Mycobacterium tuberculosis, the causative agent of tuberculosis, poses a global health challenge due to the emergence of drug-resistant strains. Recently, bacterial energy metabolism has come into focus as a promising new target pathway for the development of antimycobacterial drugs. This review summarizes our current knowledge on mycobacterial respiratory energy conversion, in particular, during the physiologically dormant state that is associated with latent or persistent tuberculosis infections. Targeting components of respiratory ATP production, such as type-2 NADH dehydrogenase or ATP synthase, is illustrated as an emerging strategy in the development of novel drugs.
The global burden of Mycobacterium tuberculosis infections causes approximately 2 million deaths per year, with an estimated one-third of the world population being latently infected (Dye et al., 1999; Check, 2007). Conventionally, tuberculosis can be treated with a cocktail of first-line antibiotics, but recently mycobacterial strains resistant to first- and/or second-line drugs have emerged, and pose a global health challenge (Check, 2007; Dye, 2009). Mycobacterium tuberculosis, an aerobic bacterium, can survive within human macrophages in a hypometabolic state with no or extremely slow growth (Wayne & Sohaskey, 2001). In this dormant metabolic state, the bacterial cell wall thickness is increased, protein and nucleic acid syntheses are significantly downregulated and lipid metabolism appears to be the primary energy source (Wayne & Sohaskey, 2001; Timm et al., 2003). These changes are accompanied by characteristic up-regulation of a set of 48 genes, referred to as the dosR regulon (Voskuil et al., 2003). This major remodeling of key metabolic pathways leads to decreased sensitivity for currently used antibiotics (Gomez & McKinney, 2004), and is thus an important factor responsible for the extended tuberculosis treatment time in patients (6–9 months). In spite of the dormant phenotype, these bacteria still have basal energy requirements to maintain critical metabolic functions (Koul et al., 2008). In recent years, significant information has been gained on the essentiality of respiratory chain components in dormant as well as in replicating bacteria. The identification of new candidate drugs targeting the ATP-producing machinery illustrates the therapeutic potential of blocking mycobacterial energy conversion (Andries et al., 2005; Weinstein et al., 2005).
Respiratory ATP generation in mycobacteria
Many bacteria, such as Escherichia coli and Bacillus subtilis, can synthesize sufficient ATP for growth using substrate-level phosphorylation of fermentable carbon sources (Friedl et al., 1983; Santana et al., 1994). However, in the case of M. tuberculosis, ATP synthase is required for optimal growth as revealed by high-density mutagenesis (Sassetti et al., 2003). Moreover, in Mycobacterium smegmatis deletion mutants indicated an essential function of ATP synthase for growth on fermentable as well as nonfermentable carbon sources (Tran & Cook, 2005). These findings suggest that mycobacteria cannot gain enough energy by substrate-level phosphorylation and need respiratory ATP synthesis for growth.
In the respiratory chain, two types of NADH dehydrogenases are present in most mycobacteria for NADH oxidation and for feeding reducing equivalents into the electron transport pathway (Fig. 1). However, the proton-transporting type-I NADH dehydrogenase (NDH-1), encoded by the nuo operon, is not essential in M. tuberculosis (Sassetti et al., 2003; Rao et al., 2008) and is largely deleted from the genome of Mycobacterium leprae (Cole et al., 2001). Alternatively, NADH can be oxidized by a non-proton-translocating, type-II NADH dehydrogenase (NDH-2), using menaquinone as an electron acceptor (Fig. 1). In M. tuberculosis, NDH-2 is present in two copies, referred to as Ndh and NdhA, whereas in M. smegmatis, only one copy is found (Weinstein et al., 2005). Mutagenesis studies in M. smegmatis indicated an essential function of NDH-2 for survival (Miesel et al., 1998). Chemical inhibition of NDH-2 was reported to be bactericidal for M. tuberculosis, whereas typical inhibitors of the NDH-1 did not have a significant effect (Rao et al., 2008). Taken together, these data indicate that NDH-2 is responsible for the bulk of NADH oxidation in mycobacteria.
