Modification of the NADH of the Isoniazid Target (InhA) from Mycobacterium tuberculosis

See allHide authors and affiliations

Science  02 Jan 1998:
Vol. 279, Issue 5347, pp. 98-102
DOI: 10.1126/science.279.5347.98


The preferred antitubercular drug isoniazid specifically targets a long-chain enoyl-acyl carrier protein reductase (InhA), an enzyme essential for mycolic acid biosynthesis in Mycobacterium tuberculosis. Despite the widespread use of this drug for more than 40 years, its precise mode of action has remained obscure. Data from x-ray crystallography and mass spectrometry reveal that the mechanism of isoniazid action against InhA is covalent attachment of the activated form of the drug to the nicotinamide ring of nicotinamide adenine dinucleotide bound within the active site of InhA.

Mycobacterium tuberculosis is particularly susceptible to isoniazid [isonicotinic acid hydrazide (INH)], the most widely used of all antitubercular drugs (1). Although isoniazid-based treatment regimens have been available since the 1950s, M. tuberculosis remains the leading cause of death worldwide from an infectious agent (2). Tuberculosis is now a disease associated with poverty and with acquired immunodeficiency syndrome (AIDS); the greatest impact is experienced in underdeveloped nations and in centers of urban decay (3). In addition, the incidence of incurable cases due to multidrug-resistant mutants is on the rise. These trends have generated renewed interest in elucidating the molecular mechanisms of action of well-established antitubercular drugs as an aid in developing new therapeutics (4.)

The inhA gene from Mycobacterium smegmatis,Mycobacterium avium, Mycobacterium bovis, andM. tuberculosis confers co-resistance to both isoniazid and ethionamide when transformed on a multicopy plasmid into M. smegmatis (5). In addition, a spontaneous laboratory-derived mutation, which results in a single amino acid substitution [Ser94 to Ala (S94A)] in the M. smegmatis and the M. bovis (100% identical to M. tuberculosis) genes, is sufficient to confer co-resistance to both isoniazid and ethionamide in M. smegmatis by allele exchange. Subsequent studies revealed that InhA is an enoyl-acyl carrier protein (ACP) reductase (6). Enoyl-ACP reductases catalyze the nicotinamide adenine dinucleotide (NADH)–dependent reduction of the double bond at position two of a growing fatty acid chain linked to ACP, an enzymatic activity common to all known fatty acid biosynthetic pathways. InhA preferentially reduces long-chain substrates (those containing 16 or more carbon atoms). Mycobacteria utilize the products of InhA catalysis to create mycolic acids, very long chain (C40 to C60) α-branched fatty acids, which are important components of mycobacterial cell walls (7). In addition to the genetic data, the fact that isoniazid inhibits mycolic acid biosynthesis (8) and that mycolic acid biosynthesis proceeds normally in cell-free extracts of isoniazid-resistant M. smegmatis, due to multicopy plasmid expression of wild-type InhA or genomic expression of the S94A mutant (5), provides compelling evidence that InhA is the drug target for both isoniazid and ethionamide in mycobacteria.

However, isoniazid itself does not directly interact with InhA. Several lines of evidence support the theory that, to inhibit InhA, isoniazid requires conversion to an activated form of the drug and that a catalase-peroxidase (KatG) participates in isoniazid activation (9): (i) isoniazid, NADH (10), Mn2+ ions, and oxygen (11) are all required for InhA inhibition; (ii) KatG is an efficient catalyst for oxidation of Mn2+ ions (Mn2+ → Mn3+) (12) and the addition of KatG accelerates isoniazid-dependent inhibition of InhA (13); (iii) deletions or mutations in the KatG gene are the single largest determinant of isoniazid resistance in clinical isolates (14) and transformation of the KatG gene into isoniazid-resistant M. tuberculosis restores isoniazid sensitivity (15). It has been proposed that the activated form of isoniazid is an intermediate in the formation of isonicotinic acid, isonicotinamide, and pyridine-4-carboxaldehyde, the products of isoniazid oxidation, none of which inhibits InhA (16).

About 25% of the clinical isolates of isoniazid-resistant M. tuberculosis contain mutations within the promoter or structural regions of the inhA gene, and no inhA mutations have been identified in isoniazid-sensitive isolates (14). Four different clinical isolate mutations (at residues 16, 21, 78, and 95) result in single amino acid substitutions within or near the NADH binding region of InhA (17). The location of the S94A mutation is also within the NADH binding region. As demonstrated by x-ray crystallography, the S94A mutant displays a reduced hydrogen bonding pattern between NADH and the enzyme (18), resulting in an increased Michaelis constant (K m) for NADH (6). Isoniazid-dependent inhibition of wild-type InhA requires the presence of NADH, and inhibition of the S94A mutant occurs only when the concentration of NADH is increased (10), which implies that there is a correlation between the ability of the enzyme to bind NADH and to become inhibited by activated isoniazid. Furthermore, acyl-ACP substrates can prevent isoniazid-dependent inhibition of InhA, which suggests that activated isoniazid interacts with the substrate binding region of InhA (13).

