By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Due to essential maintenance the subscribe/renew pages will be unavailable on Wednesday 26 October between 02:00- 08:00 BST/ 09:00 – 15:00 SGT/ 21:00- 03:00 EDT. Apologies for the inconvenience.
Present address: Laboratoire de Dynamique des Interactions Membranaires Normales et Pathologiques, Universités de Montpellier II et I, CNRS; UMR 5235, case 107, Place Eugène Bataillon, 34095 Montpellier Cedex 05, France.
Equal contribution of the first two authors.
Department of Biochemistry and Biophysics, Texas A&M University, College Station, TX, USA.
Laboratoire de Dynamique des Interactions Membranaires Normales et Pathologiques, Universités de Montpellier II et I, CNRS; UMR 5235, case 107, Place Eugène Bataillon, 34095 Montpellier Cedex 05, France.
Laboratoire de Dynamique des Interactions Membranaires Normales et Pathologiques, Universités de Montpellier II et I, CNRS; UMR 5235, case 107, Place Eugène Bataillon, 34095 Montpellier Cedex 05, France.
INSERM, DIMNP, Place Eugène Bataillon, 34095 Montpellier Cedex 05, France.
The remarkable survival ability of Mycobacterium tuberculosis in infected hosts is related to the presence of cell wall-associated mycolic acids. Despite their importance, the mechanisms that modulate expression of these lipids in response to environmental changes are unknown. Here we demonstrate that the enoyl-ACP reductase activity of InhA, an essential enzyme of the mycolic acid biosynthetic pathway and the primary target of the anti-tubercular drug isoniazid, is controlled via phosphorylation. Thr-266 is the unique kinase phosphoacceptor, both in vitro and in vivo. The physiological relevance of Thr-266 phosphorylation was demonstrated using inhA phosphoablative (T266A) or phosphomimetic (T266D/E) mutants. Enoyl reductase activity was severely impaired in the mimetic mutants in vitro, as a consequence of a reduced binding affinity to NADH. Importantly, introduction of inhA_T266D/E failed to complement growth and mycolic acid defects of an inhA-thermosensitive Mycobacterium smegmatis strain, in a similar manner to what is observed following isoniazid treatment. This study suggests that phosphorylation of InhA may represent an unusual mechanism that allows M. tuberculosis to regulate its mycolic acid content, thus offering a new approach to future anti-tuberculosis drug development.
Mycolic acids are the major and distinguishing molecular components of the Mycobacterium tuberculosis cell envelope. These very long α-alkyl β-hydroxyl fatty acids are found either unbound as esters of trehalose or glycerol that are extractable with organic solvents or attached to the terminal penta-arabinofuranosyl units of arabinogalactan, the polysaccharide that, together with peptidoglycan, forms the insoluble cell-wall skeleton (McNeil et al., 1991). These lipids play an important role in the formation of an outer membrane and in cell-wall impermeability, virulence, immune evasion (Dubnau et al., 2000; Glickman et al., 2000; Rao et al., 2006; Bhatt et al., 2007a,b; Dao et al., 2008) and the distinctive acid-fast staining of M. tuberculosis (Bhatt et al., 2007a). Recent studies have also revealed free mycolic acids in M. tuberculosis biofilms (Ojha et al., 2008). Mycolic acid synthesis is tightly related to cell division, as evidenced by the fact that isoniazid (INH) and ethionamide (ETH), two anti-tubercular drugs, inhibit mycolic acid biosynthesis, resulting in cell lysis (Takayama et al., 1972). This metabolic pathway thus represents an important reservoir of targets for new drugs (Dover et al., 2008), which are more critical than ever since the emergence of multidrug-resistant and extremely drug-resistant strains of M. tuberculosis.
InhA belongs to the family of short-chain dehydrogenases/reductases (Dessen et al., 1995; Quemard et al., 1995). Genetic studies have revealed that inhA is essential in mycobacteria (Vilcheze et al., 2000). Thermal inactivation of InhA in a Mycobacterium smegmatis strain carrying an inhA-thermosensitive allele resulted in inhibition of mycolic acid biosynthesis and cell lysis in a manner similar to that seen in INH-treated bacteria (Vilcheze et al., 2000). It is now clear that INH is activated by the catalase peroxidase KatG to form a hypothetical isonicotinoyl radical that binds to NAD. The resulting INH-NAD adduct inhibits InhA (Rozwarski et al., 1998), leading to accumulation of long-chain fatty acids, inhibition of mycolic acid biosynthesis and ultimately cell death (Vilcheze et al., 2000). INH resistance also involves several genes but only mutations in inhA have a dominant phenotype, whereas all other mechanisms of resistance (katG, ndh, msh, nat) are recessive (Vilcheze and Jacobs, 2007). The inhA gene product was originally identified as a putative target for INH and ETH in M. smegmatis (Banerjee et al., 1994). An increasing body of evidence has since pointed to InhA as the primary target of the two drugs (Vilcheze et al., 2000; Larsen et al., 2002; Kremer et al., 2003; Wang et al., 2007), culminating in the demonstration that transfer of an inhA S94A mutant allele in M. tuberculosis was sufficient for conferring resistance to both INH and ETH and establishing InhA as the clinically relevant target of INH (Vilcheze et al., 2006). InhA is one of the best-validated targets for the development of anti-tubercular agents. Further studies on INH, as well as on other small-molecule inhibitors of InhA, hold significant promise for the delivery of novel anti-tubercular agents effective against drug-resistant M. tuberculosis (He et al., 2006; 2007; Oliveira et al., 2007; Tonge et al., 2007; Vilcheze and Jacobs, 2007; am Ende et al., 2008; Freundlich et al., 2009). This prompted us to seek new alternatives for InhA inactivation, ultimately leading to mycolic acid and cell growth inhibition. Indeed, mycolic acid biosynthesis and cell division are very likely to be related (Takayama et al., 1972; Lacave et al., 1987; 1989). We hypothesized that the activity of InhA might be controlled by post-translational modification in M. tuberculosis. This idea was supported by (i) the recent demonstration that the keto-reductase activity of MabA is regulated by Ser/Thr protein kinase (STPK)-dependent phosphorylation, providing the first information on the molecular mechanism(s) involved in mycolic acid regulation through phosphorylation of a FAS-II enzyme (Veyron-Churlet et al., 2010) and (ii) the fact that genes encoding the two FAS-II reductases, MabA and InhA, are in the same operon in the M. tuberculosis genome (Banerjee et al., 1998; Cole et al., 1998).
