here, we show that inverted membrane vesicles from the slow-growing model strain Mycobacterium bovis BCG are active in ATP synthesis, but ATP synthase displays no detectable ATP hydrolysis activity and does not set up a proton-motive force (PMF) using ATP as a substrate. Treatment with methanol as well as PMF activation unmasked the ATP hydrolysis activity, indicating that the intrinsic subunit ɛ and inhibitory ADP are responsible for the suppression of hydrolytic activity. These results suggest that the enzyme is needed for the synthesis of ATP, not for the maintenance of the PMF. For the development of new antimycobacterial drugs acting on ATP synthase, screening for ATP synthesis inhibitors, but not for ATP hydrolysis blockers, can be regarded as a promising strategy.
Infections by Mycobacterium tuberculosis account for nearly 2 million deaths per year and are the predominant cause of death in HIV patients (Check, 2007). Although first line antibiotics are available for the treatment of tuberculosis, multi-drug-resistant strains of M. tuberculosis have emerged and pose a global health challenge (Mandavilli, 2007; Dye, 2009). Development of novel antibacterial compounds as well as the discovery and validation of new target proteins are of key importance to improve current tuberculosis treatment (Sassetti & Rubin, 2007; Bald & Koul, 2010).
In recent years, mycobacterial ATP synthase has been identified as the target of diarylquinolines, a new class of potent antimycobacterial drugs (Andries et al., 2005; Koul et al., 2007). Chemical inhibition of ATP synthesis by diarylquinolines strongly decreased cellular ATP levels, leading to bacterial killing (Koul et al., 2007, 2008; Rao et al., 2008). Diarylquinolines lead compound TMC207 displays pronounced target selectivity, with only an extremely low effect on ATP synthesis in the human mitochondria (Haagsma et al., 2009). In phase IIb clinical tests, TMC207 strongly decreased the count of CFUs in the sputum of patients with multi-drug-resistant tuberculosis, validating ATP synthase as a target for the treatment of tuberculosis (Diacon et al., 2009). Diarylquinolines have also been shown to effectively kill other mycobacterial strains, such as Mycobacterium leprae and Mycobacterium ulcerans (Ji et al., 2006a, b).
ATP synthase is a multisubunit complex consisting of a membrane-embedded F0 part (subunits ab2c10−15) and a cytosolic F1 moiety (α3β3γδɛ). The enzyme can utilize the proton-motive force (PMF) across the bacterial cytoplasmatic membrane for the synthesis of ATP (for a review, see Boyer, 2002). At low PMF, for example in environments with limited oxygen concentrations, this reaction can be reversed in several bacteria, which use the energy released from hydrolysis of ATP to maintain a PMF (von Ballmoos et al., 2009). However, ATP synthases from several other bacteria display only very limited ATP hydrolysis activity, for example in Paracoccus denitrificans (Harris et al., 1977), Bacillus subtilis (Hicks et al., 1994), Thermus thermophilus (Nakano et al., 2008) and Mycobacterium phlei (Higashi et al., 1975).
ATP synthase has been proven to be essential for optimal growth in M. tuberculosis (Sassetti et al., 2003) and for growth on fermentable and nonfermentable carbon sources in Mycobacterium smegmatis (Tran & Cook, 2005). However, it is not known whether the observed essentiality stems from a need for ATP synthase to produce ATP or to maintain the PMF. A number of known inhibitors of ATP synthase, for example sodium azide and aurovertin, strongly discriminate between the enzyme in ATP synthesis mode or in the ATP hydrolysis mode (Syroeshkin et al., 1995; Bald et al., 1998; Johnson et al., 2009). In order to understand diarylquinoline action and selectivity as well as for the design of improved compound derivates, an insight into the mode of action of mycobacterial ATP synthase is required.
Previous results showed only very low, ‘latent’, ATP hydrolysis activity for ATP synthase from M. phlei (Higashi et al., 1975). However, this strain is a fast-growing, saprophytic bacterium (generation time <3 h), whereas the most relevant pathogenic mycobacteria, such as M. tuberculosis, M. leprae and M. ulcerans as well as the vaccine strain M. bovis Bacillus Calmette-Guérin (BCG) are slow growers with a generation time >15 h and with significantly different energy requirements (Beste et al., 2009; Cook et al., 2009). No data on ATP synthase functioning are reported for slow-growing mycobacteria, in part due to their extremely thick cell envelope (Hoffmann et al., 2008), which makes the preparation and handling of membrane vesicles difficult.