After accepting electrons from NDH-2, menaquinol can be reoxidized via two alternative routes, ending with either a cytochrome aa3-type or a cytochrome bd-type terminal oxidase (Fig. 1, for a review, see Cox & Cook, 2007). In the energetically more efficient route, menaquinol is oxidized by the cytochrome bc1 complex (consisting of subunits QcrA-C), which then transfers the electrons to the terminal cytochrome aa3-type oxidase (CtaC-F) (Matsoso et al., 2005). The cytochrome bc1 complex and the cytochrome aa3 oxidase, thought to form a super complex in mycobacteria, are proton-translocating enzymes, assuring the high energetic yield of this route (Niebisch & Bott, 2003; Matsoso et al., 2005). Alternatively, menaquinol can be directly oxidized by a cytochrome bd-type terminal oxidase (CytA-B) (Kana et al., 2001). This reaction is not coupled to proton pumping; consequently, the cytochrome bd oxidase route is energetically less efficient. However, cytochrome bd oxidase displays a higher affinity for oxygen and is thus used at low-oxygen tensions (Kana et al., 2001), whereas the cytochrome aa3-type enzyme is the predominant terminal electron acceptor during aerobic growth (Shi et al., 2005). The energy of the proton motive force is subsequently utilized by ATP synthase for the synthesis of ATP.
During dormancy, NDH-2 was found to be upregulated, whereas NDH-1 is strongly downregulated (Schnappinger et al., 2003; Shi et al., 2005). The cytochrome bc1 and cytochrome aa3 complexes are downregulated as well; however, the cytochrome bd-type oxidase is transiently upregulated, arguably to facilitate transition to the dormant state by contributing to redox balance (Shi et al., 2005). The question of the predominant terminal electron acceptor in the dormant state is still open. It has been suggested that nitrate reductase (NarG-I) acts as an acceptor, and indeed, the enzymatic activity of nitrate reductase was found to be increased (Wayne & Hayes, 1998), and addition of nitrate increased the viability of dormant mycobacteria (Gengenbacher et al., 2010). Moreover, the nitrate transporter NarK2 is upregulated during dormancy (Schnappinger et al., 2003; Voskuil et al., 2003; Shi et al., 2005). The subunits of the ATP synthase complex were found to be downregulated using in vitro dormancy models as well as an in vivo mouse lung infection model (Shi et al., 2005; Koul et al., 2008). This considerable remodeling in dormant mycobacteria reflects reduced oxygen availability and decreased energy requirements in a state without growth.
During dormancy, cellular ATP levels are ∼10-fold lower as compared with replicating bacilli (Starck et al., 2004; Koul et al., 2008; Rao et al., 2008; Gengenbacher et al., 2010). Nevertheless, dormant M. smegmatis are active in respiratory ATP synthesis and maintain an energized membrane (Koul et al., 2008). Furthermore, both replicating and dormant M. tuberculosis were also shown to maintain an energized cytoplasmatic membrane with a proton motive force of −110 mV (Rao et al., 2008). This value is significantly lower than the values typically found for other bacteria (−180 to −200 mV). Compounds interfering with the proton motive force, such as uncouplers or ionophores, proved strongly bactericidal on dormant M. tuberculosis in vitro (Rao et al., 2008), demonstrating that the proton motive force is an essential element of life under dormant conditions. It is an open question as to which enzyme is mainly responsible for the maintenance of the proton motive force during dormancy. Conceivable candidates for this task are nitrate reductase, whose activity is upregulated in the dormant state, or succinate dehydrogenase operating in reverse as a fumarate reductase (Schnorpfeil et al., 2001; Wayne & Sohaskey, 2001; Cox & Cook, 2007; Rao et al., 2008). In contrast, NDH-2, the predominant route for oxidation of NADH and for fueling of electrons into the respiratory chain in the dormant state (Rao et al., 2008), does not translocate protons. The role of this enzyme may instead be to provide redox balance, as phenothiazine inhibition of NDH-2 resulted in elevated cellular NADH concentrations (Rao et al., 2008). Furthermore, in contrast to the situation found in most bacteria, mycobacterial ATP synthase apparently cannot efficiently invert its function to pump protons across the membrane: ATP synthase from Mycobacterium phlei showed only a very low activity in ATP hydrolysis (Higashi et al., 1975), specific inhibition of ATP synthase in replicating and dormant M. smegmatis did not decrease the proton motive force (Koul et al., 2008) and membrane vesicles of Mycobacterium bovis BCG were not able to establish a proton motive force using ATP (A.C. Haagsma & D. Bald, unpublished data). These results indicate that in dormant mycobacteria, ATP synthase is active in the production of ATP, which may provide the energy required for residual biosynthesis activity. ATP synthesis activity may also facilitate a continuous electron flow through the respiratory chain, and in this way, contribute to redox balance. Inhibition of either NADH oxidation or ATP synthesis or collapse of the proton motive force leads to killing of M. tuberculosis (Rao et al., 2008, see also Fig. 1).
