The MEP pathway: Promising drug targets in the fight against tuberculosis

Abstract

Tuberculosis (TB), caused by Mycobacterium tuberculosis (Mtb), is a severe infectious disease in need of new chemotherapies especially for drug-resistant cases. To meet the urgent requirement of new TB drugs with novel modes of action, the TB research community has been validating numerous targets from several biosynthetic pathways. The methylerythritol phosphate (MEP) pathway is utilized by Mtb for the biosynthesis of isopentenyl pyrophosphate (IPP) and its isomer dimethylallyl pyrophosphate (DMAPP), the universal five-carbon building blocks of isoprenoids. While being a common biosynthetic pathway in pathogens, the MEP pathway is completely absent in humans. Due to its unique presence in pathogens as well as the essentiality of the MEP pathway in Mtb, the enzymes in this pathway are promising targets for the development of new drugs against tuberculosis. In this review, we discuss three enzymes in the MEP pathway: DXS, DXR, and IspF, which appear to be the most promising antitubercular drug targets. Structural and mechanistic features of these enzymes are reviewed, as well as selected inhibitors that show promise as antitubercular agents.

Graphical Abstract

INTRODUCTION

Tuberculosis (TB) is now the world’s deadliest infectious disease, and 1.8 million people died from it alone in 2015.¹ In addition to high mortality, a high morbidity rate worsens the situation caused by TB: 10.4 million people fell ill from the disease in 2015.¹ TB is caused primarily by Mycobacterium tuberculosis (Mtb) and this organism is the main infectious agent in humans. TB can also be caused by the Mtb complex, consisting of Mtb, Mycobacterium bovis, and Mycobacterium africanum.² Able to spread via a single, airborne bacillus from a TB patient, Mtb has made TB a pandemic.^3–4 Around one in three individuals in the world has an established TB infection through such successful transmission of Mtb.⁵ Yet, most of these infections do not progress to active disease immediately since alveolar macrophages phagocytose the bacilli after arrival in the lung. Here, the bacilli stay inactive and do not replicate often.⁴ Although latent TB in the nonreplicating form is asymptomatic, the risk of activating it to active disease is increased by many factors including HIV infection, immunodeficiency disorders, and others.⁶ With the challenge of complete elimination of all bacilli and few treatment options against latent TB, this lifetime threat lingers.⁶ Current chemotherapy for drug-susceptible TB consists of an intensive phase for two months using the four first-line TB drugs (isoniazid, rifampicin, pyrazinamide and ethambutol) and a continuation phase for four months using isoniazid and rifampicin.⁷ While it is essential for patients to follow the regimen strictly, the pill burden over such a long duration makes treatment adherence very difficult. Misuse of drugs contributes significantly to the emergence of drug-resistant TB, which has very limited treatment options, often associated with safety and efficacy issues.^8–9 In order to improve patient compliance, to shorten the duration of treatment, and to provide more treatment options for TB, new drugs with novel modes of action are urgently required.

Blocking the biosynthesis of various essential metabolites can be bactericidal and has drawn close attention from many research groups. The TB research community has worked to validate numerous targets from several biosynthetic pathways, propelling such a strategy to a great extent. These targets include ATP synthase, the target of bedaquiline (the first antitubercular drug approved by the US FDA in over 4 decades) used to treat multi-drugresistant tuberculosis¹⁰, several enzymes related to cholesterol catabolism^11–15, isocitrate lyase^16–17, malate synthase¹⁸, and fumarase^19–20 in the glyoxylate cycle and related metabolic pathways, NAD⁺ synthetase^21–22, adenylating enzymes such as biotin protein ligase^23–24, siderophore biosynthesis^25–29, and CoaBC in the coenzyme A biosynthesis pathway³⁰. This review focuses on advances in drug discovery involving disruption of the methylerythritol phosphate (MEP) pathway (also known as the nonmevalonate pathway, NMP), and validation of the enzymes in this pathway as promising antitubercular drug targets.

It was long believed that isopentenyl pyrophosphate (IPP) and its isomer dimethylallyl pyrophosphate (DMAPP), the C₅ precursors of isoprenoids, are synthesized in all organisms via the mevalonate pathway since its discovery in eukaryotes by Conrad Bloch and Feodor Lynen.^31–34 The mevalonate pathway uses two-carbon acetate as the starting material. In 1993, an alternative pathway, the MEP pathway, was found and shown to use completely different starting materials.³⁵ The MEP pathway began with three-carbon pyruvate, rather than acetate. Mtb, among other organisms, depends solely on this pathway for the production of IPP and DMAPP, while humans exclusively utilize the alternate mevalonate pathway for such metabolites.^{34, 36–37} Isoprenoids generated from IPP and DMAPP are crucial for survival of Mtb and other microorganisms. They are used for production of many secondary metabolites including hopanoids which modify membrane properties³⁸, menaquinone or vitamin K₂ that acts in the respiratory electron transport system³⁹, and polyprenyl phosphates that are crucial for bacterial cell wall biosynthesis⁴⁰ (Figure 1). Additionally, DXP, an intermediate in the MEP pathway, feeds into the synthesis of vitamins B₁ and B₆. The importance of these and other downstream molecular products to Mtb survival, as well as the pathway’s absence in human metabolism, make the MEP pathway a promising target for developing antitubercular drugs.^36–37

Figure 1.

