Bacterial Branched-Chain Amino Acid Biosynthesis: Structures, Mechanisms, and Drugability

Abstract

The eight enzymes responsible for the biosynthesis of the three branched-chain amino acids (l-isoleucine, l-leucine, and l-valine) were identified decades ago using classical genetic approaches based on amino acid auxotrophy. This review will highlight the recent progress in the determination of the three-dimensional structures of these enzymes, their chemical mechanisms, and insights into their suitability as targets for the development of antibacterial agents. Given the enormous rise in bacterial drug resistance to every major class of antibacterial compound, there is a clear and present need for the identification of new antibacterial compounds with nonoverlapping targets to currently used antibacterials that target cell wall, protein, mRNA, and DNA synthesis.

Graphical abstract

Bacteria can synthesize all 20 proteinogenic amino acids, including the nine essential amino acids required for mammalian growth. In general, enzymes involved in the biosynthesis of amino acids are essential for the growth and survival of bacteria. In Mycobacterium tuberculosis, for example, deletions of genes involved in the synthesis of the essential branched-chain amino acid (BCAA) l-leucine generated a successful attenuated strain that was also protective of mice challenged with a virulent strain of the bacilli.¹ High-density mutagenesis and in vitro inhibitory studies have also found that the biosynthetic pathway of the three BCAAs (l-isoleucine, l-leucine, and l-valine) is vital for the growth and survival of M. tuberculosis.^2,3 These data make the enzymes involved in the BCAA biosynthetic pathway in M. tuberculosis and other pathogenic bacteria quite relevant for prospective antibacterial drug development.

Although BCAAs are structurally similar amino acids containing aliphatic side chains, their propensities for being found in protein structures are quite different. l-Valine and l-isoleucine, for example, are overrepresented in β-sheets, while l-leucine is found primarily in α-helices, loops, and leucine zippers.⁴ The small differences in size, hydrophobicity, and degree and position of branching of the side chains explains why these amino acids are not interchangeable in proteins. In fact, the substitution of one BCAA for another may in some cases lead to diseases such as alterations in the plasma lipid profile, hypocalciuria, and cardiomyopathy.⁴ The biosynthetic pathway of BCAAs is a very efficient pathway when compared to pathways leading to the synthesis of other amino acids. While other pathways require many enzymes to synthesize a single amino acid (e.g., nine enzymes are required for the conversion of l-aspartate to l-lysine), the BCAA biosynthetic pathway requires only eight enzymes for the synthesis of all three BCAAs (Figure 1).

Figure 1.

Biosynthetic pathway of the branched-chain amino acids in bacteria and its feedback regulatory points.

Four of the eight enzymes are shared for the synthesis of all three BCAAs (IlvB/N; IlvC, IlvD, and IlvE); three are responsible for only the synthesis of l-leucine (LeuA, LeuC/D, and LeuB), and one is specifically involved in l-isoleucine biosynthesis (IlvA). In this review, each enzyme is discussed individually, focusing on structure and function as well as their potential as antibacterial drug targets (Table 1).

Table 1.

Description of the Genes and Enzymes Involved in the BCAA Biosynthetic Pathway

enzyme	gene name	EC number	PDB entry	Rv number	proven drugability
l-threonine dehydratase/deaminase	ilvA	4.3.1.27	3WQE, 1TDJ, 4PB4	Rv1559	a
acetolactate (acetohydroxyacid) synthase	ilvBN	2.2.1.6/4/1.3.18	1YI1, 1OZH, 1OZF	Rv3003c (B1), Rv4370c (B2), Rv3002c (N)	herbicides
keto acid isomeroreductase	ilvC	1.1.1.86	4YPO, 1QMG, 1YRL, 1SR9	Rv3001c	herbicides
dihydroxyacid dehydratase	ilvD	4.2.1.9		Rv0189c	a
isopropylmalate synthase	leuA	2.3.3.13	3U6W, 3FIG, 4OV9	Rv3710	a
isopropylmalate isomerase	leuCD	4.2.1.33	3Q3W, 3H5E, 3H5H, 3H5J, 2HCU (C); 3H5H (D)	Rv2988c (C), Rv2987c (D)	a
isopropylmalate dehydrogenase	leuB	1.1.1.85	3UDU, 1OSJ, 1W0D	Rv2995c	a
branched-chain aminotransferase	ilvE	2.6.1.42	1A3G, 3HT5, 5U3F	Rv2210c	PLP inactivators

^aInhibition studies and compounds have not been developed or tested.

ILVA, THREONINE DEHYDRATASE/DEAMINASE

The first committed step in the biosynthesis of l-isoleucine is catalyzed by the ilvA-encoded threonine dehydratase/deaminase (EC 4.3.1.19, TD). This enzyme plays a very important role in the biosynthetic pathway of BCAAs in microorganisms and plants. The pyridoxal 5′-phosphate-dependent (PLP) enzyme is responsible for the conversion of threonine (or serine) to 2-ketobutyrate (or pyruvate) and ammonia. TD was one of the first examples of metabolic control via negative feedback in microorganisms. This was due to studies performed in the 1950s,^5,6 in which it was established that the presence of the end product, l-isoleucine, in the growth medium, reduced the activity of TD in Escherichia coli. This work led to the proposal that product or substrate analogues could inhibit an enzyme competitively and inhibition could be accomplished by a downstream product in a regulatory mechanism. The understanding of how such a structurally different molecule was capable of inhibiting TD in a competitive manner remained unclear for many years. The unusual non-Michaelis–Menten kinetics displayed by TD as a function of l-threonine concentration was noted early in the studies of this enzyme.⁶

Changeux reported completely different kinetic patterns in the presence of l-threonine and l-isoleucine, which led him to propose the existence of two separate sites in the enzyme: one site (site A) where l-threonine binds at the catalytic site and a second site (site B) for the binding of l-isoleucine.⁷ Changeux suggested that whenever l-threonine was bound to site A, the enzyme was active to perform its catalytic events. In addition, the affinity of the enzyme for l-threonine would be increased whenever l-threonine was also bound to site B, in a cooperative way. On the other hand, if l-isoleucine was bound to site B, the affinity of the enzyme for l-threonine at site A was decreased, and the enzymes’ activity was diminished. These remarkable observations led to the initial proposal of allosteric regulation.⁸ Numerous studies have confirmed the sigmoidal initial velocity kinetics of the reaction as a function of l-threonine concentration.^7,9 The sigmoidal nature of the kinetics is also altered in the presence of the end products l-isoleucine and lvaline.⁹ While l-isoleucine acts as an allosteric inhibitor, l-valine allosterically activates the enzyme. Other studies have shown that TD is competitively inhibited by aminothiols,¹⁰ including l-cysteine, which acts as an inactivator of the E. coli enzyme.¹¹ TD’s inactivation by l-cysteine is restored upon supplementation with l-threonine, and bacterial growth is further improved by l-isoleucine supplementation. The chemical mechanism of TD starts with the attack of the α-amino group of l-threonine on the Schiff base in the active site, resulting in transimination and external aldimine formation. The abstraction of a proton from C-α, followed by water elimination, leads to the formation of a product Schiff base. Reverse transimination of the external product aldimine by an enzyme lysine residue leads to the regeneration of the original internal aldimine and the release of an enamine. The tautomerization of the enamine forms 2-iminobutyrate that, upon hydrolysis, releases ammonia and 2-ketobutyrate (Scheme 1).