Crystals produced from InhA inhibited by isoniazid (19) were isomorphous to those of the native enzyme (18) and were used to collect an x-ray diffraction data set to 2.7 Å resolution (20). The resulting difference Fourier (F oF c) electron density map showed that the activated form of isoniazid was covalently linked to the NADH within the active site of InhA (21). The modified NADH consists of an isonicotinic-acyl group from isoniazid attached through its carbonyl carbon to the carbon at position four of the nicotinamide ring (Fig.1). The isonicotinic-acyl group replaces the 4S (and not the 4R) hydrogen of NADH, which is the same position in NADH involved in the hydride transfer that occurs during reduction of enoyl-ACP substrates (6). Furthermore, the covalent attachment was verified by mass spectrometry (22). The mass spectrum of isoniazid-inhibited InhA reflects the presence of a compound with a molecular mass of 770 daltons, which is in agreement with the crystallographic model for the isonicotinic-acyl-NADH (Fig. 2).

Figure 1

A portion of the crystallographic model of the isonicotinic acyl-NADH superimposed onto the final 2F oF celectron density map contoured at 1σ. Carbon atoms are green, oxygen atoms are red, nitrogen atoms are blue, and phosphorous atoms are magenta. Starting from the lower right, the two phosphates of NADH are in view. Moving toward the upper left is the nicotinamide ribose and nicotinamide ring of NADH. In the top left is the fragment derived from isoniazid, which retains a pyridine ring and a carbonyl group, referred to here as an isonicotinic acyl group. The isonicotinic acyl group is attached through its carbonyl carbon to the carbon at position four of the nicotinamide ring of NADH. The carbon at position four of the nicotinamide ring of the isonicotinic acyl-NADH inhibitor is tetrahedral and retains one hydrogen. Produced with the program O (32).

Figure 2

Mass spectra showing that the inhibitor bound to InhA is a compound with an apparent mass of 770 daltons, which is in agreement with the chemical structure of isonicotinic acyl-NADH. (A) InhA alone shows no significant components within the mass/charge (m/z) range of 650 to 850. (B) InhA in the presence of NADH shows several components related to NADH. NADH displays a peak at 666 ([M + H]+). Also present are adducts with one sodium ion (m/z = 688), with one potassium ion (m/z = 704), and with one sodium and one potassium ion (m/z = 726). Adducts with two sodium ions (m/z = 710) or two potassium ions (m/z = 742) are also present at low levels. (C) Isoniazid-inhibited InhA shows the absence of NADH-associated peaks and the presence of a new peak atm/z = 771 ([M + H]+) corresponding to isonicotinic acyl-NADH. Also present are adducts with one sodium ion (m/z = 793), with one potassium ion (m/z = 809), and with one sodium ion and one potassium ion (m/z = 831). Adducts with two sodium ions (m/z= 815) or two potassium ions (m/z = 847) are also present at low levels. Analysis of the m/z peaks with Finnigan Zoomscan confirmed that they are all singly charged species.

Earlier proposals have suggested that isoniazid can be activated to either an isonicotinic acyl anion (23) or an isonicotinic acyl radical (16). The x-ray crystallography and mass spectrometry results show that the carbon at position four of the nicotinamide ring of the isonicotinic acyl-NADH is tetrahedral and retains one hydrogen. This suggests that formation of the isonicotinic acyl-NADH consists of addition of either an isonicotinic acyl anion to NAD+ or an isonicotinic acyl radical to an NAD ·  radical (Fig. 3). Furthermore, isonicotinic acyl-NADH formation occurs within the active site of InhA (not in solution or within KatG) because an incubation mixture containing all the reaction components, except InhA, does not produce a detectable amount of isonicotinic acyl-NADH (24).

Figure 3

Proposed pathway for formation of the isonicotinic acyl-NADH inhibitor of InhA. Two possible scenarios are shown, in which an activated form of isoniazid (isonicotinic acyl anion or radical) covalently attaches to a form of NADH (NAD+ or NAD. radical) within the active site of InhA, while retaining a tetrahedral carbon at position four of the nicotinamide ring. Of the two scenarios, we favor the free radical pathway because isoniazid-dependent inhibition of InhA occurs at a faster rate with NADH than with NAD+ (13). Whether InhA catalyzes NAD. radical formation is not clear, because Mn2+ ions are known to facilitate this process (28). In addition, it is likely that Mn3+ ions facilitate the formation of isonicotinic acyl radicals and KatG participates in isoniazid activation by increasing the rate of the conversion of Mn2+ to Mn3+ ions (11,12). There are reports in the literature in which certain combinations, such as isoniazid and Mn2+ ions (29) as well as isoniazid and peroxidase enzymes (30), are known to generate free radical species. Furthermore, the final products of the KatG oxidation of isoniazid are likely to be formed through an isonicotinic acyl radical intermediate (16) and spin-trapping techniques have identified the isonicotinic acyl radical as one of the products of peroxidase-catalyzed oxidation of isoniazid (31). Produced with chem-D draw (32).