The MEP pathway is essential in Mtb. The terminal products from IPP and DMAPP include hopanoids which modify membrane properties, menaquinones or vitamin K₂ that play essential roles in the respiratory electron transport systems, and polyprenyl phosphates that are crucial for bacterial cell wall biosynthesis. DXP, the product of the first enzymatic reaction in the MEP pathway, is a precursor to vitamins B₁ and vitamin B₆.

The MEP pathway consists of seven enzymatic steps starting from pyruvate (Pyr) and Dglyceraldehyde-3-phosphate (D-GAP) to produce IPP and DMAPP (Figure 2). Thiamin diphosphate (ThDP)-dependent DXS catalyzes the first step of the biosynthesis, converting Pyr and D-GAP to DXP with release of carbon dioxide. DXP then undergoes an NADPH-dependent reduction to MEP mediated by IspC/DXR. The third step of the MEP pathway is catalyzed by IspD, transforming MEP to CDP-ME. IspE then catalyzes the ATP-dependent phosphorylation from CDP-ME to CDP-MEP. With release of CMP, the cyclic product MEcPP is generated from CDP-MEP by the enzyme IspF. In the sixth step of the pathway, HMBPP is obtained by the reductive deoxygenation of MEcPP catalyzed by IspG. Finally, IspH converts HMBPP to IPP and DMAPP, the isomeric C₅ precursors of isoprenoids.^41–44 Most of these enzymes play an essential role in MEP biosynthesis.⁴³

Figure 2.

The MEP pathway for the biosynthesis of IPP and DMAPP

While each of the seven enzymes has potential as a drug target,^{42–43, 45} many of the enzymes present particular challenges with respect to inhibitor design and drug discovery. Despite reported crystal structures,^46–47 demonstrated essentiality,⁴³ and initial inhibitors,⁴⁵ IspD and IspE suffer from other obstacles. The active site of IspD is very flexible, highly polar, and lies at a dimer interface.⁴⁶ These factors can lead to challenges when designing successful inhibitors, particularly against mycobacteria with lipophilic cell walls. While IspD may hold promise as an antimalarial target,⁴⁸ its potential for Mtb is less certain. IspE belongs to a family of ATP-binding enzymes.⁴⁹ Selectivity among bacterial and mammalian enzymes is a potential issue.⁴⁵ IspG and IspH, at the end of the MEP pathway, both contain an Fe-S cluster, notoriously difficult to screen and assay against.^{43, 50} While some crystal structures of IspG/H have been reported, none come from pathogenic organisms.⁴⁵ Furthermore, inhibitors that have been described target the Fe-S cluster, raising concerns about selectivity.^{43, 45} For these reasons, we focus in this review on the most promising MEP pathway drug targets for development of antitubercular agents: DXS, DXR, and IspF.

1-DEOXY-D-XYLULOSE 5-PHOSPHATE SYNTHASE (DXS)

The Mtb genome contains two genes encoding for DXR, Rv2682c and Rv3379c. Only Rv2682c has synthase activity due to loss of a critical residue in Rv3379c resulting from an N-terminal truncation.^51–53 DXS is the first enzyme of the MEP pathway and catalyzes one of two rate-determining reactions in the pathway.⁵⁴ With the release of CO₂, DXS catalyzes a condensation reaction joining Pyr and D-GAP to form DXP. DXS sits at a branch point in the pathway as, in addition to the production of isoprenoid precursors, it also plays an important role in the biosynthesis of vitamins B₁ and B₆ (Figure 1).^55–56 Therefore, inhibition of DXS could block multiple metabolic pathways, which could be advantageous in eliminating the bacterium.

Although X-ray crystal structures of DXS from Escherichia coli (E. coli) and Deinococcus radiodurans (D. radiodurans) have been published,⁵⁷ the Mtb DXS crystal structure has not yet been reported. Hence, design of selective inhibitors against Mtb DXS based directly on structural information of the Mtb target has been hindered to some extent. Other strategies, however, have assisted in the design of such inhibitors. Significant structural^57–58 and mechanistic ^53,57–59 understanding of DXS from E. coli, D. radiodurans, and P. falciparum have greatly impacted inhibitor design against these and other microorganisms, including Mtb.

