Insights into the catalytic mechanism of dTDP-glucose 4,6-dehydratase from quantum mechanics/molecular mechanics simulations

Abstract

dTDP-glucose 4,6-dehydratase catalyzes the biotransformation of dTDP-glucose into dTDP-4-keto-6-deoxy-glucose. We have utilized the quantum mechanical/molecular mechanical (QM/MM) approach to investigate the detailed mechanism of dTDP-glucose 4,6-dehydratase from Streptomyces venezuelae. On the basis of our calculation results, the previously proposed mechanism has been revised. The overall catalytic cycle can be divided into three sequential chemical steps: oxidation, dehydration and reduction, containing four enzymatic elementary reactions and one non-enzymatic enol–keto tautomerization reaction. The oxidation step proceeds through a concerted asynchronous mechanism with a calculated free energy barrier of 21.1 kcal mol⁻¹, in which the hydride transfer lags behind the proton transfer. The dehydration step prefers a stepwise mechanism rather than a concerted mechanism, and involves an enolate intermediate. Two highly conserved residues Glu129 and Asp128 are involved in this step. In the reduction step, NADH returns the hydride back to glycosyl C6 and the phenolic hydroxyl of Tyr151 spontaneously donates its proton to the C4-keto group, forming an enol sugar as the enzymatic product. After dissociation from the dehydratase active site and diffusion into the solution, this enol sugar will facilely rearrange to give the more favorable dTDP-4-keto-6-dexoyglucose product. Although Thr127 is not directly involved in the whole enzymatic reaction, it is responsible for promoting the catalysis by forming hydrogen-bonding interactions with glycosyl. These calculation results may provide new insight and inspiration for the catalytic mechanism of dTDP-glucose 4,6-dehydratase, even though it is not fully consistent with the previous experimental proposals.

---

1. Introduction

Deoxy sugars are an important class of carbohydrates that are frequently found in glycoproteins, lipopolysaccharides and glycolipids of bacterial cell surfaces and many secondary metabolites.¹ They are indispensable to numerous biochemical processes, such as cell–cell interactions, immune response, celhflar adhesion and target recognition of toxins and microorganisms,^2,3 and also function as single structural elements or important constituents of oligosaccharides in antibiotics and anticancer agents.⁴ The addition of unusual deoxy sugars to many macrolide antibiotics, an old and well-established class of anti-microbial agents with excellent clinical activity against Gram-positive bacteria, can enhance their biological activity. Removal of deoxy sugars from these clinically relevant drugs often decreases or even eliminates their pharmacological properties.² The deoxy sugars are usually derived from common sugars by substituting at least one hydroxyl group with a hydrogen or a non-O-linked functional group. Although our knowledge of pathways and mechanisms of deoxy sugar biosynthesis still remains limited, many significant advances have been made in the past decades in understanding several specific enzymes that catalyze the key biotransformations to yield 2-deoxyribose, 6-deoxyhexose and 2,6/3,6/4,6-dideoxyhexose.^1,3

L-Rhamnose and D-desosamine are two common deoxy sugars. L-Rhamnose is a 6-deoxyhexose found widely in the cell walls and envelopes of many pathogenic bacteria.^5,6 The metabolic precursor of L-rhamnose is dTDP-L-rhamnose, which is biosynthesized by the actions of four distinct enzymes: α-D-glucose-1-phosphate thymidylyltransferase (RmlA, EC:2.7.7.24), dTDP-D-glucose 4,6-dehydratase (RmlB, EC:4.2.1.46), dTDP-6-deoxy-D-xylo-4-hexulose-3,5-epimerase (RmlC, EC:5.1.3.13) and dTDP-6-deoxy-L-lyxo-4-hexulose-4-reductase (RmlD, EC:1.1.1.133).^7–9 Naismith and his co-workers have acquired great progress in understanding the structures and catalytic mechanisms of these four enzymes.^10–14 RmlB catalyses the second step in the dTDP-L-rhamnose biosynthetic pathway, the C6-deoxygenation of nucleotide sugar dTDP-D-glucose, which has been extensively investigated.^11,14–16 D-Desosamine is a 3-dimethylamino-3,4,6-trideoxyhexose found in some macrolide antibiotics, such as neomethymycin,¹⁷ methymycin,¹⁸ picromycin,¹⁹ and erythromycin,²⁰ which is speculated to interact with the bacterial ribosome and specifically with the peptidyl transferase center.^21,22 In Streptomyces venezuelae, the D-desosamine is derived from dTDP-D-desosamine, and seven genes, namely desI, desII, desIII, desIV, desV, desVI, and desVIII are required for the biosynthesis of dTDP-D-desosamine. The desIV gene codes for a dTDP-glucose 4,6-dehydratase, also referred to as DesIV, catalyzing the second step in the dTDP-D-desosamine biosynthetic pathway by oxidizing the C4-hydroxyl group and removing the C6-hydroxyl group of the dTDP-glucose substrate.^23–25

