Inhibition of Shikimate Kinase and Type II Dehydroquinase for Antibiotic Discovery : Structure-Based Design and Simulation Studies

The loss of effectiveness of current antibiotics caused by the development of drug resistance has become a severe threat to public health. Current widely used antibiotics are surprisingly targeted at a few bacterial functions cell wall, DNA, RNA, and protein biosynthesis and resistance to them is widespread and well identified. There is therefore great interest in the discovery of novel drugs and therapies to tackle antimicrobial resistance, in particular drugs that target other essential processes for bacterial survival. In the past few years a great deal of effort has been focused on the discovery of new inhibitors of the enzymes involved in the biosynthesis of aromatic amino acids, also known as the shikimic acid pathway, in which chorismic acid is synthesized. The latter compound is the synthetic precursor of L-Phe, L-Tyr, L-Phe, and other important aromatic metabolites. These enzymes are recognized as attractive targets for the development of new antibacterial agents because they are essential in important pathogenic bacteria, such as Mycobacterium tuberculosis and Helicobacter pylori, but do not have any counterpart in human cells. This review is focused on two key enzymes of this pathway, shikimate kinase and type II dehydroquinase. An overview of the use of structure-based design and computational studies for the discovery of selective inhibitors of these enzymes will be provided. A detailed view of the structural changes caused by these inhibitors in the catalytic arrangement of these enzymes, which are responsible for the inhibition of their activity, is described.


INTRODUCTION
The increasing development and spread of resistance to current antibiotics have turned ordinary bacterial infections into illnesses that cannot be controlled.Infections from resistant bacteria are now too common and some pathogens have even become resistant to multiple types of antibiotics.The lack of effective antibiotics will undermine our ability to fight infectious diseases and manage the infectious complications common in vulnerable patients undergoing chemotherapy for cancer, dialysis for renal failure, and surgery, especially organ transplantation, for which the ability to treat secondary infections is crucial.In addition, antibacterial drug resistance also has a huge economic impact on individuals as well as healthcare systems and societies [1,2].The European Centre for Disease Prevention and Control (ECDC) estimates that antimicrobial resistance results each year in 25,000 deaths and related costs of over €1.5 billion in healthcare expenses and productivity losses.However, despite this alarming trend, research into the discovery of new antibiotics by large pharmaceutical companies has dwindled dramatically [3,4].Therefore, there is great interest in the discovery of novel drugs and therapies to treat antibiotic-resistant infections and, in particular, of drugs with new mechanisms of action.The most widely used strategy to combat bacterial infections is based on the disruption of the growth cycle by preventing the synthesis and assembly of key components of bacterial processes [5,6].Many drugs that are highly successful in human clinical use mimic a substrate, a transition state or a product of essential enzymes.A detailed knowledge of the catalytic mechanism and the binding determinants of those enzymes can be valuable for the rational design of these mimetics (inhibitors) that can be used as drugs.The increasing availability in recent years of crystal structures of enzyme/inhibitor complexes has undoubtedly been very important to gain a deeper understanding of the enzymatic mechanisms and for the development of potent inhibitors of the selected targets.This success has also benefitted extensively from the recent advances in computational methods that have consolidated their value as important complementary tools that can assist in the elucidation of how these reactions are catalyzed, the binding requirements and the essential motions required for catalysis.
On the other hand, the increasing availability of bacterial genome sequences has created new opportunities for drug discovery.The identification of essential genes and the proteins encoded by them provides an excellent starting point for uncovering novel and important biological processes in microorganisms [7].In fact, current widely used antibiotics are surprisingly targeted at a very small number of key bacterial functions, such as cell wall, DNA, RNA, and protein biosynthesis, and resistance to these antibiotics is widespread and well known [8].Targeting of new pathways will likely play an important role in the discovery of new antibiotics to combat the growing problem of antibiotic-resistant bacteria.For this purpose, much attention has been devoted to the inhibition of the enzymes involved in the shikimic acid pathway in recent years.This is a biosynthetic route through which erythrose-4-phosphate (1) and phosphoenol pyruvate (2) are converted to chorismic acid (9), the precursor in the synthesis of aromatic amino acids, L-Phe, L-Tyr and L-Thr, and folate cofactors, ubiquinone and vitamins E and K (Scheme 1) [9].The absence of this pathway in mammals, combined with its essential nature in certain microorganisms, makes it an attractive target for the development of new antimicrobial agents [10].The important features of the shikimic acid pathway have not gone unnoticed by the scientific community and in the past few years a great deal of effort has been focused on the discovery of new inhibitors of the enzymes involved in this route.In this review, we summarize examples of the use of structure-based design and computational studies in the development of selective inhibi-tors of two key enzymes of the shikimic acid pathway that are essential for bacterial survival.The review is focused on the structural changes (binding, motion, conformation) caused by these compounds, which are responsible for their inhibitory activity.This information helps to rationalize the determinants of binding affinity for inhibitors and could provide guidelines for future inhibitor design.