Many of the stimuli encountered by M. tuberculosis are transduced via sensor kinases in the membrane, allowing the pathogen to adapt to survive in hostile environments. In addition to the classical two-component systems, M. tuberculosis contains 11 eukaryotic-like STPKs (Cole et al., 1998; Av-Gay and Everett, 2000). There is now an increasing body of evidence suggesting that, in M. tuberculosis, many STPKs are involved in regulating metabolic processes, transport of metabolites, cell division or virulence (Molle and Kremer, 2010). Signalling through Ser/Thr phosphorylation has recently emerged as a key regulatory mechanism in pathogenic mycobacteria (Wehenkel et al., 2008; Molle and Kremer, 2010).
As a first step to decipher an original molecular mechanism for future drug development by specifically targeting InhA, this study was undertaken to determine whether InhA represents a new substrate of M. tuberculosis SPTKs and to investigate whether phosphorylation negatively regulates InhA activity and consequently mycolic acid biosynthesis and mycobacterial growth.
InhA is phosphorylated in vitro by multiple Ser/Thr kinases (STPK)
By analogy with the mode of action of INH, we reasoned that inhibition of mycolic acid biosynthesis might occur in vivo via post-translational modifications of InhA capable of reducing the enoyl reductase (ENR) activity of the enzyme. This prompted us to examine whether InhA could be modified by phosphorylation, a reaction that changes the physicochemical properties of defined Ser or Thr residues by introducing negative charges, which can ultimately affect the overall activity of the protein. This was first investigated in vitro in the presence of purified STPKs (PknA to PknL). The kinase domains of several transmembrane kinases from M. tuberculosis were expressed as GST-tagged fusion proteins and purified from Escherichia coli as reported earlier (Molle et al., 2006). The purified kinases were incubated with InhA and [γ-33P]-ATP, resolved by SDS-PAGE and their phosphorylation profiles analysed by autoradiography. The presence of intense radioactive signals indicated that InhA was phosphorylated by multiple kinases, including PknA, PknB, PknE and PknL (Fig. 1A). No signal was observed in the presence of PknF, PknH or PknK, all of which displayed various autokinase activities as reported earlier (Molle et al., 2006). These results clearly indicate that InhA is a specific substrate and interacts with various STPKs in vitro, suggesting that this key protein of the mycolic acid biosynthetic pathway might be regulated in mycobacteria by multiple extracellular signals.
InhA is phosphorylated on a unique threonine residue
Mass spectrometry was used to identify the number and nature of the phosphorylation sites on InhA. Such a method has been successfully used to elucidate the phosphorylation sites in a sequence-specific fashion for several M. tuberculosis STPK substrates (Barthe et al., 2009; Canova et al., 2009; Veyron-Churlet et al., 2009; 2010). InhA was incubated with unlabelled ATP in the presence of PknB (one of the most active kinases for InhA, Fig. 1A) and subjected to mass spectrometric analysis after tryptic and chymotryptic digestion. Spectral identification and phosphorylation determination were achieved with the paragon algorithm from the ProteinPilot® 2.0 database-searching software (Applied Biosystems) using the phosphorylation emphasis criterion against a homemade database that included the sequences of InhA and derivatives. The phosphopeptides identified by the software were then validated by manual examination of the corresponding MS/MS spectra. Manual validations were performed based on neutral loss of H3PO4 from the precursor ion and the assignment of major fragment ions to b- and y-ion series or to the corresponding neutral loss of H3PO4 from these ions. The sequence coverage of the protein was 94% and phosphorylation occurred only on peptide (241–269) with an 80 Da mass increment from 3071.53 Da (theoretical MW) to 3151.52 Da (monoisotopic mass). The MS/MS spectra unambiguously confirmed the presence of a phosphate group on Thr-266 (Figs 1B and S1A). Then, in order to prevent in vitro phosphorylation on Thr-266, this residue was changed to alanine by site-directed mutagenesis. The corresponding phosphoablative InhA_T266A mutant was expressed and purified as a His-tagged protein in E. coli BL21(DE3)Star harbouring pETPhos_inhA_T266A. The resulting InhA_T266A mutant protein was purified and incubated with PknB in the presence of [γ-33P]-ATP. Following separation by SDS-PAGE and analysis by autoradiography, total abrogation of the phosphorylation signal was observed compared with InhA_WT (Fig. 1C). Similar results were obtained when InhA_T266A was incubated in the presence of either PknA, PknE or PknL (data not shown), indicating that Thr-266 represents the phosphoacceptor for all four kinases.
