Mechanisms of Action and Resistance of the Antimycobacterial Agents

Abstract

This chapter will review the mechanisms of action and resistance of the antimycobacterial agents, with an emphasis on the four first-line antituberculosis drugs isoniazid, rifampin, pyrazinamide, and ethambutol. The mechanisms of action of other drugs used to treat mycobacterial infections, including the fluoroquinolones, aminoglycosides, and the macrolides, will be reviewed elsewhere in this book and this chapter will focus on specific mutations associated with resistance to these agents in M. tuberculosis and M. avium complex. This chapter includes mechanistic studies carried out in M. smegmatis, which, because of its relatedness to M. tuberculosis, its fast-growing nature and lack of pathogenicity, and its relative genetic tractability, is widely used as a model system to study mycobacterial physiology. The phenomenon of M. tuberculosis phenotypic drug tolerance will not be addressed in this section, and discussion will be limited to genotypic mechanisms of drug resistance.

1 Introduction

Historically, mycobacterial infections have been associated with significant morbidity and mortality worldwide. In particular, Mycobacterium tuberculosis is a highly successful human pathogen, causing ~8.7 million cases of active tuberculosis (TB) and ~1.4 million deaths annually [1]. The organism is unique in its ability to establish persistent infection, requiring prolonged treatment with antimicrobials in order to achieve clinical cure. In general, the goals of antituberculosis therapy include rapid reduction of the massive numbers of actively multiplying bacilli in the diseased host, prevention of acquired drug resistance, and sterilization of infected host tissues to prevent clinical relapse. In order to achieve these goals, currently accepted guidelines recommend administration of multiple active drugs for a minimum duration of 6 months [2]. In areas where drug resistance is prevalent and resources permit, M. tuberculosis clinical isolates should be routinely tested for susceptibility to first-line antituberculosis agents in order to optimally guide therapy. The emergence of multidrug-resistant TB (MDR-TB) [3], defined as resistance to the first-line drugs isoniazid and rifampicin, extensively drug-resistant TB (XDR-TB), defined as MDR-TB with additional resistance to fluoroquinolones and at least one of the injectable second-line drugs (capreomycin, kanamycin, and amikacin), and totally drug-resistant tuberculosis (TDR-TB), loosely used for TB strains resistant to a wider range of drugs than those classified as XDR-TB, poses formidable challenges to global TB control efforts [4–6]. The global incidence of MDR-TB is estimated to be ~500,000 cases annually, of which 5–7 % represent XDR-TB.

Prior to the advent of highly active antiretroviral therapy (HAART), disseminated infection with M. avium complex was the most common bacterial opportunistic infection in adults infected with HIV-1 in the developed world, occurring annually in 10–20 % of individuals with AIDS [7, 8]. The availability of HAART, as well as the use of effective prophylaxis with azithromycin or clarithromycin, has reduced the annual incidence of disseminated M. avium complex infection among individuals with advanced HIV disease to less than 1 % per year [9]. Nevertheless, M. avium complex continues to cause disseminated disease in persons with HIV and advanced immunosuppression not receiving or unable to tolerate HAART. In addition, M. avium complex is an important cause of pulmonary infection, particularly in HIV-negative persons with underlying lung disease or other immunosuppression [10]. Infections with other mycobacteria, including M. kansasii, M. genavense, M. hemophilum, M. fortuitum, M. xenopi, M. chelonae, have been reported with increasing frequency, particularly in the setting of HIV infection [11–13].

This chapter will review the mechanisms of action and resistance of the antimycobacterial agents (Table 25.1), with an emphasis on the four first-line antituberculosis drugs isoniazid, rifampin, pyrazinamide, and ethambutol. The mechanisms of action of other drugs used to treat mycobacterial infections, including the fluoroquinolones, aminoglycosides, and the macrolides, will be reviewed elsewhere in this book and this chapter will focus on specific mutations associated with resistance to these agents in M. tuberculosis and M. avium complex. This chapter includes mechanistic studies carried out in M. smegmatis, which, because of its relatedness to M. tuberculosis, its fast-growing nature and lack of pathogenicity, and its relative genetic tractability, is widely used as a model system to study mycobacterial physiology. The phenomenon of M. tuberculosis phenotypic drug tolerance will not be addressed in this section, and discussion will be limited to genotypic mechanisms of drug resistance.

Table 25.1 Mechanisms of action and resistance of the antimycobacterial agents

2 Isoniazid

Isoniazid (isonicotinic acid hydrazide [INH]) has been the most commonly used drug in the armamentarium against M. tuberculosis since recognition of its clinical activity by Robitzek and Selikoff in 1952 [14]. Consisting of a pyridine ring and a hydrazide group (Fig. 25.1), INH is a nicotinamide analog structurally related to the antituberculosis drugs ethionamide and pyrazinamide [15]. Because of its significant bactericidal activity, it has become a critical component of first-line antituberculous chemotherapy. However, in the last two decades, resistance to INH has been reported with increasing frequency, ranging from 3 % to as high as one-quarter of all M. tuberculosis isolates from previously untreated individuals [16–20], with the highest rates of resistance reported from southeast Asia and the Russian Federation [19, 21].

Fig. 25.1

Chemical structures of frontline TB drugs

2.1 Mechanism of Action

Despite the widespread use of INH for more than half a century, its mechanism of action has only recently begun to be elucidated. The drug appears to penetrate host cells readily [22, 23], and diffuses across the M. tuberculosis membrane [24, 25]. INH is a pro-drug, requiring oxidative activation by the M. tuberculosis katG-encoded catalase-peroxidase enzyme [26]. The resulting isonicotinoyl radical reacts nonenzymatically with oxidized NAD⁺ to generate several different 4-isonicotinoyl-NAD adducts [27]. Although the active metabolites of INH have been reported to inhibit multiple essential cellular pathways, including synthesis of nucleic acids [28] and phospholipids [29], and NAD metabolism [30, 31], the primary pathway inhibited by the drug appears to be the synthesis of mycolic acids [32–34], manifesting as a loss of acid-fast staining of the organisms following INH treatment [35].

Mycolic acids are high-molecular-weight α-alkyl, β-hydroxy fatty acids, which are unique outer cell wall components of mycobacteria and other Actinomycetales [36]. Mycolic acids are covalently attached to arabinogalactan and, together with other lipids of the outer leaflet, constitute a very hydrophobic barrier [37] responsible for resistance to certain drugs [38, 39]. Disruption of this hydrophobic barrier is believed to result in a loss of cellular integrity [40]. INH interrupts mycolic acid synthesis by binding tightly to the NADH-dependent enoyl acyl carrier protein (ACP) reductase InhA [41], a component of the fatty acid synthase II system of mycobacteria, which is essential for fatty acid elongation [42]. Genetic, biochemical, and structural data provide compelling evidence that InhA is the primary target for INH in the mycolic acid synthesis pathway. When transferred on a multicopy plasmid, the wild-type inhA gene of M. tuberculosis or M. smegmatis confers INH resistance to M. smegmatis and M. bovis BCG [41], as well as to M. tuberculosis [43]. A missense mutation within the mycobacterial inhA gene leading to the amino acid substitution S94A confers INH resistance to M. smegmatis [41] and M. bovis [44]. In addition, the same single point mutation in inhA (S94A) was sufficient to cause fivefold increased resistance to INH and inhibition of mycolic acid biosynthesis in M. tuberculosis [45]. Interestingly, overexpression of or mutation within inhA also confers resistance to the structurally related second-line antituberculosis drug ethionamide in M. tuberculosis, M. smegmatis, and M. bovis, suggesting that inhA encodes the target of both INH and ethionamide in these mycobacteria [41]. In addition, enoyl reductases, and specifically mycobacterial InhA, have been shown to be targets for the widely used topical disinfectant triclosan, and particular M. smegmatis mutants in inhA are cross-resistant to INH and triclosan [46]. However, although it affects InhA function, INH does not directly interact with InhA. Biochemical and structural studies have shown that InhA catalyzes the NADH-specific reduction of 2-trans-enoyl-ACP, and that the INH-resistant phenotype of S94A mutant InhA is related to reduced NADH binding [42, 47]. X-ray crystallographic and mass spectrometry data revealed that the activated form of INH covalently attaches to the nicotinamide ring of NAD bound within the active site of InhA, causing NADH to dissociate from InhA [47, 48].