THE SHIKIMIC ACID PATHWAY ENZYMES -ATRACTIVE TARGETS FOR DRUG DISCOVERY
The enzymes involved in the shikimic acid pathway are present in bacteria, fungi and higher plants, but they are absent in mammals.Some of these enzymes have also been detected in Toxoplasma gondii (which causes malaria) and Plasmodium falciparum extracts, indicating that a complete shikimic acid pathway is present in this parasite [11,12].However, only a single gene encoding the last enzyme of the pathway, chorismate synthase, has been identified conclusively in the genome annotation.It has been suggested that these missing enzymes are either substituted by nonhomologous enzymes that catalyze the same reactions, or that the enzymes are homologous but are too divergent to be identified readily [13].A good example of the utility of this Scheme 1.The shikimic acid pathway.The reaction catalyzed by the dehydroquinase enzyme is also part of the quinic acid pathway that converts quinic acid (10) into protocatechuate (11).The genes encoding the enzymes involved in both pathways are also indicated in brackets.
With the exception of the qa-3 gene that encodes for quinate dehydrogenase, all of the gene-encoding enzymes involved in the quinic and shikimic acid pathways have been sequenced.DAHP synthase has three isoenzymes, one for each aromatic amino acid, and three different genes, aroF, aroG, and aroH, specifically encode each one of these enzymes, which allows control of the production of each amino acid, Tyr, Phe and Thr, respectively.The enzymes involved in the shikimic acid pathway are particularly attractive targets for the development of new anti-tubercular therapies because six of the seven enzymes, specifically, the ones encoded by aroF, aroG, aroB, aroD, aroE, aroK and aroA genes, are essential for Mycobacterium tuberculosis, which is the causative agent of tuberculosis [17].Hence, it is not surprising that many developed inhibitors have focused on the disruption of the growth cycle of this bacterium, which will be discussed below.Also remarkable is the potential of these enzymes for the treatment of Helicobacter pylori infections.This bacterium is the causative agent of gastric and duodenal ulcers and has also been classified as a type I car-cinogen.For this bacterium four enzymes of this route are essential, specifically dehydroquinase, shikimate dehydrogenase, shikimate kinase and chorismate synthase (aroQ, aroE, aroK and aroC genes).This review will provide an overview of the most important inhibitors that have been developed against two key enzymes for M. tuberculosis and H. pylori survival, specifically shikimate kinase (aroK) and type II dehydroquinase (aroD/aroQ), which are the fifth and the third enzymes of the shikimic acid pathway.

TARGETING SHIKIMATE KINASE
Shikimate kinase (SK) catalyzes the stereospecific phosphorylation of the C3 hydroxyl group of shikimic acid (6) by transferring the γ-phosphate group of ATP to the hydroxyl group to provide shikimate 3-phosphate (7) and ADP.In Escherichia coli there are two types of enzymes, SK I, encoded by the aroK gene, and SK II, encoded by the aroL gene.However, most bacteria only have one type of SK.
The SK enzyme is essential in Mycobacterium tuberculosis, Helicobacter pylori, Acinetobacter baylyi, Haemophilus influenzae, Francisella novicida and Pseudomonas aeruginosa.In addition, it has also been shown that the antimicrobial properties of glyphosate are not only due to EPSP inhibition, as mentioned previously, but also to the high concentrations of shikimic acid and protocatechuate that it causes [18,19].These important features have encouraged diverse research groups to study in detail the substrate binding requirements of this enzyme [20][21][22][23][24][25].As a result, several crystal structures of SK, mainly from M. tuberculosis, have been reported, either as binary complexes with ATP, ADP or shikimic acid (6) in the active site, or as tertiary complexes with ADP/shikimic acid (6) or ADP/shikimate 3-phosphate (7) [20][21][22][23].These structures have been an excellent starting point for the structure-based design of inhibitors, which will be discussed below.

Substrate Recognition
SK is a magnesium-dependent enzyme that has two recognition centers, one for shikimic acid (6) and another for the cofactor (ATP), with different key interactions in both cases.SK has three domains (Fig. 1A) [20]: (1) the CORE domain containing five-stranded parallel β-sheets and the P-loop [residues 9-17 in SK from M. tuberculosis (Mt-SK)], which forms the binding site for ATP and ADP; (2) the LID domain (residues 112-124 in Mt-SK), which closes over the active site and has residues that are essential for the binding of ATP; and (3) the substrate binding (SB) domain (residues 33-61 in Mt-SK), which is responsible for the recognition and binding of shikimic acid (6).The SB domain is highly conserved and involves a number of charged residues, specifically Arg136, Arg58, Glu61, Asp34, Arg117 and Lys15, and a lipophilic pocket formed by Phe49, Phe57, Pro118, Gly79, Gly80 and Gly81 (in Mt-SK).Circular dichroism [21], fluorescence [22] and structural [20][21][22][23][24] studies suggested that ATP first binds to the enzyme and induces a large movement of the LID domain over the active site.Shikimic acid (6) subsequently binds to the active site and the LID domain closes over the active site and brings Arg110 and Arg117 into the ATP and SB domains, respectively [20].The available crystal structures of SK with shikimic acid (6) in the active site show that the substrate is strongly anchored to the active site by a salt bridge with the guanidinium group of Arg136 and by hydrogen bonding between the side chain of the essential Asp34 residue and the C4 hydroxyl group (Fig. 1B).Remarkably, shikimic acid (6) binds to the enzyme in an unusual and unstable conformation for a cyclohexene ring with two of the three hydroxyl groups (C4 and C5) in an axial arrangement.By forcing the axial disposition of the C4 and C5 hydroxyl groups, the enzyme controls the equatorial disposition of the C3 hydroxyl group and allows its selective phosphoryl-transfer reaction.The exquisite control of the axial position of the C4 hydroxyl group is achieved by hydrogen-bonding interactions with the essential Asp34 [24], the conserved Gly80 residues and a water molecule, the position of which is frozen by the conserved residues Glu61, Arg58 and Gly80.The axial arrangement of the hydroxyl group in C5 is achieved by two water molecules that form a bridge between the carboxylate group and the C5 hydroxyl group.These water molecules are always located in a nonpolar pocket generated by the conserved Phe49 and Phe57 residues located in α-helices α3 and α4, which -together with Arg117 -helps to seal the active site from the solvent environment.