Sequence alignment of DXS homologs shows that Mtb DXS shares 45% and 40% sequence identity compared with D. radiodurans and E. coli DXS, respectively.⁵⁴ Based on crystal structures from the E. coli and D. radiodurans DXS homologs, it is known that the enzyme exists as a homodimer of three-domain monomers, similar to mammalian transketolase (TK).⁵⁷ Interestingly, however, the domain arrangements of the two enzymes are significantly different, affecting the location of each enzyme’s active site. The active site of transketolase lies at the dimer interface, with active site residues stemming from two subunits. The active site of DXS, however, lies between domains of a single monomer. Residues comprising the DXS active site belong to a single subunit.⁵⁷ The location difference noted here is impactful in terms of DXS inhibitor design. As also noted in this report, the ThDP and pyruvate binding sites are conserved across DXS homologs as well as residues outside these binding sites including Arg420 and Arg478, part of the DRAG sequence and suggested to play an important role in the mechanistic features of DXS.^59–60

In the DXS active site, ThDP is activated by deprotonation of the C-2 carbon on the thiazolium ring to generate the ThDP ylide (Figure 3).⁶¹ The ThDP ylide, acting as a nucleophile, then attacks C-2 of Pyr to form the predecarboxylation intermediate, lactyl thiamin diphosphate (LThDP). Following a random sequential order, DXS, LThDP, and D-GAP form a ternary complex, which is required for decarboxylation.^62–64 With the release of CO₂, the product is a carbanion, in equilibrium with its enamine form. The carbanion attacks the aldehyde carbon of D-GAP bound to DXS. The generated tetrahedral intermediate is deprotonated, releasing the ketone product (DXP) as well as the regenerated ThDP ylide for the next round of catalysis.

Figure 3.

Proposed mechanism for the DXP formation catalyzed by DXS

Extensive studies have been conducted on DXS from D. radiodurans, E. coli, and Plasmodium falciparum (P. falciparum) clarifying the enzyme’s unique mechanism.^{62, 65} While the overall chemical reaction catalyzed by DXS combines features of a carboligase (like TK) and a decarboxylase (like pyruvate dehydrogenase, PDH), DXS follows a unique mechanism, impacting inhibitor development.^{42, 53, 58, 62, 66–67} The pyruvate decarboxylation catalyzed by DXS has a unique mechanism. Most ThDP-dependent enzymes such as TK and acetolactate synthase follow a canonical ping-pong kinetic mechanism.⁶⁸ Contrary to this, a ternary complex is formed during DXS catalysis, involving D-GAP, LThDP, and DXS (Figure 3). This complex is formed through random sequential binding, distinguishing it from other enzymes in this class.^{53, 62–63, 65, 67, 69–70} As reported in recent studies, DXS, in the absence of D-GAP, stabilizes the predecarboxylation intermediate, LThDP (Figure 3), presumably via a closed conformation of the region near the active site.^63–64 D-GAP-binding is thought to cause a conversion of the enzyme to an open conformation, which greatly accelerates the rate of decarboxylation by at least 600-fold.^63–64 The unusually large active site volume present in DXS allows for the bulky ternary complex to form. This striking mechanistic feature of DXS, in contrast to TK and PDH, highlights the possibility of selective inhibitor design. A possible strategy would be using pyruvate mimics that selectively inhibit formation of the unique ternary complex, which is indispensable for DXS catalysis.

In addition to the structural and mechanistic features of DXS, feedback inhibition of recombinant DXS cloned from Populus trichocarpa by the ultimate products in the pathway, IPP and DMAPP, has been reported.⁷¹ IPP and DMAPP were found to compete with ThDP when binding to DXS, with K_i values in micromolar range (60–80 μM). Similar feedback regulation was also recently confirmed on DXS from poplar trees.⁷² Despite the species difference, the report indicates that the products from the MEP pathway could downregulate DXS, limiting production of isoprenoids as a common regulatory mechanism.⁷³

A variety of DXS inhibitors have been reported (Figure 4). The first reported DXS inhibitor was ketoclomazone which demonstrated good activity against DXS from Haemophilus influenza (K_i = 23.3 μM) and modest activity against DXS from Chlamydomonas (IC₅₀ = 0.1 mM).^74–76 Freel Meyers and co-workers have reported a series of alkylacetylphosphonates (Figure 4) as DXS inhibitors against a variety of pathogens including Mtb.^77–79 These alkylacetylphosphonates are mechanism-based pyruvate mimics that are unable to undergo decarboxylation, and the size of the alkyl substitution regulates the selectivity for DXS over other ThDP-dependent enzymes. Specifically, the butylacetylphosphonate analog shows single-digit micromolar activity against Mtb DXS (K_i = 4.0 μM), as well as an outstanding selectivity profile over mammalian ThDP-dependent enzymes.⁷⁸ While these phosphonates at first showed modest antimicrobial activity against pathogenic bacteria (e.g., E. coli MIC₉₀ = 2500 μM), activity improved 500-fold by altering growth medium (E. coli MIC₉₀ = 5 μM).⁷⁹ Additionally, the antimicrobial effects of butylacetylphosphonate were shown to be synergistic with fosmidomycin against E. coli.⁷⁸ Finally, Hirsch et al. reported a series of thiamin derivatives as Mtb DXS inhibitors (Figure 4). The active compounds are closely related to ThDP and have sub-micromolar activity against the enzyme.⁸⁰ Taken together, these compounds serve as a proof-of-concept that DXS inhibitors could be viable antitubercular agents.