Although inhibition of DNA synthesis by INH had been observed long ago [28], only relatively recently was a mechanism of action for this phenomenon proposed. Argyrou and colleagues cloned and overexpressed the M. tuberculosis gene encoding dihydrofolate reductase (DHFR), dfrA, in M. smegmatis and demonstrated a twofold increase in MIC [49]. M. tuberculosis DHFR was shown to selectively bind and co-crystallize with an active INH metabolite which is distinct from that which binds InhA [48]. However, this work requires further biochemical and genetic confirmation. Mutations in dfrA have yet to be reported among INH-resistant clinical isolates of M. tuberculosis.

Despite the identification of specific cellular targets in the last 15 years, the precise mechanism by which INH kills M. tuberculosis remains elusive. Interestingly, depletion of mycolic acids does not necessarily result in loss of viability in other mycobacteria in vitro [50, 51]. However, inhibition of mycolic acid synthesis may more severely compromise the intracellular survival of M. tuberculosis in vivo. It remains to be shown that inhibition of mycolic acid synthesis is both necessary and sufficient for the highly potent in vivo bactericidal activity of INH against M. tuberculosis.

2.2 Mechanisms of Drug Resistance

Spontaneous INH resistance may be observed at a rate of 10⁻⁵ to 10⁻⁶ per bacterium per generation in M. tuberculosis cultures grown in vitro [52]. Because INH is the most commonly used antituberculosis drug, resistance to INH occurs more frequently among clinical isolates than to any other agent [53]. INH resistance varies geographically [54], and may be as high as 20–30 % in some parts of the world [19, 21]. Mutations are most commonly detected in the katG gene, occurring in 50–80 % of INH-resistant clinical isolates, or in the inhA gene, accounting for 15–34 % of INH resistance [53]. Depending on the mutation, the degree of INH resistance may vary from low (0.2 μg/mL) to high (100 μg/mL) [55]. Interestingly, a report from South Africa noted a higher frequency of mutations in the inhA region among patients with XDR-TB than among those with MDR-TB. In this study, mutations in the inhA promoter region accounted for ∼50 to 60 % of cases of MDR-TB and 85–90 % of XDR-TB [56].

2.3 katG

INH resistance among clinical isolates of M. tuberculosis has long been associated with loss of catalase and peroxidase enzyme activities [57]. In general, there is a strong inverse correlation between degree of INH resistance and catalase activity [58]. Zhang and colleagues first demonstrated that deficiency of katG, which encodes the M. tuberculosis catalase-peroxidase enzyme, accounts for the observed resistance to INH in drug-resistant clinical isolates of M. tuberculosis [26, 59]. Mutations in KatG reduce the ability of the enzyme to activate the pro-drug INH, thus leading to resistance. The M. tuberculosis katG gene is situated in a highly variable and unstable region of the genome, perhaps because of the presence of repetitive DNA sequences [60], thereby potentially predisposing to a high frequency of katG mutations. Point mutations in katG are more common than deletions in INH-resistant clinical isolates, and a single point mutation resulting in substitution of threonine for serine at residue 315 (S315T) accounts for the majority of INH-resistant clinical isolates [61–63]. The S315T mutation is associated with a 50 % reduction in catalase and peroxidase activity and high-level INH resistance (MIC = 5–10 μg/mL) [58, 64]. The availability of the crystal structure for M. tuberculosis KatG [65] has provided greater insight into the process of INH activation and may permit a more accurate interpretation of structural and functional effects of mutations implicated in causing INH resistance in clinical isolates. Down-regulation of katG expression has also been shown recently to be associated with resistance to INH [66]. Three novel mutations in the furA-katG intergenic region were identified in 4 % of 108 INH-resistant strains studied; none of these was present in 51 INH-susceptible strains. Reconstructing these mutations in the furA-katG intergenic region of isogenic strains decreased the expression of katG and conferred resistance to INH.

2.4 inhA

INH resistance may arise either from mutations in inhA, resulting in reduced affinity of the enzyme for NADH without affecting its enoyl reductase activity [67], or in the promoter region of the mabAinhA operon [63], resulting in overexpression of the wild-type enzyme. Mutations in the mabA promoter region appear to be more frequent, but overexpression of MabA alone does not confer INH resistance in mycobacteria [68]. Unlike mutations in katG, which can confer low-level or high-level INH resistance, depending on the extent to which catalase-peroxidase enzyme activity is affected, mutations in inhA or in the promoter region of its operon usually confer low-level resistance (MIC = 0.2–1 μg/mL) [69]. Similarly, mutations in the intergenic region oxyR-ahpC can reduce the level of expression of inhA and have been associated with resistance to INH. A study by Dalla Costa et al. [70] found mutations in the intergenic region oxyR-ahpC in 8.9 % of 224 INH-resistant strains studied, confirming its less frequent involvement as a cause of resistance to INH. However, the precise role of these genes in INH resistance has not been completely elucidated.

2.5 Other Genes

The role of mutations in kasA, which encodes a β-ketoacyl ACP synthase of the type II fatty acid synthase system, with respect to INH resistance is controversial. Initial reports identified an association between clinical INH resistance and four independent mutations in kasA [71], but subsequent studies reported the presence of three of these mutations in INH-sensitive M. tuberculosis strains [72, 73]. In addition, although one group reported a fivefold increase in the MIC of INH following kasA overexpression in M. tuberculosis [74], another group found that overexpression of kasA conferred resistance to thiolactomycin, a known KasA inhibitor, but no increased resistance to INH in M. smegmatis, M. bovis BCG, and M. tuberculosis [43]. Using radioactive INH, Mdluli and colleagues reported KasA to be covalently associated with INH and ACP in M. tuberculosis [71], but Kremer and colleagues used anti-KasA antibodies to show that INH treatment in mycobacteria does not result in significant KasA sequestering, and also demonstrated that activated INH does not inhibit KasA activity in an in vitro assay [75]. Although the preponderance of evidence suggests that InhA, and not KasA, is the primary target of INH in the mycolic acid synthesis pathway, the role of kasA mutations in INH resistance of clinical M. tuberculosis isolates requires further investigation.

Mutations in ndh, encoding a NADH dehydrogenase, were first shown in M. smegmatis to confer resistance to INH and ethionamide, as well as to exhibit other phenotypes, including thermosensitive lethality and auxotrophy [76]. Subsequently, ndh mutations were detected in almost 10 % of INH-resistant M. tuberculosis clinical isolates, which did not contain mutations in katG, inhA, or kasA [72]. Defective NADH dehydrogenase, which normally oxidizes NADH and transfers electrons to quinones of the respiratory chain, could lead to an increased ratio of NADH/NAD, which may interfere with KatG-mediated peroxidation of the drug, or displace the INH/NAD adduct from the InhA active site [76].

Mutations in the promoter region of ahpC, leading to overexpression of an alkylhydroperoxide reductase, have been observed in INH-resistant M. tuberculosis [77]. Although rarely found in some INH-resistant strains with apparently intact KatG [78], the ahpC mutation is usually found in KatG-negative INH-resistant M. tuberculosis, presumably as a compensatory mechanism for the loss of catalase-peroxidase activity in such strains [79–81]. AhpC does not appear to play a direct role in INH resistance, since ahpC overexpression in a wild-type reference strain of M. tuberculosis does not appreciably increase the MIC of INH, but mutations in the ahpC promoter region may serve as a useful marker for detection of INH resistance [78].