In this study, we investigated ATP synthase in membrane vesicles of a slow-growing Mycobacterium, M. bovis BCG, as well as in a fast-growing model strain, M. smegmatis.
Materials and methods
Bacterial strains and growth conditions
Mycobacterium bovis BCG Copenhagen and M. smegmatis mc2155 were kindly provided by B.J. Appelmelk, Department of Molecular Cell Biology & Immunology, VU University Medical Center Amsterdam, the Netherlands. Replicating cultures of M. bovis BCG and M. smegmatis were grown in Middlebrook 7H9 broth (Difco) with 10% Middlebrook albumin dextrose catalase enrichment (BBL) and 0.05% Tween-80 at 37 °C to the late exponential phase.
For growth under low-oxygen conditions, M. bovis BCG was cultured in a gradual oxygen-depletion model (Wayne & Hayes, 1996) using Middlebrook 7H9 broth (Difco) with 10% Middlebrook oleic BBL and 0.05% Tween-80 at 37 °C. Cells were harvested after 7 days in nonreplicating persistence-1 phase (Wayne & Hayes, 1996)
Preparation of inverted membrane vesicles (IMVs)
Cells of M. bovis BCG were pelleted by centrifugation at 6000 g for 20 min and washed once with phosphate-buffered saline (PBS, pH 7.4). Five grams of cells (wet weight) were resuspended in 10 mL of 50 mM MOPS-KOH (pH 7.5), 2 mM MgCl2 including protease inhibitors (complete, EDTA free; protease inhibitor cocktail tablets from Roche). Lysozyme (10 mg mL−1), 1500 U of deoxyribonuclease I (Invitrogen) and 15 mM MgCl2 were added and cells were incubated with stirring at 37 °C for 1 h. Separation of this cell envelope digestion procedure into a lysozyme preincubation step (1 mM MgCl2) and a subsequent DNase I digestion step (17 mM MgCl2) did not improve the results. The cells were broken by four passages through a precooled French pressure cell at 20 000 psi (Thermo Electron, 40 K). The lysate was centrifuged at 6000 g and 4 °C for 20 min to remove unbroken cells. Two additional centrifugation steps at 6000 g and 4 °C for 20 min were carried out to remove additional cell wall components. The supernatant was centrifuged at 370 000 g and 4 °C for 1 h and the pellet of IMVs was washed with 50 mM MOPS-KOH (pH 7.5), 2 mM MgCl2. After the second centrifugation step, the inverted membrane fraction was resuspended in an appropriate volume of 50 mM MOPS-KOH (pH 7.5), 2 mM MgCl2.
ATP-driven proton translocation into IMVs of M. bovis BCG and M. smegmatis was measured by a decrease of 9-amino-6-chloro-2-methoxyacridine (ACMA) fluorescence using a Cary Eclipse Fluorescence spectrophotometer (Varian Inc., Palo Alto). IMVs (0.18 mg mL−1) were preincubated at 37 °C in 10 mM HEPES-KOH (pH 7.5), 100 mM KCl, 5 mM MgCl2 containing 2 μM ACMA and a baseline was monitored for 5 min. The reaction was then started by adding 2 mM ATP, 5 mM succinate or 5 mM NADH. After 20 min, any proton gradient was collapsed by the addition of 1 μM SF6847. The excitation and emission wavelengths were 410 and 480 nm, respectively. Other fluorophores reported for PMF detection in bacteria, such as 9-aminoacridine (9AA) (Yoshimura & Brodie, 1981) or Oxonol X (Bashford et al., 1979), did not yield interpretable signals with either succinate or NADH as a substrate (data not shown).