Figure 4.

Selected inhibitors of DXS in Mtb

The risk of developing resistance is an important consideration when validating a new drug target. Drug resistance in Mtb can be developed intrinsically or extrinsically. As the product of a single gene copy, drug resistance due to a single point mutation in Mtb DXS is less likely to lead to resistance compared with enzymes that stem from several gene copies, offering possible redundancy.⁵⁴ In addition, the high conservation of the active site residues gives confidence that the risk of endogenous resistance through active site mutation is low.⁶⁰ The overexpression of DXS, as a second form of intrinsic drug resistance, could be downregulated by the downstream metabolites IPP and DMAPP, as mentioned above.^{71, 81} Interestingly, spontaneous mutations in E. coli were found to rescue strains depleted of DXS.⁸² In this report, the loss of DXS activity could be bypassed via a single point mutation in the gene encoding E. coli PDH complex E1 subunit.⁸² As with all microorganisms, resistance may develop extrinsically via many avenues including drug efflux.⁸³

DXS is a promising drug target against tuberculosis due to its unique structural and mechanistic features that encourage selective inhibitor development. Further, with its unique mechanism, synergism remains a possible strategy against drug resistance.^{78, 84–85} Essentiality of DXS for bacterial viability, as well as its important role in the MEP and vitamin B1/B6 pathways encourage further pursuit of inhibitors. Report of an Mtb DXS crystal structure could greatly improve our understanding of the roles of active site residues and the ternary complex binding pocket, accelerating design of Mtb DXS inhibitors working through this novel mode of action.

1-DEOXY-D-XYLULOSE 5-PHOSPHATE REDUCTOISOMERASE (IspC/DXR)

DXR is the second enzyme in the MEP pathway and catalyzes the second rate-determining reaction that converts DXP to MEP with the assistance of NADPH.^86–87 As DXP is also a biosynthetic intermediate for the biosynthesis vitamins B₁ and B₆, DXR catalyzes the first committed step in the MEP pathway. The DXR reaction product is unique in bacterial metabolism, giving the possibility to selectively block the biosynthesis of isoprenoids.⁴³ By screening ~20,000 E. coli mutant colonies auxotrophic for 2-C-methylerythritol (ME), the free alcohol of MEP that E. coli can use as a permeable precursor to MEP, Seto and co-workers first identified the gene that encodes DXR and successfully expressed the recombinant protein.⁸⁸ Homologs of this gene were detected in various bacterial genomes as well as in plants, accounting for the production of plastidic isoprenoids.^88–89 The locus of Mtb IspC/DXR is Rv2870c and was shown to be essential for bacterial vitality.⁹⁰ Interestingly, this result is in contrast to the transposon site hybridization (TraSH) analysis that predicted Rv2870c to be non-essential in Mtb.^90–91 Thus, the individual analysis of Rv2870c helped to validate DXR as a promising drug target against Mtb.

DXR catalyzes a two-step reaction. In the first step, with the assistance of redox cofactor NADPH, DXR catalyzes the isomerization of DXP to 2C-methyl-D-erythrose 4-phosphate (MEsP) where a divalent metal cation is absolutely required.⁹² In the second step, MEsP is reduced to MEP.⁴³ Significant effort has been spent understanding the mechanism of DXR catalysis. Two mechanisms have been proposed for the initial isomerization phase of DXR (Figure 5).⁹³ In both mechanisms, the divalent metal cation acts as a Lewis acid to activate the carbonyl of substrate DXP and coordinate its C-3 hydroxyl group. In the first hypothesis, an α-ketol rearrangement occurs via a 1,2-alkyl shift. The C-2 carbonyl is reduced to a tertiary alcohol. Abstraction of the proton from the C-3 hydroxyl group yields a carbonyl at C-3 to produce isomerization intermediate MEsP.^94–95 A second mechanistic hypothesis for the isomerization is a sequential retro-aldol/aldol mechanism.^94–96 Upon deprotonation of the C-4 hydroxyl group, formation of a C-4 carbonyl leads to cleavage of the C-3-C-4 carbon-carbon bond. Two intermediates would be generated: hydroxylacetone enolate and glycoaldehyde phosphate.⁹⁷ An aldol reaction with C-2 as the nucleophile then recombines these intermediates and forms intermediate MEsP. It remains controversial which mechanism is actually at work for the isomerization step. Kinetic isotope effect studies using selectively deuterated substrate (DXP) demonstrated the change in hybridization of both the C-3 and C-4 carbon atoms, arguing against the a-ketol rearrangement mechanism.^{95, 97} On the other hand, NMR studies aiming to detect the hydroxyacetone enolate intermediate were not successful, and exogenously added hydroxyacetone enolate or glycoaldehyde phosphate were not converted to MEP by the enzyme. These data argue against the retro-aldol/aldol mechanism.^{94, 98} It is, however, possible that the putative fragments in the retro-aldol/aldol mechanism are strictly held in the active site, thereby being recalcitrant to detection or being replaced by exogenous sources. The two putative mechanisms were also examined via kinetic studies, where a C-1 mono-fluorinated DXP analog was synthesized.⁹⁹ Thus, the carbonyl group on C-2 is made more electron deficient, which should accelerate a 1,2-migration if the reaction proceeds through the α-ketol rearrangement mechanism.⁹⁹ However, the fluorinated DXP analog appeared to be a poor substrate with a higher K_m value (100 μM) and a lower k_cat value (4.5 s⁻¹), when compared to DXP (K_m = 61μM, k_cat = 21.3 s⁻¹).⁹⁹ Hence, a retro-aldol/aldol mechanism is probably more likely to occur.