The M. tuberculosis iniA gene (Rv0342), part of a three-gene operon (Rv0341, Rv0342, Rv0343) induced in the presence of INH, appears to contribute to the development of tolerance to both INH and ethambutol, perhaps functioning through an MDR-pump-like mechanism, although IniA does not appear to directly transport INH from the cell [82]. INH also induces several other genes, including an operon cluster of five genes that code type II fatty acid synthase enzymes and fbpC, which encodes trehalose dimycolyl transferase. Other genes also induced are efpA, fadE23, fadE24, and ahpC, which mediate processes linked to the toxic activity of the drug and efflux mechanisms [83].

Despite the identification of several genetic mutations associated with resistance to INH, as many as a quarter of all clinical INH-resistant isolates do not have mutations in any of the above genes, suggesting alternative mechanisms of INH resistance. Tessema et al. reported 8 % of phenotypically defined isoniazid-resistant strains had no mutations in codon 315 of the katG gene and in the regulatory region of the inhA gene, demonstrating that other mechanisms or mutations in other codons of the katG gene may be responsible for the development of INH resistance in M. tuberculosis strains [84].

3 Rifampin and Other Rifamycins

The rifamycins were first isolated in 1957 from Amycolatopsis (formerly Streptomyces) mediterranei as part of an antibiotic screening program in Italy [85]. Their discovery and widespread use has revolutionized antituberculosis therapy, allowing for the reduction of the duration of treatment from 18 months to 9 months [86]. Although the early bactericidal activity of the rifamycins is inferior to that of INH [87–89], the former are the most potent sterilizing agents available in TB chemotherapy, continuing to kill persistent tubercle bacilli throughout the duration of therapy [90, 91]. Rifampin is a broad-spectrum antibiotic and the most widely used rifamycin to treat tuberculosis. Rifabutin, another rifamycin with reduced induction of the hepatic cytochrome P-450 enzyme system, was originally shown to be effective for the prophylaxis [92] and treatment [93] of M. avium-intracellulare complex infection in persons with advanced HIV disease. Rifapentine is a rifamycin with favorable pharmacokinetic properties [94, 95], including substantially greater maximum serum concentration and extended half-life, which permits highly intermittent therapy for HIV-negative patients who do not have cavitation on chest radiograph and who are sputum culture-negative after 2 months of therapy [96].

3.1 Mechanism of Action

The rifamycins are characterized by a unique chemical structure consisting of an aromatic nucleus linked on both sides by an aliphatic bridge (see figure) [97]. Although structural changes at positions C-21, C-23, C-8, or C-1 may significantly reduce microbiological activity, modifications at C-3 do not alter antituberculous activity. Rifampin is a 3-formyl derivative of rifamycin S; rifabutin is a spiropiperidyl derivative of rifamycin S; and rifapentine is a cyclopentyl-substituted rifampin [97].

The rifamycins are highly protein-bound in plasma, but easily diffuse across the M. tuberculosis cell membrane due to their lipophilic nature [69]. The bactericidal activity of the rifamycins has been attributed to their ability to inhibit mRNA synthesis by binding with high affinity to bacterial DNA-dependent RNA polymerase [98]. The core structure of RNA polymerase, comprising the subunits α₂ββ′ω, is evolutionarily conserved among prokaryotes [99], explaining the antimicrobial activity of the rifamycins against a broad range of bacteria. X-ray crystallographic data examining the interaction of rifampin and RNA polymerase from Thermus aquaticus revealed that rifampin exerts its effect by binding in a pocket between two structural domains of the RNA polymerase β subunit and directly blocking the path of the elongating RNA transcript at the 5′ end beyond the second or third nucleotide [100].

Although the molecular target of rifampin has been well characterized, the precise mechanism by which this interaction leads to mycobacterial killing remains unclear. Interestingly, transcriptional inhibition of the toxin-antitoxin mazEF module by rifampin was shown to trigger programmed cell death in Escherichia coli by reducing cellular levels of the labile antitoxic protein MazE, allowing the unrestrained lethal action of the long-lived toxic protein MazF [101]. Although M. tuberculosis contains homologous toxin-antitoxin gene modules [102, 103], it appears that these modules may play a role in M. tuberculosis growth arrest and persistence under adverse conditions, rather than in programmed cell death, as originally suggested [103].

3.2 Mechanism of Resistance

Although resistance to INH alone is common in M. tuberculosis, resistance to rifampin alone is rare, and more than 90 % of rifampin-resistant isolates are also resistant to INH. Therefore, rifampin resistance has been used as a surrogate marker for multidrug-resistant tuberculosis [104]. Resistance to rifampin develops in a single step at a frequency of 10⁻⁷ to 10⁻⁸ organisms in M. tuberculosis [105].

As in E. coli [106–108], resistance to rifampin in M. tuberculosis arises from mutations in rpoB, which encodes the β-subunit of RNA polymerase [109]. Over 90 % of rifampin-resistant clinical isolates contain point mutations clustered in an 81-base pair region between codons 507 and 533 of the rpoB gene [110, 111]. Although at least 35 distinct rpoB mutant allelic variants have been described [110], amino acid substitutions at one of two positions (Ser₅₃₁ and His₅₂₆) account for the great majority of mutations conferring clinical resistance to rifampin [109, 111–113]. Consistent with the clinical data, selection of spontaneous rifampin resistance in vitro in the M. tuberculosis laboratory reference strain H37Rv yields rpoB mutations only at Ser₅₃₁ and His₅₂₆, with the Ser₅₃₁Leu mutation predominating [114]. Strains with the point mutations CAC → TAC (His → Tyr) at codon 526 and TCG → TTG (Ser → Leu) at codon 531 account for 30 % and 25 %, respectively, of rifampin-resistant clinical isolates in the USA [112], while the same mutations represent 12 % and 47 %, respectively, of predominantly foreign rifampin-resistant isolates [109], suggesting that there may be geographic variation in the frequency of occurrence of particular rpoB mutations [110, 115]. Unlike mutations in codons 531 and 526, which confer high-level resistance to rifampin (MIC > 32 μg/mL) and cross-resistance to all rifamycins [69], mutations in codons 511, 516, and 522 are associated with low- or high-level resistance to rifampin and rifapentine (MIC 2–32 μg/mL), but preservation of susceptibility to rifabutin and the new rifamycin rifalazil [116–118]. In particular, MDR strains with the rpoB point mutation Asp516Val were almost always identified as rifabutin-susceptible [119]. Rare mutations in M. tuberculosis also have been reported in the 5′ region of the rpoB gene, and one such mutation at V176F confers intermediate- to high-level resistance to rifampin [120–123].

Several fast-growing strains of mycobacteria, including M. smegmatis, M. chelonae, M. flavescens, and M. vaccae, are able to inactivate rifampin by ribosylation, leading to inherent resistance to this antibiotic [124, 125]. However, this mechanism of rifampin resistance has not been described in M. tuberculosis. Nevertheless, a small percentage of rifampin-resistant isolates (<5 %) do not contain any mutations in the rpoB gene, suggesting additional molecular mechanisms of rifampin resistance in M. tuberculosis, such as altered rifampin permeability or mutations in other RNA polymerase subunits [110].

An important finding related to resistance to rifampicin is that almost all rifampicin-resistant strains also show resistance to other drugs, particularly to isoniazid. For this reason, rifampicin resistance detection has been proposed as a surrogate molecular marker for MDR [126].

4 Pyrazinamide

The use of pyrazinamide (PZA) in combination with rifampin in modern antituberculosis regimens has permitted shortening the duration of therapy from the previous 9–12 months to the current 6 months [127]. PZA is one of the key components of primary drug therapy against TB, especially when MDR has been diagnosed [128, 129]. Although its bactericidal activity is inferior to that of isoniazid and rifampin [130], the reduction of relapse rates associated with the addition of PZA in 6-month regimens is attributed to the drug’s unique ability to target semi-dormant populations of bacilli residing within an acidic environment [131]. Consistent with this hypothesis, the drug was shown to be more active against old non-growing tubercle bacilli than against young, actively replicating organisms [132]. Interestingly, despite its established activity in vivo [133–136], PZA is inactive against M. tuberculosis grown under normal conditions in vitro [137], and requires acidification of the medium pH to demonstrate antituberculosis activity [138].