Essential Motions for Catalytic Turnover
Molecular dynamics (MD) simulations performed with the Michaelis complex (Mt-SK/ADP/7 ternary complex) revealed that the strong salt-bridge interaction between the phosphate group in 7 and the guanidinium group of Arg117 triggers the release of the product from the active site [25] (Fig. 1C).By pulling on the phosphate group in 7, Arg117 causes the breakdown of the favorable interactions of the product with the active site residues, particularly the saltbridge interaction between the C1 carboxylate group and the guanidinium group of Arg136.As a consequence, a conformational change in the product and a large movement of the SB domain takes place (up to 10 Å).During product release, the LID domain also undergoes large movements (up to 6 Å) to allow the Arg117 and Arg136 side chains to expel the product from the active site.
MD simulation studies also highlighted that there are four key structural factors required for the catalysis [25]: (1) the C4/C5 diaxial conformation of shikimic acid (7); (2) a closed form of the LID domain to allow activation of the γphosphate of ATP and stabilization of the transition state by the guanidinium group of the Arg117 side chain; (3) a closed form of the SB domain in order to bring the substrates together for the phosphoryl-transfer reaction and to isolate the substrate within an apolar cavity; (4) flexible LID and SB domains for product release.These key points have been the starting point for the structure-based design of mimetics of the natural substrate, which will be discussed below.

Substrate Mimetics
Considering that the P-loop, which forms the binding site for ATP and ADP, is rather conserved in many ATP-or GTP-binding proteins, several mimetics of shikimic acid (6) were designed for specific binding to the SB binding site of this kinase [25].Based on the aforementioned structural requirements, compounds 12-13 were designed to fix the diaxial conformation of the C4 and C5 substituents in the natural substrate (Fig. 2).Both compounds proved to be reversible competitive inhibitors of the Mt-SK with K i values of 62 and 46 µM, respectively -in both cases below the K m of the enzyme (544 µM).

Fig. (2). Competitive reversible inhibitors of Mt-SK.
The crystal structure of Mt-SK in complex with ADP and compound 12 (PDB entry 4BQS, 2.15 Å) reveals that the cyclohexene ring of inhibitor 12 occupies approximately the same site and has similar binding interactions as the natural substrate (Fig. 3A).Inhibitor 12 provides a tidy closed form of the active site and prevents the opening of the SB binding domain, which is required for turnover.MD simulation studies suggested that the reduction of the double bond of the C3-C5 ether bridge in 12 enhances the lipophilic interactions of this bridge with the conserved residues Arg117 and Pro118.This more effective reduction of the flexibility of the LID domain might also explain its higher inhibitory potency.
The (3R)-( 15) and (3S)-3-aminoshikimic acids (16) were also designed as mimetics of the natural substrate.Both compounds proved to be reversible competitive inhibitors of the Mt-SK enzyme with K i values of 62 and 65 µM, respectively.Docking and MD simulation studies showed that both compounds would interact with the γ-phosphate of ATP through a strong electrostatic interaction and they would cause a dramatic reduction in the flexibility of the LID and the SB domains (Fig. 3B and C).
6-Hydroxy derivatives 14 and 18 proved to be weak competitive reversible inhibitors of the enzyme with K i values of 1050 and 792 µM, respectively.The MD simulation studies carried out with (6R)-6-hydroxyshikimic acid (18) showed that this compound caused significant changes in the water network and caused an opening of the SB domain by nearly 6 Å.During the simulation the ligand also undergoes a conformational change to the thermodynamically more stable conformation, i.e., with C4 and C5 hydroxyl groups in the pseudo-equatorial conformation.This binding conformation of 18 and the greater flexibility of the SB domain seem to be responsible for the lower inhibitory potency of this compound.
In summary, the inhibitory properties and binding mode of compounds 12-18 reveal that the fixation of the diaxial conformation of the C4 and C5 hydroxyl groups recognized by the enzyme is a promising strategy for inhibition because it causes a dramatic reduction of the flexibility of the LID and shikimic acid binding domains.Moreover, the (3R)-( 15) and (3S)-3-hydroxyshikimic acids ( 16) are also good scaffolds for the future generation of inhibitors.