Taken together, these results indicate that Thr-266 represents the unique phosphorylation site in InhA and suggest that phosphorylation of this residue is likely to play a critical role in the regulation of InhA activity.
In vivo phosphorylation of InhA
To corroborate the in vitro results, it was necessary to confirm the phosphorylation state of the InhA protein in vivo. The inhA gene was cloned into the pMK1 mycobacterial expression vector under the control of the strong promoter hsp60 (Table S1). The resulting construct was used to transform Mycobacterium bovis BCG Pasteur in order to allow over-production of recombinant His-tagged InhA, which was purified and subjected to 2D-gel electrophoresis. As shown in Fig. 2A, two spots, presumably corresponding to the non-phosphorylated form followed in the acidic direction by a mono-phosphorylated form of InhA, were clearly detected. Definitive identification and localization of Thr-266 as the unique phosphorylation site in InhA in vivo was achieved by mass spectrometric analysis using InhA purified from M. bovis BCG cultures. LC-MS/MS identified an 80 Da mass increase corresponding to the peptide (241–269), supporting the conclusion that InhA is phosphorylated in vivo in M. bovis BCG (Figs 2B and S1B). Together, these results confirm that Thr-266 corresponds to the primary phosphorylation site both in vitro and in vivo.
ENR activity is strongly reduced for the InhA_T266D and InhA_T266E phosphomimetic mutants
Earlier studies showed that the acidic Asp or Glu amino acids qualitatively mimic the effect of phosphorylation with regard to functional activity. Following a strategy successfully used to demonstrate the role of phosphorylation on the condensing activity of mtFabH (Veyron-Churlet et al., 2009) and the β-ketoacyl-ACP reductase MabA (Veyron-Churlet et al., 2010), we first expressed and purified the phosphomimetics InhA_T266D and InhA_T266E. Next, we determined the in vitro ENR activity of these mimetics and compared their enzymatic activity with those of InhA_WT in the presence of increasing concentrations of trans-2-dodecenoyl-CoA (DD-CoA). Figure 3A clearly shows a strongly reduced activity of the two phosphomimetic mutants. The effect of mutation at T266 on the enzymatic activity was then calculated as the remaining per cent activity. The phosphoablative InhA_T266A protein showed very similar catalytic activity to the wild-type enzyme, indicating that this residue is not involved in catalysis (Fig. 3A and B, Table 1). InhA_T266E and InhA_T266D mutants showed only ∼30% residual enzymatic activity. Subsequent enzymatic studies were focused on Asp and Glu mutants to determine the reason behind this significant loss of function. Initial velocity studies with Asp and Glu mutants suggested that both followed a random order reaction mechanism with preferred NADH binding prior to DD-CoA binding, which was reported previously for wild-type InhA (Quemard et al., 1995; Parikh et al., 1999) (Fig. 3A and C). This preferred-ordered pathway towards NADH binding resulted in sigmodial initial velocity curves as the NADH concentrations were varied at a fixed DD-CoA concentration (Fig. 3A, right panel). On the other hand, when the concentration of DD-CoA was varied at fixed concentrations of NADH, a pattern similar to substrate inhibition was observed (Fig. 3A, left panel). The steady-state kinetic parameters, kcat and Km, calculated for DD-CoA for InhA_WT, InhA_T266A, InhA_T266D and InhA_T266E mutants are presented in Table 1. Both Asp and Glu mutants exhibited Km values for DD-CoA very similar to those of the wild-type and T266A mutant enzymes but kcat and Vmax values for the wild type and Ala mutant showed three- to fourfold higher catalytic turnover and maximum velocity compared with the phosphomimetic mutants. kcat/Km, which defines the substrate specificity, is almost the same for all four enzymes. This suggests that the overall observed decrease in the activity (100% wild type versus 30% mutants) is not due to the deficiency in the DD-CoA substrate binding.
Table 1. Steady-state kinetic parameters Km, kcat, Vmax and kcat/Km for DD-CoA, for wild-type and mutant InhA enzymes.
Vmax (µM min−1 mg−1)
kcat/Km (min−1 µM−1)
Kd (NADH) (µM)
Dissociation constants for NADH are presented. The values were determined in triplicate and the average numbers calculated from three different experiments are represented. For each InhA enzyme, kcat/Km was calculated from three independent experiments and the mean kcat/Km is presented. Values are mean ± standard error.
40.9 ± 15.4
320.4 ± 52.7
11.4 ± 1.9
9.9 ± 2.7
1.5 ± 0.2
102.6 ± 7.8
52.3 ± 3.6
427.4 ± 34.4
15.3 ± 1.2
8.2 ± 0.2
3.5 ± 0.4
31.4 ± 6.3
19.6 ± 5.6
87.3 ± 4.4
3.1 ± 0.1
6.0 ± 2.7
4.7 ± 0.6
29.5 ± 1.2
20.3 ± 4.9
149.4 ± 11.9
4.1 ± 0.8
6.9 ± 2.3
6.0 ± 0.6
Fluorescence experiments clearly indicated that both Asp and Glu mutants had three- to fourfold decreased binding affinity for NADH while InhA_T266A displayed approximately twofold reduction compared with InhA_WT as judged by the Kd values (Table 1). This suggests that T266 is an important residue for NADH binding and mutation of this residue can affect the binding of the enzyme to NADH. However, the overall effect on enzymatic activity is not linearly related to the fold decrease in NADH binding affinity since T266A is catalytically as active as the wild-type enzyme. Because phosphomimetic mutant enzymes kinetically prefer NADH binding first, the overall activity for these mutant proteins were affected significantly by the decreased affinity towards NADH. It is therefore very likely that the change in the NADH binding affinity is responsible for the decreased activity of the phosphomimetic InhA mutants.