4.1 Mechanism of Action

PZA is an amide derivative of pyrazine-2-carboxylic acid and a nicotinamide analog (see figure) [139]. Despite recognition of its antituberculosis activity more than half a century ago [133], the mechanism of action of PZA remains poorly understood. Because of the strict requirement for an acidic microenvironment, it was originally hypothesized that the site of action of PZA was in the macrophage phagolysosome [140], where intracellular M. tuberculosis resides. However, the interior pH of these organelles may be neutral or only slightly acidic [141, 142], well above the range where PZA is active [143]. In addition, although older studies suggested otherwise [140, 144], more recent studies have demonstrated that PZA has neither bacteriostatic nor bactericidal activity against intracellular M. tuberculosis in human monocyte-derived macrophages [145]. An alternative hypothesis is that PZA acts against bacilli residing in acidified compartments of the lung that are present during the early inflammatory stages of infection [131], which is consistent with the clinical observation that the potent sterilizing activity of PZA is limited to the first 2 months of therapy [146–148]. Anaerobic and microaerophilic conditions in vitro have been shown to enhance the activity of PZA against M. tuberculosis, suggesting an alternative explanation for the higher sterilizing activity of PZA against in vivo organisms residing within oxygen-deprived granulomas as compared to bacilli grown under in vitro conditions with ambient oxygen tension [149].

PZA enters M. tuberculosis through passive diffusion and via an ATP-dependent transport system [150]. The drug accumulates intracellularly because of an inefficient efflux system unique to M. tuberculosis [151]. Similar to INH, PZA is a pro-drug, which requires activation to its active form, pyrazinoic acid (POA), by the enzyme pyrazinamidase (PncA) [152, 153]. PncA is expressed constitutively in the cytoplasm of M. tuberculosis [154]. After conversion of PZA into POA, the drug exerts its cidal effect on tubercle bacilli by destabilizing the membrane potential and affecting membrane transport function [155]. The uptake and accumulation of POA in M. tuberculosis is enhanced when the extracellular pH is acidic [151]. The inhibitory effects of POA accumulation initially were attributed to direct inhibition of the mycobacterial fatty acid synthase I (FAS-I) enzyme [156], which is responsible for de novo synthesis of C₁₆ fatty acids from acetyl-CoA primers and their elongation to C_24–26 fatty acyl-CoA derivatives [157, 158]. However, subsequent studies showed that, although the PZA analog 5-chloro-pyrazinamide irreversibly inhibits fatty acid synthesis through inhibition of FAS-I, POA does not directly inhibit purified mycobacterial FAS-I, suggesting that the enzyme is not the immediate target of PZA [159]. It has been proposed that the antituberculosis activity of PZA is not attributable to inhibition of a specific cellular target, but rather may reflect disruption of the proton motive force required for essential membrane transport functions by POA at acidic pH [160]. These findings could explain the enhanced susceptibility to PZA of old, non-replicating bacilli, which have a relatively low membrane potential [151] and reduced ability to maintain membrane energetics [161], as compared to young, actively replicating organisms [160]. Alternatively, the accumulation of POA or other weak organic acids has been hypothesized to lower the intracellular pH sufficiently to inactivate FAS-I or other vital enzymes required for cellular metabolism [159].

Recently, the ribosomal protein S1 (RpsA), a vital protein involved in protein translation and the ribosome-sparing process of trans-translation, was identified as a target of POA [162]. RpsA overexpression in M. tuberculosis conferred increased PZA resistance, and POA was shown to bind RpsA, thereby inhibiting trans-translation. Since the latter process is essential for freeing scarce ribosomes in nonreplicating organisms, inhibition of RpsA by POA may explain the sterilizing activity of PZA against persistent bacilli.

4.2 Mechanisms of Resistance

It has been known for some time that PZA resistance in M. tuberculosis is associated with loss of PZase activity [152]. More recently, pyrazinamide resistance has been attributed to mutations in pncA, the gene encoding PZase [153]. Consistent with these findings, integration of wild-type pncA into a pyrazinamide-resistant pncA mutant of M. tuberculosis is sufficient to restore susceptibility to PZA [163]. M. bovis, another member of the M. tuberculosis complex, is inherently resistant to PZA, most frequently because of a point mutation at codon 169 of the pncA gene, which renders the enzyme nonfunctional [164]. In contrast, studies of PZA-resistant clinical isolates of M. tuberculosis revealed that 72–97 % of these strains may contain various missense mutations, insertions, deletions, or termination mutations throughout the pncA gene or its promoter [165–167]. Recent data indicate that three mutations (D8G, S104R, and C138Y) in PncA confer excessive hydrogen bonding between PZA-binding residues and their neighboring residues, creating a rigid binding cavity, which in turn abolishes conversion of PZA into POA [168].

Resistance to PZA is also mediated by mutations in rpsA [162], which encodes a protein required for trans-translation in nonreplicating bacilli. A rare type of PZA-resistant isolate was found containing a deleted alanine at the C terminus of RpsA, preventing binding of the protein to tmRNA [162, 169, 170].

A small percentage of isolates with high-level PZA resistance contain no mutation in pncA or its promoter, suggesting other potential mechanisms of resistance to the drug [165], including perhaps deficient uptake [150], enhanced efflux, or altered pncA regulation. Alternatively, these findings may reflect the intrinsic problems associated with PZA susceptibility testing, since PZA resistance may be reported erroneously when the culture medium contains excessive bovine serum albumin, or a high inoculum of M. tuberculosis, as both of these conditions may raise the pH of the medium and falsely elevate the MIC of the drug [132]. In fact, reliable methods for susceptibility testing of PZA have only recently been developed, using media with slightly higher pH (6.0–6.2) and higher concentrations of PZA (ranging from 300 μg/mL to as high as 1200 μg/mL, depending on the culture broth) [171].

5 Ethambutol

Ethambutol (EMB; dextro-2,2′-(ethylenediimino)-di-1-butanol), a synthetic compound structurally similar to D-arabinose (see figure) [172], was initially reported to have antituberculosis activity in 1961 [173]. In addition to its role as a first-line agent against M. tuberculosis, EMB is an important component of combination therapy against M. avium complex [9], and the drug exhibits activity against other mycobacteria, including M. kansasii, M. xenopi, and M. marinum [174]. EMB kills only actively multiplying bacilli [175], although its early bactericidal activity is not as potent as that of INH [176, 177]. EMB has poor sterilizing activity, as its addition to a regimen of INH, rifampin, and streptomycin does not improve culture conversion rates after 2 months of therapy [178], and its substitution for PZA increases clinical relapse rates [179]. Because of its modest contribution to the standard regimen of INH, rifampin, and PZA, the principal role of EMB is in the empiric treatment of individuals who are deemed at increased risk for harboring INH-resistant or multidrug-resistant M. tuberculosis, until drug susceptibility results become available.

5.1 Mechanism of Action

The mechanism of action of EMB remains incompletely understood. EMB has been reported to inhibit numerous mycobacterial cellular pathways, including RNA metabolism [175, 180], transfer of mycolic acids into the cell wall [181], phospholipid synthesis [182, 183], and spermidine biosynthesis [184]. However, the primary pathway affected by EMB appears to be that of arabinogalactan biosynthesis [185], through inhibition of cell wall arabinan polymerization [186].

Initial studies showed that treatment of M. smegmatis with EMB results in rapid bacterial disaggregation and morphological changes, consistent with alterations in cell wall composition [187]. A potential explanation for this phenomenon was provided by the observations that EMB inhibits transfer of mycolic acids to the cell wall in M. smegmatis [181], leading to rapid accumulation of trehalose monomycolate, trehalose dimycolate, and free mycolic acids in the culture medium [188]. Subsequently, EMB was shown to inhibit arabinogalactan synthesis, since MIC levels of the drug immediately inhibited the transfer of label from D-[¹⁴C] glucose into the D-arabinose residue of arabinogalactan in EMB-susceptible M. smegmatis, but not in a drug-resistant strain [185]. In addition to inhibiting the synthesis of the arabinan component of the mycobacterial cell wall core polymer arabinogalactan, EMB inhibits biosynthesis of the arabinan of lipoarabinomannan, a lipoglycan noncovalently associated with the cell envelope [189, 190]. The observations that the latter effect is delayed relative to the former [186], and that EMB treatment results in rapid accumulation of β-D-arabinofuranosyl-1-monophosphoryldecaprenol (decaprenol phosphoarabinose) [191], an intermediate in arabinan biosynthesis, suggested that the primary site of EMB action is not on de novo synthesis of D-arabinose or on its activation, but rather in the final polymerization steps [186].