Although RmlB and DesIV are derived from different bacterial sources, both of them catalyze the biotransformation of dTDP-glucose into dTDP-4-keto-6-deoxy-glucose (Scheme 1), and therefore have the same name of dTDP-glucose 4,6-dehydratase. In the past decades, several X-ray structures of dTDP-glucose 4,6-dehydratase from Salmonella enterica, Streptococcus suis, Escherichia coli, and Streptomyces venezuelae have been determined.^11,15,26,27 Notably, the structure of DesIV from Streptomyces venezuelae was crystallized by Holden et al. with the highest resolution of 1.44 Å.²⁷ It is a wild-type enzyme complexed with cofactor NAD⁺ and dTDP. In addition, a double site-directed mutant protein (D128N/E129Q) was also crystallized with resolution of 1.35 Å as a complex with cofactor NAD⁺ and the substrate dTDP-glucose.²⁷ These three-dimensional structures provide meaningful insights into the binding modes of coenzyme and substrate, and catalytic mechanism of dTDP-glucose 4,6-dehydratase. It has been demonstrated that dTDP-glucose 4,6-dehydratase function as a homodimer, and each monomer emerges an mixed α/β structure that can be divided into two domains.^11,15,26,27 The smaller C-terminal domain binds the substrate dTDP-glucose, which is dominated by mixed β-sheets. The larger N-terminal domain is responsible for binding the cofactor NAD⁺ and contains seven strands of parallel β-sheet connected by α-helices, as presented in Fig. 1a. On the basis of amino acid sequence analysis, dTDP-glucose 4,6-dehydratase belongs to the short chain dehydrogenase/reductase superfamily (SDR), which contains a highly conserved Tyr-XXX-Lys catalytic couple and binds the cofactor NAD(P)⁺.^28,29

Scheme 1 Schematic diagram of mechanism of C6-deoxygenation catalyzed by dTDP-glucose 4,6-dehydratase.

Fig. 1 (a) Crystal structure of dTDP-glucose 4,6-dehydratase in complex with NAD⁺ and dTDP-glucose (PDB code: 1R6D), in which the Asn128 and Gln129 have been mutated to Asp128 and Glu129, respectively. (b) Structure of the active pocket taken from the crystal.

Up to now, great strides have been made in understanding the structure and reaction mechanism of dTDP-glucose 4,6-dehydratase.^{11,14–16,27} Based on the structural analyses and mutagenesis studies of dTDP-glucose 4,6-dehydratase, a three-step sequential mechanism, including oxidation, dehydration and reduction, has been proposed, as shown in Scheme 1. The oxidation step requires the hydride abstraction from glycosyl C4 in concert with the general base catalysis to remove the proton of C4-hydroxyl group. By mutagenesis and steady-state kinetic analysis, Frey et al. proposed that a conserved tyrosine residue is the catalytic base for the hydride abstraction step, and a threonine residue may relay a proton from the glycosyl to the tyrosine residue.³⁰ The dehydration step involves the proton abstraction at C5 and elimination of hydroxyl group at C6 of the sugar, in which two conserved aspartic and glutamic residues are the acid and base catalysts, respectively.³¹ The previous experimental studies about the dehydration step mainly focused on whether the elimination of water proceeds through a concerted or stepwise mechanism. Gassman et al. proposed that the enzymatic β-elimination reaction follows a stepwise mechanism, in which an enol or enolate intermediate was involved.³² Frey and his co-workers investigated the dehydration mechanism in the wild-type and mutated Escherichia coli dTDP-glucose 4,6-dehydratases by using a simultaneous kinetic characterization of glycosyl C5(¹H/²H) solvent hydrogen and C6(¹⁶OH/¹⁸OH) solvent oxygen exchanges. They proposed that the reaction of wild-type enzyme follows a concerted dehydration mechanism, but mutation of Asp135 to Asn or Ala switches the concerted mechanism to a stepwise one due to the loss of control over the C5–C6 bond rotation of glycosyl.³³ As for the final reduction step, mechanistic studies are still controversial. Most of the experimental results suggested that the reduction step proceeds through a concerted mechanism, in which the NADH returns the hydride back to glycosyl C6 and the glutamic residue involved in the dehydration step acts as a general base to protonate the glycosyl C5 to give the final product.^{11,14–16,27} In contrast, ab initio electronic structure calculations indicated that the final reduction reaction might follow a stepwise mechanism, involving an enol intermediate. But it was not clear whether the enol rearranges in the dehydratase active site to give the final keto sugar product or whether this occurs in solution after the enol dissociated from the enzyme.³⁴