Local structural changes generated by Thr-266 mutations
Recombinant InhA_WT, InhA_T266D and InhA_T266E proteins were purified from E. coli and subjected to crystallographic studies. In the InhA_WT structure, the OG1 atom of T266 interacts with the NE2 atom of Q267 (3.08 Å) and carbonyl oxygen of G263 (2.74 Å) (Fig. S2). For the T266D/E mutants, due to the loss of the Thr residue, this direct H-bonding interaction between residues T266 and Q267 is lost. Instead, a new water molecule (H2O #9 and #5 for InhA_T266D and InhA_T266E mutants respectively) is introduced between residues 266 and 267 (Fig. S2). Regarding the InhA_T266D structure, this makes H-bonding interactions with the NE atom of Q267 (3.01 Å), the carbonyl oxygen of G263 (2.74 Å) and the carbonyl oxygen of T254 from the neighbouring subunit (2.77 Å). In InhA_T266E these interaction distances are 3.0 Å, 2.71 Å and 2.79 Å respectively. Notably, the direct interaction between the side-chains of residues 266 and 267 is completely abrogated in the mutant structures. On the other hand, OD1 and OD2 atoms of D266 interact with NE2 and ND1 of H265 (3.07 Å and 3.45 Å respectively), causing a ∼15 degree flip in the H265 ring. For E266, the distances are: OE1 of E266 with NE2 of H265 4.4 Å; ND1 of H265 3.04 Å; OE2 of E266 with ND1 H265 3.5 Å and NE2 of H265 3.8 Å. OE2 of E266 and OD2 of D266 also H-bonds with the carbonyl oxygen of T254 from the neighbour subunit (3.0 Å), which is not present in the wild-type InhA:NADH structure (3.7 Å) (Fig. S2). In summary, mutation of Thr→Asp or Thr→Glu led to local structural changes limited to the surrounding of the mutated residue.
Growth defect of fast- and slow-growing mycobacteria overexpressing InhA phosphomimetic proteins
Mycobacterium smegmatis mc2155 was transformed with pMK1 derivatives allowing constitutive expression of the different inhA alleles under the control of the strong hsp60 promoter (Table S1): inhA_WT, the phosphoablative inhA_T266A or the phosphomimetic inhA_T266D and inhA_T266E. Transformed mycobacteria were selected on Middlebrook 7H10 plates. As shown in Fig. 4A, which illustrates the morphology and size of M. smegmatis colonies after 4 days' incubation at 37°C, it is clear that overexpression of InhA_T266A did not impair growth of M. smegmatis as compared with the strain overproducing InhA_WT. In contrast, overexpression of either InhA_T266D or InhA_T266E was accompanied by a decrease in the growth rate compared with the strains overexpressing InhA_WT or InhA_T266A (Fig. 4A). Similar results were also observed in M. bovis BCG strains harbouring these plasmids (Fig. S3). Taken together, these results indicate that overexpression of the phosphomimetic InhA alleles severely altered mycobacterial growth in both fast- and slow-growing mycobacterial species. Our results show that the impaired ENR activity of InhA_T266D or InhA_T266E is likely to be responsible for this growth defect, presumably by inhibiting FAS-II activity, either interacting with other FAS-II system enzymes or forming non-functional oligomers. Because InhA acts as a tetramer in mycobacteria, overproduction of InhA_T266D or InhA_T266E may lead to unproductive InhA multimers, which may coexist with functional InhA multimers. The hypothesis that the impaired in vitro activity of InhA_T266D or InhA_T266E is linked to a different oligomerization state than InhA_WT can, however, be excluded since both mutant enzymes formed only tetramers in solution as demonstrated by gel-filtration chromatography (data not shown), and by the fact that we could crystallize and determine the structures of the tetrameric InhA_T266D and T266E mutants (Fig. S4). Western blot analysis from M. smegmatis and M. bovis BCG cultures revealed both the endogenous and the recombinant InhA proteins in each strain, which can be clearly distinguished by the presence (recombinant InhA) or absence (endogenous InhA) of a His-tag (Figs 4B and S3). Although the recombinant InhA forms are produced in large excess compared with the endogenous InhA, the presence of significant amounts of endogenous protein may explain the partial growth defect in the phosphomimetic strains. The growth impairment also hampers further examination and evaluation of the possible link between growth defect and mycolic acid biosynthesis inhibition in the mycobacterial strains overproducing the phosphomimetic mutants. We then explored whether this could be demonstrated by generating isogenic strains carrying either the phosphoablative or the phosphomimetic inhA alleles in M. tuberculosis.