Using target overexpression by a plasmid vector as a selection tool, Belanger et al. demonstrated that the translationally coupled embA and embB genes of M. avium are both necessary and sufficient to render a naturally susceptible M. smegmatis strain resistant to EMB [192]. Subsequently, the embCAB gene cluster encoding the homologous arabinosyl transferase enzymes EmbA, EmbB, and EmbC was cloned, sequenced, and characterized in M. tuberculosis [193]. Although it has been proposed that these genes constitute an operon, there is evidence to suggest that the embB gene can be expressed from a unique promoter [194], the location of which remains unknown. The Emb proteins are thought to be integral membrane proteins with 12 transmembrane domains and a large carboxyl-terminal globular region of approximately 375 amino acids [193, 195]. Genetic and biochemical studies have shown that the EmbA and EmbB proteins are involved in the formation of the proper terminal hexaarabinofuranoside motif during arabinogalactan synthesis [196], while EmbC is involved in lipoarabinomannan synthesis [197]. Since the majority of EMB-resistant clinical isolates contain mutations in embB (see below) [193, 195, 198], the EmbB protein has been proposed as the main target of EMB, although X-ray crystallographic data supporting this interaction are lacking.

5.2 Mechanisms of Resistance

In M. smegmatis, high-level resistance to EMB appears to require multiple steps, including overexpression of the Emb proteins, as well as mutations in the conserved region of EmbB or further increases in protein expression levels [193]. Resistance to EMB in M. tuberculosis is usually associated with point mutations in the embCAB operon, commonly involving amino acid substitutions at codon Met306 of the embB gene [193, 195, 198]. EmbB mutations have been identified in 30–70 % of EMB-resistant isolates of M. tuberculosis [193, 198–200]. Mutations in the embB gene were reported to be associated with high-level EMB resistance [201], with the mutations Met306Leu or Met306Val yielding a higher MIC (40 μg/mL) than the Met306Ile substitution (20 μg/mL) [198]. However, a study of 183 epidemiologically unlinked M. tuberculosis isolates collected in St. Petersburg, Russia, detected the presence of embB mutations at codon 306 in 48 % of EMB-resistant isolates, but also in 31 % of EMB-susceptible isolates, suggesting that embB mutations may not be sufficient to confer resistance to EMB, or the presence of a compensatory mutation that reverses the EMB resistance phenotype of embB mutants [202]. Interestingly, the discrepancy in phenotypic and genotypic EMB resistance tests was restricted to strains already resistant to other antituberculosis drugs; specifically, embB mutations at codon 306 were noted in 40 of the 69 (60 %) of EMB-susceptible strains resistant to isoniazid, rifampin, and streptomycin but none of the 43 pan-susceptible strains [202]. Huang et al. identified several novel mutations in embB, including at codon 319 and codon 497 [203]. Interestingly, Lacoma et al. found that two EMB-susceptible strains harbored a mutation at codon 306 [204].

Nucleotide polymorphisms in the embC-embA intergenic region have been reported in association with resistance-associated amino acid replacements in EmbA or EmbB, suggesting that these intergenic mutations represent secondary or compensatory changes [195]. Other potential mutations involved in EMB resistance include a Gln379Arg replacement in M. tuberculosis embR, a homologue of the synonymous gene encoding a putative transcriptional activator of embAB in M. avium [192], as well as mutations in rmlD and rmlA2, which encode proteins involved in rhamnose modification [195]. In addition, mutations associated with EMB resistance have been described in Rv0340 [195], a gene transcribed in the same orientation and upstream of the iniBAC operon, which is significantly upregulated following exposure to EMB in vitro [205]. As many as one-quarter of all EMB-resistant M. tuberculosis isolates do not harbor mutations in any of the genes described above, suggesting alternative mechanisms of EMB resistance [206].

6 Aminoglycosides

The discovery of streptomycin (see figure) in the early 1940s represented the first breakthrough in the chemotherapy of tuberculosis [207]. Patients treated with streptomycin and bed rest improved initially compared to those assigned to bed rest alone, but streptomycin monotherapy led inevitably to relapses with streptomycin-resistant M. tuberculosis [208]. Although relapse rates are comparable when streptomycin is substituted for ethambutol as the fourth drug in addition to INH, rifampin, and PZA, the poor oral absorption of streptomycin, which necessitates parenteral administration, as well as the toxicity profile of the aminoglycosides have favored the use of ethambutol in first-line antituberculosis therapy [209]. Other aminoglycosides with significant antimycobacterial activity include kanamycin and amikacin [210]. The detailed mechanisms of action of the aminoglycosides will be addressed elsewhere, and this section will cover mechanisms of aminoglycoside resistance identified specifically in M. tuberculosis.

As in other bacteria, the mode of action of the aminoglycosides against mycobacterial species is through their binding to the 30S ribosomal subunit, which affects polypeptide synthesis and ultimately results in inhibition of translation [211]. In clinically relevant bacteria, resistance to the aminoglycosides most often results from modification of the aminoglycoside molecule. Although genes encoding aminoglycoside-modifying enzymes have been identified in the chromosome of slow-growing mycobacteria [212, 213], and disruption of aminoglycoside 2′-N-acetyltransferase genes has been correlated with increased aminoglycoside susceptibility in M. smegmatis [214], this mechanism of resistance has not been described for M. tuberculosis [105]. Instead, resistance to streptomycin and the other aminoglycosides in M. tuberculosis usually develops by mutation of the ribosome target binding sites. Interestingly, although cross-resistance is observed between amikacin and kanamycin [215], these drugs are not cross-resistant with streptomycin [216], suggesting distinct mechanisms of resistance. Amikacin is a derivative of kanamycin, and the two drugs are structurally related, each containing a 2-deoxystreptamine moiety, while streptomycin is structurally distinct, containing a streptidine moiety. High-level resistance to amikacin and kanamycin with preserved susceptibility to streptomycin has been reported in M. abscessus and M. chelonae [217], and in M. tuberculosis [218] in association with a point mutation at position 1400 (corresponding to position 1408 in E. coli) of the rrs gene, which encodes 16S rRNA [216, 217]. On the other hand, streptomycin resistance in mycobacteria is most commonly associated with mutations in the rpsL gene, which encodes the ribosomal protein S12 [219–224]. Specifically, a missense mutation resulting in a substitution of an arginine for a lysine at codon 43, as well as point mutations in codon 88 account for the majority of rpsL mutations in M. tuberculosis [224]. As in E. coli, streptomycin resistance in M. tuberculosis also commonly arises from rrs mutations, which are usually clustered in the regions surrounding nucleotides 530 or 912 [219, 220, 225]. Unlike most other bacteria, which have multiple copies of the rrs gene, M. tuberculosis and other slow-growing mycobacteria have a single copy of the gene, making it an easily selected mutation site. Thus, alterations in the drug target arising from reduced association of the 16S rRNA with the S12 ribosomal protein lead to an inability of aminoglycosides to disrupt translation of mycobacterial mRNA, thereby resulting in antibiotic resistance. Mutations in rpsL and rrs, which occur in about 50 % and 20 %, respectively, of streptomycin-resistant M. tuberculosis clinical isolates, are usually associated with intermediate-(MIC 64–512 μg/mL) or high-level resistance (MIC > 1000 μg/mL) [105]. The mechanisms responsible for streptomycin resistance in other M. tuberculosis isolates, particularly those with low-level resistance (MIC 4–32 μg/mL), are unknown, but may involve changes in cell envelope permeability and diminished drug uptake [219, 221].