As discussed above, a rough picture of the catalytic mechanism of dTDP-glucose 4,6-dehydratase has been obtained. But many key fundamental questions still remain unresolved. In particularly, how the multistep proton and hydride transfer proceed, whether the dehydration and reduction steps follow a concerted or stepwise mechanism, and whether the final step completes inside or outside the active site still have no definite conclusion yet. Answering the above questions is vital for the understanding of dTDP-glucose 4,6-dehydratase mechanism and even the deoxy sugar biosynthesis.

In this paper, a combined quantum mechanics and molecular mechanics (QM/MM) method^35–38 has been used to study the detailed reaction pathway of dTDP-glucose 4,6-dehydratase. This methodology has been testified to be successful in studying enzyme active sites and reaction mechanisms.^39–42 Based on our calculations, the key stable intermediates and transition states as well as the energy barriers of all elementary steps were obtained. Also, the specific roles of some key residues have been clarified. To our knowledge, this is the first study of the reaction mechanism of dTDP-glucose 4,6-dehydratase with the QM/MM method.

2. Models and methods

2.1 System preparation

The initial coordinates were taken from the crystal structure of double site-directed mutant dTDP-glucose 4,6-dehydratase (D128N/E129Q) in complex with cofactor NAD⁺ and substrate dTDP-glucose (PDB code: 1R6D, resolution: 1.35 Å) obtained from Streptomyces venezuelae.²⁷ The mutated residues Asn128 and Gln129 were changed back to Asp128 and Glu129 to recover the wild type enzyme by using the VMD program.⁴³ The obtained model is presented in Fig. 1a. The protonation states of all ionizable residues were determined on the basis of their pK_a values calculated by the empirical PROPKA program⁴⁴ and verified by the VMD program.⁴³ The missing hydrogen atoms in the crystal structure were added via the HBUILD facility in the CHARMM package,⁴⁵ and the crystallographic water molecules were kept at their original positions. Force field parameters for substrate dTDP-glucose were obtained using SwissParam,⁴⁶ and NAD⁺ adopts its default parameters.

The system was hydrated using the droplet model with a 35 Å sphere of pre-equilibrated TIP3 water molecules, and then it was neutralized by a total of ten sodium ions at random positions. To equilibrate the prepared system, a series of energy minimizations were first carried out using CHARMM program.⁴⁵ Subsequently, the minimized system was heated to 300 K for 200 ps, and then followed by equilibration for 200 ps at 300 K. Finally, a total of 15 ns MD simulation was performed with stochastic boundary condition in the NVT ensemble using the CHARMM22 all-atom force field⁴⁷ as implemented in the CHARMM program.⁴⁵ During the MD simulation, the temperature was maintained at 300 K and the pressure was set to 1 atm. The integration time step is 1 fs. A group-based switching scheme was applied to treat non-bond interactions. The final equilibrated system was composed of 19 057 atoms, including 4675 TIP3 water molecules.