Transfer of T266D or T266E point mutations in InhA is lethal to M. tuberculosis
Using specialized linkage transduction (Vilcheze et al., 2006) we could transfer a single-point mutant allele (T266A) into inhA in M. bovis BCG and M. tuberculosis H37Rv (Fig. 5A), indicating that introduction of the phosphoablative inhA allele was not lethal. In addition, the growth curves of M. tuberculosis bearing either the inhA_WT or the inhA_T266A allele were similar, indicating that introduction of the T266A mutation did not alter growth of M. tuberculosis (data not shown). In contrast, and despite several attempts, we could not transfer the T266D or T266E alleles (Fig. 5A) into a wild-type genetic background, suggesting that these single mutations were lethal to M. bovis BCG and M. tuberculosis strains. To test this hypothesis, we performed allelic replacements using M. tuberculosis inhA merodiploid strains constructed by integrating a functional copy of inhA into the chromosome of M. tuberculosis H37Rv using the integrative pMV361 vector, which inserts at the attB chromosomal site. Gene replacements were carried out using specialized linkage transduction as described above. Interestingly, only in the presence of a complementing copy of the wild-type inhA were we able to introduce the T266D or T266E mutation into both the M. tuberculosis H37Rv and the M. tuberculosis CDC1551 merodiploid strains, as shown by Southern blot (Fig. 5B) and PCR analyses (data not shown). Sequence analysis confirmed the presence of the expected inhA alleles in the various strains (data not shown).
Taken together, these results indicate that a single T266D or T266E point mutation within InhA confers lethality on M. tuberculosis. They also strongly suggest that, in the absence of wild-type InhA, transfer of a phosphomimetic allele leads to cell death.
Lack of ENR activity complementation with the InhA phosphomimetic mutants in InhA-thermosensitive mutant strain
To investigate whether cell death occurring in a phosphomimetic strain is associated with inhibition of mycolic acid biosynthesis, we took advantage of using an inhA-thermosensitive (Ts) mutant of M. smegmatis (m22359) carrying an inhA_V238F allele (Vilcheze et al., 2000). At a non-permissive temperature (42°C), thermal inactivation of InhA occurred, resulting in mycolic acid biosynthesis inhibition and leading to a drastic morphological change and cell death in a manner similar to the result in INH-treated cells (Vilcheze et al., 2000). M. smegmatis m22359 was therefore transformed with either the empty pMK1 vector or the pMK1 derivatives, allowing constitutive expression of the different inhA alleles, and selected for kanamycin resistance. Cultures were grown at the permissive temperature (30°C), then plated and grown at either 30°C or 42°C for 4–5 days. Figure 6A shows that cells transformed with constructs carrying either the inhA_WT or inhA_T266A alleles could grow on Middlebrook plates at the non-permissive temperature, indicating that functional complementation occurred. In sharp contrast, neither the inhA (T266D) nor the inhA (T266E) alleles restored growth at 42°C, indicating that introduction of these mutations was lethal in the absence of functionally active InhA (Fig. 6A). These phenotypes cannot be attributed to an eventual alteration of the expression levels of the various InhA mutants, as demonstrated by Western blot analysis (Fig. 6B). Comparable expression levels of the various InhA variants were found at either 30°C or 42°C. Moreover, due to the presence of an additional His-tag in the pMK1_InhA-derived constructs, it became possible to distinguish the recombinant variants (InhA_WT, T266A, T266D and T266E) from the endogenous protein (InhA_V238F) in the various strains. Importantly, following a temperature shift to 42°C, only the strains overexpressing InhA_WT or InhA_T266A could grow in broth medium, whereas those harbouring either the empty pMK1 plasmid or the pMK1-inhA_T266D/E constructs failed to grow (Fig. 6C). Overall, these results indicate that the T266D or T266E phosphomimetic proteins (even when present in large excess compared with the endogenous protein) cannot complement thermal inactivation of InhA_V238F at the non-permissive temperature.
The usefulness of these strains prompted us to examine whether this lethal phenotype was related to de novo mycolic acid biosynthesis inhibition. This was achieved by shifting the temperature of the various strains to 42°C prior to labelling with [1-14C]-acetate. After a further 3 h incubation at 42°C, cells were harvested and fatty acid methyl esters (FAMEs) and mycolic acid methyl esters (MAMEs) were extracted and separated by thin layer chromatography (TLC). As shown on the autoradiogram, and consistently with a previous study (Vilcheze et al., 2000), mycolic acid biosynthesis was almost completely abrogated in M. smegmatis mc22359 carrying the empty pMK1 at 42°C (Fig. 6D). In contrast, M. smegmatis mc22359 transformed with either pMK1_inhA_WT or pMK1_inhA_T266A displayed a wild-type mycolic acid profile at both 30°C and 42°C. Importantly, synthesis of mycolic acids in the phosphomimetic strains was severely impaired at the non-permissive temperature. Synthesis of α- and epoxy-mycolates was almost completely abrogated. The shorter-chain α′-mycolates, only present in this particular mycobacterial species, appeared less affected, as could be observed also when M. smegmatis cultures were treated with INH (Fig. 6D). This observation is consistent with previous studies reporting that α′-mycolates are less affected than the other mycolic acid subtypes after treatment with FAS-II-inhibitory drugs such as INH (Kremer et al., 2003) or thiolactomycin (Slayden et al., 1996).
Overall, these results clearly indicate that, in the absence of a functional endogenous protein, the T266D or T266E mutants cannot restore mycolic acid production, leading to mycobacterial growth inhibition. Importantly, this mycolic acid profile was reminiscent of that observed in INH-treated cultures (Vilcheze et al., 2000), leading to the conclusion that phosphomimetic mutants of InhA inhibit mycolic acid biosynthesis in a way similar to INH treatment.