Inhibitors Identified by Screening
Compounds that target the SK enzyme have also been discovered by screening.For example, Simithy et al. used LC-MS and a library of about 400 antimycobacterial compounds, which were previously found by the NIH Tuberculosis Antimicrobial Acquisition and Coordination Facility (TAACF), to identify several compounds that inhibit Mt-SK [26].The most active compounds were 19, 20 and 21, with IC 50 values between 1.9 and 3.8 µM, and these contain an oxadiazole-amide or a 2-aminobenzothiazole moiety in their structure (Fig. 4).The authors used sulfonamide 22, with an IC 50 value of 0.2 µM, previously described by Bandodkar et al. [27], as a positive control for the assay.naphthalene-2-sulfonate 25, which had an IC 50 value of 4.9 µM (Fig. 4) [29].The authors also solved the crystal structure of E114A Hp-SK variant enzyme in complex with 25 (PDB entry 3N2E, 2.53 Å) (Fig. 5).This structure reveals that the ligand prevents the closure of the active site for catalysis by causing a large conformational change in the LID domain.Specifically, the ligand establishes two strong cation-π interactions, one with the conserved residue Arg132, which is responsible for the C1 carboxylate recognition of the substrate/product, and another with the essential Arg116, which is key for ATP binding and product release and it has been suggested that this acts as a Lewis acid during the catalysis [25].In addition, a strong electrostatic interaction between the guanidinium moiety of Arg116 and one of the sulfonate groups fixes the ligand between the LID and the SB domain.
Moreover, the availability of several crystal structures of the SK enzymes, mainly from M. tuberculosis and H. pylori, have encouraged diverse virtual screening studies for the identification of potential inhibitors of these enzymes [30,31].

TARGETING DEHYDROQUINASE
Dehydroquinase (3-dehydroquinate dehydratase, DHQ, EC 4.2.1.10)catalyzes the reversible dehydration of 3dehydroquinic acid (4) to form 3-dehydroshikimic acid (5) (Scheme 1).There are two distinct types of enzymes, known as type I (DHQ1, aroD gene) and type II (DHQ2, aroD/aroQ gene), which have different biochemical and biophysical properties and show no sequence similarity [32,33].DHQ1, which is found in plants, fungi and bacterial species such as Escherichia coli, Staphylococcus aureus and Salmonella typhi, is exclusively biosynthetic and is a heat-labile dimer with a subunit of around 27 KDa.DHQ1 is a class I aldolase enzyme that catalyzes the syn elimination of water in 4 by a multi-step mechanism that involves the formation of a Schiff  base.It has been suggested that DHQ1 may act as a virulence factor in vivo as the deletion of the aroD gene affords satisfactory live oral vaccines, that provides monkeys with protection against oral challenge with live S. flexneri 2457T [34][35][36][37].In contrast, DHQ2, which has both biosynthetic and catabolic roles and is a heat-stable dodecamer (tetramer of trimers) with a subunit of 16-18 KDa, catalyzes the anti elimination of water in 4. In contrast to DHQ1, the reaction mechanism of DHQ2 does not involve any covalent attachment of the substrate to the enzyme through Schiff base formation.DHQ2 is essential in important pathogenic bacteria such as Mycobacterium tuberculosis (aroD gene) and Helicobacter pylori (aroD/aroQ gene) [17].
The fact that two enzymes have evolved to catalyze the same reaction by entirely different mechanisms and stereochemical courses [38] has been used for the design of compounds that specifically inhibit either DHQ1 or DHQ2 enzymes.Of the two, particular attention has been paid to the inhibition of the DHQ2 enzyme as a potential target for the development of anti-tubercular drugs as well as for the treatment of infectious diseases caused by H. pylori.As a result, numerous potent inhibitors of this target have been developed.The resolution of crystal structures of several DHQ2/inhibitor binary complexes has undoubtedly been very important for this success, not only for the development of inhibitors of improved activity but also to gain a deeper understanding of its anti dehydration mechanism and the binding requirements of the DHQ2 enzyme.