2.2 QM/MM calculations

The QM/MM calculations were carried out using the ChemShell package,⁴⁸ which incorporates Turbomole module⁴⁹ used for treating QM region and DL_POLY program⁵⁰ for MM region. The QM region contained 89 atoms, including the sugar ring of dTDP-glucose, nicotinamide ribose moiety of NAD⁺ and the side chains of Thr127, Asp128, Glu129 and Tyr151 in the active site, as shown in Fig. 2. The remaining atoms of the solvated model were assigned to MM region. The QM region carried a charge of −1 and a spin multiplicity of 1, which was treated quantum mechanically at the B3LYP/6-31G(d,p) level. The MM part was described at CHARMM22 force field as included in DL_POLY program. During the subsequent QM/MM geometry optimization, an active region was selected including the QM atoms and all residues and water molecules in the MM region within 10 Å around the C4 atom of substrate glycosyl, whereas the remaining MM atoms were kept frozen for simplification. The electronic embedding scheme⁵¹ was used to incorporate the MM point charge into the one-electron Hamiltonian of QM calculation to avoid hyperpolarization of the QM wave function. No electrostatic cutoff was introduced for the MM and QM/MM interactions. Hydrogen link atoms in combination with the charge shift model were applied to treat the QM/MM boundary.⁵² Geometry optimizations were carried out using the hybrid delocalized internal coordinates (HDLC) optimizer,⁵³ where the quasi-Newton limited memory Broyden–Fletcher–Goldfarb–Shanno (L-BFGS) algorithm^54,55 was used to search for minima and the partitioned rational function optimization (P-RFO) algorithm⁵⁶ was used for transition state searches. On the basis of optimized geometries, the entropic effects were calculated at B3LYP/6-31G(d,p)//CHARMM22 level. Further, high-level single point energy calculations were performed with a larger basis set of B3LYP/6-31++G(d,p) to obtain accurate energies. The calculated QM/MM energetic data for all reaction species are summarized in Table S1 of ESI.†

Fig. 2 Arrangement of active site. The QM region is colored red. Dash lines represent possible hydrogen-bonding interactions.

Considering the initial structure may greatly influence the QM/MM calculations,^57–59 a reasonable snapshot from MD trajectories is crucial to explore the reaction mechanism. By analyzing the time dependence of the root-mean-square deviation (RMSD) for the backbone atoms of the dehydratase from 15 ns simulation (Fig. S1 of ESI†), we found that the RMSD value tends to be flat after 8 ns, which indicated that the system was basically equilibrated. Therefore, a series of snapshots were taken as the QM/MM models from the MD trajectories at intervals of 200 ps from 13 ns to 15 ns. These eleven models were further optimized at B3LYP/6-31G(d,p)//CHARMM22 level, and the superposition of the optimized active site structures are presented in Fig. S2.† One can see that the substrate, cofactor and key residues are superposed very well. From these geometries, a representative structure was chosen as the reactant model to start the following QM/MM calculations, and its QM/MM optimized solvation model and active site structure are shown in Fig. 3.

Fig. 3 The selected solvation model (a) and its active site structure (b) optimized by QM/MM method. The QM region is marked with a stick model. Distances are given in Å.

3. Results and discussion

3.1 Protonation states of active site residues

The pK_a value of Tyr151 was calculated to be 12.6 by using PROPKA method, indicating this residue should be protonated. However, a deprotonated Tyr151 is strictly necessary to initiate the catalytic reaction. Inspection of the active site structure reveals that the 3′′-hydroxyl group of nicotinamide ribose lies within hydrogen-bonding distance to the side chains of Tyr151 and Lys155 (see Fig. 1b and 3b). Besides, it has been speculated that the ammonium ion of Lys155 (Lys171 in S. enterica and Lys165 in S. suis) creates a positive electrostatic field around Lys155 and the positively charged nicotinamide ring, thereby effectively lowering the pK_a value of Tyr151 (Tyr167 in S. enterica and Tyr161 in S. suis).^15,27 The pK_a value of Tyr160 in E. coli dTDP-glucose 4,6-dehydratase was measured to be 6.41.³⁰ Similar to Tyr160 in E. coli, the Tyr151 should also have a lower pK_a value than the empirical calculation. To initiate the enzymatic reaction, the conserved Tyr151 was assigned to its deprotonated state here. It should be noted that Lys155 was not included in QM region for simplifying the calculations. Experimental results have determined that Asp135 and Glu136 in E. coli (Asp128 and Glu129 in S. venezuelae) act as the acid and base catalysts in the elimination of water, respectively.³¹ Herein, Asp128 was set to be protonated and Glu129 was modeled in its deprotonated state.

3.2 Reaction pathways

In this section, we have explored the detailed catalytic mechanism of dTDP-glucose 4,6-dehydratase using QM/MM calculations. On the basis of earlier proposals^{14–16,30–34} and our calculation results, we revised the catalytic mechanism, as outlined in Scheme 2. Three highly conserved residues Asp128, Glu129 and Tyr151 have been confirmed to be directly responsible for dehydrate reaction. Although Thr127 is not involved in enzymatic reaction, it may play important role in promoting catalysis by forming hydrogen bonds with substrate sugar ring. According to our proposal, the whole catalytic process contains four elementary steps and one non-enzymatic enol–keto tautomerism outside the active site. Here, we also divide the reaction into three distinct chemical steps: oxidation, dehydration and reduction, and discuss them in the following sections. The optimized structures of reactant, transition states, intermediates and product are shown in Fig. 4, and the corresponding energy profile is presented in Fig. 5.

