The Mycobacterium tuberculosis FAS-II condensing enzymes: their role in mycolic acid biosynthesis, acid-fastness, pathogenesis and in future drug development

Authors

  • Apoorva Bhatt,

    1. School of Biosciences, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK.
    Search for more papers by this author
  • Virginie Molle,

    1. Institut de Biologie et Chimie des Protéines, CNRS UMR 5086, Université de Lyon I, IFR128 BioSciences, Lyon-Gerland, 7 passage du Vercors, 69367 Lyon Cedex 07, France.
    Search for more papers by this author
  • Gurdyal S. Besra,

    1. School of Biosciences, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK.
    Search for more papers by this author
  • William R. Jacobs Jr,

    1. Howard Hughes Medical Institute, and
    2. Department of Microbiology and Immunology, Albert Einstein College of Medicine, Bronx, NY 10461, USA.
    Search for more papers by this author
  • Laurent Kremer

    Corresponding author
    1. Laboratoire de Dynamique des Interactions Membranaires Normales et Pathologiques, Université de Montpellier II et I, CNRS; UMR 5235, case 107, Place Eugène Bataillon, 34095 Montpellier Cedex 05, France.
    2. INSERM, DIMNP, Place Eugène Bataillon, 34095 Montpellier Cedex 05, France.
    Search for more papers by this author

E-mail: laurent.kremer@univ-montp2.fr; Tel. (+33) 4 67 14 33 81; Fax (+33) 4 67 14 42 86.

Summary

Mycolic acids are very long-chain fatty acids representing essential components of the mycobacterial cell wall. Considering their importance, characterization of key enzymes participating in mycolic acid biosynthesis not only allows an understanding of their role in the physiology of mycobacteria, but also might lead to the identification of new drug targets. Mycolates are synthesized by at least two discrete elongation systems, the type I and type II fatty acid synthases (FAS-I and FAS-II respectively). Among the FAS-II components, the condensing enzymes that catalyse the formation of carbon-carbon bonds have received considerable interest. Four condensases participate in initiation (mtFabH), elongation (KasA and KasB) and termination (Pks13) steps, leading to full-length mycolates. We present the recent biochemical and structural data for these important enzymes. Special emphasis is given to their role in growth, intracellular survival, biofilm formation, as well as in the physiopathology of tuberculosis. Recent studies demonstrated that phosphorylation of these enzymes by mycobacterial kinases affects their activities. We propose here a model in which kinases that sense environmental changes can phosphorylate the condensing enzymes, thus representing a novel mechanism of regulating mycolic acid biosynthesis. Finally, we discuss the attractiveness of these enzymes as valid targets for future antituberculosis drug development.

Introduction

Mycolic acids are a homologous series of C60–C90α-alkyl β-hydroxy fatty acids produced by all mycobacteria, and similar but shorter-length mycolic acids are found in related genera like Corynebacterium and Nocardia. They are found, primarily, as esters of the non-reducing arabinan terminus of arabinogalactan (Brennan and Nikaido, 1995) but also present as extractable ‘free’ lipids within the cell wall, mainly associated to trehalose to form trehalose dimycolate (TDM), also known as the ‘cord factor’. TDM represents the most abundant, granulomagenic, and significant toxic lipid extractable from the cell surface of virulent Mycobacterium tuberculosis (Hunter et al., 2006a).

Mycolates comprise a meromycolate chain (up to C56) and a long saturated α-branch (C24–C26) and can be distinguished according to the chemical modifications that decorate the meromycolate. α-Mycolic acids do not contain any oxygen functionality other than the β-hydroxyl group, but possess two cyclopropanation rings and therefore are different from oxygenated mycolic acids such as keto- and methoxy-mycolates (Brennan and Nikaido, 1995). The oxygenated mycolates possess polar modifi cations containing oxygen functions in the distal portion of the meromycolate chain, whereas non-polar modifications, i.e. cyclopropane rings and unsaturations, are found in both the proximal (closest to the β-hydroxy group) and distal positions of the chain (Kremer et al., 2000a). Mycolic acids give rise to important characteristics, including resistance to chemical injury, resistance to dehydration, low permeability to hydrophobic antibiotics, virulence (Dubnau et al., 2000; Glickman et al., 2000; Glickman and Jacobs, 2001), acid-fast staining (Bhatt et al., 2007), biofilm formation (Ojha et al., 2005) and the ability to persist within the host (Daffe and Draper, 1998; Yuan et al., 1998; Glickman et al., 2000; Bhatt et al., 2007).

In addition, the enzymes involved in mycolate biosynthesis are essential for mycobacterial survival and thus represent excellent drug targets.