Reeves et al. recently identified aminoglycoside cross-resistance in M. tuberculosis due to mutations in the 5′ untranslated region of whiB7. These mutations led to an increase in the number of whiB7 transcripts and increased expression of both eis (Rv2416c) and tap (Rv1258c) [226]. An association has been reported between M. tuberculosis clinical isolates harboring a variety of mutations in the gidB gene (Rv3919c) and low-level streptomycin resistance [227–229]. These data show that a mutation at either the whiB7 or gidB locus leads to the acquisition of high-level streptomycin resistance at an elevated frequency, which may partly explain why streptomycin resistance can develop so quickly in the host [226, 229]. However, recent results suggest that for gidB this may remain problematic both because the number of mutations required to accurately assess gidB status is large and also because the impact of specific mutations in gidB on the resistance level of the isolate remains unclear [230]. A recent systematic review described additional mutations in the rrs, tlyA, eis promoter, and gidB genes appear to be associated with resistance to the injectable agents amikacin, kanamycin, and/or capreomycin [231]. Mutations in the gene tlyA encoding a 2′-O-methyltransferase of 16S rRNA and 23S rRNA have been implicated in resistance to capreomycin and viomycin [232]. Based on our understanding of aminoglycoside cross-resistance, the best order in which to introduce an injectable agent with the hope of preventing cross-resistance to other injectables would be streptomycin first, then capreomycin, then kanamycin, and finally amikacin [233].

7 Fluoroquinolones

The fluoroquinolones demonstrate excellent activity against several mycobacterial species, including M. tuberculosis, M. kansasii, and M. fortuitum, but not against others, such as M. avium, M. marinum, M. chelonae, and M. abscessus [234]. In particular, drugs of the fluoroquinolone class are highly active against M. tuberculosis both in vitro [235, 236] and in animal models [237–239]. In descending order of activity, fluoroquinolones active against M. tuberculosis include moxifloxacin, sparfloxacin, levofloxacin, ofloxacin, and ciprofloxacin [240]. The 8-methoxy-fluoroquinolone moxifloxacin has bactericidal activity similar to that of INH against M. tuberculosis both in vitro and in the murine model of TB [239, 241, 242], as well as early bactericidal activity comparable to INH in patients with pulmonary TB [243–245]. Unlike gatifloxacin, which appears to lack sterilizing activity against stationary-phase cultures of M. tuberculosis [246], moxifloxacin, when substituted for INH, is able to shorten the duration of therapy needed to effect stable cure in murine TB [247, 248], suggesting that the drug has significant sterilizing activity. Until recently, the fluoroquinolones have been recommended primarily as second-line agents in the treatment of multidrug-resistant tuberculosis [2]. However, the use of a fluoroquinolone as the only active agent in a failing regimen for treatment of multidrug-resistant M. tuberculosis constitutes the most frequent cause of fluoroquinolone resistance [240]. Resistance to fluoroquinolones also may arise extremely rapidly following use of these drugs for other infections [249, 250]. Despite the widespread use of fluoroquinolones to treat a variety of bacterial infections, fluoroquinolone resistance is detected in fewer than 2 % of M. tuberculosis isolates in the United States and Canada [251]. Because of its potent bactericidal and sterilizing activities, moxifloxacin is currently under investigation as a first-line agent in the treatment of tuberculosis. The mechanism of action and detailed mechanisms of resistance to this class of drugs will be discussed in another chapter, and this section will highlight specific mutations identified in fluoroquinolone-resistant M. tuberculosis.

Fluoroquinolones exert their powerful antibacterial activity by trapping gyrase and topoisomerase IV on DNA as ternary complexes and blocking the movement of replication forks and transcription complexes [252]. Unlike most other bacterial species, M. tuberculosis lacks topoisomerase IV but does contain the genes gyrA and gyrB, which encode the A and B subunits, respectively, of DNA gyrase [212]. Consequently, fluoroquinolone resistance in M. tuberculosis is most commonly associated with mutations in the quinolone resistance-determining region (QRDR) of gyrA and gyrB, conserved regions involved in the interaction between the drug and DNA gyrase [240]. Spontaneous fluoroquinolone resistance develops in laboratory strains of M. tuberculosis at frequencies of 2 × 10⁻⁶ to 1 × 10⁻⁸ [253]. The most frequent mutations associated with high-level fluoroquinolone resistance involve substitutions at codons 88, 90, 91, and 94 of the gyrA gene [254–256].

The degree of resistance to fluoroquinolones depends on the specific amino acid substitution in the QRDR, and on the number of resistance mutations present. Therefore, while single mutations in gyrA may confer low-level resistance (MIC > 2 μg/mL) [257], high-level resistance to fluoroquinolones usually requires a stepwise process of at least two mutations in gyrA or the combination of mutations in gyrA and gyrB [255, 257]. Mutations in the QRDR of gyrA do not occur following exposure of M. tuberculosis to low concentrations of fluoroquinolones in vitro, and the selection pressure for mutants in gyrA increases when M. tuberculosis is exposed to high concentrations of fluoroquinolones in vitro [258]. However, mutations in the QRDR region of gyrA are identified in only 42–85 % of fluoroquinolone-resistant clinical isolates, suggesting alternative mechanisms of resistance.

Mutations in the QRDR of the gyrB gene in the absence of gyrA mutations have been identified in some laboratory isolates [257, 258], but are rare in fluoroquinolone-resistant M. tuberculosis clinical isolates [259]. Pantel et al. reported four GyrB substitutions in fluoroquinolone-resistant M. tuberculosis clinical strains (D500A, N538T, T539P, and E540V) suggesting that the GyrB QRDR may extend from positions 500–540 [260]. Functional genetic analysis and structural modeling of GyrB suggest that N538D, E540V, and R485C/T539N conferred resistance to four different fluoroquinolones in at least one genetic background. The GyrB D500H and D500N mutations conferred resistance only to levofloxacin and ofloxacin while N538K and E540D consistently conferred resistance to moxifloxacin only. These findings indicate that certain mutations in gyrB may be sufficient to confer fluoroquinolone resistance, but the level and pattern of resistance varies among different mutations [261].

The lfrA gene, which encodes a multidrug efflux pump, has been shown to confer low-level resistance to fluoroquinolones when expressed on multicopy plasmids in M. smegmatis [262, 263]. Furthermore, expression of MfpA, a member of the pentapeptide repeat family of bacterial proteins [264], which includes McbG in E. coli and Qnr in K. pneumoniae, confers low-level resistance (four- to eightfold increase in the MIC) in M. smegmatis to ciprofloxacin and sparfloxacin [265]. Fluoroquinolone resistance related to MfpA has been attributed to DNA mimicry, as MfpA can directly bind to and inhibit DNA gyrase, thus preventing the formation of the DNA gyrase-DNA complex required for fluoroquinolone binding [266]. High-level resistance of M. smegmatis to ciprofloxacin (MIC = 64 2 μg/mL) also has been associated with overexpression and chromosomal amplification of the pstB gene, which encodes a putative ATPase subunit of the phosphate-specific transport (Pst) system, and disruption of this gene in M. smegmatis results in a twofold increase in sensitivity to fluoroquinolones relative to the isogenic wild-type strain [267]. Although homologues of lfrA, mfpA, and pstB appear to be present in M. tuberculosis [212], mutations or amplifications of these genes have not been identified in fluoroquinolone-resistant clinical isolates.

8 Macrolides

Clinical outcomes of patients with AIDS and disseminated M. avium complex have improved substantially since the introduction of the extended-spectrum macrolides, which are now considered the cornerstone of any potent regimen [268–270]. However, combination therapy with at least one other antimycobacterial agent, usually ethambutol, is necessary to prevent the emergence of macrolide resistance [271–273]. Although clarithromycin and azithromycin are both effective against disseminated M. avium complex infection, several studies directly comparing these two drugs when used in combination with ethambutol suggest trends toward more rapid clearance of bacteremia with clarithromycin [271, 272]. The mechanism of action of the macrolide antibiotics will be covered elsewhere in this book, and this section will focus on known macrolide resistance mutations occurring in M. avium complex.