Figure 5.

Two proposed mechanisms for the MEsP formation catalyzed by DXR

Contrary to the isomerization step, the reduction mechanism of the DXR catalyzed reaction is more straightforward. The Mtb DXR catalyzed reduction reaction likely follows a random, sequential reaction mechanism where DXP and NADPH bind to the enzyme in a random order.⁹⁶ This mechanism is in slight contrast to the sequential mechanism suggested for E. coli ortholog where NADPH binds before DXP.⁹² Mtb DXR has an ordered product release of NADP⁺ first, followed by MEP dissociation, which is rate-limiting for enzyme turnover.¹⁰⁰ Monitoring NADPH oxidation has been a useful means of studying DXR enzyme kinetics as well as inhibitory activity of novel ligands.^101–105

Fosmidomycin (FR-31564) was isolated from Streptomyces lavendulae and showed activity against many Gram-negative bacteria as well as the protozoan parasite P. falciparum.^{48, 106–109} Fosmidomycin was evaluated for the treatment of urinary tract infections and in Phase I and II clinical trials in 1985, before it was demonstrated to inhibit DXR specifically and block isoprenoid biosynthesis a decade later.^{87, 110–114} Because of the significance of fosmidomycin, research interest in DXR greatly increased and thus DXR has become the most studied enzyme in the MEP pathway. The first apo crystal structure was reported in 2002 by Reuter et al. using E. coli DXR.¹¹⁵ Soon after in 2006, the crystal structure of Mtb DXR was solved by Henriksson et al.¹¹⁶ These early structures revealed that the enzyme is a homodimer, with each subunit consisting of three domains including an N-terminal domain for cofactor binding, a central catalytic domain, and a C-terminal helical domain. Sequence alignments of DXR homologs from Mtb, E. coli, and Zymomonas mobilis showed that the central catalytic domain of the enzymes is highly conserved with 45–50% sequence identity, and the residues interacting with substrate DXP are even more strictly conserved.¹¹⁷ Thus, DXR inhibitors resembling DXP analogs are expected to have great potential against a broad spectrum of bacteria. Nonetheless, when ligands bind strongly to DXR, a dramatic conformational change occurs where a loop migrates, covering the active site. This is known as the “loop-open” to “loop-closed” conformation change.¹¹⁸ Thus, the complexity in predicting the binding mechanism of designed DXR inhibitors has made novel development of such inhibitors challenging despite the availability of sufficient crystallographic information.

Fosmidomycin is a potent inhibitor of E. coli DXR (IC₅₀ = 8.2 nM),¹¹² and its binding is competitive with substrate DXP and noncompetitive with cofactor NADPH.⁹⁶ As would be expected given such high sequence identity, fosmidomycin is also a potent inhibitor of Mtb DXR, with an IC₅₀ value of 80 nM, and a quaternary co-crystal structure with the enzyme, NADPH, and Mn²⁺ was resolved with high resolution.¹¹⁶ Despite the efficacy of fosmidomycin against various pathogens including E. coli (MIC₉₀ = 0.78 μg/mL)¹⁰⁹, Proteus mirabilis (MIC₉₀ = 1.56 μg/mL),¹⁰⁹ and P. falciparum (IC₅₀ = 1 μM),¹¹⁹ as well as its potent inhibition against Mtb DXR, it is completely inactive against Mtb.¹²⁰ As shown by Brown and Parish,⁹⁰ its lack of whole cell Mtb activity is because the highly polar fosmidomycin cannot penetrate the waxy mycobacterial cell wall, a common cause of mycobacterial resistance to many antibiotics. In addition, Mtb lacks a glycerol 3-phosphate transporter (GlpT) used by fosmidomycin and other highly polar compounds to enter nonmycobacteria.^120–121 Thus, the design of novel TB drug candidates via inhibition of DXR presents the following challenge: A polar inhibitor is required for binding to the hydrophilic substrate binding pocket. Such an inhibitor must also be lipophilic enough to penetrate through the greasy bacterial cell wall. The balance between these two physical characteristics represents the backbone of successful design of antitubercular DXR inhibitors.