The biosynthetic pathway of mycolates involves two types of fatty acid-synthesizing systems, the type I and type II fatty acid synthases, FAS-I and FAS-II respectively. The eukaryotic-like FAS-I catalyses the de novo synthesis of fatty acids from acetyl-CoA. In contrast, FAS-II is similar to systems found in bacteria, apicomplexa parasites and plants, and is composed of four dissociable enzymes that act successively and repetitively to elongate the growing acyl chain. Acyl-primers are continually activated via a thioester linkage to the prosthetic group of coenzyme A (CoA) for FAS-I, or of an acyl carrier protein (ACP) for FAS-II. The biosynthesis involves five distinct stages, which have been detailed in a recent review (Takayama et al., 2005) and are illustrated in Fig. 1: (i) the synthesis of the C26 saturated straight chain fatty acids by FAS-I to provide the α-alkyl branch of the mycolic acids, (ii) the synthesis of the C56 fatty acids by FAS-II providing the meromycolate backbone, (iii) the introduction of functional groups to the meromycolate chain by numerous cyclopropane synthases, (iv) the condensation reaction catalysed by the polyketide synthase Pks13 between the α-branch and the meromycolate chain before a final reduction to generate the mycolic acid and (v) the transfer of mycolic acids to arabinogalactan and other acceptors such as trehalose via the Antigen 85 complex.

Figure 1.

Fatty acid/mycolic acid biosynthesis in mycobacteria. FAS-I is involved in the synthesis of C16 and C26. The C16 acyl-CoA product acts as a substrate for the synthesis of meromycolic acids by FAS-II, whereas the C26 fatty acid constitutes the α-branch of the final mycolic acid. MtFabH has been proposed to be the link between FAS-I and FAS-II by converting C14-CoA generated by FAS-I to C16-AcpM, which is channelled into the FAS-II cycle. This latter comprises four enzymes which will act successively and repeatedly to ensure fatty acid elongation, ultimately leading to meromycolates (C56). These enzymes are the condensing enzymes KasA and KasB, the keto-reductase MabA, an unidentified dehydratase, and the enoyl-reductase InhA. Finally, the polyketide synthase Pks13 catalyses the condensation of the α-branch and the meromycolate to produce mycolic acids. Targets for the action of activated isoniazid (INH), ethionamide (ETH), triclosan (TRC), or thiolactomycin (TLM) are indicated. FAS-II enzymes are labelled in black, excepted the condensing enzymes, which are indicated in red. The relative contribution of FAS-I and FAS-II activities in fatty acid/mycolic acid biosynthesis is represented in green and purple respectively.

The condensation reaction that leads to carbon-carbon bond formation during acyl-group elongation is an extremely important step that is catalysed by condensing enzymes (Heath and Rock, 2002). These are of three different types, depending on the step of the elongating process in which they are involved. The initiation condensing enzyme mtFabH, which has been proposed as the pivotal link between FAS-I and FAS-II, uses acyl-CoA primers, whereas the elongating condensing enzymes KasA and KasB exclusively use acyl-ACP thioesters. The termination condensing enzyme Pks13 uses both a meromycolyl-AMP and an acyl-CoA as a substrate. Although they all catalyse carbon-carbon bond formation, they are defined by differences in their active sites, chain-length specificity and sensitivity to inhibitors.

This review reports the recent advances made in the current understanding of the condensing enzymes participating in mycolic acid synthesis in M. tuberculosis. Recent data shedding light on the structure and mechanisms of this important family of enzymes are covered, so is their role in mycobacterial growth and virulence. This review also discusses potential of the condensing enzymes as attractive drug targets for future drug development against tuberculosis.

Type I fatty acid synthase system (FAS-I) and short-chain fatty acid biosynthesis

In M. tuberculosis, the fas gene (Rv2524c) encodes the multifunctional FAS-I polypeptide that contains all the functional domains required for de novo fatty acid synthesis (Smith et al., 2003). These domains are organized in the following order: acyltransferase, enoyl reductase, dehydratase, malonyl/palmitoyl transferase, acyl carrier protein, β-ketoacyl reductase and β-ketoacyl synthase. All intermediates generated remain enzyme-bound during the process of elongation and undergo transacylation to other catalytic sites within the enzyme. FAS-I generates short-chain fatty acyl-CoA primers that are further elongated by FAS-II. It is thought that FAS-I also synthesizes the hexacosanoyl-S-CoA (α-branch) for the final condensation reaction catalysed by Pks13 (Fig. 1). Although two studies suggest that the drug pyrazinamide targets FAS-I (Zimhony et al., 2000; Zimhony et al., 2007) it was also demonstrated that pyrazinoic acid (the active form of pyrazinamide) also targets the membrane and interferes with the energetics and function of the membrane (Zhang et al., 2003).