Scheme 2 Proposed reaction mechanism of dTDP-glucose 4,6-dehydratase by QM/MM calculations.

Fig. 4 Optimized structures of reactant, transition states, intermediates and product. Distances are given in Å.

Fig. 5 The QM/MM energy profile for the dehydratase catalytic process. ^aRelative single-point energies. ^bRelative Gibbs free energies.

3.2.1 Oxidation step. In the first step, the phenolate group of Tyr151 acts as the general base to deprotonate the C4-hydroxyl of sugar ring while the hydride at C4 position is transferred to C4′ atom of NAD⁺ nicotinamide ring, yielding NADH and a dTDP-4-keto glucose intermediate (IM1). Based on the calculation results, the oxidation step occurs through a concerted asynchronous mechanism in which the proton transfer is prior to the hydride transfer.

As can be seen from Fig. 3b and 4, in reactant complex (R), the substrate glycosyl and NAD⁺ nicotinamide ring are parallel to each other and the C4′–H4 distance is 2.62 Å, which is ideal for hydride transfer. The phenolate group of Tyr151 forms three hydrogen bonds with the C3- and C4-hydroxyls of sugar ring and 3′-hydroxyl of nicotinamide ribose ring with distances of 1.74, 1.51 and 1.63 Å, respectively, which in turn stabilize the negatively charged Tyr151. Besides, Thr127 and Asp128 form hydrogen bonds with the C4- and C6-hydroxyl groups, respectively, which may play a role in fixing the substrate. Based on the optimized reactant complex, we scanned the departure of the H_O4 atom from C4-hydroxyl group to Tyr151 and the departure of H4 atom from C4 to C4′ atom to locate the transition state TS1. In TS1, the hydrogen atom of C4–OH has been transferred to the phenolate group of Tyr151 with a distance of 1.01 Å. On the other hand, the distance of C4′–H4 decreases from 2.62 Å in R to 1.28 Å in TS1, and finally to 1.10 Å in IM1. In IM1, the nicotinamide ring of NADH is slightly puckered, and the C4-keto group of sugar ring is in an ideal orientation for hydrogen bonding to both side chains of Thr127 and protonated Tyr151 with distances of 1.83 and 1.90 Å, respectively. The calculated Gibbs free energy barrier of this oxidation step is 21.1 kcal mol⁻¹, which is slightly higher than the calculated energy barrier (20.7 kcal mol⁻¹), as shown in Fig. 5. The C4-keto group of sugar ring has been demonstrated to acidify the C5 proton, lowering the pK_a value to the range of 18–19,⁶⁰ thereby activating the subsequent syn elimination of water across glycosyl C5 and C6.

As mentioned above, Lys155 was not included in the QM region. On the basis of optimized enzyme–substrate complex (R), we further added the residue Lys155 into the QM part to explore the rationality of selection of QM region. To simplify the calculation, we only re-calculated the oxidation step, which corresponds to the highest barrier of all enzymatic elementary reactions. Optimized structures of reactant complex R′, transition state TS1′ and intermediate IM1′, as well as the energy profile for the oxidation step are shown in Fig. 6. Compared with reactant complex R, the conformation of R′ only exhibits a minor change. Ammonium ion of Lys155 forms two hydrogen bonds with the 2′′- and 3′′-hydroxyl groups of nicotinamide ribose. The calculated free energy barrier of this step is 20.9 kcal mol⁻¹, which is almost the same as the original free energy barrier (21.1 kcal mol⁻¹). The calculation results suggest that the addition of Lys155 to the original QM region has no obvious influence on calculation of energy barrier, indicating the original selection of QM region is appropriate.

Fig. 6 Optimized structures of reactant complex R′, transition state TS1′ and intermediate IM1′, as well as the energy profile for the oxidation step. ^aRelative single-point energies. ^bRelative Gibbs free energies.

Taking the entropic contribution into consideration, except for TS1, the relative Gibbs free energy for each species is a little lower than the calculated relative energy (see Table S1† and Fig. 5). This demonstrates that the entropic contribution only has minor effect on this enzymatic reaction.