The crystal structure of isoniazid-inhibited InhA provides an explanation for the exquisite specificity of activated isoniazid for InhA. The location and orientation of the isonicotinic acyl group are complementary to those of the surrounding InhA side chains, which create a specific binding pocket for the isonicotinic acyl group (Fig.4). In addition, the size and shape of the pocket could accommodate the isoniazid analog ethionamide. Although KatG is not the activator of ethionamide (13), ethionamide also requires activation and, by analogy to isoniazid, we propose that activated ethionamide inhibits InhA by becoming covalently attached to position four of the nicotinamide ring of NADH by a 2-ethyl isonicotinic thioacyl group.

Figure 4

Molecular contacts between isonicotinic acyl-NADH and the active site of InhA. The isonicotinic acyl group derived from isoniazid is red, the NADH portion of the analog is blue, the side chains of InhA are green, and Ser94, the residue that causes isoniazid resistance when converted into Ala, is magenta. Numbers represent the distance in angstroms between selected atoms. The orientation of the isonicotinic acyl group with respect to the NADH portion of the inhibitor is such that its carbonyl oxygen is positioned about halfway between two hydrogen bond donors, the amide nitrogen of the nicotinamide ring, and the 2′-hydroxyl oxygen of the nicotinamide ribose ring. In addition, the nitrogen atom of the isonicotinic acyl group is within hydrogen-bonding distance of a buried water molecule held by the side chain of Met155. The pyridine ring of the isonicotinic acyl group is surrounded by hydrophobic residues, which include Phe149, Gly192, Pro193, Leu218, Tyr158, and Trp222.

Comparison of the crystal structures of InhA with bound NADH (18) and with bound isonicotinic acyl-NADH reveals that the only significant difference in the protein is location of the side chain of Phe149 (Fig. 5). When isonicotinic acyl-NADH is bound, the side chain of Phe149 has rotated ∼90° and forms an aromatic ring-stacking interaction with the pyridine ring of the isonicotinic acyl group. Although a binding constant that describes the affinity of InhA for isonicotinic acyl-NADH has not been determined, this new structural arrangement would increase the affinity over NADH alone. Similarly, mutations with decreased affinity for NADH (such as S94A) are likely to possess decreased affinity for isonicotinic acyl-NADH.

Figure 5

Superposition of the active sites of InhA with bound NADH (yellow) and with bound isonicotinic acyl-NADH (red). The only conformational difference between the active sites is displacement of the side chain of Phe149. In the native enzyme, NADH binds at the bottom of a large open cavity, and the side chain of Phe149 lies immediately above the nicotinamide ring, appearing to protect the reactive portion of NADH from solvent. In the presence of the isonicotinic acyl group, the side chain of Phe149 has rotated away from the nicotinamide ring, creating space for the isonicotinic acyl group. In addition, the side chain of Phe149 is now oriented adjacent to the pyridine ring of the isonicotinic acyl group, allowing it to participate in an aromatic ring-stacking interaction. Produced with the program Insight (32).

The existence of the isonicotinic acyl-NADH inhibitor, the affinity of InhA for NADH, and the order in which NADH and acyl-ACP substrates bind InhA now become critical to explaining InhA-related isoniazid susceptibility and resistance in M. tuberculosis. Kinetic isotope analysis of InhA has demonstrated that the binding sequence of NADH and long-chain acyl-ACP substrates is not strictly ordered, but there is a preference for NADH binding first (6). This preference would leave most of the wild-type enzyme in the NADH-bound form, available for attack by activated isoniazid. If wild-type InhA cannot release significant amounts of isonicotinic acyl-NADH, this will effectively create permanent inhibition of the enzyme and prevent mycolic acid biosynthesis. In contrast, the decreased affinity of the S94A mutant for NADH would promote acyl-ACP substrates binding before NADH, thereby protecting most of the enzyme from activated isoniazid. When the isonicotinic acyl-NADH is formed on the mutant enzyme, the lowered affinity for NADH promotes release of isonicotinic acyl-NADH, allowing normal substrate catalysis to proceed and resulting in isoniazid-resistant tuberculosis.

The mechanisms of drug action and drug resistance presented here for isoniazid are quite different from those predicted for isoniazid by analogy to diazaborine attachment to the 2′-hydroxyl oxygen of the nicotinamide ribose of the NAD+ of the Escherichia coli enoyl-ACP reductase (FabI) (25). The crystal structure of the complex between isonicotinic acyl-NADH and InhA provides a basis for the design of agents that inhibit InhA without the need for KatG drug activation. The pathway of mycolic acid biosynthesis is essential to mycobacteria and therefore InhA is a logical choice for the design of drugs that control growth of M. tuberculosis.


Table 1

Data collection and model refinement statistics: space group P6222; lattice constants, a =b = 100.53 Å, c = 138.96 Å, α = β = 90.0°, and γ = 120.0°; rms, root mean square. Completeness = (number of F observed/number ofF expected) × 100;R sym = [Σ|(I − 〈I〉)|/Σ(I)) × 100; average II = Σ(II)/number ofI; R value = [Σ(|F observedF calculated|)/Σ(F observed)] × 100; R free = R value of 10% of the data omitted at random.

View this table:
View Abstract

Navigate This Article