In this study, a combination of genetic, biochemical and structural approaches was used to provide, for the first time, evidence that phosphorylation of InhA negatively regulates de novo biosynthesis of mycolic acids in M. tuberculosis. We found that phosphorylation occurred both in vitro and in vivo on a unique residue, identified as Thr-266, although InhA was not identified in the recent phosphoproteomic study performed on M. tuberculosis (Prisic et al., 2010). Moreover, enzymatic analyses clearly indicated reduced ENR activity of InhA phosphomimetic mutant proteins, suggesting that phosphorylation may negatively regulate mycolic acid production. This hypothesis is supported by the following three facts: transfer of single T266D or T266E mutations into M. tuberculosis was lethal, unless an additional copy of inhA was present in a merodiplid strain; overexpression of InhA_T266D/E severely impaired growth in both fast- and slow-growing mycobacterial species; and T266D or T266E phosphomimetic proteins could not complement thermal inactivation of InhA_V238F at the non-permissive temperature in an inhA(Ts) mutant of M. smegmatis, leading to strong inhibition of α- and epoxy-mycolic acid production and growth arrest. It is noteworthy that, during review of our manuscript, an independent study has been accepted for publication (Khan et al., 2010), presenting very similar results and conclusions by demonstrating that (i) InhA was phosphorylated in vivo at Thr-266, (ii) phosphorylation of InhA resulted in decreased enzymatic activity, and (iii) an InhA_T266E mutant was unable to rescue a M. smegmatis conditional inhA gene replacement mutant. Therefore, from both studies it can be inferred that STPK-dependent phosphorylation of InhA may represent an important mechanism controlling mycolic acid biosynthesis and largely supports and extends the emerging theme that Ser/Thr phosphorylation plays a critical role in the regulation of cell-wall biosynthesis and cell division in mycobacteria (Wehenkel et al., 2008; Molle and Kremer, 2010).
One hypothesis could be STPK-dependent phosphorylation of InhA is used by M. tuberculosis during the non-replicating dormant state, known to correlate with a slowdown of cell division, energy metabolism and lipid biosynthesis (Betts et al., 2002). However, further studies are required to investigate whether InhA (hyper)phosphorylation participates in the inhibition of mycolic acid metabolism during the ‘persistent’ state.
From a purely mechanistic perspective, our kinetic studies revealed that both InhA phosphomimetics followed a preferred-order pathway with NADH binding preceding DD-CoA binding. Both mutant proteins exhibited a three- to fourfold decreased binding affinity to NADH even though the mutation site was not at the cofactor binding pocket of InhA. Since there is a kinetic advantage when InhA binds to NADH first, it appears very likely that the overall activity for the mutant proteins was affected significantly by the decreased affinity towards NADH. T266 is very close to the C-terminal end of the protein (266 of 269 residues) and crystallographic studies indicated that mutation at this residue was not accompanied by significant differences in the secondary structure and did not alter the overall backbone conformation of the protein. Both T266E and T266D complexed with NAD+/NADH crystallized in the same space group (P6222) as the wild-type protein, with one molecule in the asymmetric unit (Table S3). As for InhA_WT, all mutant proteins formed tetramers in solution as judged by the gel-filtration chromatography profile (data not shown), indicating that mutation of the Thr-266 residue does not alter the oligomeric state of the protein. Moreover, the comparison of the tetrameric structures generated by crystallographic symmetry-related molecules for the wild type and Asp/Glu mutants revealed that the Rmsd value between the superimposed structures was 0.3 Å (Fig. S4). Structural differences due to Thr→Asp or Thr→Glu replacements were limited to the surroundings of the mutated residue and introduced only subtle and local structural changes, perturbing the H-bonding network and introducing water molecules (Fig. S2). The active site of InhA protein was ∼15–20 Å away from the mutation site in InhA_T266 (Fig. S5). Although no significant structural difference or conformational change was observed at either the active site residues or the substrate/cofactor binding pockets between native and mutants structures, it remains possible that mutations cause changes in the network of interactions and affect the interaction and communication between the subunits of the functional tetramer. This in turn may affect NADH binding affinity and the overall enzymatic activity of the phosphomimetic InhA mutants. Based on our enzymatic analyses, it is noteworthy that NADH binding is affected even though the mutation site is ∼16 Å away from the cofactor binding pocket.
InhA is one of the best-validated targets for anti-tubercular agents and several recent studies reported the development of new InhA inhibitors (He et al., 2006; 2007; Oliveira et al., 2007; Tonge et al., 2007; Vilcheze and Jacobs, 2007; am Ende et al., 2008; Freundlich et al., 2009). Our work suggests that displacement of the unphosphorylated/phosphorylated InhA balance in favour of the phosphorylated isoform rapidly leads to mycolic acid cessation, as happens following INH treatment, and to mycobacterial growth inhibition. Therefore, an elegant hypothesis arising from the present work is that, by increasing the activity of the kinases, it may be possible to directly alter mycobacterial growth, opening new and original perspectives for future anti-tuberculosis drug development. Indeed, small molecules that modulate the activity of STPK may be of great therapeutic value in inhibiting M. tuberculosis growth. Bryostatin, a natural product synthesized by a marine bacterium, which activates eukaryotic intracellular STPKs (Hale et al., 2002), is one of these molecules. Interestingly, bryostatin acts directly on the Bacillus subtilis Ser/Thr kinase PrkC, which contains an extracellular domain able to bind to peptidoglycan fragments and this signals the bacteria to exit dormancy by stimulating germination (Shah et al., 2008). PrkC, like M. tuberculosis PknB, possesses the PASTA (Penicillin And Ser/Thr kinase Associated) domains, which are found in the extracellular portion of membrane-associated STPKs and which have been proposed to bind to peptidoglycan and act as signalling molecules. In this context, bryostatin or other STPK-activating molecules, along with the recent structural determination of the M. tuberculosis PknB PASTA domains (Barthe et al., 2010), may provide new therapeutic strategies to be developed against tuberculosis. From our results, it can also be inferred that STPK-induced phosphorylation of InhA would be active on M. tuberculosis clinical isolates carrying the inhA alleles (I21, S94 or I194) that confer resistance to INH (Ramaswamy et al., 2003; Hazbon et al., 2006), since phosphorylation occurs exclusively on Thr-266, a residue that has never been linked to INH resistance.