Type II fatty acid synthase II (FAS-II) and long-chain fatty acid biosynthesis

The M. tuberculosis FAS-II system is analogous to other bacterial FAS-II systems, except that it is not capable of de novo synthesis. Instead, mycobacterial FAS-II elongates acyl-CoA primers generated by FAS-I to a mixture of homologous fatty acids of longer chain lengths (meromycolic acids). The basic FAS-II cycle is illustrated in Fig. 1. Elongation requires the substrate malonyl-AcpM (synthesized from malonyl-CoA and phosphopantothenylated holo-AcpM by the malonyl-CoA : AcpM transacylase mtFabD (Kremer et al., 2001). The function of AcpM is to shuttle acyl intermediates between enzymes. Acyl-CoA primers undergo a condensation reaction with malonyl-AcpM in a reaction catalysed by mtFabH. Next, the acyl-AcpM precursor undergoes a cycle of keto-reduction, dehydration and enoyl-reduction catalysed by a β-ketoacyl-AcpM reductase (MabA, FabG1) (Marrakchi et al., 2002), a β-hydroxyacyl-AcpM dehydratase and an enoyl-AcpM reductase (InhA) (Quemard et al., 1995) respectively. The AcpM-bound acyl chain (which is now two-carbon units longer) then undergoes several iterative cycles of reduction except that mtFabH is replaced by the β-ketoacyl-AcpM synthases KasA or KasB (Bhatt et al., 2005; Bhatt et al., 2007; Kremer et al., 2002; Schaeffer et al., 2001; Slayden and Barry, 2002).

Elongation by the FAS-II system is achieved by a condensation reaction that occurs in three distinct steps (Fig. 2): (1) transfer of the pantotheine-bound acyl primer to a cysteine residue within the active site of mtFabH, KasA or KasB, (2) decarboxylation of the donor malonyl-AcpM substrate to yield an acetyl-AcpM carbanion and (3) nucleophilic attack of the carbanion on the carbonyl group on the enzyme-bound primer to yield the elongated β-ketoacyl-AcpM product.

Figure 2.

Condensation reaction in fatty/mycolic acid biosynthesis. The elongation condensing enzymes (KasA or KasB) catalyse a three-step reaction. The first substrate is an acyl-AcpM, covalently transferred to the active site cysteine, with a concomitant release of reduced AcpM. Malonyl-AcpM then binds and is decarboxylated, allowing the condensation reaction to occur. The final product, a β-ketoacyl-AcpM with two additional carbons, is then released. For the initiating condensing enzyme mtFabH, the reaction is identical excepted that the first substrate is an acyl-CoA.

The FAS-II condensing enzymes

The initiation β-ketoacyl-ACP synthase III (mtFabH)

The role of mtFabH (Rv0533c) is to catalyse the condensation of long-chain acyl-CoA primers, formed by FAS-I, and malonyl-AcpM and to funnel the generated precursors into the FAS-II system. In doing so, mtFabH represents the pivotal link between the two FAS systems thus initiating meromycolic acid biosynthesis (Choi et al., 2000). The proposed function of mtFabH is supported by the presence of a conserved catalytic triad (Cys112, His244 and Asn274) characteristic of β-ketoacyl-ACP synthase III enzymes, and by its ability to catalyse the condensation of acyl-CoA with malonyl-AcpM to generate a β-ketoacyl-AcpM product (Choi et al., 2000). Purified recombinant mtFabH exhibits a preference for acyl-CoA as a substrate, rather than acyl-ACP primers, and condenses lauroyl-CoA (C12) and myristoyl-CoA (C14) to generate myristoyl-ACP and palmitoyl-ACP (C16), respectively, using malonyl-ACP as the cosubstrate. Compared with the general Escherichia coli FabB/FabH type of β-ketoacyl synthases, mtFabH exhibits high substrate specificity, consistent with structural studies (Scarsdale et al., 2001). Like the E. coli enzyme, mtFabH contains a CoA/malonyl-ACP binding channel that runs from the surface of the protein to the cysteine residue within the active site. MtFabH also contains a second hydrophobic pocket leading from the active site, which is blocked by Thr87. The difference in these amino acid residues is postulated to account for the difference in substrate specificities between the two enzymes; Phe87 constrains specificity to acetyl-CoA in E. coli while Thr87 allows binding of long-chain acyl CoAs in mtFabH. This second hydrophobic pocket in mtFabH is capped by an α helix, which restricts the bound acyl chain length to 16 carbons, thus excluding longer chain acyl-CoA products (C24–C26) from chain elongation. In addition, several residues influencing catalysis and substrate specificity of mtFabH have recently been assigned through the combination of structural studies and site-directed mutagenesis (Brown et al., 2005).

These observations are consistent with the proposed role of mtFabH as the initiator of mycolic acid elongation and clearly distinguish it from the chain extension steps catalysed by KasA/KasB.

The elongation β-ketoacyl-ACP synthases (KasA and KasB)

Two genes, kasA (Rv2245) and kasB (Rv2246), which are present in the same operon, encode two distinct β-ketoacyl-AcpM synthases (KasA and KasB) that catalyse the initiation of subsequent rounds of acyl extension by FAS-II (Fig. 1). Both KasA and KasB catalyse the condensation of acyl-AcpM and malonyl-AcpM, hence elongating the growing meromycolate chain by a further two carbon units (Kremer et al., 2000b; Schaeffer et al., 2001). KasA and KasB share many similarities, including the specificity for long-chain acyl-AcpM primers (Schaeffer et al., 2001). The Kas enzymes, ecFabB/ecFabF and KasA/KasB share substantial sequence similarity (Kremer et al., 2000b): KasA and KasB are 67% identical, and are 92% and 91% identical with their respective homologues in Mycobacterium leprae. KasA and KasB share also 28% and 29% sequence identity with the keto-synthase domain of Pks13 respectively. A three-dimensional model of KasA, based on the crystallographic co-ordinates of the E. coli FabF protein, reveals the putative residues of the catalytic triad in the active site (Kremer et al., 2000b). Participation of Cys171, His311, Lys340 and His345 in catalysis was subsequently confirmed by replacing these residues with alanine, which abolishes the overall elongation activity of KasA (Kremer et al., 2002).