Scheme 1.

Chemical Mechanism of Threonine Dehydratase/Deaminase

Site-directed mutagenesis studies of the tetrameric enzyme consisting of 514-residue monomer chains⁹ confirmed what was previously proposed with regard to its effector sites.¹² Using these data, a more complex model to explain the homotropic cooperativity observed in TD was developed. The cooperativity profile is a consequence of the greater affinity of the substrates and analogues for the regulatory sites than for the catalytic sites,¹³ suggesting that the allosteric change observed from the low- to high-activity state happens synchronously and progressively throughout the range of l-threonine concentrations.

The 2.8 Å resolution crystal structure of TD from E. coli¹⁴ revealed detailed implications for the allosteric mechanism. The 56 kDa enzyme belongs to the type II fold of PLP-dependent enzymes, as a result of its similarities in sequence and structural to the enzymes belonging to this family. The chemical reaction catalyzed by TD, β-elimination, is also a common feature shared among type II fold PLP-dependent enzymes. TD is organized into two different domains: a larger N-terminal catalytic domain, which contains the PLP cofactor bound to Lys62 as a Schiff base,¹⁴ and a smaller C-terminal regulatory domain. Each of the C-terminal regulatory domains in E. coli TD has nonequivalent effector-binding sites, and the allosteric regulation is proposed to be dependent on Ile/Val concentration¹⁵ (Figure 2).

Figure 2.

2.8 Å resolution three-dimensional structure of threonine dehydratase/deaminase (ilvA) from E. coli (PDB entry 1TDJ). The structure is colored by secondary structure: β-sheets in light magenta, α-helices in teal, and loops in yellow. The catalytic (or site A as described by Changeux)7 and allosteric (or site B as described by Changeux)7 domains are labeled, and a close-up of the active site containing the PLP cofactor bound as a Schiff base to Lys62 is shown.

The catalytic product of l-serine dehydration by TD, α-aminoacrylate, inhibits the common, final enzyme in the pathway, the PLP-dependent enzyme, branched-chain amino-transferase (BCAT) IlvE,^16,17 through a mechanism-based type of inhibition.¹⁷ Studies have shown that RidA (YjgF/YER057c/UK114) proteins can, however, prevent this toxic enamine from building up in the cell and protect against BCAT inhibition.^17,18

In M. tuberculosis and Bacillus subtilis, the essentiality of this enzyme has been demonstrated. The deletion of the ilvA gene in B. subtilis generates isoleucine auxotrophy,¹⁹ while the downregulation of the M. tuberculosis enzyme leads to growth impairment and increased susceptibility to stress. The ilvA knockdown strain of M. tuberculosis H37Ra is also more sensitive to antibiotic treatment, which may point to a synergistic potential for targeting this enzyme in combination with other therapeutic targets in the treatment of tuberculosis (TB).

ILVB/N, ACETOLACTATE (ACETOHYDROXYACID) SYNTHASE

The second enzyme in the biosynthesis of isoleucine is also the first universally shared enzyme in the biosynthetic pathway of BCAAs. The ilvBN-encoded acetohydroxyacid synthase (EC 2.2.1.6, AHAS), also known as acetolactate synthase, catalyzes the formation of 2-acetolactate or 2-aceto-2-hydroxybutyrate from the decarboxylation of pyruvate and its condensation with either pyruvate (on the valine pathway branch) or 2-ketobutyrate (on the leucine pathway branch), respectively.²⁰ AHAS requires three different cofactors to catalyze its reaction: thiamin diphosphate (ThDP), flavin adenine dinucleotide (FAD), and a magnesium ion (Mg²⁺).

AHAS is present solely in autotrophic organisms, and in some such as E. coli and Salmonella typhimurium, three different isozymes may be expressed: AHAS I (encoded by the ilvBN genes),^21-23 AHAS II (encoded by the ilvGM genes),^24-27 and AHAS III (encoded by the ilvIH genes).^26,28 However, because of different chromosomal genetic mutations, AHAS II from E. coli²⁹ and AHAS III from S. typhimurium³⁰ are inactive proteins. In the M. tuberculosis genome, four catalytic AHAS subunits (ilvB1, Rv3003c; ilvB2, Rv3470c; ilvG, Rv1820; and ilvX, Rv3509c) and one regulatory subunit (ilvN, Rv3002c) have been annotated and the genes are dispersed along the chromosome.^31,32 A putative small regulatory subunit has also been identified (ilvH) and is believed to be the regulatory subunit of ilvB2, based on similarities with the E. coli enzymes.³³ The ilvB1–ilvN pair displays the enzymatic characteristics expected for the AHAS involved in the biosynthesis of BCAAs, including cofactor requirements. On the other hand, ilvB2, ilvG, and ilvX are implicated in a catabolic pathway leading to the production of 3-hydroxy-2-butanone and/or 2,3-butanediol³² and will not be discussed here.

AHAS belongs to the pyruvate oxidase (PO)-like subfamily, displaying structural similarities with enzymes of this class.^34,35 Several crystal structures of AHAS have been deposited in the PDB. The enzyme is a heterodimer, comprised of two subunits, a large catalytic subunit (ilvB) of approximately 60–70 kDa and a small regulatory subunit (ilvN) with an estimated molecular weight of 10–54 kDa.³⁶ The large subunit of AHAS, ilvB, has activity by itself and binds ThDP in a V conformation³⁶ that leads to a close contact between C2 of the thiazolium ring and N4′ of the pyrimidine moiety. This is a common feature of ThDP-dependent enzymes. The active site of the yeast AHAS, with a sequence that is 42% identical with that of M. tuberculosis AHAS, is located at the dimer interface. ThDP is oriented by hydrogen bonds and van der Waals interactions, including the highly conserved l-glutamate residue, which is Glu85 in M. tuberculosis. This residue is implicated in catalysis as demonstrated by several mutagenesis studies.^37-39 Substitution of Glu85 with alanine and with isosteric or isofunctional amino acid residues, e.g., l-glutamine or l-aspartate, respectively, led to a dramatic decrease in the activity of the enzyme in comparison to that of the wild type.³⁹ Amino acid substitutions of conserved residues located in the immediate proximity of Glu85, His84, and Gln86 also led to a decrease in AHAS activity, suggesting that these residues are involved in the stabilization of the Glu85 side chain, keeping it interacting with N1′ of the ThDP.³⁹ Mutagenesis studies of Arg318 led to complete inactivation of the protein,⁴⁰ while the substitution of another highly conserved residue, Pro126, demonstrates its importance for ThDP binding.³⁶ The FAD-binding site in AHAS is located in one monomer and seems not to be implicated in dimer stabilization⁴¹ but is required for structural integrity.⁴² The adenine ring of FAD is solvent-exposed, while the isoalloxazine ring is buried adjacent to the active site.^43–45 Several crystal structures of AHAS from Klebsiella pneumoniae have been deposited (PDB entries SDX6, SD6R, 1OZF, 1OZG, and 1OZH)⁴¹ (Figure 3).