The present study provides a foundation for further investigation of a seemingly important functional linkage between STPKs and the FAS-II system. It also provides conceptual advances in our understanding of the mycolic acid metabolic adaptation and regulatory events exploited by M. tuberculosis to adapt its mycolic acid cell-wall content. Although very challenging, future studies should now help to identify extracellular cues sensed by the different kinases and leading to InhA phosphorylation. This will not only allow us to understand how M. tuberculosis senses its environment and mediates its response in a co-ordinated manner to regulate mycolic acid biosynthesis, but also to possibly link InhA phosphorylation to the establishment of the non-replicating persistent state. One can also anticipate that similar strategies involving STPK-dependent mechanisms will be found to be used by pathogenic mycobacteria to regulate expression of other cell-wall lipids/glycolipids to respond to the various signals encountered during infection or latency.
Bacterial strains, plasmid, phage and growth conditions
Strains used for cloning and expression of recombinant proteins were E. coli TOP10 (Invitrogen) and BL21(DE3)pLysS (Novagen) or BL21(DE3)Star (Novagen) grown in LB medium at 37°C. Media were supplemented with ampicillin (100 µg ml−1) or kanamycin (25 µg ml−1), as required. Mycobacterial strains used were M. smegmatis mc2155, M. bovis BCG Pasteur 1173P2 and M. tuberculosis H37Rv, Erdman and CDC1551. Mycobacteria were grown on Middlebrook 7H11 agar plates with OADC enrichment (Difco) or in Sauton's medium containing 0.05% tyloxapol (Sigma) supplemented with kanamycin (25 µg ml−1) or hygromycin (50 µg ml−1) when required. The shuttle phasmid phAE159 (Bardarov et al., 2002) and the integrative plasmid pMV361 (Stover et al., 1991) were reported earlier.
Cloning, expression and purification of recombinant InhA and mutant proteins
For in vitro kinase assays, the inhA gene was amplified by PCR using M. tuberculosis H37Rv genomic DNA as a template and a set of primers containing NdeI and NheI restriction sites (Table S2). This amplified product was digested with NdeI and NheI and ligated into pETPhos (Canova et al., 2008), generating pETPhos_inhA_WT (Table S1). Site-directed mutagenesis was directly performed on this vector using inverse-PCR amplification with the self-complementary primers (Table S2). All constructs were verified by DNA sequencing. Recombinant InhA proteins were overexpressed in E. coli BL21(DE3) and purified as described (Veyron-Churlet et al., 2010). Fractions containing pure InhA proteins were pooled, dialysed when required and stored at −20°C until further use. For enzymatic or crystallographic studies, inhA_T266A, inhA_T266D and inhA_T266E were amplified from H37Rv genomic DNA using the inhA_forward primer (containing an NdeI site) with the appropriate reverse primer (containing a HindIII site) (Table S2). The amplified products were digested with NdeI and HindIII and ligated into pET30b (Novagen). Mutation sites were verified by DNA sequencing. Expression and purification of InhA_WT (Quemard et al., 1995), InhA_T266A, InhA_T266D and InhA_T266E proteins were performed in E. coli BL21(DE3) as described (Freundlich et al., 2009).
Construction of the integrative pMV361_inhA and cosmid p004_inhA_hemZ
inhA was amplified from H37Rv genomic DNA using primers inhAE1 and inhAH1 (Table S2). The PCR fragment was cut with EcoRI and HindIII and cloned into EcoRI–HindIII-cut pMV361 (Table S1), generating pMV361_inhA. To produce p004_inhA_hemZ, the hemZ gene was PCR-amplified from H37Rv genomic DNA using the RR and RL primers (Table S2). inhA was then PCR-amplified using upstream LL primers and downstream LR-266A, LR-266D or LR-266E primers (Table S2). PCR fragments were cut with Van91I, cloned into Van91I-cut p004 (Table S1). All constructs were sequenced before use.
The transducing mycobacteriophages were prepared as described (Bardarov et al., 2002). Briefly, the recombinant cosmids p004_inhA_hemZ were cut with PacI and ligated to the PacI-cut shuttle phasmid phAE159. The resulting phasmids were packaged in vitro (GigapackII, Strattagene) and transduced into E. coli HB101. Phasmid DNA was electroporated into M. smegmatis mc2155 and the resulting transducing phages were amplified to obtain high-titre phage lysates. M. bovis BCG and M. tuberculosis H37Rv, Erdman and CDC1551 strains (100 ml) were grown to log phase (OD600 about 1.0), spun down, washed twice with mycobacteriophage (MP) buffer (50 mM Tris, 150 mM NaCl, 10 mM MgCl2, 2 mM CaCl2; 100 ml) and finally resuspended in MP buffer (10 ml). The cell suspension (1 ml) was mixed with high-titre phage lysate (1010–1011 pfu ml−1, 1 ml) and incubated at 37°C for 4 h. After centrifugation, the cell pellet was resuspended in Middlebrook 7H9 medium (0.4 ml) and plated onto Middlebrook 7H10 plates containing 75 µg ml−1 hygromycin. Plates were then incubated at 37°C for 4 weeks and transductants screened by PCR using the primer pairs TW361 and TW398 (Table S2) to verify that the inhA–hemZ region had been disrupted and TW 361 and p004R (Table S2) to check the integration of the sacB–hyg cassette between inhA and hemZ. Confirmation of allelic exchange was by Southern blotting on genomic DNA of HygR transductants digested with BamHI and probed with inhA.