The respiratory chain of M. tuberculosis may show special adaptations for survival under dormant conditions and/or low proton motive forces. The activity of ATP synthase significantly depends on the proton motive force, with considerable variation between different organisms (Kaim & Dimroth, 1999). ATP synthase of M. tuberculosis may turn out to be active at lower membrane potential as compared with most bacteria or mitochondria. The molecular basis for this variation between species is obscure, although a role for the intrinsic inhibitory subunit ɛ and for the oligomeric, proton-translocating subunit c has been implied (Turina et al., 2006, see also Fig. 2). In the alkaliphilic Bacillus sp. strain TA2.A1, which has to cope with low proton motive force conditions as well, the subunit c complex is composed of 13 monomers, compared with 10 monomer complexes found in E. coli and Bacillus PS3 (Jiang et al., 2001; Mitome et al., 2004; Meier et al., 2007). A larger number of monomers per subunit c oligomer may increase the H+/ATP ratio and thus facilitate proton flow and the synthesis of ATP under low proton motive force conditions (Meier et al., 2007). Biochemical investigations and bioinformatics studies will help to answer this question and may also clarify why mycobacterial ATP synthase cannot invert its function to set up a proton motive force.
Only very little information is available on energy and metabolic fluxes in dormant mycobacteria, for example on the cellular rates of ATP production and consumption and on the most prominent ATP sinks. Quantitative analyses of metabolic fluxes can provide information on the minimal ATP requirements for survival during dormancy.
It appears that respiratory ATP synthesis is a key metabolic pathway in replicating as well as in dormant mycobacteria. In the next paragraph, the approach of utilizing respiratory ATP production as the target of novel antibacterial drugs is illustrated.
Respiratory ATP production as a new drug target
As described above, inhibition of NADH oxidation, interference with the proton motive force or blocking ATP synthase all have a pronounced bactericidal effect on replicating and dormant M. tuberculosis. Whereas compounds interfering with the proton motive force tend to be nonselective and toxic, for the other two prospective targets, small-molecule drug candidates have been reported: the phenothiazines inhibit NDH-2 (Boshoff & Barry, 2005; Weinstein et al., 2005) and the diarylquinolines block ATP synthase (Andries et al., 2005; b28).
Phenothiazines and phenothiazine analogues efficiently killed M. tuberculosis in vitro and were shown to be effective in a mouse infection model (Weinstein et al., 2005). Phenothiazines inhibited both homologues of NDH-2 in M. tuberculosis, Ndh and NdhA, and strongly suppressed oxygen consumption by mycobacterial membrane vesicles energized with NADH (Weinstein et al., 2005; Yano et al., 2006). Based on kinetic data, it has been suggested that phenothiazines do not compete with NADH or menaquinone binding, but block the formation or the reaction of an intermediate species of the catalytic cycle (Yano et al., 2006). NDH-2 is a membrane-associated, single-subunit enzyme, which carries one flavin–adenine dinucleotide (FAD) cofactor (Kerscher et al., 2008; Fisher et al., 2009). Homology studies suggest the presence of two domains for binding of NADH and FAD, respectively (Schmid & Gerloff, 2004). As such, NDH-2 differs significantly from the NDH-1 in the human mitochondria, which is a membrane-bound, multisubunit protein complex carrying additional iron–sulfur redox centers (Kerscher et al., 2008). NDH-2 represents an essential component of the mycobacterial respiratory chain, is upregulated during dormancy and has no homologue in the human mitochondria. These features make NDH-2 a promising target for the development of new drug candidates. High-resolution structural data and deeper understanding of phenothiazine action may facilitate structure-based design of small-molecule NDH-2 inhibitors with improved efficacy and selectivity.