Figure 3.

2.3 Å resolution three-dimensional structure of acetolactate (acetohydroxyacid) synthase (IlvBN) from K. pneumoniae (PDB entry 1OZF).41 The structure is colored by monomer. A close-up shows the active site containing the ThDP-bound cofactor, with the green sphere representing the Mg²⁺ ion.

Steady-state kinetic studies of AHAS from M. tuberculosis and other eubacteria⁴⁶ reveal nonhyperbolic behavior. Saturation curves with varying levels of pyruvate revealed that the large catalytic subunit of AHAS (ilvB1) alone displays positive cooperativity (Hill coefficient of 2.0), while the heterodimeric holoenzyme displays negative cooperativity (Hill coefficient of 0.6).⁴⁷ Titrations of the small regulatory subunit (ilvN) increased the specific activity of the large catalytic subunit (ilvB1) to values corresponding to that observed for the holoenzyme.⁴⁷ The M. tuberculosis AHAS, as well as its orthologues from other bacteria, is regulated by negative feedback by branched-chain amino acids. In the presence of l-valine and, to a lesser extent, l-isoleucine, the holoenzyme is partially inhibited while l-leucine has no effect on the activity of the enzyme.⁴⁷ The catalytic mechanism of AHAS starts with the binding of pyruvate and ionization of ThDP at the active site, followed by the addition of ThDP to C2 of pyruvate. The covalent tetrahedral intermediate, 2-lactyl-ThDP (LThDP), is decarboxylated in the following step. The decarboxylated product, hydroxyethylthiamine diphosphate (HEThDP), attacks the carbonyl of the second substrate, which can be either pyruvate or 2-ketobutyrate, to form the acetohydroxyacid–ThDP (AHAThDP) adduct. The final step involves release of ThDP from either product, acetolactate or acetohydroxybutyrate. FAD does not play a catalytic role in the mechanism of AHAS; however, during the catalytic process, it can undergo reduction as a consequence of an oxygen-dependent side reaction⁴⁸ (Scheme 2).

Scheme 2.

Chemical Mechanism of Acetolactate (acetohydroxyacid) Synthase

AHAS is a very attractive target for drug development and inhibition, because of its absence in mammals and, thus, reduced potential for toxicity.³ Since the serendipitous discovery of AHAS as the target of sulfonylurea (SU) herbicides in plants,⁴⁹ the potency of these sulfonylureas as antibacterial agents has been tested.²⁴ The M. tuberculosis AHAS is also inhibited by sulfonylureas and other AHASspecific inhibitors;³ however, their inhibitory activities are inferior to that of the standard antibiotics used in the treatment of TB.⁴⁷ However, monosubstituted sulfonylureas displayed potent inhibitory activity against clinical tuberculosis strains, including multidrug-resistant (MDR) and extensively drugresistant (XDR) TB isolates.⁵⁰ The druggability of the M. tuberculosis AHAS has led to the screening and design of compounds targeting the enzyme of the bacilli, including ssDNA aptamers.^3,40,50-55 Many of these inhibitors are active against both the enzyme and resistant strains, with minimal inhibitory concentration (MIC) values similar to those of the standard antibiotics.^40,53 Molecular docking experiments revealed that most of these inhibitors likely bind outside of the active site,⁵³ in agreement with what was previously reported for SU herbicides.^40,43,47,53

ILVC, KETO ACID ISOMEROREDUCTASE

The second enzyme in the BCAA biosynthesis pathway is the ketol-acid isomeroreductase (EC 1.1.1.86, KARI). The enzyme converts the products of AHAS, 2-acetolactate or 2-aceto-2-hydroxybutyrate, to their respective 2,3-dihydroxy products. The reaction catalyzed by KARI was first thought to be performed by two different enzymes, the first catalyzing the alkyl migration followed by keto acid reduction. However, in 1961, it was reported that purified KARI from S. typhimurium was the only enzyme involved in the conversion of either 2- acetolactate or 2-aceto-2-hydroxybutyrate to products.⁵⁶ The properties of the enzyme were also investigated and shown to be similar to those reported for the E. coli,^57,58 Neurospora crassa,⁵⁷ and Saccharomyces cerevisiae⁵⁹ enzymes. KARI from all sources absolutely required Mg²⁺ and used NADPH (reduced nicotinamide adenine dinucleotide phosphate) as a reductant, and both NADP⁺ and 2,3-dihydroxyacid exhibited product inhibition.⁵⁶ The chemical bifunctionality, isomerization and reduction, of KARI has also been investigated through sitedirected mutagenesis.⁶⁰

There are two different classes of KARI enzymes. Class I enzymes are composed of ~340 amino acids, while class II enzymes are larger at approximately 490 amino acids. The class II KARIs are present in some bacteria, including E. coli, and all plants.⁶¹ M. tuberculosis KARI belongs to class I,⁶¹ as do those of Pseudomonas aeruginosa,⁶² Spodoptera exigua,⁶³ and other bacteria.^64,65 In all cases, two different domains, N-terminal and C-terminal, are present. The N-terminal domain folds into a mixed β-sheet flanked on both sides by helices in a nucleotidebinding Rossmann fold. The C-terminal domain is composed of eight α-helices forming a knotted structure.⁶¹ The M. tuberculosis KARI (MtKARI) crystal structure was determined at 1.0 Å resolution and is a dimer in solution⁶¹ (Figure 4). The active site of MtKARI is formed upon dimerization of the protein, a characteristic feature of the class I KARI enzymes.^61-65 Two Mg²⁺ ions are present at the active site and are separated by approximately 5 Å. The first metal ion is coordinated by Asp188 and Glu192 along with four water molecules, while the second one is also coordinated by Asp188, Glu224, Glu228, and three water molecules.⁶¹ The Mg²⁺ ions, as well as the active site of M. tuberculosis KARI, are solventexposed and allow easy access for substrate binding. The NADPH-binding site consists primarily of residues in the Nterminal Rossmann fold, although some contacts from the Cterminal domain may influence NADPH binding.⁶¹ As opposed to the larger class II KARI enzymes, M. tuberculosis KARI does not undergo significant conformational changes upon NADPH binding.⁶¹

Figure 4.

1.0 Å resolution three-dimensional structure of keto acid isomeroreductase from M. tuberculosis (PDB entry 4YPO). The homodimer is colored by monomer, and the loops are colored yellow. The domain containing the Rossmann fold is highlighted. A close-up of the active site is shown highlighting the main residues coordinating the Mg²⁺ ions (green spheres). The active site is at the juncture of the N-terminus of one monomer and the C-terminus of the other monomer (circle).