Although there is some redundancy, KasA might catalyse the initial elongation reactions while KasB extends the elongation to full-length mycolates. Cell lysates of Mycobacterium smegmatis overproducing KasA generate C40-monounsaturated fatty acids, whereas cell lysates of M. smegmatis overproducing both KasA and KasB generate C54-multiunsaturated fatty acids (Slayden and Barry, 2002). Disruption of kasB by transposon mutagenesis in Mycobacterium marinum results in a mutant producing meromycolate chains that are two to four carbons shorter than wild-type mycolates, with a significant reduction of keto-mycolates (Gao et al., 2003). Furthermore, recent work has demonstrated that while disruption of kasB in M. tuberculosis generates similar shortened mycolates, an unexpected outcome of kasB deletion was the loss of keto-mycolic acid trans-cyclopropanation and a drastic reduction in methoxy-mycolic acid trans-cyclopropanation, activities that are usually associated with the trans-cyclopropane synthase CmaA2 (Bhatt et al., 2007).

Together, these results indicate that KasA and KasB function independently on separate sets of substrates and that KasB is required for the full extension of the mycolic acids.

The termination polyketide synthase (Pks13)

The condensation of C26-S-CoA and meromycolyl-AMP has been used to describe the final step in the synthesis of mycolic acids in M. tuberculosis (Takayama and Qureshi, 1984; Trivedi et al., 2004; Takayama et al., 2005; Gokhale et al., 2007). The participation of Pks13 (Rv3800) in this final step in mycolate assembly has been demonstrated (Gande et al., 2004; Portevin et al., 2004; Gokhale et al., 2007). In M. tuberculosis, Pks13 is a member of the type I polyketide synthase family. Sequence analysis reveals the presence of two non-equivalent phosphopantothein-binding (PPB), ketoacyl synthase (KS), acyl transferase (AT) and thioesterase (TE) domains. A high degree of control is placed over the entire condensation reaction. A specific fatty acyl-AMP ligase (FadD32) converts each meromycolyl-S-AcpM derived from the FAS-II system to meromycolyl-AMP (Trivedi et al., 2004). Following release from FAS-I, the C26-S-CoA is carboxylated by the acyl-CoA carboxylases AccD4 and AccD5 to yield 2-carboxyl-C26-CoA. The first reaction is the binding of the C52-meromycolyl (from its AMP derivative) and the 2-carboxyl-C26-acyl group (from its CoA derivative) to the N-terminus of the PPB domain and the near C-terminus PPB domain of Pks13, respectively, as thioesters (Takayama et al., 2005). The second reaction involves the transfer of the meromycolyl group from the N-terminal PPB domain to the KS domain. The third reaction corresponds to condensation of the two fatty acyl groups, characterized by a nucleophilic attack on the carbonyl group of the meromycolyl-S-KS by the carboxylate of the 2-carboxyl-C26-S-PPB. This results in the formation of 3-oxo-C78-mycolate bound to the near C-terminal and the release of CO2. Finally, a reduction step, proposed to be catalysed by the reductase Rv2509 converts the 3-oxo intermediate to a secondary alcohol to generate the mature C78-mycolate (Lea-Smith et al., 2007).

Disruption of pks13 in Corynebacterium glutamicum leads to a mutant unable to synthesize corynomycolic acids which accumulates large amounts of fatty acid precursors (Gande et al., 2004). The functional role of Pks13 has been investigated through the use of a conditional mutant of M. smegmatis. When grown at a non-permissive temperature, mycolic acid synthesis is severely impaired (Portevin et al., 2004). Because pks13 is essential in mycobacteria, it represents an attractive drug target. However, development of an in vitro assay useful for high throughput screening will be very challenging, given the complexity of the substrates required.

The multiple interconnected FAS-II systems and their specific condensing enzymes

An attractive model based on the interaction between the various FAS-II components, in which different specialized FAS-II complexes are interconnected, has been proposed (Veyron-Churlet et al., 2004). In particular, this model predicts the occurrence of four FAS-II systems involved in mycolic acid biosynthesis represented in Fig. 3: (i) the ‘initiation FAS-II’ (I-FAS-II) formed by a core and mtFabH representing the pivotal link between FAS-I and FAS-II; (ii) two ‘elongation FAS-II’ complexes, the first ‘elongation-1 FAS-II’ (E1-FAS-II) consisting of a core and KasA capable of elongating acyl-AcpM originating from the I-FAS-II complex to produce immature meromycolates, and the ‘elongation-2 FAS-II’ (E2-FAS-II) characterized by a core and KasB that converts the acyl-AcpM from E1-FAS-II to full-length meromycolyl-AcpM; and finally (iii) the ‘termination FAS-II’ (T-FAS-II) that involves Pks13 and which condenses the α-branch with the meromycolate. This model also opens new perspectives for the search of drugs against mycobacterial infections. Indeed, through the design of point mutations of MabA, it was demonstrated that protein–protein interactions within the FAS-II system of M. tuberculosis are essential for mycobacterial viability (Veyron-Churlet et al., 2004). Inhibitory peptide aptamers capable of disrupting specific and essential interactions have been developed and optimized in different cellular processes (Colas et al., 1996). Thus, aptamers capable of preventing protein–protein interactions within the various FAS-II systems represent a novel approach of inhibiting M. tuberculosis growth.