Mechanism and Substrate Binding
The DHQ2 mechanism involves the loss of the more acidic pro-S hydrogen from C2 in 4 (Scheme 2).An arginine [Arg17/Arg19 in DHQ2 from H. pylori (Hp-DHQ2) and from M. tuberculosis (Mt-DHQ2), respectively] and a tyrosine (Tyr22/Tyr24 in Hp-DHQ2 and Mt-DHQ2, respectively) have been identified by chemical modification and site-directed mutagenesis studies as being essential for enzyme activity [39,40].Both residues are on the flexible loop that closes over the active site upon substrate binding.In the absence of any crystal structure and based on kinetic isotope effects and pH profile studies carried out on Mt-DHQ2 and Aspergillus nidulans DHQ2 (An-DHQ2), Harris et al. [41] suggested that the elimination proceeds by a stepwise E 1 CB mechanism via enolate intermediate 26 (Scheme 2).Subsequent resolution of the DHQ2 from Streptomyces coelicolor (Sc-DHQ2) with an inhibitor in the active site (PDB entry 1GU1, 1.8 Å) by Roszak et al. [42] allowed a detailed description of the active site and provided a good knowledge of the specific functions of individual amino acid residues.The substrate is strongly bound to the active site by a series of hydrogen bonding interactions through the carboxylate group and the three hydroxyl groups.In particular, the carboxylate binding pocket is a key part in the recognition and this gives rise to four strong hydrogen bonding interactions involving the main chain amide NH groups of Ile102 and Ser103, the side chain OH group of Ser103 and the side chain amide of the conserved Asn75 (in Mt-DHQ2).The latter residue also acts by correctly positioning the C1 hydroxyl group to accept a proton from the conserved His101.In addition, the C5 hydroxyl group interacts by hydrogen bonding with the conserved His81 and Arg112 residues and Asp88* forms the major interaction with the C4 hydroxyl group.The reported crystal structure reveals two key facts: (a) the absence of any suitable residue that could stabilize an enolate intermediate; (b) the presence of a conserved water molecule (henceforth WAT1) in close proximity to the position that would be occupied by the enolate intermediate 26.On the basis of these structural considerations, the authors suggested that the reaction is initiated by the essential tyrosine, which removes the pro-S hydrogen from C2 of 4 to afford an enol intermediate.The final step is the acid-catalyzed elimination of the C1 hydroxyl group -a reaction mediated by the conserved histidine His102/His101 (in Hp-DHQ2 and Mt-DHQ2, respectively) acting as a proton donor.However, the suggested formation of an enol rather than an enolate intermediate did not explain the observed solvent isotope effects and proton inventory of Mt-DHQ2, in which there is a single proton contributing to this effect [41].On the basis of results from quantum mechanical (QM) calculations using different Hamiltonians on Sc-DHQ2, Blomberg et al. [43] subsequently suggested again the formation of enolate intermediate 26 because of its significantly lower energy compared to that of the enol intermediate.
More recently, hybrid QM/MM (quantum mechanics/molecular mechanics) calculations and biochemical studies have provided a good understanding of the catalytic mechanism in atomic detail and provided details of the required motions for the catalytic turnover [44].Remarkably, these studies clarified how the catalytic form of the essential tyrosine (tyrosinate) is formed and the role of the essential arginine in the enzymatic catalysis.These studies showed that an aspartate from the neighboring enzyme subunit (Asp89*/Asp88* in Hp-DHQ2 and Mt-DHQ2, respectively; residues from this subunit are marked with an asterisk) is the residue responsible for the deprotonation of the essential tyrosine to afford the catalytic tyrosinate, which triggers the enzymatic process (Fig. 6).The essential nature of this residue was corroborated by site-directed mutagenesis studies.For Hp-DHQ2, this reaction takes place with the assistance of a water molecule (Fig. 6B), while for Mt-DHQ2 the tyrosine is directly deprotonated by the aspartate residue (Fig. 6A).These results are consistent with the solvent isotope effects and proton inventory [44], which shows that there are at least two protons involved in the reaction, one associated with the elimination step and another/others involved in tyrosine deprotonation.These studies also show that the guanidinium group of Arg17/Arg19 must be close to the essential tyrosine side chain for catalysis.In this arrangement, the arginine controls the appropriate orientation of the tyrosine, which is roughly perpendicular to the cyclohexane ring of the natural substrate, through a cation-π interaction and contributes to decreasing the pKa of the phenol group.The tyrosinate removes the axial hydrogen from C2 of 4 to give an enolate intermediate 32 (Fig. 6C), which is stabilized by the positively charged side chain of the essential Arg17/Arg19 and by a hydrogen bond between the neutral aspartate Asp89*/Asp88* and the conserved water molecule WAT1.
The presence of WAT1 in many crystal structures and its role in the enzymatic mechanism have been widely discussed.WAT1 interacts through hydrogen bonding with a conserved asparagine (Asn10/Asn12 in Hp-DHQ2 and Mt-DHQ2, respectively), the carbonyl group of a conserved proline (Pro9/Pro11 in Hp-DHQ2 and Mt-DHQ2, respectively) and the main-chain amide of a glycine or an alanine (Ala79/Gly78 in Hp-DHQ2 and Mt-DHQ2, respectively).Initially, it was proposed that WAT1 would be involved in the enzymatic mechanism [42].However, QM/MM studies suggest that WAT1 would be mainly involved in the stabilization of the substrate and enol intermediate 32 [44].

Reversible Competitive Inhibitors
Based on the reaction mechanism, a large number of competitive reversible inhibitors have been developed [45,46].In general, these compounds are either analogs of the natural substrate or mimetics of the enolate intermediate 26.The structural changes (binding, motion, conformation) caused by these compounds, which are responsible for their inhibition activity, are discussed below.

Substrate Analogs
Reasoning that the reaction catalyzed by DHQ2 involves the abstraction of the axial hydrogen from C2 of 4, a good strategy to inhibit the enzyme proved to be the substitution of the axial hydrogen by a group that is able to interact with the base responsible for this process, i.e., the tyrosine.Nevertheless, we have shown that substrate analogs equatorially substituted at C2 by benzyl groups are also good reversible competitive inhibitors of the enzyme, since the methylene group of the benzyl group allows the aromatic ring of the inhibitor to be located close to the tyrosine ring.Indeed, the substitution of either the pro-S or pro-R hydrogen at C2 of 4 by a functionalized benzyl group, compounds 27 and 28, respectively, provides good reversible competitive inhibitors of the DHQ2 enzyme (Fig. 7) [47,48].most potent inhibitors of the 2R-substituted derivatives 28, i.e., 4-methoxybenzyl derivative 28b, which has K i values of 26 nM and 170 nM, respectively, has been very important in understanding the role of the aromatic moiety for inhibition (Fig. 8A and C) [49].These crystal structures reveal that these compounds inhibit DHQ2 because their aromatic functional groups block the catalytic arrangement of the tyrosine and the arginine, i.e., it avoids the cation-π interaction between the guanidinium group of Arg17/Arg19 and the phenol group of the tyrosine side chain.In particular, the aromatic moiety of 28b interacts with the catalytic tyrosine by π-stacking and, more importantly, expels the arginine sidechain from the active site.Moreover, MD simulations suggest that the benzyl groups of ligands 28 inactivate the enzyme by causing a significant conformational change of the substrate-covering loop and a dramatic reduction of its flexibility, which is essential for the function of the enzyme [49].In fact, this approach allowed the resolution of the M. tuberculosis loop containing residues 20-25 for the first time (Fig. 8A, PDB entry 2XB8, [49]).It is important to highlight that the 2Rsubstituted derivatives 28 still contain the axial hydrogen that could be removed by the essential tyrosine, the residue that triggers the dehydration reaction.