KARI enzymes exhibit high stereospecificity for the S isomers of their substrates.^66,67 Kinetic constants for the MtKARI enzyme have been determined⁶¹ and compared with values reported for the E. coli enzyme.^60,68,69 KARI’s reaction can be monitored spectrophotometrically by observing the oxidation of NADPH at 340 nm. As a general feature of both class I and class II KARI enzymes, the activity of KARI is higher with 2- aceto-2-hydroxybutyrate than with 2-acetolactate.^61,69,70 At saturating concentrations of Mg²⁺, the E. coli and S. typhimurium⁷¹ enzymes follow an ordered kinetic mechanism where NADPH binds first followed by 2-acetolactate or 2-aceto-2-hydroxybutyrate. The binding order of Mg²⁺ and NADPH, however, is random.⁶⁹ The isomerization chemistry involves base-assisted deprotonation of C2 hydroxyl and migration of a methyl or ethyl group to C3 to generate the α-keto-β-hydroxyacid. NADPH transfers the pro-S hydrogen as a hydride ion to reduce the keto acid to the C2 hydroxyl product.^61,69 Solvent kinetic isotope effect data suggest that reduction of the intermediate is not the rate-limiting step, but more likely the isomerization/alkyl migration step preceding the reduction reaction⁶⁹ (Scheme 3). In the presence of 2- acetolactate, E. coli KARI specifically requires Mg²⁺ ion for activity; however, Mn²⁺ can substitute for Mg²⁺ when 2-aceto-2- hydroxybutyrate is utilized as the substrate.⁶⁹ There are no data concerning the activity of M. tuberculosis KARI with different divalent metal ions.

Scheme 3.

Chemical Mechanism of Keto Acid Isomeroreductase

The potential of KARI as an antibacterial drug target has been established. Some herbicidal compounds, such as the KARI transition state analogue N-isopropyl oxalylhydroxamate (IpOHA), also display antibacterial activity⁷² and are very potent inhibitors of the E. coli KARI. The intermediate analogue binds to the active site of the enzyme^40,43,47,53 and inactivates the enzyme in a time-dependent fashion, leading to the formation of an irreversible enzyme–inhibitor complex.⁶⁸ MtKARI is also inhibited by the tight-binding inhibitor IpOHA and displays a K_I of approximately 98 nM,⁶¹ validating the targetability of this enzyme in M. tuberculosis. The analogue also displayed good inhibitory activity against clinical drug-resistant strains of M. tuberculosis; however, its effect was not superior to those of the current drugs used to treat the disease.³ Several compounds have been designed for herbicidal purposes^73-76 and could also be explored in the inhibition of KARI from pathogenic bacteria, including M. tuberculosis.

ILVD, DIHYDROXYACID DEHYDRATASE

Preceding the final transamination step that leads to the synthesis of l-isoleucine and l-valine, and the fifth of the seven steps leading to the production of l-leucine, the ilvD-encoded dihydroxy acid dehydratase (EC 4.2.1.9, DHAD) is the enzyme responsible for the synthesis of 2-keto-3-methylvalerate and 2-ketoisovalerate. DHAD is also required for the synthesis of pantothenate, because one of its products, 2-ketoisovalerate, is a precursor to this pathway. The importance of DHAD was first investigated by using a partially purified enzyme from E. coli extracts,⁷⁷ where the first evidence for the requirement of a ferrous ion or other divalent cation was identified.⁷⁷

DHAD has been studied in several organisms, including bacteria,^59,78-83 fungi,^59,84 and plants.^80,85-88 The most wellstudied DHAD from a bacterial source is the E. coli enzyme.^89-91 Stereospecificity studies of S. typhimurium DHAD⁷⁹ as well as transcriptional and post-transcriptional studies of B. subtilis DHAD⁹² have also been performed. S. typhimurium DHAD demonstrates absolute stereospecificity because only 2(R)-keto-3(R)-methylvalerate and 2(R)-ketoisovalerate support bacterial growth.⁷⁹

DHADs are homodimeric iron–sulfur cluster enzymes with monomer molecular weights ranging from 60 to 70 kDa.^83,90 Although plants⁸⁸ have been shown to contain a [2Fe-2S] cluster, bacterial DHADs contain a [4Fe-4S] cluster, but in both cases, the iron–sulfur cluster is absolutely required for catalysis.^83,88,90,91 Enzymes having iron–sulfur clusters as cofactors are highly sensitive to oxygen, and the oxidative disruption of the cluster in bacterial DHAD leads to the complete inactivation of the protein.^89,90 The activity of E. coli DHAD under aerobic conditions decreased by 100% in 2 h.⁹⁰ When the effects of nitric oxide (NO) were tested on E. coli DHAD, the rapid reaction of NO with the iron–sulfur cluster of the enzyme was synchronous with the formation of a DHAD–dinitrosyl–iron complex, which completely inactivates the enzyme even under anaerobic conditions. The rate of NO reaction with the iron–sulfur cluster of DHAD is faster than the rate of its oxidation by oxygen, and the addition of glutathione to the reaction mixture, performed under anaerobic or aerobic conditions, does not prevent the enzyme from being inactivated by NO.⁹³ The sensitivity of DHAD enzymes to oxygen is likely the reason why, to date, no three-dimensional crystal structures have been reported from any organism.

The catalytic mechanisms of both S. typhimurium and E. coli DHADs have been reported, and in both cases, catalysis is dependent on the [4Fe-4S] cluster. The cluster acts as a Lewis acid, and the C3-hydroxyl group of the 2,3-dihydroxy-valerate substrates binds as a ligand to the cluster, activating it for β-elimination of a water molecule upon abstraction of a C2 proton. This reaction results in the formation of an enol intermediate that tautomerizes with stereospecific C3 protonation to generate the keto acid product.⁷⁹ The stereospecificity of the tautomerization/protonation strongly suggests that this step occurs in the proteins’ active site prior to product release (Scheme 4).^90,94

Scheme 4.

Chemical Mechanism of the [4Fe-4S] Cluster Protein Dihydroxyacid Dehydratase

M. tuberculosis DHAD (MtDHAD) is a 118 kDa homodimer in solution and like the E. coli enzyme contains a bacterial type [4Fe-4S] cluster. The enzyme is also sensitive to oxygen and NO, and the growth of M. tuberculosis treated with NO can be restored only upon supplementation with the three BCAAs.⁸³ Downregulation of the ilvD gene in M. tuberculosis generated a partially auxotrophic strain, which was unable to grow in mice but still capable of persisting in the tissue.⁸³ The downregulation of the gene allows for the suboptimal synthesis of branched-chain amino acids, and therefore, a knockout study could clarify the potential of the ilvD gene for auxotrophy. The essentiality of this enzyme for growth and survival of mycobacteria makes this enzyme a very interesting potential antibacterial target, because its inhibition impairs not only the synthesis of BCAAs but also pantothenate and consequently the synthesis of CoA, for which pantothenate is a precursor.