Some time ago, lipophilic prodrugs of fosmidomycin and its acetyl analog FR900098 were synthesized and examined as antimalarial agents.^{45, 122–123} We adopted this strategy to ask if, as a proof-of-concept, these prodrug esters could display antitubercular activities better than that of the parent compounds. Lipophilic esters of the phosphonate moiety on fosmidomycin/FR900098 were subsequently synthesized, and tested against the cell growth of Mtb.¹⁰¹ These esters mask the polar phosphonate of fosmidomycin with nonpolar groups, gaining the lipophilicity required to penetrate the bacterial cell wall. The esters then undergo hydrolysis to yield the polar phosphonates. Regenerated fosmidomycin/FR900098 then inhibits DXR inside the bacilli to block the biosynthesis of essential isoprenoids and ultimately result in killing the bacterial cell. Examination of several prodrug esters of fosmidomycin/FR900098 showed inhibition of Mtb cell growth, albeit with weak MIC values (25–500 μg/mL, Figure 6).¹⁰¹ Research efforts were then dedicated to the development of more effective Mtb DXR inhibitors.

Figure 6.

Selected fosmidomycin analogs as Mtb DXR inhibitors

Because of the relatively clear mechanism of action of fosmidomycin, its low toxicity, its progress in clinical trials, as well as its ability to work synergistically with other drugs yielding enhanced bactericidal activities,⁸⁴ it is not surprising that abundant research has focused on the synthesis of fosmidomycin analogs (Figure 6). These analogs aim at providing more options as novel Mtb DXR inhibitors, as well as improving antibacterial activity against Mtb. Conspicuously, most of the fosmidomycin analogs reported retained the retro-hydroxamate or hydroxamate moiety to mimic the crucial interaction of fosmidomycin with the divalent metal ion, as well as the phosphonate moiety, which favorably anchors the inhibitor in the active site, enabling potent enzymatic inhibition.

Since the adjacent adenosine-binding pocket of NADPH is druggable,^{50, 116} fosmidomycin analogs with aromatic groups extending from the N-acyl or N-hydroxyl moieties have been synthesized. This bisubstrate strategy aims to occupy both the DXP and NADPH sites in an attempt to gain more potent inhibition.^{102–103, 105, 124} While data supports the bisubstrate binding behavior of this strategy,¹⁰⁵ fosmidomycin so far remains the most potent Mtb DXR inhibitor among these analogs.

Fosmidomycin analogs with modifications on different positions of the scaffold were also examined as Mtb DXR inhibitors. These changes primarily involve substitution on the alpha or beta carbon atom, relative to the phosphorus.^{104, 124–125} Some of these analogs showed more potent inhibition against Mtb Dxr when compared with the parent compound (e.g., an α-aryl-β-thia analog has an IC₅₀ value of 9.2 nM,¹²⁵ Figure 6). However, prodrug modifications to improve Mtb cell penetration have been reported for only a few of these analogs. The best of these prodrug compounds, bearing unsaturation along the propyl chain, shows significant of Mtb cell growth with an MIC value of 9.4 μg/mL.^{103, 105} Additional analogs, with low micromolar activity against the enzyme have unreported Mtb whole cell activity.^{104, 124} Hence, poor uptake of fosmidomycin analogs by Mtb cells remains a challenge, and the corresponding intracellular delivery/fate of these compounds is still underexplored.

Similar to DXS, DXR is encoded by a single gene copy, and thus has a lower risk of developing intrinsic drug resistance clinically due to a single point mutation.⁹⁰ Endogenous resistance through active site mutation is improbable due to high conservation of the central catalytic domain.^{88, 117} Although mutation in E. coli dxr has been reported,⁸⁷ the likelihood of this mutation occurring clinically is unknown. Fosmidomycin resistance was found in strains of bacteria deficient in GlpT,^126–128 the transporter used to take up the drug. As noted above, GlpT is absent in Mtb. Further, lipophilic prodrugs of fosmidomycin analogs are active against GlpT mutants, bypassing this mechanism of resistance.¹²⁶ Fosmidomycin resistance in Burkholderia was also noted due to upregulated efflux,¹²⁷ a potential issue for an antitubercular agent. More recently, P. falciparum strains without Pfhad1 were found to be resistant to fosmidomycin.^{119, 129} The had1 gene encodes for a sugar phosphatase that regulates the MEP pathway.