Figure 3.

Model for the regulation of mycolic acid initiation, elongation and termination by phosphorylation of the condensing enzymes. Changes in cell wall and mycolic acid composition to various environmental stimuli are central to M. tuberculosis adaptation during infection. Because most mycobacterial Ser/Thr protein kinases (STPKs) contain a transmembrane domain, it is very likely that environmental stimuli are transduced within the bacteria by the extracellular sensor domain. When sensing external stimuli, STPKs (such as PknA) are then phosphorylated and transfer the signal to the FAS-II condensing enzymes by means of transphosphorylation, which inhibits KasA activity, or increases KasB activity. KasA/KasB can also be dephosphorylated by the Ser/Thr phosphatase PstP. In addition to KasA/B, recent work has revealed that mtFabH and Pks13 are also phosphorylated (V. Molle and L. Kremer, unpublished data). Whether STPK-dependent phosphorylation of mtFabH and Pks13 induces either positive or negative signalling to the different interconnected specialized FAS-II complexes remains under investigation. I-FAS-II, initiation complex; E1-FAS-II, elongation 1 complex; E2-FAS-II, elongation 2 complex and T-FAS-II, termination complex.

The possible existence of such multiple interconnected FAS-II systems may also offer an explanation for the loss/reduction of trans-cyclopropanation in the M. tuberculosis kasB mutant. It is likely that the trans-cyclopropane synthase CmaA2 interacts preferentially with the KasB-containing E2-FAS-II and thus does not function efficiently in a kasB null mutant.

Phosphorylation of the condensing enzymes by mycobacterial Ser/Thr protein kinases

Reversible protein phosphorylation is a key mechanism by which environmental signals are transmitted to cause changes in protein expression or activity in both eukaryotes and prokaryotes. Protein phosphorylation is an important mechanism by which extracellular signals are translated into cellular responses. Genes encoding functional serine/threonine protein kinases (STPKs) are ubiquitous in prokaryotic genomes, and signalling through Ser/Thr phosphorylation has emerged as a critical regulatory mechanism in various microorganisms. In bacteria, the presence of several STPKs suggests a central role of protein phosphorylation in regulating various biological functions, ranging from environmental adaptive responses to bacterial pathogenicity. Genome sequence analysis predicts the presence of 11 different STPKs in M. tuberculosis (Cole et al., 1998; Av-Gay and Everett, 2000) and some of them have been investigated for their physiological roles. Moreover, recent studies have demonstrated that kinases PknG and PknH are implicated in the pathogenesis and survival of M. tuberculosis within the host (Walburger et al., 2004; Papavinasasundaram et al., 2005). Consequently, deciphering STPK-dependent phosphosignalling systems has been the focus of many recent studies (Greenstein et al., 2005). Therefore, regulation of cell wall biosynthesis by STPKs might be important to understand several physiological processes such as those related to the slow growth rate of M. tuberculosis and its ability to survive within the phagosome, or to enter into dormancy.

In a recent study, Molle et al. (Molle et al., 2006) proposed a model for mycolic acid synthesis regulation based on the phosphorylation of the β-ketoacyl AcpM synthases KasA and KasB (Fig. 3). Both enzymes were found to be phosphorylated at high rates in vivo in Mycobacterium bovis BCG, mainly at two positions, and could be dephosphorylated by the mycobacterial Ser/Thr phosphatase PstP. Interestingly, phosphorylation of KasA and KasB differentially modulates their condensing activities (Molle et al., 2006). While phosphorylation reduces the condensation activity of KasA, it increases the activity of KasB in the presence of either malonyl-AcpM or C16-AcpM. The differential effect of phosphorylation of two highly similar enzymes sharing the same enzymatic activity, but with different substrate specificities, is rather unexpected. Figure 3 presents an unusual mechanism of regulation where damping KasA might allow time to produce immature mycolates. Conversely, boosting KasB activity might ensure that mycobacteria produce only full-length mycolates required for intracellular survival and virulence (Bhatt et al., 2007). Thus, the phosphorylation model of regulation suggests that STPK-dependent phosphorylation can induce either positive or negative signalling to the different interconnected FAS-II complexes (Fig. 3). This is supported by our recent observation that, in addition to KasA and KasB, the two other condensases mtFabH and Pks13 are also substrates of mycobacterial STPKs (V. Molle and L. Kremer, unpublished data). The differential expression of the mycobacterial STPKs in response to stress conditions might directly affect the phosphorylation profile of these substrates, and as a consequence modulate mycolic acid biosynthesis in order to promote adaptation to environmental changes. In addition, interfering with these regulatory processes, such as selective inhibition of FAS-phosphorylation, might pave the way for future drug development against tuberculosis. This notion is supported by the fact that specific inhibitors of protein kinases have been successfully developed for therapeutic use against a variety of diseases (Shawver et al., 2002).