The resolution of the crystal structures of the binary complexes between Mt-DHQ2 and Hp-DHQ2 and one of the
For the Hp-DHQ2 enzyme and regardless of the type of substituent of the aromatic ring, the binding mode of the 2S-27 and 2R-2-benzyl derivatives 28 are quite similar, which could explain the similar inhibition constants obtained (Fig. 8C and D) [49,50].In all cases, the essential Arg17 is expelled from the active site and faces away from the essential Tyr22.In contrast, for the Mt-DHQ2 enzyme there are significant differences in the binding mode of these compounds depending on the type of substituent on the aromatic ring.Thus, all of the inhibitors cause a dramatic reduction of the loop flexibility, albeit by a different mechanism (Fig. 8C and D) [49,50].For instance, the perfluorobenzyl derivative 27d causes the complete closure of the active site through the formation of a strong π -π stacking interaction with the aromatic ring of Tyr24 and a favorable electrostatic interaction with the guanidinium group of Arg19.In contrast, the 4-methoxybenzyl derivative 27b favors a more open conformation of the loop, with Arg19 facing out of the active site.
The idea of adding substituents at the C3 position arose from the X-ray crystal structure of the Sc-DHQ2/29 binary complex (PDB entry 1GU1) reported by Roszak et al. [42].In the active site of this structure it was observed that, in addition to 29, a molecule each of glycerol and tartrate, originating from the enzyme storage buffer, were present at distances of 3.7 Å and 4.6 Å from the inhibitor 29, respectively.This suggested the existence of additional positions in the vicinity of 29 that could be used to enhance the binding affinity of the enzyme by the inhibitor.In particular, the incorporation at the C3 position of 29 of groups that could interact with the residue triggers the formation of the enolate intermediate 26, the tyrosine.This possibility was first studied by incorporation of aromatic rings at the C3 position in 29 in order to establish π-stacking interactions with the tyrosine ring.These studies led to the discovery of the most potent inhibitor at that time against any DHQ2, the 3nitrophenyl derivative 32.7, with a K i value of 54 nM (against Mt-DHQ2) [54,55].At that time a crystal structure of Mt-DHQ2 with a full description of the substrate-covering loop in which the essential tyrosine and arginine are located was not available and NMR studies [Saturation Transfer Difference (STD) and 1D and 2D-TR-NOESY experiments) were performed to determine its binding mode [56].These studies showed that only one of the conformations existing in solution for this inhibitor is selected when bound to the active site of Mt-DHQ2.In the bound state, the 3-nitrophenyl unit of the potent inhibitor 32.7 is frozen in one orientation relative to the cyclohexene ring, with the nitro group orientated towards double bond side.In this arrangement, the nitro group establishes a strong π-stacking interaction with the phenol group of the essential tyrosine.These initial results were very important in highlighting that the incorporation of aromatic rings at C3 improves dramatically the inhibitory potency and resulted in numerous subsequent structure-based studies to improve this activity further.
Another result of these investigations was the resolution of several crystal structures of diverse enzyme-inhibitor complexes that have been instrumental in gaining a deep understanding of the role of the substituents in the cyclohexene core of 29.Among them, it is worth noting two crystal structures of Mt-DHQ2 solved at 1.5 Å.The first one is the structure with a 2,3-disubstituted compound 39 (39.8, Ar = benzo[b]thiophen-2-yl; 2 R = methylthien-2-yl) (PDB entry 2Y77) bound in the active site (Fig. 10A) [66].This structure revealed that the 2,3-disubstituted compounds might inhibit the enzyme by a double effect.Thus, as for 2-benzyl derivative 28b in PDB entry 2XB8, the aromatic moiety at C3 in the 2,3-disubstituted compounds blocks the entry of the essential arginine side chain into the active site.In addition, the benzyl substituent at C2 prevents the hydrogenbonding interaction of Arg108 with the essential Tyr24 of the substrate-covering loop, i.e., the residue that initiates catalysis (Fig. 10A).These results suggest that compounds that are able to block the alignment between Tyr24 and Arg108 would also be good competitive inhibitors of DHQ2 enzymes.
The second crystal structure that is worth highlighting is the one in complex with an O-alkyl derivative 38 (38.2, Ar = 5-methylbenzo[b]thiophen-2-yl) (PDB entry 2Y71) bound in the active site [66].This compound has K i values of 42 nM [66] and 130 nM [63] against Mt-DHQ2 and Hp-DHQ2, respectively.As in PDB entry 2Y77, the essential arginine is out of the active site.More importantly, this structure reveals that the oxygen atom of the methyleneoxy spacer of the inhibitor is located 3.1 Å away from WAT1 (Fig. 10B).This suggests that an important contribution to the high potency of inhibitors of these types is the interaction between the oxygen atom of the methyleneoxy spacer and WAT1 and this aspect should therefore be considered in the structure-based design of inhibitors.Indeed, the replacement of the oxygen atom of the methyleneoxy spacer of 38 by a carbon atom (ethylene spacer, compounds 36, [68]) leads to a decrease in the inhibition potency by up to 20-fold.The inhibition data, together with the results of MD simulation studies, show that, in general, effects of the geometry and size of the alkyl spacer are more pronounced in the inhibition efficiency of the Hp-DHQ2 enzyme.For both enzymes, compounds with a flexible spacer proved to be more potent than compounds with a more rigid one because it provides the necessary conformational flexibility for the aryl moiety to be accommodated in the active site in an extended conformation.Docking studies suggest that compounds with a three-carbon spacer fit more efficiently into the active site than the corresponding two-carbon spacer analogs because they locate the aromatic ring closer to the aliphatic residues of the enzyme active site.For instance, replacement of the ethylene spacer in 36.1 by a propylene one, compound 36.4,leads to a decrease in the inhibition potency by up to 4-fold (Table 1).Taking into account the important stabilizing contribution that the interaction between WAT1 and either enolate mimetics like compound 38, the natural substrate or the enolate intermediate 32 seem to have in their binding with the DHQ2 enzyme, the WAT1 binding pocket was also explored in the structure-based design of inhibitors of these enzymes [70].This aspect was explored with compounds 40-43 (Fig. 11).
The crystal structures of Mt-DHQ2 in complex with compounds 40 (PDB entry 4CIV, 2.9 Å) and 41 (PDB entry 4CIW, 2.2 Å) show that, in the absence of the benzyl group, the hydroxyl group of the C3 side chain of these compounds does not interact with WAT1 and does not occupy the WAT1 binding pocket.However, in the presence of the benzyl group in the C3 side chain, important differences in the binding mode interactions with WAT1 and the WAT1 binding pocket are observed (Fig. 12).The crystal structures of Mt-DHQ2/42a (PDB entry 4CIX, 2.9 Å) and Mt-DHQ2/43a (PDB entry 4CIY, 2.1 Å) binary complexes, along with the inhibition data and results of MD simulation studies with the corresponding chlorides 42b and 43b, reveals that analogs with the R configuration on the C3 side chain, which also bind to the active site with an extended conformation of this chain, are more potent because, in this arrangement, they promote interactions with WAT1 and the residues located within this pocket.More importantly, the binding affinity is reduced if the ligands occupy the WAT1 binding pocket.The binding requirements of the DHQ2 active site, particularly regarding the carboxylate binding pocket, which is a key part in the enzyme recognition, has been a major challenge in order to achieve optimal in vitro antibacterial activity, especially in the anti-tubercular case.However, an ester prodrug approach proved to improve permeability of the mycobacterial cell and therefore the antibacterial activity [66].The stability of the ester proved to be crucial to achieve improved in vitro activities.In general, the activity dramatically increases with the stability of the ester.Propyl esters were the most potent of the series, with MIC values of 5 µg/mL.These initial results have been very important to prove that the DHQ2 enzyme is a good target for the development of novel antibiotics.However, further improvement of the prodrug approach is required to achieve the nanomolar activity obtained with the isolated enzyme.
In this context, and in the search for drug-like aromatic inhibitors of Mt-DHQ2, several 3-nitrobenzylgallate-based analogs 44-46 with inhibitory activities in the micromolar range have recently been described [71,72] (Fig. 13).