The macrolides exert their antibacterial effect by binding to the bacterial 50S ribosomal subunit and inhibiting RNA-dependent protein synthesis [274]. However, these drugs have limited activity against wild-type M. tuberculosis [275]. This intrinsic resistance is believed to be associated with expression of the erm gene [276, 277], which is induced upon exposure of M. tuberculosis to clarithromycin [278]. Interestingly, disruption of the pks12 gene, which encodes a polyketide synthase required for synthesis of the major cell wall lipid dimycocerosyl phthiocerol, results in increased susceptibility of M. tuberculosis to clarithromycin relative to its parent strain, but no change in susceptibility to ciprofloxacin or penicillin [279].

In M. avium, spontaneous resistance to clarithromycin has been estimated to occur at a rate of 10⁻⁸ to 10⁻⁹ organisms [280, 281]. Clarithromycin resistance in M. avium isolated from patients with pulmonary disease has been associated with point mutations in the generally conserved loop of domain V of 23S rRNA [282], corresponding to position 2058 in E. coli 23S rRNA, which confer resistance to erythromycin and the macrolides-lincomide-streptogramin B antibiotics [283]. Similarly, clarithromycin-resistant M. avium isolates obtained from patients with AIDS and disseminated M. avium infection contained point mutations in the domain V sequences of 23S rRNA at position 2274 [284]. Mutations in the M. avium 23S rRNA gene are associated with high-level resistance (MIC ≥ 128 μg/mL) [285]. As in M. avium, clarithromycin resistance in M. chelonae and M. abscessus has been associated with point mutations in the 23S rRNA peptidyltransferase region at positions 2058 or 2059 in strains with a single chromosomal copy of the rRNA operon [286, 287]. However, a few clarithromycin-resistant M. avium isolates, particularly with low-level resistance, have been described in which no mutation can be identified in the peptidyltransferase region of the 23S rRNA [281, 288], suggesting alternative mechanisms of drug resistance.

8.1 Cross-Resistance of Antimycobacterial Agents

In general, there is low cross-resistance among most antituberculosis drugs. When present, the degree of cross-resistance depends on the particular mutations and mechanism of drug resistance. Although the most commonly observed INH resistance mutations, i.e., those involving katG, do not generate cross-resistance to other agents, mutations in inhA itself or in its promoter region confer resistance to the second-line antituberculosis drug ethionamide [41, 43, 289]. Mutations in ethA, which confer ethionamide resistance, also yield cross-resistance to thiacetazone and thiocarlide [290].

Mutations in the rpoB gene of M. tuberculosis, particularly in codons Ser531 and His526, have been associated with high-level resistance (MIC > 32 μg/mL) to rifampin and cross-resistance to all the rifamycins. On the other hand, the rpoB mutations L511P, D516Y, D516V, or S522L, which are associated with low- to high-level resistance to rifampin and rifapentine, do not significantly alter susceptibility to rifabutin (MIC 0.5 μg/mL) or rifalazil (MIC 0.01–0.04 μg/mL) [118, 291]. In one study of 25 rifampin-resistant M. tuberculosis isolates (MIC > 2 μg/mL), three of these isolates (12 %) retained susceptibility to rifabutin [292]. Another study of 112 M. tuberculosis clinical isolates detected 73 % cross-resistance between rifabutin and rifampin [293], suggesting that rifabutin may have a role in the therapy of multidrug-resistant tuberculosis in cases where the isolate retains susceptibility to rifabutin.

Cross-resistance among the aminoglycosides is variable. Thus, cross-resistance is usually seen between the 2-deoxystreptamine aminoglycosides amikacin and kanamycin [218], but not between these two drugs and the streptidine aminoglycoside streptomycin [216]. In addition, cross-resistance may be observed between kanamycin and capreomycin or viomycin [294, 295]. Although cross-resistance has not been reported between fluoroquinolones and other classes of antituberculosis agents, mutations associated with individual fluoroquinolone resistance appear to confer cross-resistance to the entire class of drugs [240]. Similarly, resistance to clarithromycin or azithromycin in M. avium complex is usually associated with class-wide resistance to the macrolides [275, 284].

8.2 Mechanism of Spread of Resistance

Although drug resistance may be spread by plasmids or transposons among many bacterial species, including the fast-growing M. fortuitum [262], these mobile genetic elements are not known to cause drug resistance in M. tuberculosis [105]. As described above, drug resistance in M. tuberculosis is caused by mutations in specific chromosomal genes. In general, genetic resistance of M. tuberculosis to specific antimycobacterial drugs does not alter the fitness or virulence of the organism [105], suggesting that drug-resistant isolates may spread to previously uninfected individuals and cause disease equivalent to that caused by drug-susceptible isolates. One notable exception to this rule is in the case of certain INH-resistant M. tuberculosis isolates with reduced catalase activity, which demonstrate decreased virulence in the guinea pig model of tuberculosis [57]. Reduced catalase activity in these isolates correlates well with increased INH resistance, as well as decreased virulence [105]. Molecular genetic studies have shown that integration of a functional katG gene into the genome of INH-resistant, catalase-defective M. bovis restores INH susceptibility, as well as virulence in the guinea pig model [44]. Consistent with these findings, KatG-deficient M. tuberculosis is attenuated relative to a wild-type strain during infection of immunocompetent mice and mouse-derived macrophages, as a result of exposure to the peroxides generated by the phagocyte NADPH oxidase [296]. Although M. tuberculosis clinical isolates containing the S315T mutation appear to retain full virulence and transmissibility in humans [297], it is unknown if other katG mutants, with more greatly reduced catalase activity, are less transmissible or virulent in humans. On the other hand, restoration of virulence may be associated with promoter-up mutations in the ahpC gene, which may compensate for loss of catalase activity resulting from mutations in katG [79]. Full transmissibility and virulence are expected among M. tuberculosis strains in which INH resistance is mediated by mutations in genes other than katG, such as inhA or ndh [105].

The efficient spread of drug-resistant isolates certainly may occur from person to person, as evidenced by the ecologically successful strain W. This strain, which is resistant to as many as 11 antimycobacterial drugs, caused a multidrug-resistant outbreak of tuberculosis in New York City and spread across the United States [298]. However, the emergence of drug resistance in a particular individual is most often not due to primary infection with a drug-resistant isolate, but rather a result of human error. Thus, a prior history of tuberculosis and antituberculosis therapy has been implicated strongly in the causation of multidrug-resistant M. tuberculosis [299]. Factors associated with acquisition of drug resistance include incomplete and inadequate treatment, such as the use of a single drug to treat tuberculosis, the addition of a single drug to a failing regimen and the failure to identify preexisting resistance, as well as inadequate treatment adherence on the part of the patient [300]. Mathematical models predict that the future of the MDR and XDR-TB epidemic will depend to a large extent on the transmission efficiency or relative fitness of drug-resistant M. tuberculosis compared to drug-susceptible strains [301].

8.3 Alternative Agents

M. tuberculosis strains that are resistant to either isoniazid or rifampin may be treated effectively with other first-line drugs. However, strains that are resistant to both drugs, termed “multidrug-resistant” strains, require the use of “second-line drugs,” which are generally less effective and more toxic [299]. These drugs include ethionamide, capreomycin, cycloserine, and paraaminosalicylic acid. Promising new antituberculosis drugs [302, 303], such as the nitroimidazoles (PA-824, OPC-67683) [304, 305], a diarylquinoline (TMC207/bedaquiline) [306], an ethylene diamine (SQ-109) [307], oxazolidinones (Linezolid, PNU-100480/sutezolid) [308, 309], benzothiazinones [310], clofazimine, and thioridazine [311–313], are currently being tested in preclinical or clinical trials [314]. Known mechanisms of action and resistance for each of these drugs will be discussed briefly in this section.