Role of KasA in biofilm formation

Mycobacterium smegmatis, which can form biofilms when grown in specialized biofilm media, shows a prevalence of C56–C68 fatty acids during biofilm growth (Ojha et al., 2005), suggesting a role for FAS-II enzymes in biofilm formation. The study of a biofilm-deficient M. smegmatis GroEL1 mutant (Ojha et al., 2005) revealed that KasA plays a role in biofilm formation: when the ΔgroEL1 mutant was seeded into biofilm medium, levels of KasA reduced drastically after 4 days of culture, accompanied by a decrease in all classes of mycolic acids. The authors also showed that in 2D gels, two isoforms of KasA can be detected and the levels of the more basic isoform varies under different growth conditions. Additionally, overproduction of KasA in wild-type M. smegmatis delays biofilm formation, indicating that transition to a ‘biofilm phase’ involves reduction in levels of the KasA protein. Furthermore, KasA interacts physically with GroEL1 in biofilms, but not in planktonic cultures, and this association seems specific as no interaction was observed between GroEL1 and InhA (Ojha et al., 2005). It has therefore been proposed that GroEL1 promotes a biofilm-specific FAS-II complex. The role of KasA and other FAS-II enzymes in biofilm formation in slow growers, M. bovis and M. tuberculosis remains to be investigated.

Role of KasB in mycobacterial virulence and dormancy

Deletion of kasB in M. marinum and M. tuberculosis resulted in the loss of cording, a property often associated with virulence, and acid-fast staining, which remains the cornerstone of tuberculosis diagnosis (Gao et al., 2003; Bhatt et al., 2007). The effect of loss of KasB function on mycobacterial virulence was first assessed in M. marinum (Gao et al., 2003). The authors first showed that the transposon-disrupted M. marinum kasB mutant is more sensitive, in vitro, to lysozyme and human neutrophil defensin peptide 1. Furthermore, infection of bone marrow-derived macrophages with the mutant strain results in partial loss of lysosome–phagosome fusion inhibition that is normally observed with wild-type M. marinum. Nearly 30–40% of intracellular kasB mutants colocalize with the lysosome, while > 95% of wild-type bacteria remained in the phagosome. The M. tuberculosis CDC1551 ΔkasB mutant displays a number of interesting phenotypes following aerosol infection into immunocompetent C57BL/6 mice (Bhatt et al., 2007) (Fig. 4). Although initially successful in colonizing the lungs, liver and spleen, the ΔkasB mutant fails to replicate to the levels normally observed in these tissues, and does not cause the pathology usually associated with M. tuberculosis infection. Furthermore, strikingly, the ΔkasB mutant is able to persist at constant low levels in lungs and spleen for 450 days post aerosol infection. It also fails to cause active disease as ΔkasB-infected mice appear healthy even 600 days post infection. In contrast to immunocompetent mice, SCID mice succumb to infection by the ΔkasB mutant (Bhatt et al., 2007). TDM, also known as the ‘cord factor’, is believed to play a key role in colony morphology of virulent M. tuberculosis and in cording (Hunter et al., 2006b). It is a highly granulomagenic and bioactive lipid that is extractable from the surface of virulent M. tuberculosis (Hunter et al., 2006a). TDM has also recently been shown to modulate the host immune response (Rao et al., 2005; Rao et al., 2006). Loss of cording is often correlated with decreased virulence (Glickman et al., 2000), and this is also the case with the ΔkasB mutant. Although it remains to be demonstrated, it is likely that the observed attenuation of the ΔkasB mutant is due to changes in modulation of both innate and adaptive immune responses by altered mycolic-acid-containing TDMs.

Figure 4.

Comparison of the biochemical, structural and biological features of KasA and KasB in M. tuberculosis. The figure shows the kasA operon and highlights the major characteristics of KasA and KasB. The chemical structure of TLM which inhibits both enzymes is also shown. The three insets are light microscopy pictures of M. tuberculosis strains fixed on glass slides. Acid-fast staining was performed on the fixed smears using the Kinyoun stain kit. The M. tuberculosis CDC1551 wild-type and complemented kasB mutant strains form serpentine cords (cording) and are stained red (Basic Fuschin). In contrast, the kasB null mutant (ΔkasB) fails to form cords and retains the primary Basic Fuschin stain following acid-alcohol decolourization (Magnification = 400×).

Proteomic analysis has shown that KasB is upregulated in M. tuberculosis cultures growing under anaerobic conditions (Starck et al., 2004). This finding emphasizes the fact that metabolic activity related to mycolic acid biosynthesis occurs under anaerobic conditions. This might result in modulation of mycolic acid chain length during a dormant or persistent anaerobic state. However, the role and participation of KasB in mycobacterial dormancy requires to be clearly addressed.