QSAR Studies
The large number of inhibitors described and the available structural data have allowed the development of QSAR models for the Hp-DHQ2 and Mt-DHQ2 enzymes and these aid in rationalizing the determinants of binding affinity and in understanding the differences between the two enzymes.Thus, comparative binding energy (COMBINE) analysis allowed quantification of the contribution of each active site residue to inhibition of the two enzymes [73].
For Hp-DHQ2, these QSAR studies identified the van der Waals interactions with Asn10, Met13, Leu14, Tyr22, Leu103, His102 and Gly78, and Gln63* and Ala90* of a symmetry-related neighboring subunit, as well as the electrostatic and hydrogen bonding interactions with the WAT1 water molecule, as those that contribute most significantly to explaining the differences in inhibitory potency [74].For Mt-DHQ2, the QSAR model is dominated by van der Waals interactions with active site residues Tyr24, Leu13, Leu16, Ile102, Gly77, Asn75 and His101, and with Ala91* and Glu92* of the neighboring subunit, as well as by interactions with the WAT1 water molecule and the crystallographic water molecule close to Arg108 and Tyr24.In general, the key binding interactions of the inhibitors with the enzymes can be organized into five key pockets: (1) the substrate-covering loop (Tyr); (2) the interface between chains; (3) the WAT1 binding pocket; (4) the C2 binding pocket; and (5) the C1 binding pocket.For both enzymes, it is worth noting the crucial contribution of the ligand interactions with the WAT1 water molecule and with the interface pocket close to the active site, which are responsible for significant differences between the two enzymes.In particular, for Hp-DHQ2 these studies suggested that enzyme inhibition might be enhanced by promoting interactions between the ligand and residues located at the 'bottom' of the interface, and avoiding contacts with residues located at the 'upper part' of the interface.For Mt-DHQ2, improving the van der Waals interactions with residues located at the entrance to the interface might provide more potent inhibitors.For both enzymes, extending the contacts with the essential aspartate would be unfavorable.decades.Besides this alarming trend, research into the discovery of new drugs by large pharmaceutical companies has dwindled.Hence, there is great interest in the development of alternative therapies, in particular the identification of new therapeutic targets and the discovery of compounds that target them.In fact, many antibiotics that are highly successful in human clinical use inhibit only a few essential processes for bacterial survival.In this context, the enzymes involved in the biosynthesis of aromatic amino acids are recognized as attractive targets for antibiotic drug discovery because they are essential in many pathogenic bacteria but are absent in mammals.It is therefore not surprising that, in recent years, the enzymes involved in this pathway have attracted a great deal of attention, particularly shikimate kinase and type II dehydroquinase, for the discovery of new anti-tubercular drugs and novel treatments for infections caused by Helicobacter pylori bacterium.Several selective inhibitors of these enzymes have been developed based on the mechanism of action of these enzymes and the available crystal structures.Most of the developed inhibitors are either analogs of the natural substrate recognized by the enzyme or mimetics of the reaction intermediate catalyzed by them.Some examples of inhibitors identified by screening have also been described.This success has benefitted from very important biochemical and structural studies and from the computational calculations performed with the solved crystal structures.These studies have provided an in-depth knowledge of the reaction mechanism and the determinants of binding affinity for these inhibitors.The examples summarized in this review show that, in both cases, those compounds that are able to reduce the flexibility of the substrate-covering loop, which is essential for catalytic turnover, is a good strategy for inhibition.For the DHQ2 enzyme, those compounds that are able to disturb the catalytic arrangement of the essential tyrosine and arginine are very potent inhibitors.Thus, potent inhibitors are afforded by avoiding the entrance into the active site of the arginine side chain or blocking its cation-π interaction with the tyrosine phenol group.Another efficient strategy is either preventing the hydrogen-bonding interaction of the tyrosine with the conserved Arg108 (in Mt-DHQ2) of the active site or establishing a π-stacking interaction with the tyrosine side chain.Moreover, the interaction of the aromatic group of the inhibitors with the interface between chains is responsible for the different inhibition potency against the two DHQ2 enzymes and this is crucial to achieve high affinity ligands.