LEUA, ISOPROPYLMALATE SYNTHASE

The first enzyme in the branch that leads to l-leucine biosynthesis is the leuA-encoded isopropylmalate synthase (EC 2.3.3.13, IPMS). IPMS catalyzes the first committed step in the synthesis of l-leucine with the conversion of 2- ketoisovalerate and acetyl-CoA to 2-isopropylmalate and CoA, respectively. The enzyme was first purified from S. typhymurium in 1969, and some kinetic properties of this enzyme were characterized,⁹⁵ such as the optimum pH of the enzyme (pH 8.5) and evidence of an allosteric mechanism.

IPMS has been extensively studied in M. tuberculosis (MtIPMS). MtIPMS was crystallized, and the structure was determined at 2.0 Å resolution (Figure 5).⁹⁶ The homodimer is composed of two 70 kDa monomers containing 644 residues. Each monomer is folded into two major domains, an Nterminal domain and a C-terminal domain, connected by a linker domain that is further divided into two smaller domains. The N-terminal domain is composed of an (α/β)₈ TIM barrel where the catalytic site is located. The (α/β)₈ TIM barrel has N- and C-terminal extensions that are highly important and involved in the dimerization of the protein. The active site is located at the C-terminal extension of the N-terminal domain and binds the divalent cation Mn²⁺ and 2-ketoisovalerate. This metal-binding center has a pair of His residues in addition to an Asp residue. The presence of these amino acid residues allows for a lack of discrimination toward divalent metals that will be discussed below. The connecting linker domain is composed of two smaller domains, where one consists of an α-helix and two β-strands, while the other has three α-helices. The residues linking the two small subdomains of the linker domain are disordered and flexible. The C-terminus of MtIPMS consists of a regulatory domain composed of two identical βββα units, built as a three-layer β–α–β sandwich.⁹⁶

Figure 5.

2.21 Å resolution three-dimensional structure of isopropylmalate synthase from M. tuberculosis (PDB entry 3U6W). The homodimer is colored by monomer, and the loops are colored yellow. A close-up of the active site displays α-ketoisovalerate and the metal-binding center coordinated by the two His residues and Asp.

The C-terminal regulatory domain of MtIPMS is the binding site for l-leucine and is responsible for the control of the activity of this enzyme.⁹⁶ IPMS is regulated via a product inhibition mechanism by l-leucine in many organisms, including M. tuberculosis. l-Leucine was shown to be a reversible, slow-onset inhibitor of MtIPMS, and its affinity for l-leucine is independent of substrate binding.^97,98 The binding of l-leucine does not affect the quaternary structure of MtIPMS; however, binding of l-leucine results in an increase in the stability of the N- and C-terminal protein domains. l-Leucine acts as a V-type inhibitor by binding in a noncooperative fashion at the interface of the dimer approximately 50 Å from the active site.^99-102 MtIPMS displays modest activity with small keto acid substrates, including pyruvate, 2-ketobutyrate, and 2-ketovalerate. However, the efficiency of the reaction with 2-ketoisovalerate is much higher.¹⁰¹ The enzyme is extremely specific with regard to the acyl donor, with only acetyl-CoA showing any activity. MtIPMS follows a non-rapid equilibrium random bi-bi kinetic mechanism.¹⁰¹

The chemical mechanism of the MtIPMS-catalyzed reaction consists of the enolization of acetyl-CoA, followed by an aldol condensation with a rapid protonation of the 2-hydroxyl group of 2-isopropylmalyl-CoA by an active site acid. The hydrolysis of the thioester carbon of the intermediate yields the formation of a tetrahedral intermediate whose breakdown releases the two products, 2-isopropylmalate and CoA (Scheme 5).¹⁰¹

Scheme 5.

Chemical Mechanism of Isopropylmalate Synthase

MtIPMS can use a broad range of divalent metals to catalyze its reaction, including Mg²⁺, Mn²⁺, Co²⁺, Ni²⁺, and Ca²⁺, with Mg²⁺ being the preferred metal. Metal binding induces contacts between the N-terminal catalytic domain and the C-terminal regulatory domain of MtIPMS, thus establishing structural cooperativity.¹⁰³ Both Zn²⁺ and Cd²⁺ inhibit the enzyme¹⁰¹ by inducing a partial denaturation and/or unfolding of the domain.¹⁰³ Monovalent cations also play an essential role in the activity of MtIPMS. The enzyme can utilize a large number of monovalent cations; however, K⁺ and Rb⁺ are the preferred activators.¹⁰¹ There is no direct interaction of the K⁺ with the substrates 2-ketoisovalerate and acetyl-CoA; instead, K⁺ plays an allosteric effector role and alters the surroundings of the Mg²⁺-binding site without changing the enzyme structure.^101,103 The rate-limiting step of the reaction catalyzed by MtIPMS in the absence of l-leucine has been proposed as being the release of products; however, when l-leucine binds, the rate-limiting step shifts to the hydrolysis of the thioester carbon of the intermediate isopropylmalyl-CoA.¹⁰⁰

The essentiality of this enzyme in bacteria, and its absence in mammals, makes IPMS a very interesting drug target to be exploited. So far, only in silico inhibition studies have been reported for the mycobacterial enzyme.¹⁰⁴ Therefore, the development of compounds targeting IPMS remains to be explored.

LEUC/D, ISOPROPYLMALATE ISOMERASE

The second exclusive enzyme in the branch of l-leucine biosynthesis is the leuCD-encoded isopropylmalate isomerase (EC 4.2.1.33, IPMI), which catalyzes the isomerization of 2-isopropylmalate to 3-isopropylmalate.^105,106 In some organisms, such as S. typhimurium and E. coli, the genes encoding enzymes involved specifically in the leucine biosynthesis branch are organized in a single operon leuABCD and co-expressed. In 1981, S. typhimurium IMPI was identified as a multimeric enzyme formed by two separate genes, leuC and leuD.¹⁰⁷ The enzyme was purified, and the products of the leuC and leuD genes were present in a 1:1 ratio. The activity of S. typhimurium IPMI was analyzed in bacterial crude extracts, and the enzyme displayed very low activity. Attempts to purify the protein to homogeneity resulted in a complete loss of activity.¹⁰⁷

The chemical reaction catalyzed by IPMI presumably involves a base-assisted dehydration of 2-isopropylmalate to generate a cis-vinylogous intermediate. The rotation of the intermediate in the active site followed by a trans addition of water results in the formation of the 3-isopropylmalate product^108,109 (Scheme 6). This reaction can be continuously measured by observing the decrease in the absorbance of citraconate, an isopropylmalate analogue, at 253 nm.¹¹⁰

Scheme 6.