Together, these findings demonstrate that KasB function is crucial for virulence, and highlight the potential of the ΔkasB mutant as a model for investigating latent infection and to study drugs acting during dormancy (Fig. 4). Moreover, this attenuated strain might represent a valuable vaccine candidate against tuberculosis. In addition, the kasB mutant was found to be more sensitive to lipophilic antibiotics, thus emphasizing the attractiveness of KasB as an important secondary drug target in combination therapy.

Both KasA and KasB are candidate targets for drug development

The inability to delete kasA, but not kasB, in M. smegmatis indicates that the former is an essential gene. This has been proven using an improvised gene essentiality testing tool termed Conditional Expression Specialized Transduction Essentiality Test or CESTET (Bhatt et al., 2005). Transductants (gene knockouts of the native chromosomal copy of kasA) could be obtained in a merodiploid strain containing a second integrated copy of kasA under the control of the acetamide-inducible acetamidase promoter. Furthermore, the effects of KasA depletion could be monitored by culturing the transductants in the medium devoid of acetamide. Reduction in the levels of KasA causes a rapid decrease in mycolic acid biosynthesis and leads to bacterial lysis (Bhatt et al., 2005). Notably, the lytic phenotype that occurs following KasA depletion is different from that observed in INH-treated cultures (Bhatt et al., 2005). These attributes of KasA highlights its potential as a candidate drug target.

Although KasB is not essential in M. smegmatis, M. marinum or M. tuberculosis, it can still be used as a secondary drug target for two reasons. First, the ‘hypovirulence’ of the kasB mutant in mice suggests that targeting the second β-ketoacyl AcpM synthase would attenuate bacteria in the infected host and allow for rapid and efficient clearance of the bacteria when used in combination with current drug therapies. Second, the kasB mutant was more sensitive to lipophilic antibiotics, which means that inhibiting KasB would lead to increased susceptibility to antibiotics like rifampicin. Thus, inhibitors of KasB could be envisioned as inhibitors of full-length meromycolates synthesis, which would attenuate M. tuberculosis, while at the same time rendering the bacterium even more susceptible to lipophilic drugs such as rifampicin.

Drugs inhibiting the condensation reaction steps

Inhibition of mycolic acid biosynthesis is effective in the combat against mycobacterial infection and, given the importance of the elongation step in mycolic acid biosynthesis, the condensing enzymes represent an attractive and potent drug target. The fact that several natural products target the elongation step in fatty acid biosynthesis clearly emphasizes the condensing enzymes as desirable targets for drug development (Heath and Rock, 2004). Cerulenin, produced by Cephalosporium caerulens, inhibits both KasA and KasB activity (Schaeffer et al., 2001; Kremer et al., 2002). It has a 12-carbon acyl chain that associates with the hydrophobic channel that accommodates the hydrocarbon chain of the acyl enzyme intermediate (Price et al., 2001). However, the natural molecule receiving the most interest is thiolactomycin (TLM), composed of a thiolactone ring, which is produced by Nocardia (Hayashi et al., 1984) (Fig. 5). In contrast to cerulenin, TLM binds to the other side of the condensing enzyme active site in a position occupied by malonyl-ACP. Initial work has shown that TLM blocks mycobacterial growth by inhibiting mycolic acid biosynthesis (Slayden et al., 1996). It was subsequently demonstrated that TLM targets both M. tuberculosis KasA and KasB. These enzymes also share significant differences in sensitivity to TLM, with KasA being the most sensitive (Kremer et al., 2000b; Schaeffer et al., 2001). Although TLM itself has only modest antibacterial activity, the intense interest in TLM derives from its favourable physical properties, its effectiveness in mouse infection models, and its broad spectrum of activity against important pathogens such as M. tuberculosis and Plasmodium falciparum. TLM analogues with different side chains at 5-position of the thiolactone ring (Fig. 5) show increased potency and more desirable pharmacokinetic properties (Kremer et al., 2000b; Senior et al., 2003; 2004; Kim et al., 2006). 4-O-alkyl analogues (Fig. 5) have also been reported and exhibit useful whole cell growth inhibition of drug resistant strains of M. tuberculosis (Kamal et al., 2005). However, because the chemical synthetic routes to make TLM are difficult, the discovery of new chemical scaffolds is needed to develop potent condensing enzyme inhibitors.

Figure 5.

Chemical structures of the major mycobacterial condensing enzyme inhibitors. TLM and its related molecules with substitutions at position 4 (4-O-alkyl analogues), or 5 (acetylene and biphenyl-based analogues) in the thiolactone ring are shown as well as OSA.

The X-ray structure of KasB from M. tuberculosis revealed a thiolase fold, like other KAS-II enzymes. However, it contains unique structural features in the capping region that might explain its higher preference for longer fatty acyl chains compared with its counterparts from other bacteria (Sridharan et al., 2007). Modelling of TLM binding in KasB shows that the drug fits the active site poorly. The structure of KasB highlights the potential of TLM as an antitubercular lead compound, and will be hoped to provide further exploration of the TLM scaffold towards the design of more efficient compounds. A structural model of KasA based on the KasB co-ordinates, shows the overall structural identity of the two enyzmes. However, a larger entrance to the active site tunnel was predicted in KasA, which might contribute to its greater sensitivity to TLM (Sridharan et al., 2007).