CONCLUSION
The most challenging issue in targeting the enzymes of this pathway is the high polarity of the inhibitors, mainly due to the carboxylate recognition that is required for good binding affinity.This makes it difficult to achieve good delivery of the inhibitors inside the cells.However, the fact that an ester prodrug approach proved to improve permeability into the mycobacterial cell, which is a Gram-positive bacterium but with characteristics of the Gram-negative ones due to the peptidoglycan layer in its cell wall, to provide optimal MIC values is highly encouraging.This finding has opened the door for the future development of new and more effective ways to deliver these inhibitors inside the cell.More importantly, it has proved that these enzymes are good targets for the development of inhibitors that can be used as drugs.

Fig. ( 1
Fig. (1).A) Michaelis complex of Mt-SK obtained by MD simulation studies [25].The three domains of SK are shown: CORE (green), LID (blue) and SB (magenta).B) Detail view of the relevant binding interactions of shikimic acid (6) and ATP in the Mt-SK active site.Relevant side chain residues are shown and labeled.Hydrogen bonding interactions between the substrates and the enzyme are shown as dashed red lines.C) Representative snapshots of the shikimate-3-phosphate (7) release from the active site of Mt-SK obtained by MD simulations (10 ns).Note the large conformational changes of the LID and the α-helices α3 and α2 of the SB binding domains during product release.

Fig. ( 5
Fig. (5).A) Comparison of the complexes Hp-SK/ADP/7 (PDB entry 3MUF) vs Hp-SK/25 (PDB entry 3N2E).The LID and SB domains of both crystal structures are highlighted (yellow and pink, respectively).Note how compound 25 causes a large conformational change in the LID domain, thus avoiding its closure for catalysis.B) Binding interactions of compound 25 (green) in the binary complex Hp-SK/25 (yellow, PDB entry 3N2E).L-Tartaric acid molecules (blue) from crystallization conditions are also shown.Polar (red) interactions between 25 and the Hp-SK are shown as dashed lines.Relevant residues are shown and labeled.

Scheme 2 .
Scheme 2. Proposed mechanism for the reaction catalyzed by Mt-DHQ2.Relevant residues are indicated and labeled.Hydrogen bonds are shown as dashed lines.Note how Asp88* from a neighboring enzyme subunit is responsible for the generation of the catalytic form of the essential tyrosine (tyrosinate) and Arg19 stabilizes the enolate intermediate 26.

Fig. ( 6 )
Fig. (6).Detail view of the DHQ2 Michaelis complexes and enolate intermediate 26 geometries obtained by QM/MM studies: A) Mt-DHQ2 Michaelis complex.B) Hp-DHQ2 Michaelis complex.C) Hp-DHQ2 enolate intermediate 26 geometry.Note how Asp88*/Asp89* from a neighboring enzyme subunit is responsible for the generation of the catalytic form of the essential tyrosine (tyrosinate).For Hp-DHQ2, this reaction takes place with the assistance of a water molecule.Relevant residues and water molecules are indicated and labeled.Hydrogen bonds and key distances are shown.