Chemical Mechanism of the Isopropylmalate Isomerase

Because of the similarity of the reaction catalyzed by IPMI and the mitochondrial enzyme aconitase (ACN),¹⁰⁹ and sequence comparisons,¹¹¹ IPMI is known to be a member of the ACN superfamily.¹¹² ACNs are well-characterized monomeric enzymes that require an intact [4Fe-4S] cluster for activity, catalyzing the isomerization of citrate to isocitrate with the production of the intermediate, cis-aconitate. Studies of an E. coli mutant strain lacking peroxidases, including katG, katE, and ahp, and treated with H₂O₂ found that supplementation of the medium with 2-ketoisocaproate rescued the growth of the bacteria in the presence of the peroxide. By employing a plasmid that overexpressed IPMI in the H₂O₂-treated culture, the growth defect was corrected, confirming the sensitivity of IPMI to reactive oxygen species.¹¹⁰ Inactivation of IPMI by H₂O₂ is due to the abstraction of an electron from the [4Fe-4S] cluster, altering it to [4Fe-4S]³⁺, which is an unstable valence for the cluster, and leads to the release of Fe²⁺, which results in the inactivity of the remaining [3Fe-4S]⁺ cluster.¹¹⁰ The oxygen sensitivity of [4Fe-4S] cluster enzymes such as IPMI makes it difficult to study because an anaerobic environment is required.

IPMIs are classified into two different groups depending on their subunit composition. The first group consists of fungal IPMIs, which are 80–90 kDa monomeric proteins, while in bacteria and archaea, the enzymes are heterodimers composed of a 45–50 kDa large leuC subunit and a 15–20 kDa small leuD subunit. The IPMIs from the second group are active only when the two subunits come together to form the heterodimer. The large subunit presents three conserved Cys residues that are proposed ligands for the [4Fe-4S]³⁺ cluster, while 2- isopropylmalate is predicted to bind the sequence motif GSSR, located on the small subunit.¹¹³ To date, only one structure of the large subunit of IPMI has been determined and reported.¹¹⁴ The large subunit of Methanococcus jannaschii IPMI has been crystallized under aerobic conditions, and the structure has been determined at 1.8 Å resolution.¹¹⁴ The monomer is composed of 18 α-helices and 17 β-strands distributed in three domains organized in a triangular manner. The active site contains two disulfide bonds formed as a result of the oxidation of the presumed Cys ligands to the cluster and is surrounded by the three domains of the large leuC subunit of Me. jannaschii IPMI (Figure 6). No [4Fe-4S] cluster was present in the crystallized structure. A larger number of crystal structures have been reported for the small subunit of IPMI (PDB entries 3Q3W, 3H5E, 3H5H, 3H5J, and 2HCU),¹¹⁵ including that of M. tuberculosis¹¹³ (Figure 6).

Figure 6.

(A) Three-dimensional structure at 1.8 Å resolution of ilvC from Me. jannaschii colored by domain. Mg²⁺ is represented as a green sphere (PDB entry 4KP1). (B) Three-dimensional structure at 2.5 Å resolution of ilvD from M. tuberculosis colored by secondary structure (β-sheets in light magenta and α-helices in teal) (PDB entry 3H5H).

M. tuberculosis IPMI (MtIPMI) has been purified to homogeneity as a homodimer (1:1 ratio), and the enzyme was more stable in solution than when each subunit was purified individually. When assayed, MtIPMI was completely inactive, probably because of the oxidation reaction leading to an inactive [3Fe-4S]⁺ cluster. Three variants of the MtIPMI leuD subunit, differing in the length of the protein, were crystallized at different resolutions. The overall fold of the small subunit is a twisting β/β/α three-layer sandwich. Alignments of the MtIPMI leuD subunit with other homologues and a portion of aconitase led to the proposal that MtIPMI leuD Arg32 (ACN Arg580) plays an important role in substrate recognition by making important hydrogen bonds with the γ-carboxylate of 2- isopropylmalate. Alignment of MtIPMI leuC with mitochondrial ACN revealed that the sequences of the enzymes are 28 and 43% identical and similar, respectively. The residues involved in the binding of substrates, catalysis, and cofactor coordination are all well conserved between the two enzymes. Therefore, it is likely that the heterodimer complex of MtIPMI is structurally similar to that of ACN.

The importance of the leuCD genes and therefore of IPMI in M. tuberculosis growth and virulence has been recognized for more than decades. Disruption of the leuD gene in the Mycobacterium bovis strain (BCG) causes a growth and infection impairment in mouse models.¹¹⁶ The auxotrophic profile is attributed to the inability of BCG ΔleuD to replicate and grow inside macrophages, because the bacteria cannot scavenge intracellular leucine.¹¹⁷ The same profile was observed when the leuD gene was disrupted in M. tuberculosis, and the successfully attenuated strain was also protective of mice challenged with a virulent strain of the bacilli. A double auxotroph strain (ΔpanCDΔleuCD) was even more protective than ΔleuD⁻ alone, resulting in the protection of macaques coinfected with TB/SIV (simian immunodeficiency). These data support the essentiality of IMPI in the growth and survival of M. tuberculosis and confirm the potential of this enzyme as a potential drug target.

LEUB, ISOPROPYLMALATE DEHYDROGENASE

The leuB-encoded isopropylmalate dehydrogenase (EC 1.1.1.85, IPMDH) is responsible for converting 3-isopropylmalate to 2-ketoisocaproate via oxidation of the second alcohol and decarboxylation. This reaction is NAD⁺-dependent and requires the presence of a divalent metal such as Mg²⁺ or Mn²⁺ and the monovalent cation K⁺ for activation.^118–120 IPMDH was first purified and characterized in S. typhimurium in 1969.¹²¹ Starting with 19 g of crude extract, four purification steps yielded 120 mg of >95% pure enzyme. S. typhimurium IPMDH is a 70 kDa homodimer in solution.¹²¹ The optimum pH found for this enzyme was 9, and the activity of S. typhimurium IPMDH was measured using a continuous spectrophotometric assay in which the increase in absorbance at 340 nm due to NAD⁺ reduction was observed over time.¹²¹ The enzyme followed Michaelis–Menten kinetics for both substrates.

Crystal structures of IPMDH from a number of different organisms have been determined,^122–127 including M. tuberculosis. ¹²⁸ Sequences of IPMDH from E. coli and S. typhimurium are 94.5% identical, while the sequences of these enzymes are 51% identical with that of Thermus thermophilus IPMDH.¹²⁷ The overall topology of all three proteins is however very similar.¹²⁷ Most IMPDHs are homodimeric proteins, and the structure is generally composed of two (α/β) domains organized in a 10-stranded β-sheet where each monomer consists of 300–400 amino acid residues. The crystal structure of M. tuberculosis IPMDH (MtIPMDH) was determined at 1.65 Å resolution in the absence of any bound substrates or cofactors, and in solution, it is a 70 kDa homodimer (Figure 7). The sequence of the enzyme is approximately 40% identical to those of other bacterial orthologues.¹²⁸

Figure 7.

1.65 Å resolution three-dimensional structure of the asymmetric unit of isopropylmalate dehydrogenase from M. tuberculosis (PDB entry 1W0D). The tetrameric enzyme is colored by monomer.