Biphenyl-based 5-substituents as well as acetylene-based analogues of TLM (Fig. 5) exhibit increased potency in vitro against mtFabH, compared with TLM (Senior et al., 2003, 2004), although no whole cell activity is reported. However, no existing drugs or new compounds have been shown to specifically target mtFabH. A key feature in developing novel antimycobacterial agents is the understanding of distinct characteristics that separate mtFabH from other condensing enzymes. It is noteworthy that several species, including M. smegmatis or M. leprae, lack a fabH-like gene. This raises the question of the importance of mtFabH, which is supported by the fact that this gene is not essential, as revealed by Himar-1-based transposon mutagenesis in H37Rv and CDC1551 (Lamichhane et al., 2003; Sassetti et al., 2003). Therefore, mtFabH still remains to be validated as a drug target.

β-Sulfonylcarboxamide compounds were designed as potential inhibitors of β-ketoacyl-ACP synthases of pathogenic mycobacteria by acting as mimics of the putative transition state in the condensation reaction (Jones et al., 2000). These compounds were particularly specific, showing no activity against other bacteria or even non-pathogenic fast-growing mycobacteria. One of them, n-octanesulfonylacetamide (OSA; Fig. 5) inhibits the growth of a range of slow-growing pathogenic and multi-drug resistant M. tuberculosis strains (Parrish et al., 2001). Mycobacterial lipid analysis reveals a marked reduction of all mycolic acid subtypes without affecting the panoply of polar or non-polar extractable lipids. Moreover, OSA-treated bacteria are characterized by a dysfunction in cell wall biosynthesis and incomplete septation as shown by electron microscopy (Parrish et al., 2001).

Through genetic and biochemical approaches, two genes of the FAS-II system, inhA and kasA, have been postulated to encode the primary target of INH (Banerjee et al., 1994; Mdluli et al., 1998). Due to conflicting reports, the mode of action of INH has recently been re-examined. First, in vivo studies demonstrated that overexpression of InhA, but not KasA, in M. smegmatis, M. bovis BCG and M. tuberculosis confers increased resistance against INH (Larsen et al., 2002). Second, in vitro assays using purified KasA or InhA, showed that KatG-activated INH inhibits InhA activity but not KasA activity (Kremer et al., 2003). Finally, replacement of the wild-type inhA gene by an inhA(S94A) allele within M. tuberculosis is sufficient to confer clinically relevant levels of resistance to INH killing and inhibition of mycolic acid biosynthesis (Vilcheze et al., 2006). In contrast, no resistance to INH is observed when introducing various kasA mutant genes within M. tuberculosis. Together, these findings clearly establish that InhA, but not KasA, is the primary target of INH in M. tuberculosis. In fact, the mycobactericidal agents are INH-NAD(P) adducts, which are generated in vivo after INH activation and which bind to, and inhibit InhA. Recently, the M. tuberculosis proteome was profiled using both the INH-NAD and INH-NADP adducts, leading to the discovery of 16 other proteins that bind these adducts with high affinity and appear to participate in many cellular processes (Argyrou et al., 2006).

Concluding remarks

Considerable strides have been made during the last decade in determining the role of mycolic acids in mycobacterial physiology and virulence, as well as in identifying genes participating in mycolic acid biosynthesis. Genetic strategies have now clearly established that some of these genes are essential for mycobacterial growth, whereas others are required for persistence and play a critical role in the physiopathology of tuberculosis (Fig. 4). In this regard, the M. tuberculosis kasB mutant, which persists in infected immunocompetent mice for up to 600 days without causing disease or mortality, represents an attractive vaccine candidate. In addition, the long-term persistence of ΔkasB represents a novel model for studying latent M. tuberculosis infections. As they are absent from humans, the FAS-II condensing enzymes represent valuable drug targets for future drug development. The use of TLM and its analogues combined with genetic studies validate these enzymes as interesting targets. The determination of the three-dimensional structures of mtFabH and KasB will be hoped to lend a hand to the design of new lead compounds with potent antitubercular activity. The development of molecules acting simultaneously against the different condensing enzymes is warranted by the fact that they would hinder the emergence of resistant strains. Among the key features that remain to be resolved is the exact role of phosphorylation of the condensing enzymes in the regulation of mycolates in vivo, and whether this process is linked to virulence and survival of M. tuberculosis within the infected host.

Acknowledgements

The authors wish to thank the National Research Agency (ANR-06-MIME-027–01 to V.M. and L.K.). L.K. is supported by a Grant from the Centre National de la Recherche Scientifique (CNRS) (Action Thématique Incitative sur Programme ‘Microbiologie Fondamentale’). A.B. acknowledges support from the Medical Research Council in the form of a Career Development Award. G.S.B. acknowledges support in the form of a Personal Research Chair from Mr James Bardrick, the Medical Research Council and the Wellcome Trust.

Ancillary