Assay of ATP synthesis
ATP synthesis was measured as described by Haagsma et al. (2009). Briefly, IMVs (0.5 mg mL−1) from M. bovis BCG or M. smegmatis were incubated in 10 mM HEPES-KOH (pH 7.5), 100 mM KCl, 5 mM MgCl2, 2 mM ADP, 20 mM KH2PO4, 100 μM P1,P5-di(adenosine-5′) pentaphosphate (Ap5A), 25.4 mM glucose, 11.8 U mL−1 hexokinase (Sigma) and protease inhibitors (complete, EDTA-free; protease inhibitor cocktail tablets from Roche). Samples (0.25 mL) were incubated at 37 °C with vigorous stirring in 18-mL flasks. For the DCCD control, samples were preincubated with 100 μM DCCD at room temperature for 30 min. The reaction was initiated with 5 mM succinate. After 2 h, each reaction was stopped with 25 mM EDTA, followed by transfer to ice. Samples were transferred to Eppendorf tubes, boiled for 5 min and centrifuged (10 000 ×g, 20 min) to remove denatured protein. In the supernatants, the synthesized glucose-6-phosphate was oxidized by 2.5 mM NADP in the presence of 3 U mL−1 of glucose-6-phosphate dehydrogenase (Roche). NADPH formation was monitored using a spectrophotometer at 340 nm.
Assay of ATP hydrolysis
ATP hydrolysis activity was measured by quantifying the amount of phosphate released (Bell & Doisy, 1920). IMVs (0.5 mg mL−1) from M. bovis BCG or M. smegmatis were incubated in 10 mM HEPES-KOH (pH 7.5), 100 mM KCl, 5 mM MgCl2 at 37 °C. For the DCCD control, samples were preincubated with 100 μM DCCD at room temperature for 30 min. The reaction was initiated by 2 mM ATP. After 30 min, the reaction was quenched by the addition of 2.4% (w/v) trichloroacetic acid and the membranes were pelleted by centrifugation at 20 800 g and 4 °C for 15 min.
Activation of ATP hydrolysis activity
Activation by methanol: IMVs (0.5 mg mL−1) were incubated with 17% or 25% methanol. ATP hydrolysis was assayed as described earlier.
Activation by PMF: IMVs (2.5 mg mL−1) were incubated in the presence of 10 mM succinate to establish a PMF at 37 °C for 10 min. A mixture of malonate and ATP (final concentrations are, respectively, 50 and 2 mM) was added and the incubation was quenched by the addition of 2.4% (w/v) trichloroacetic acid after 2.5 min. ATP hydrolysis was assayed as described earlier.
Activation by trypsin: IMVs (0.5 mg mL−1) were incubated with trypsin at 30 °C for 10 min. Mycobacterium bovis BCG was treated with 90 or 750 U mL−1 of trypsin, while M. smegmatis was treated with 90 U mL−1 of trypsin. The reaction was terminated by the addition of trypsin inhibitor (1.5 mg of inhibitor per 1.0 mg of trypsin). ATP hydrolysis assay was performed as described earlier.
Activation by sulfite: IMVs (0.5 mg mL−1) were incubated with 10 mM sodium sulfite. ATP hydrolysis was assayed as described earlier.
Results and discussion
Functional IMVs from M. bovis BCG
To investigate the role of mycobacterial ATP synthase, we prepared functionally coupled IMVs from the slow-growing M. bovis BCG. This strain shares >99.9% DNA sequence identity with M. tuberculosis and strongly resembles M. tuberculosis in terms of sensitivity to diarylquinolines (Mattow et al., 2001; Huitric et al., 2007). For comparison, we carried out the same set of experiments with the fast-growing saprophyte M. smegmatis.
To cope with the extremely thick cell envelope of M. bovis BCG, we optimized the preparation of IMVs in terms of the time and temperature of cell envelope digestion by lysozyme, number of French Press passages and subsequent centrifugation steps (cf. Materials and methods). To confirm the functionality of the IMVs produced, we tested their ability to set up a PMF with succinate or NADH as an external electron donor using ACMA as a fluorescence indicator. Both succinate and NADH caused fluorescence quenching, which was eased after the addition of an uncoupler, proving that the observed quenching was indeed caused by the PMF (Fig. 1a). Quenching reached a maximum after ∼10 min (Fig. 1a), significantly slower than IMVs from Escherichia coli under identical conditions (data not shown). This slow quenching may be caused by a larger percentage of leaky IMVs. The lower PMF observed with NADH (11% quenching) compared with succinate (39%) might be due to the partial detachment of the membrane-associated type-II NADH-dehydrogenase (NDH-II), the main NADH-oxidizing enzyme of the respiratory electron transport chain in M. bovis BCG (Boshoff et al., 2004; Weinstein et al., 2005). From these results, it can be concluded that the IMVs are functional. Similarly, IMVs from the fast-growing M. smegmatis accepted both NADH and succinate as electron donors (Fig. 1b).