Some of the structurally characterized IPMDHs display different conformations, varying from open to partially closed and fully closed active sites. The superimposition of MtIPMDH with the E. coli, S. typhimurium, and T. thermophilus structures revealed that MtIPMDH presents a closed conformation; however, the open conformation can also be observed in some of the subunits of this enzyme.¹²⁸ The active site of MtIPMDH was inferred to be located at the cleft between the two domains of the enzyme, based on a sequence alignment with the previously characterized IPMDH from Thiobacillus ferrooxidans, ¹²³ which was crystallized in the presence of isopropylmalate. ¹²⁸ The amino acids involved in isopropylmalate and Mg²⁺ binding are conserved between the two structures.¹²⁸ A similar alignment strategy was used to compare the NAD⁺-binding site of MtIPMDH with that of T. thermophilus IPMDH.¹²²

IPMDH is a member of the metal ion-dependent, β-hydroxyacid oxidative decarboxylase family. Enzymes such as malic enzyme and isocitrate dehydrogenase are also part of this family.¹²⁹ In general, these enzymes catalyze a reaction consisting of a pyridine nucleotide-dependent reversible secondary alcohol oxidation followed by an irreversible decarboxylation that leads to the formation of an enol intermediate; which is tautomerized and results in the final product, which in the case of IPMDH is 2-ketoisocaproate^119,129 (Scheme 7).

Scheme 7.

Chemical Mechanism of Isopropylmalate Dehydrogenase

IPMDH follows a random steady-state kinetic mechanism in common with other members of the β-hydroxy acid oxidative decarboxylase family.^118,129 Transient kinetic studies proposed that in T. thermophilus IPMDH,¹¹⁸ NAD⁺ binds rapidly, whereas the binding of isopropylmalate is slower and induces a conformational change that brings the protein into a closed state. According to their studies, the closed conformation is a requirement and determines the rate of isopropylmalate oxidation and formation of NADH (reduced nicotinamide adenine dinucleotide) in the presteady state. The decarboxylation and tautomerization processes are spontaneous and occur faster and without conformational changes. However, product release requires the opening of the protein domains and is the rate-limiting step in catalytic turnover.¹¹⁸ Structural studies and quantum mechanics/molecular mechanics (QM/MM) calculations resulted in the following proposed mechanism for T. thermophilus IPMDH. A general base, Lys185, deprotonates the OH group of isopropylmalate via a water molecule. The oxidation of isopropylmalate and hydride transfer display a higher energy barrier and therefore are the rate-limiting steps in this model. However, K⁺ can increase the rate of hydride transfer by 1000–2000-fold. Decarboxylation is a spontaneous step, has a very low energy barrier, and drives the formation of the final products.¹¹⁹ In this case, the hydride transfer step seems to account for rate limitation.

IPMDH also has the potential for herbicidal and antibacterial drug development. O-Isobutenyl oxalylhydroxamate (OIbOHA) was a compound initially designed to display increased herbicidal activity against KARI enzymes.¹²⁰ However, through supplementation studies with l-leucine and 2-ketoisocaproate, IPMDH was found to be the target of this compound in plants and in S. typhimurium.¹²⁰ Besides O-IbOHA, several other inhibitors have been designed and assayed against the T. thermophilus enzyme. In general, these compounds act as competitive inhibitors and display inhibition constants in the nanomolar range.^120,130,131 To date, no experimental data have been published on the activity of O-IbOHA against MtIPMDH. However, small molecule docking of the inhibitor identified that the potential binding site of O-IbOHA in MtIPMDH is very likely to be the same binding site as that of isopropylmalate. O-IbOHA likely inhibits MtIPMDH by mimicking the enol intermediate of the reaction.¹²⁸ These data can be used as a foundation for the future design of MtIPMDH inhibitors with potential antimycobacterial activity.

ILVE, BRANCHED-CHAIN AMINOTRANSFERASE

The final step in the synthesis of all three BCAAs involves the transfer of the α-amino group of l-glutamate to the α-carbon of 2-ketoisocaproate, 2-ketoisovalerate, and 2-keto-3-methylvalerate, the keto acid precursors of l-leucine, l-valine, and l-isoleucine, respectively. Transamination reactions were first identified in the late 1930s and observed to be biologically relevant in many organisms.¹³² The first aminotransferases to be discovered were l-glutamate aminotransferase and l-aspartate aminotransferase.¹³² Snell and co-workers,¹³³ who had previously characterized chemical transaminations in the presence of pyridoxal, made the association that the enzymatically catalyzed transamination reactions were dependent on the presence of this cofactor. Feldman and Gunsalus¹³⁴ reported in 1950 that bacteria have a variety of distinct transaminases and confirmed the need for pyridoxal as a cofactor/coenzyme for enzyme activity.

The ilvE-encoded branched-chain aminotransferase (EC 2.6.2.41, BCAT) was first isolated in 1952 from E. coli,¹³⁵ and two different aminotransferases were identified. The first one had higher activity in the presence of aromatic amino acids and was named Transaminase A, and the second, Transaminase B, displayed higher activity in the presence of the BCAA’s isoleucine, leucine, and valine.¹³⁵ Between the 1950s and 1976, studies of BCATs from several Gram-negative bacteria were reported. These studies are covered in a review on the catabolism of BCAAs in bacteria.¹³⁶ We will instead examine the anabolic reaction, or BCAA synthesis, as well as what is known of BCATs in terms of their mechanism, structures, and potential for drug development.

In the anabolic direction, BCATs are responsible for the transfer of an amino group from l-glutamate to the α-keto acid form of the respective amino acid to be synthesized.¹³⁷ This reaction is reversible and depends on the coenzyme PLP being covalently bound to the enzyme through a Schiff base with a protein lysine residue.¹³⁸ The first three-dimensional structure of a bacterial BCAT was reported in 1997.¹³⁹ The homohexameric E. coli structure was reported at 2.5 Å resolution and displayed an interesting triangular prism shape arranged as a doublet of trimers. Numerous structures of BCATs from different eubacteria, as well as human BCATs, have been determined, including the M. tuberculosis enzyme (MtIlvE), and all belong to the type IV fold class of PLP-dependent transaminases.¹³⁸ This classification is due to the presence of two domains connected by an interdomain loop, where the chemical reaction occurs at the re face of the Schiff base to the PLP cofactor.¹³⁸ The MtIlvE structure was determined at 1.9 Å resolution and is a homodimer of 40 kDa with the pyridoxamine 5′-phosphate (PMP) molecule present at the active site of both monomers.¹³⁸ Each monomer is composed of two domains that together form the active site. The first domain is composed of a core eight β-strands surrounded by three α-helices, and the second domain consists of two β-sheets surrounded by three α-helices. A particular Cys residue is found in MtIlve but absent from any other orthologue. In this crystal structure, the thiol groups of Cys196 are 3.6 Å apart and therefore do not form a disulfide bond.