2C-METHYL-D-ERYTHRITOL 2,4-CYCLODIPHOSPHATE SYNTHASE (IspF)

IspF is the fifth enzyme in the MEP pathway, catalyzing the cyclization and dephosphorylation reactions converting CDP-MEP to MEcPP and releasing CMP concomitantly.⁴³ IspF is encoded by the Rv3581c gene, which was reported to be essential for the viability of Mtb, showing its potential as a valuable target.^{91, 130} Despite over 50 crystal structures of IspF appearing in the literature from different organisms, including many bacteria and P. falciparum, the structure of Mtb IspF remains undetermined.^130–131 Nevertheless, most IspF orthologs share high sequence and structural identity, especially regarding key interactions in the active site.⁴⁵ Moreover, the known crystal structure of IspF from Mycobacterium smegmatis (M. smegmatis) is a reliable homology model to adopt for structure-based inhibitor design against Mtb IspF because of the superior 73% sequence identity between the two enzymes.¹³⁰

The reported crystal structures revealed that IspF, in general, forms a bell-shaped homotrimer, which is tightly associated and contains three active sites at each monomer interface.¹³² Each monomer consists of a four-stranded β-sheet and a flexible two-stranded β-sheet on one side, as well as three α- and two 3₁₀ helices on the other.¹³² This flexible two-stranded β-sheet caps the pocket involved in binding the phosphate moiety of the substrate, and becomes ordered upon substrate binding. Interestingly, this conformational change leaves the remaining substrate-binding pockets unaffected.¹³³ Notably, two divalent ions are also present as they contribute to substrate alignment and are required for facilitating catalysis. They include Zn²⁺, with tetrahedral coordination, and Mn²⁺ or, under physiological conditions, Mg²⁺, with octahedral coordination.⁴³ These metal cofactors act as Lewis acids to activate the diphosphate group of CDP-MEP. This activation facilitates nucleophilic attack by the terminal phosphate (from MEP) onto the β-phosphorus atom (Figure 7). A pentacoordinate transition state is subsequently generated, stabilized by the metal cations. This intermediate then collapses by (re)formation of a phosphoryl bond to produce the product MEcPP, with concomitant release of CMP.³⁷

Figure 7.

Proposed mechanism for the MEcPP formation catalyzed by IspF

The flexibility of IspF complicates inhibitor design. Despite this, IspF has been predicted to be the most druggable target in the MEP pathway due to the lipophilic character of its active site compared with many other enzymes in the pathway.⁵⁰ IspD, for example, has an active site with low lipophilicity and is rather solvent-exposed.⁵⁰ Even with very potent activity against their target enzymes, polar inhibitors can rapidly lose efficacy when introduced to the lipid-rich cell wall of mycobacteria.¹²¹ Thus, the lipophilic character of the IspF active site renders this enzyme notably advantageous as a drug target against Mtb. Thus, rational design of nonpolar Mtb IspF inhibitors is feasible and could show promise in eliminating the bacillus.

IspF is an interesting target against Mtb also because of its involvement in metabolic regulation. Downstream products of the MEP pathway, such as IPP/DMAPP, geranyl diphosphate (GDP), and farnesyl diphosphate (FDP), are reported to bind to a hydrophobic cavity of IspF, indicating a possible feedback regulatory role of this enzyme.^{132, 134} In addition, IspF is stabilized and activated by MEP, the upstream intermediate produced by DXR.^{67, 127} This data outlines an MEP-specific regulatory mechanism using feedforward activation.^{73, 135} MEcPP, the product of IspF, appears to play an important role in general bacterial metabolism as well. It has been suggested that MEcPP acts as an antistressor signal in bacteria that accumulates when bacteria are under oxidative stresses,^136–138 and has a resuscitating effect on non-culturable M. smegmatis.¹³⁹ Efflux of MEcPP in E. coli was measured when certain enzymes were overexpressed.¹⁴⁰ These results suggest the possibility of MEcPP being a branch point in bacterial metabolism, a role that necessitates more elucidation. In terms of drug resistance, the risk of a single point mutation is low in Mtb IspF, which is encoded by a single gene copy.¹³⁰ High sequence and structural identities of the active site residues of most IspF orthologs decreased the chance of endogenous resistance via active site mutation.⁴⁵ Further studies on drug resistance related to IspF remains unreported to date.