ATP hydrolysis is blocked in M. bovis BCG and M. smegmatis
We then investigated whether the IMVs can establish a PMF with ATP as a substrate. No significant quenching was detected either for M. bovis BCG or for M. smegmatis, even after an extended (>30 min) incubation time (Fig. 1a and b). The very small intensity decrease directly after ATP addition is due to sample volume increase and is not reverted with the addition of an uncoupler.
Neither variation of the ATP/Mg2+ ratio (from 0.4 : 1 to 2 : 1), or variation of the pH value (pH 5.5–8.0) nor preparation of IMVs from M. bovis BCG cultured in an oxygen depletion model (Wayne system) led to detectable quenching upon ATP addition. These results indicate that mycobacterial ATP synthase is not carrying out ATP-hydrolysis-driven proton transport.
To exclude the possibility that the observed lack of ATP-hydrolysis-driven proton transport is caused by an extremely low number of ATP synthase molecules in the mycobacterial membrane or by of the detachment of the extrinsic F1 part of ATP synthase, we compared the DCCD-sensitive activities in ATP synthesis and ATP hydrolysis. As shown in Table 1, the IMVs from both M. bovis BCG and M. smegmatis were active in ATP synthesis with specific activities of 0.27 and 0.96 nmol min−1 mg−1, respectively. In contrast, we could not detect any significant DCCD-sensitive ATP hydrolysis activity in IMVs from M. bovis BCG. For M. smegmatis IMVs, DCCD-sensitive ATP hydrolysis activity was detectable, but >4-fold lower as compared with ATP synthesis (Table 1). For an enzyme working with equal speed in both directions, the ATP hydrolysis activity is expected to be higher than the synthesis activity, for example ∼10-fold for ATP synthase from Bacillus PS3 (Bald et al., 1998, 1999). This effect is due to the presence of enzymes in leaky vesicles, unavoidably present in IMV preparations, which can split ATP, but are unable to synthesize it.
Table 1. DCCD-sensitive ATP synthesis and ATP hydrolysis activities of mycobacterial membrane vesicles
Activity (nmol min−1 mg−1)
ATP synthesis and ATP hydrolysis activities of inverted membranes from Mycobacterium bovis BCG and Mycobacterium smegmatis were measured at 37°C by, respectively, the glucose-6-phosphate dehydrogenase method and inorganic phosphate release as described under Materials and methods. The results shown represent the mean ± SD from three different preparations.
M. bovis BCG
0.27 ± 0.17
0.96 ± 0.15
0.21 ± 0.32
Our results thus show that ATP synthase in both M. bovis BCG and M. smegmatis is blocked in the ATP hydrolysis mode and is not able to generate a PMF by hydrolyzing ATP. The essentiality of ATP synthase is thus based on a function in the synthesis direction, most likely either for the production of ATP, pH homeostasis, or for contributing to the NAD+/NADH redox balance. The task of PMF maintenance under low oxygen tensions is most probably fulfilled by other membrane–protein complexes, such as by nitrate reductase or by fumarate reductase acting in reverse (Schnorpfeil et al., 2001; Wayne & Sohaskey, 2001).
Mechanism of ATP hydrolysis blockage in mycobacterial ATP synthase
In order to gain an insight into the mechanism of ATP hydrolysis blockage in mycobacteria, we tested the effect of four different treatments reported to activate ‘latent’ ATP hydrolysis activity in bacteria. Limited trypsin proteolysis is reported to cleave the inhibitory intrinsic subunit ɛ and in this way activate ATP hydrolysis (Bogin et al., 1970; Keis et al., 2006), while the addition of methanol is thought to compromise hydrophobic interactions within ATP synthase (Hisabori et al., 1997). Moreover, oxy-anions, for example sulfite, are reported to remove inhibitory ADP and to uncouple ATP synthase function (Bakels et al., 1994; Cappellini et al., 1997; Pacheco-Moisés et al., 2002). Finally, membrane energization is known to relieve ADP inhibition and to switch the conformation of subunit ɛ to a noninhibitory state (Suzuki et al., 2003).