In vitro kinase assay
In vitro phosphorylation was performed as described (Molle et al., 2003) with 4 µg of InhA in 20 µl of buffer P (25 mM Tris-HCl, pH 7.0; 1 mM DTT; 5 mM MgCl2; 1 mM EDTA) with 200 µCi ml−1[γ-33P]-ATP corresponding to 65 nM (PerkinElmer, 3000 Ci mmol−1), and 0.2–1.0 µg of kinase in order to obtain for each specific kinase its optimal autophosphorylation activity for 30 min at 37°C. Cloning, expression and purification of the eight recombinant GST-tagged STPKs from M. tuberculosis were described previously (Molle et al., 2006).
Mass spectrometry analysis
Purified InhA_WT and mutant derivatives were subjected to in vitro phosphorylation by GST-tagged PknB as described above, except that [γ-33P]-ATP was replaced with 5 mM unlabelled ATP. Purified InhA from M. bovis BCG cultures was subjected to mass spectrometry without further treatment. Subsequent mass spectrometric analyses were performed as previously reported (Fiuza et al., 2008).
Overexpression of InhA proteins in mycobacteria
inhA was amplified by PCR using M. tuberculosis H37Rv chromosomal DNA as template and a set of primers containing an NdeI and an EcoRI restriction site (Table S2). This amplified product was then digested by NdeI and EcoRI and ligated into the pMK1 expression vector (Table S1). The resulting construct was electroporated into M. bovis BCG and transformants were grown in broth medium and harvested for immunoblotting analysis as described previously (Kremer et al., 2003).
Enoyl reductase activity was monitored spectrophotometrically by following the change in absorption at 340 nm via the oxidation of NADH to NAD+. All reactions were performed with a Cary 100 spectrophotometer at 25°C in 100 mM phosphate buffer pH 7.5 using 80 nM of either wild-type or mutant enzyme. Kinetic parameters Km, kcat and Vmax for DD-CoA were determined at a fixed saturating concentration of NADH (150 µM) while varying the concentration of DD-CoA (25–100 µM) using the equation v = kcat[E]0[S]/(Km + [S]). The percentage of remaining activity was calculated by comparing the rate of the reaction of the mutant enzyme to the rate of the reaction of the wild-type enzyme in the presence of 50 µM DD-CoA and 100 µM NADH with wild-type enzyme activity set at 100%.
Binding constants of NADH to the wild-type, Asp and Glu mutant enzymes were obtained by measuring the protein fluorescence using a Cary Eclipse spectrofluorometer at 25°C. Excitation and emission wavelengths were 280 nm and 335 nm respectively. Emission was first monitored between 300 and 450 nm and 335 nm was selected as λmax. Approximately 0.7–0.76 µM protein solutions in 100 mM Pipes buffer pH 7.0 were titrated with various concentrations of NADH (0–20 µM). The contribution of NADH fluorescence to final fluorescence intensity was checked with a control experiment by titrating NADH into buffer in the absence of protein. The fluorescence intensities at λmax for a corresponding NADH concentration were corrected for dilution factors and NADH contributions before any calculations. At the concentrations used for InhA and NADH, the fluorescence intensity change due to inner filter effect was negligible. Kd values were calculated using the equation ΔF = ΔFmax − Kd*(ΔF/[NADH]), where F represents fluorescence intensity (Li and Lin, 1996).
Complementation studies of the inhA(Ts) M. smegmatis strain
Cultures of M. smegmatis strains mc22359 (InhA(Ts), INH-resistant) were grown at 30°C. Competent bacteria were prepared and transformed with pMK1, pMK1_inhA_WT, pMK1_inhA_T266A, pMK1_inhA_T266D or pMK1_inhA_T266E. Clones selected on kanamycin were grown in Sauton medium at 30°C to mid-log phase and plated at 30°C or 42°C for 3–5 days.
Mycolic acid biosynthesis
Cultures of M. smegmatis mc22359 (InhA(Ts), INH-resistant) were grown to mid-log phase in Sauton medium at 30°C. The temperature was then shifted to 42°C for 3 h and cultures were labelled by adding [1-14C]-acetate (1 µCi ml−1). After a further 3 h incubation at 42°C, cells were harvested and FAMEs and MAMEs were extracted as reported (Kremer et al., 2002). Equal amounts of counts were subjected to TLC using petroleum ether/acetone (95/5, v/v) and exposed overnight to a Kodak X-Omat film. As a control of mycolic biosynthesis inhibition, M. smegmatis mc2155 (InhA_WT, INH-sensitive) was treated with 50 µg ml−1 INH for 3 h and labelled by adding [1-14C]-acetate (1 µCi ml−1) for another 3 h at 37°C prior to mycolic acid extraction and analysis.
We thank M. Becchi and A. Cornut (Institut de Biologie et Chimie des Protéines, Lyon) for excellent technical assistance in mass spectrometry analysis and Sir David Hopwood for critical reading of the manuscript. This work was supported by grants from the National Research Agency (ANR-09-MIEN-004) to V.M. and L.K.