Despite the promise of IspF as a drug target against Mtb and some inhibitors reported against the orthologs from different organisms,^141–144 inhibitor design targeting Mtb IspF is relatively underexplored. Part of the challenge toward discovery of an IspF inhibitor stems from the instability of the enzyme substrate, CDP-MEP, the lack of an assay for direct measurement of IspF substrate turnover,¹⁴⁵ and the instability/insolubility of the enzyme itself.¹⁴⁴ To the best of our knowledge, only a few non-cytidine-like thiazolopyrimidine derivatives¹⁴⁴ were tested against Mtb IspF (Figure 8). These show potent activity against the enzyme with the most active compound displaying an IC₅₀ value of 2.1 μM.¹⁴⁴ While the Mtb MIC value of this compound has not been reported, its calculated clogP value is 5.96, indicating high lipophilicity that generally favors cell penetration in mycobacteria.¹⁴⁶ This analysis supports the potential druggability of IspF as a target against Mtb as mentioned above, encouraging more research in rational inhibitor design via this promising antitubercular drug target.

Figure 8.

Selected inhibitors of IspF in Mtb

CONCLUSIONS

Various metabolic biosynthetic pathways have been explored to fight the pandemic threat caused by tuberculosis. The MEP pathway, which is used by Mtb to produce five carbon isoprenoids, stands out as a promising target toward developing antitubercular agents. MEP enzymes DXS, DXR, and IspF appear to be promising drug targets for tuberculosis in our view. Each of these enzymes is reported to be essential for the bacterial viability in Mtb and possesses druggable binding pockets. DXS catalyzes the first reaction in the MEP pathway to synthesize DXP, which is a branch point that is also involved in the biosynthesis of vitamins B₁ and B₆. Inhibitor design against Mtb DXS is attractive due to its importance in Mtb metabolism and unique mechanistic/structural features. With X-ray crystal structures available for Mtb DXR, the design of Mtb DXR inhibitors is greatly facilitated. The Mtb DXR inhibitor fosmidomycin, with an unambiguous mode of action and outstanding safety profile, is a promising parent structure in design of Mtb DXR inhibitors and antitubercular agents. IspF could be the most druggable target in the pathway, due to the lipophilic character of its active site. Reported inhibitors with favorable Mtb IspF inhibition possess clogP values that appear promising for mycobacterial cell wall penetration. These three enzymes in the MEP pathway show great potential as drug targets in fighting tuberculosis. Designing potent and selective inhibitors against these enzymes in Mtb should be of great significance and may facilitate the development of new TB drugs via a novel mode of action.

ABBREVIATIONS

ATP

adenosine triphosphate

CoA

coenzyme A

CoaBC

the bifunctional enzyme catalyzing the second and third steps of CoA biosynthesis

CDP-ME

4-diphosphocytidyl-2C-methyl-D-erythritol

CDP-MEP

4-diphosphocytidyl-2C-methyl-D-erythritol 2-phosphate

CMP

cytidine monophosphate

D. radiodurans

Deinococcus radiodurans

DMAPP

dimethylallyl pyrophosphate

DXP

1-deoxy-D-xylulose-5-phosphate

DXS

1-deoxy-D-xylulose-5-phosphate synthase

E. coli

Escherichia coli

FDP

farnesyl diphosphate

D-GAP

D-glyceraldehyde-3-phosphate

GDP

geranyl diphosphate

H. influenza

Haemophilus influenza

HIV

human immunodeficiency virus

HMBPP

1-hydroxy-2-methyl-2-(E)-butenyl 4-diphosphate

IC₅₀

inhibitory concentration where the response (or binding) is reduced by half

IPP

isopentenyl pyrophosphate

IspC/DXR

1-deoxy-D-xylulose-5-phosphate reductoisomerase

IspD

4-diphosphocytidyl-2C-methyl-D-erythritol synthase

IspE

4-diphosphocytidyl-2C-methyl-D-erythritol kinase

IspF

2Cmethyl-D-erythritol 2,4-cyclodiphosphate synthase

IspG

1-hydroxy-2-methyl-2(E)butenyl-4-diphosphate Synthase

IspH

1-hydroxy-2-methyl-2(E)-butenyl-4-diphosphate reductase

K_m

Michaelis constant

k_cat

turnover number

LThDP

lactyl thiamin diphosphate

M. smegmatis

Mycobacterium smegmatis

MEcPP

2C-methyl-D-erythritol-2,4-cyclodiphosphate

MEP

methylerythritol phosphate

MEsP

2C-methyl-D-erythrose 4-phosphate

MIC₉₀ or MIC₉₉

the lowest concentration of the antibiotic at which 90% or 99% of the isolates were inhibited

Mtb

Mycobacterium tuberculosis

NAD⁺

nicotinamide adenine dinucleotide (oxidized)

NADP⁺

nicotinamide adenine dinucleotide phosphate (oxidized)

NADPH

nicotinamide adenine dinucleotide phosphate (reduced)

P. falciparum

Plasmodium falciparum

PDH

pyruvate dehydrogenase

Pyr

pyruvate

TB

tuberculosis

ThDP

thiamin diphosphate

TK

transketolase

TraSH

transposon site hybridization