BCATs, as well as other aminotransferases, display a pingpong kinetic mechanism. In the ping half-reaction, the α-amino group from the donor amino acid reacts with the Schiff base PLP form of the enzyme, and then chemistry takes place followed by an α-keto acid release. The enzyme is now in the PMP form, and the second half-reaction can take place. The pong half-reaction starts with the binding of a different α-keto acid to the PMP form of the enzyme, followed by the transfer of a proton from C4′ of the cofactor to the carbon of the ketimine to generate the new BCAA and regenerate the enzyme in its PLP form. The detailed chemical mechanism of the M. tuberculosis BCAT¹⁴⁰ revealed that transamination occurs via an unusual 1,3-prototropic shift mechanism,¹⁴⁰ where α-C–H bond cleavage from l-glutamate occurs simultaneously with the protonation of C4′ of the PLP cofactor in the same transition state (Scheme 8).

Scheme 8.

Chemical mechanism of IlvE from M. tuberculosis

The absence of an observable quininoid intermediate, which is a common feature of PLP-dependent enzyme reactions,^141,142 confirmed the concerted mechanism of MtIlvE. Some PLPdependent enzymes have reported ketamine hydrolysis as the rate-limiting step of the reaction.^143,144 Despite kinetic, chemical, and structural similarities, BCATs from bacteria exhibit different substrate specificities. E. coli BCAT and MtIlvE can utilize a large number of amino acid substrates, ranging from BCAAs to methionine and aromatic amino acids.^140,145 On the other hand, the mammalian isoforms have no activity in the presence of aromatic amino acids and are active with aspartate.¹⁴⁶ These small variances play an important role in selectivity and therefore can be used for rational drug design.

The potential of this enzyme as a drug target has been evaluated for many years in humans because of its selective inhibition by gabapentin, used in the treatment of epilepsy. ^146-148 The sequences of the two human BCAT isoforms (cytosolic BCAT and mitochondrial BCAT) are 58% identical, yet gabapentin does not inhibit the mitochondrial isoform. The sequence of MtIlvE is 31% identical with that of mitochondrial BCAT, and the tuberculosis enzyme is not inhibited by gabapentin,^138,145 showing that MtIlvE has the potential to be specifically targeted. Inhibition studies of MtIlvE with aminooxy compounds were performed and demonstrated the druggability of this enzyme.¹⁴⁵ The best mycobacterial inhibitor was Oallylhydroxylamine, which displayed a K_I of 22 μM and a MIC value of 156 μM. The difference between the O-allylhydroxylamine MIC value in M. tuberculosis and the K_I for MtIlvE may be the result of inhibitory effects of the compound on multiple PLP-dependent enzymes. A similar feature was observed when MtIlvE was inhibited by d- and l-cycloserine.¹³⁷ d-Cycloserine is a second-line drug used in the treatment of MDR-TB and inhibits two sequential reactions of the peptidoglycan biosynthetic pathway: the PLP-dependent alanine racemase and d-alanine-d-alanine ligase.¹⁴⁹ MtIlvE is inhibited by both cycloserine isomers in a time- and concentration-dependent fashion. l-Cycloserine is, however, 40-fold more potent as an inhibitor of the enzyme than d-cycloserine is, displaying a K_I of 88 μM. The MIC values also show a 10-fold better inhibitory effect with the l-isomer, and in all cases, supplementation with BCAAs either individually or in combination did not rescue the growth of the bacteria. It was suggested that the cycloserine isomers, as well as the aminooxy compounds, are generalized inhibitors of PLP-dependent enzymes.¹³⁷ The mechanism of inactivation of MtIlvE by the cycloserine isomers was shown to be the result of the aromatization of the cycloserine ring to form a stable PLP adduct.¹³⁷ The structure of the inhibited MtIlvE–d-cycloserine complex was determined at 1.7 Å resolution and reveals an intact and planar d-cycloserine ring (Figure 8).

Figure 8.

1.7 Å resolution three-dimensional structure of the branched-chain aminotransferase IlvE from M. tuberculosis (PDB entry 5U3F). The homodimeric protein is colored by secondary structure: β-sheets in light magenta and α-helices in teal. Both monomers contain the bound irreversible PMP–d-cycloserine complex at the active site.

CONCLUSIONS

The enzymes belonging to the BCAA biosynthetic pathway in bacteria are an excellent potential source of targets to be explored for the development of new antibacterial agents. Of particular interest, in M. tuberculosis, all the enzymes in this pathway are essential for the growth and survival of the bacilli, and the disruption of any enzyme of the pathway may have a serious consequence for the survival of the bacteria. M. tuberculosis is relatively poor at scavenging BCAAs from the host cell, and therefore as observed for many enzymes of the BCAA pathway, it can be used as a strategy for the development of auxotrophic strains to be used as vaccines or drug inhibition candidates. In addition, the advantages of targeting this pathway in bacteria are evident, due to the lack of a similar pathway in mammals (only BCATs are present in mammals), which would reduce related toxicity. Several compounds, including herbicides, have been identified as potent inhibitors of the enzymes from this pathway in different organisms. Leading compounds can be found through the screening of the already existing molecules or used as a foundation for better and more specific inhibitors. In addition, depending on the enzyme targeted, BCAA synthesis is affected and some of the products of this pathway, such as α- ketoisovalerate, are precursors for the pantothenate biosynthetic pathway. The inhibition of AHAS, KARI, DHAD, and BCAT in bacteria affects the synthesis and/or recycling of essential amino acids and metabolites (BCAAS, methionine pantothenate, and CoA), which can be explored as a “death by a thousand cuts” strategy against pathogenic organisms. In summary, the enzymes of the BCAA biosynthetic pathway in pathogenic bacteria seem to represent mechanistically and structurally well-characterized targets for further exploitation for antibacterial drug design.

ABBREVIATIONS

ACN

aconitase

AHAS

acetohydroxyacid synthase

AHAThDP

acetohydroxyacid thiamin diphosphate

BCAA

branched-chain amino acid

BCAT

branched-chain aminotransferase

BCG

Bacille Balmette-Guerin

CoA

coenzyme A

DHAH

dihydroxyacid dehydratase

FAD

flavin adenine dinucleotide

HEThDP

hydroxyethylthiamine diphosphate

IPMDH

isopropylmalate dehydrogenase

IPMI

isopropylmalate isomerase

IpOHA

N-isopropyl oxalylhydroxamate

IPMS

isopropylmalate synthase

KARI

keto/ketol-acid isomeroreductase

LThDP

2-lactyl-thiamin diphosphate

MDR

multidrug-resistant

MIC

minimal inhibitory concentration

NADH

nicotinamide adenine dinucleotide

NADPH

nicotinamide adeninde nucleotide phosphate

NO

nitric oxide

O-IbOHA

O-isobutenyl oxalylhydroxamate

PDB

Protein Data Bank

PLP

pyridoxal 5′-phosphate

PMP

pyridoxamine 5′-phosphate

PO

pyruvate oxidase

QM/MM

quantum mechanics/molecular mechanics

SU

sulfonylurea

TB

tuberculosis

TD

threonine dehydratase

ThDP

thiamin diphosphate

XDR

extensively drug-resistant