Ethionamide, a synthetic compound structurally related to INH, was shown to have antituberculosis activity in the late 1950s [315]. Although less potent than INH, ethionamide also inhibits mycolic acid synthesis [41, 316]. Ethionamide is a pro-drug requiring activation by the monooxygenase EthA [290, 317, 318], which itself is negatively regulated by the transcriptional repressor EthR. [317] Similarly to INH, ethionamide inhibits mycolic acid synthesis by binding the ACP reductase InhA [41]. Using a cell-based activation method, Wang et al. recently showed that the thioamide drugs ethionamide and prothionamide form covalent adducts with NAD, which are tight-binding inhibitors of M. tuberculosis and M. leprae InhA [319]. Approximately three-quarters of M. tuberculosis isolates with high-level ethionamide resistance (MIC > 50 μg/mL) have mutations in ethA or inhA [289]. Recently, other potential mechanisms of resistance have been identified, as M. tuberculosis mshA deletion mutants were found to be defective in mycothiol biosynthesis and resistant to ethionamide, likely due to defective activation of the drug [320].

Although often grouped together with the aminoglycosides because of similar activity and toxicities, capreomycin is a macrocyclic polypeptide antibiotic isolated from Streptomyces capreolus [210]. Like streptomycin and kanamycin, capreomycin inhibits protein synthesis through modification of ribosomal structures at the 16S rRNA [69]. Recent studies using site-directed mutagenesis have identified the binding site of capreomycin on 16S rRNA helix 44 [321]. In M. smegmatis, mutations in vicA and vicB, which encode components of the 50S and 30S ribosomal subunits, confer resistance to capreomycin and viomycin [294, 295]. In M. tuberculosis, mutations in the rrs gene encoding 16S rRNA have been associated with resistance to capreomycin as well as kanamycin [218, 322]. The rrs mutation A1401G causes high-level amikacin/kanamycin and low-level capreomycin resistance. C1402T is associated with capreomycin (and viomycin) resistance and low-level kanamycin resistance. G1484T has been linked to high-level amikacin/kanamycin and capreomycin/viomycin resistance [323–325]. Various single-nucleotide polymorphisms (SNPs) in the tlyA gene have been also associated with capreomycin resistance [326].

Cycloserine interrupts peptidoglycan synthesis by inhibiting the enzymes D-alanine racemase (AlrA) and D-alanine:alanine ligase (Ddl) [327, 328]. Overexpression of M. tuberculosis AlrA and Ddl on a multicopy vector results in resistance to D-cycloserine in M. smegmatis and M. bovis BCG [327, 328], and M. smegmatis alrA mutants lacking D-alanine racemase activity display hypersusceptibility to D-cycloserine [329]. In E. coli, cycloserine resistance has been attributed to mutations in cycA, which encodes a permease responsible for uptake of the drug [330], but such a mechanism of resistance has not been described for mycobacteria. In addition, mutations in a gene homologous to that encoding E. coli penicillin binding protein 4 (PBP4) were shown to confer resistance to D-cycloserine, as well as to vancomycin in M. smegmatis [331]. However, the mechanism of cycloserine resistance in M. tuberculosis remains unknown.

Paraaminosalicylic acid (PAS) was introduced in 1945 [332, 333]. Although its activity was inferior to that of streptomycin when used alone, the combination of PAS with streptomycin significantly reduced the emergence of streptomycin-resistant organisms [334]. The mechanisms of action and resistance to PAS have not been well characterized, but it has been suggested that the drug may inhibit folic acid biosynthesis and uptake of iron [69]. Recently, PAS-resistant transposon mutants of M. bovis BCG were found to harbor insertions in the thyA gene, which encodes the enzyme thimidylate synthesis in the folate biosynthesis pathway [335]. In addition, mutations in the thyA gene resulting in diminished enzymatic activity were identified in PAS-resistant M. tuberculosis clinical isolates, suggesting that PAS may act as a folate antagonist and that thyA mutations may mediate clinical PAS resistance [335]. However, only slightly more than a third of the evaluated PAS-resistant strains had mutations in thyA, suggesting the existence of additional mechanisms of PAS resistance. Thr202Ala has been reported as the most common mutation associated with PAS resistance, although this mutation has also been identified in several PAS-susceptible isolates [336].

PA-824, a small molecule nitroimidazopyran related to metronidazole, was recently shown to have bactericidal activity against replicating and static M. tuberculosis cultures in vitro, as well as in murine and guinea pig models of tuberculosis [337]. In the mouse model, PA-824 has bactericidal activity comparable to that of INH [338, 339]. However, unlike INH, but like metronidazole, the drug also has potent activity against nonreplicating bacilli exposed to microaerophilic conditions [337, 338]. In addition, PA-824 is highly active against multidrug-resistant clinical isolates of M. tuberculosis (MIC < 1 μg/mL), suggesting no cross-resistance with current antituberculosis drugs [338]. Like metronidazole, PA-824 is a pro-drug which requires bioreductive activation of an aromatic nitro group in order to exert an antitubercular effect [337]. Although the precise mechanism by which PA-824 exerts its lethal effect is unknown, the drug appears to inhibit the oxidation of hydroxymycolates to ketomycolates, a terminal step in mycolic acid synthesis [337]. Similar to INH, resistance to PA-824 is most commonly mediated by mutations which lead to loss of pro-drug activation. Mutations in fgd1 and fbiC result in the loss of a specific glucose-6-phosphate dehydrogenase and its deazaflavin cofactor F₄₂₀, respectively, which together provide electrons for the reductive activation of PA-824 [340]. In addition, resistance to PA-824 has been associated with mutations in Rv3547, a gene encoding a conserved hypothetical protein which appears to be involved in PA-824 activation [340, 341]. Among laboratory strains, the frequency of resistance to PA-824 is slightly less than that to INH, approximately 9.0 × 10⁻⁷ [337].

As in the case of PA-824, mutations in the Rv3547 gene have been identified in strains resistant to OPC-67683, indicating defective drug activation [342]. Resistance to TMC207 is mediated by mutations in the atpE gene encoding the transmembrane and oligomeric C subunit of ATP synthase, typically at positions 63 or 66 [343]. However, more recent studies have shown that a majority of in vitro-generated mutants resistant to TMC207 lacked mutations in atpE, indicating alternative mechanisms of drug resistance [344].

Whether upregulation of ahpC expression, observed in strains resistant to INH, EMB, and SQ109, plays a role in resistance to SQ109 or merely reflects a compensatory metabolic mechanism remains to be determined [345].

While resistance to linezolid in M. tuberculosis clinical isolates is rarely reported, in vitro-selected mutants with high-level resistance to linezolid (MIC = 16–32 mg/L) have been found to contain mutations at G2061T and G2576T in the 23S rRNA gene [346]. On the other hand, mutants with lower level linezolid resistance (MIC = 4–8 mg/L) lack mutations in the 23S rRNA gene, implicating other possible mechanisms of resistance, such as the possible involvement of efflux pumps or other non-ribosomal alterations, as has been shown in M. smegmatis mutants [347, 348]. Sutezolid (PNU-100480) [349] is undergoing Phase I studies and resistance mechanisms are expected to be similar to those of linezolid.

Although spontaneous benzothiazinone-resistant laboratory mutants were found to have a Ser or Gly substitution at codon Cys387 of dprE1, resistance to benzothiazinones has not been reported in clinical M. tuberculosis isolates [350].

Several existing drugs used for other medical conditions have been “repurposed” for the treatment of tuberculosis. A recent study showed that the antitubercular activity of clofazimine, including against MDR-TB, is due to the generation of reactive oxygen species formed as a consequence of Ndh-mediated reduction of clofazimine [351]. Xu et al. reported that clofazimine mutants resistant to 0.48 and 1.92 μg/mL were not observed using the indirect method in mice during 90 days of treatment [352]. The mechanism of antitubercular activity of thioridazine is likely multifactorial [353, 354], as the drug appears to act on enzymes involved in fatty acid metabolism and membrane proteins, particularly efflux pumps, in addition to inhibiting type II NADH:menaquinone oxidoreductase as a phenothiazine [355]. Mechanisms of M. tuberculosis resistance to the phenothiazines remain to be elucidated.