The ATP hydrolysis activity of IMVs of M. smegmatis was indeed activated >30-fold by trypsin (Table 2), indicating that subunit ɛ is an important determinant for ATP hydrolysis blockage in this fast-grower. However, in the case of M. bovis BCG, trypsin treatment did not lead to significant activation (Table 2). This lack of activation can be explained either by inaccessibility of the trypsin cleavage site or by the presence of alternative inhibitory mechanisms. To further investigate ATP hydrolysis in M. bovis BCG, we tested the effect of methanol, sodium sulfite and PMF activation. Whereas sulfite slightly activated ATP hydrolysis activity, both addition of methanol and membrane energization by succinate led to more significant activation for M. bovis BCG, with the resulting activity ∼10-fold higher than the ATP synthesis activity (Table 2).
Table 2. Activation of ATP hydrolysis in membrane vesicles from Mycobacterium bovis BCG and Mycobacterium smegmatis
Activity (nmol min−1 mg−1)
M. bovis BCG
Inverted membranes were preincubated with 90 U mL−1 trypsin or 10 mM succinate at, respectively, 30 and 37°C. Subsequently, 2 mM ATP was added and the initial rate of ATP hydrolysis was measured by inorganic phosphate production as described under Materials and methods. The results shown represent the mean ± SD from three different preparations.
Trypsin (90 U mL−1)
4.30 ± 0.90
Sulfite (10 mM)
0.80 ± 0.35
1.17 ± 0.75
3.24 ± 0.52
0.33 ± 0.22
Succinate (10 mM)
1.82 ± 1.0
1.30 ± 0.20
The results suggest that ATP hydrolysis in both slow-growing as well as fast-growing mycobacteria is regulated in a PMF-dependent manner, preventing excess ATP consumption under low oxygen tensions. Suppression of activity appears to be more pronounced in the slow-grower, which may be an adaptation to environments with a low energy supply and/or decreased oxygen tensions, for example in remote parts of the mammalian lungs. mycobacteria, requiring oxygen for growth, but able to persist under anaerobic conditions, thus utilize a similar mechanism of ATP hydrolysis inhibition as reported for the obligate aerobic bacteria P. denitrificans (Zharova & Vinogradov, 2003, 2006) and Micrococcus luteus (Grüber et al., 1994) and for the alkaliphilic bacteria Bacillus firmus OF4 (Hicks & Krulwich, 1990) and Bacillus sp. TA2.A1 (Keis et al., 2006). Whereas in alkaliphilic bacteria subunit ɛ has been pinpointed as the PMF-dependent regulator of ATP hydrolysis activity, in P. denitrificans and related Alphaproteobacteria, recently, a new intrinsic inhibitor protein, termed subunit ζ, was found (Morales-Ríos et al., 2010). However, as database search did not reveal any homologue of subunit ζ in mycobacteria, we regard subunit ɛ as the most likely candidate for this regulatory task.
Our results show that mycobacterial ATP synthase is blocked in the ATP hydrolysis direction and also suggest that any potential small-molecule inhibitor acting on mycobacterial ATP synthase should interfere with the ATP synthesis reaction in order to be considered as a drug candidate. An approach as used for the development of antiischemia drugs blocking ATP hydrolysis (Harmann et al., 2004) is thus not expected to be a promising strategy for the development of new antimycobacterial drugs.
However, activation of the latent ATP hydrolysis activity may lead to depleted cellular ATP levels and decrease the bacteria's viability. Compounds that can specifically relieve the blockage of ATP hydrolysis may thus be potential drug candidates. Experiments to clarify this point and to understand the molecular mechanism of ATP hydrolysis blockage in slow-growing mycobacteria are under way in our laboratory.
A.C.H. and D.B. gratefully acknowledge financial support from the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO-ECHO grant 700.55.017).