Diarylquinolines represent a novel class of antimycobacterial drugs with strong in vitro and in vivo activity against different mycobacterial species (Andries et al., 2005; Ji et al., 2006). Diarylquinolines block ATP synthesis and cause a decrease of cellular ATP levels (Koul et al., 2007). As the bacterial ATP stores are depleted over a period of time, subsequently pronounced bacterial killing is observed (Koul et al., 2008). Diarylquinolines specifically interact with the oligomeric transmembrane subunit c of mycobacterial ATP synthase (Koul et al., 2007, see also Fig. 2). During enzymatic catalysis, this oligomeric subunit, together with subunits ɛ and γ, rotates relative to subunits α3β3δab and in this way couples proton flow to the synthesis of ATP (Boyer, 1993; Junge et al., 1997). Protons enter from the periplasmic space via an entry channel in subunit a and are then transferred to an essential acidic residue in the membrane-spanning part of subunit c (Fig. 2). After a close to 360° rotation of the cylindrical subunit c oligomer relative to subunit a, the protons are released on the cytosolic side of the membrane via an exit channel in subunit a (Vik & Antonio, 1994; Diez et al., 2004). Mutagenesis studies indicate that diarylquinoline lead compound TMC207 binds to the central region of subunit c, close to the essential acidic residue (Koul et al., 2007). TMC207 may compete with protons for binding to subunit c or may alternatively interfere with the extensive conformational changes of this subunit during catalysis. Whereas typical inhibitors of ATP synthase subunit c, such as dicyclohexyl-carbodiimide and oligomycin, are not selective and highly toxic (Matsuno-Yagi & Hatefi, 1993; Wallace & Starkov, 2000; Amacher, 2005), TMC207 displays a surprising selectivity, with only an extremely low effect on human ATP synthesis (Haagsma et al., 2009). Although several residues of subunit c are reported to modulate diarylquinoline sensitivity (Koul et al., 2007), the molecular basis for the observed selectivity needs to be further investigated. No high-resolution structure is available for mycobacterial ATP synthase or its subunits, and structural models for mycobacterial subunit c have only been built based on the known structure of the homologous subunit from E. coli, Ilyobacter tartaricus or Bacillus PS3 (de Jonge et al., 2007; Koul et al., 2007). High-resolution structural data for mycobacterial subunit c and biochemical investigations on drug/target interaction would help to explain drug selectivity and would provide input for docking studies to design new compound derivates. Based on the structure of diarylquinolines, novel quinoline derivates with significant in vitro bactericidal activity on M. tuberculosis have been synthesized (Lilienkampf et al., 2009; Upadhayaya et al., 2009).
Diarylquinolines were also shown to kill dormant M. tuberculosis as effectively as replicating bacilli and to inhibit ATP synthesis in dormant M. smegmatis (Koul et al., 2008). This unique dual bactericidal activity, with equal potency on replicating and dormant bacilli, distinguishes diarylquinolines from all the currently used antituberculosis drugs, such as isoniazid and rifampicin. These front-line drugs show significantly less activity on dormant mycobacteria as compared with replicating bacilli (Koul et al., 2008; Rao et al., 2008). Thus, although ATP synthase is significantly downregulated during dormancy, its residual activity appears to be essential for mycobacteria irrespective of their physiological state. This makes ATP synthase an efficient drug target to tackle both replicating as well as dormant bacilli.
In vivo experiments using mouse models indicated that diarylquinolines have bactericidal activity exceeding the effect of current first-line antibiotics (Andries et al., 2005; Lounis et al., 2006). Diarylquinolines, in particular when applied in combination therapy with the first-line antibiotic pyrazinamide, have a strong potential for shortening the duration of tuberculosis treatment (Lounis et al., 2006; Ibrahim et al., 2007). The physiological basis for this observed synergy remains obscure. In phase IIb clinical tests, the addition of TMC207 to standard therapy strongly decreased the count of CFU in the sputum of patients with multi-drug-resistant tuberculosis as compared with an active-placebo group (Diacon et al., 2009). TMC207 also accelerated conversion to a negative sputum culture, as compared with a placebo. These findings validate ATP synthase as a target for the treatment of tuberculosis.
Respiratory ATP production is not only essential for growth, but also represents a critical weak point in dormant mycobacteria. Although most enzymes involved in respiratory ATP synthesis are conserved between prokaryotes and eukaryotes, targeting ATP production may be a highly efficient approach for the development of antibacterial drugs. The strategy may be to target enzymes, which do not have homologs in human metabolism, as in the case of NDH-2. Alternatively, as applied for ATP synthase, small differences in the structure between a bacterial enzyme and a human homologue may be utilized for selective inhibition. Understanding respiratory ATP production in replicating and dormant mycobacteria will not only fuel development of novel drugs but also shed light on how these bacteria perform their intriguing task of extreme persistence without significant growth.
The authors wish to thank Prof. Dr H. Lill (VU Amsterdam) and Prof. Dr K. Andries (Johnson and Johnson) for critically reading the manuscript, and Dr J. Guillemont, Dr E. Arnoult (Johnson & Johnson) and Ms A. Haagsma (VU Amsterdam) for assistance with the design of figures.