3.2.2 Dehydration step. Based on the QM/MM calculation results, the elimination of water proceeds through a stepwise (E1cb) mechanism, as outlined in Scheme 2. Residue Glu129 firstly acts as a general base to remove the C5 proton of sugar ring, leading to an enolate intermediate (IM2). After the formation of IM2, the C6-hydroxyl is eliminated as a water molecule by abstracting the proton of Asp128, thereby forming a dTDP-4-ketoglucose-5,6-ene intermediate (IM3). The free energy barriers of these two elementary steps are calculated to be 13.7 and 6.0 kcal mol⁻¹, respectively (see Fig. 5), indicating the dehydration reaction is facile to occur.

Structural analysis shows that Glu129 is perfectly positioned to abstract the C5-proton. Transition state TS2 corresponds to the proton transfer from glycosyl C5 atom to Glu129. The distance of O(Glu129)–H5 decreases from 2.29 Å in IM1 to 1.21 Å in TS2, and then to 0.99 Å in IM2. In IM2, the negatively charged sugar ring is stabilized by an oxyanion hole formed by the side chains of Thr127 and Tyr151. Additionally, the C6–OH forms two strong hydrogen bonds with the protonated Asp128 and Glu129 with distances of 1.66 and 1.79 Å, respectively. Compared with the reactant complex, IM2 is not very stable and the C6-hydroxyl of sugar ring can be easily eliminated as a water molecule by accepting the proton of the side chain of Asp128 via a low-energy transition state (TS3). TS3 was located by scanning two variables: the cleavage of C6–O6 bond and the proton transfer from Asp128 to O6 atom. In TS3, the bond length of C6–O6 elongates to 2.12 Å, and simultaneously the distance of O6–H(Asp128) decreases to 1.34 Å from 1.66 Å in IM2. The elimination of water leads to the formation of a key α,β-unsaturated-ketone, dTDP-4-ketoglucose-5,6-ene intermediate (IM3). The existence of IM3 has been experimentally confirmed by rapid mix quench mass spectroscopy.⁶¹

3.2.3 Reduction step. Comparison of the optimized structures of R, IM1, IM2 and IM3 indicates that, during the reaction the glycosyl not only changes its conformation, but also apparently deviates from its original position, as shown in Fig. 7a. In IM3, the strong hydrogen bonds between water molecule and the side chains of Asp128 and Glu129 prevent the water from moving freely. As the water is not free to move, the van der Waals repulsion between glycosyl C6 and water make the C6 away from water molecule and close to NADH by spontaneously adjusting the position of sugar moiety, which has been pointed out in the earlier work.¹⁵ On the other hand, the hydrogen-bonding interactions between the glycosyl C4-keto group and the side chains of Thr127 and Tyr151 play crucial role in enhancing the electrophilicity of C6, thereby inducing the hydride transfer from NADH to the methylene at C6. As shown in Fig. 4 (IM3), the glycosyl C6 locates in an ideal position to abstract the hydride from NADH with a distance of 2.59 Å. In consideration of these factors, it should not be a difficult challenge for NADH to transfer the hydride to glycosyl C6. The real challenge is where the proton for adding to glycosyl C5 comes from if the reduction step proceeds through a concerted mechanism. Structural analysis of active pocket and previously proposed mechanism suggest that the only possibility is the protonated Glu129, which denotes a proton to acidify the C5 while the hydride of NADH transfers to C6.²⁷ However, the strong hydrogen-bonding interactions between the side chains of Glu129 and water molecule make the proton of Glu129 lies in an inactive orientation to denote its proton to glycosyl C5. Furthermore, the van der Waals repulsion between glycosyl C6 and water increases to some extent the distance between the proton of Glu129 and C5. Based on the above considerations, the experimentally proposed concerted mechanism is unlikely to occur for the final reduction step.

Fig. 7 (a) Superposition of the optimized active sites of reactant and intermediates (IM1-IM3); (b) optimized active site structure of enol product (P′). Distances are given in Å.

Different from the experimental proposal that the reduction step follows a concerted mechanism,²⁷ both the previously ab initio electronic structure calculation³⁴ and our QM/MM calculation results suggest that the final reduction reaction occurs stepwisely, as outlined in Scheme 2. We scanned the hydride transfer from C4′ atom of NADH to glycosyl C6 atom of dTDP-4-ketoglucose-5,6-ene intermediate (IM3) to locate the transition state TS4. In TS4, the distance of C6–H4 decreases to 1.35 Å from 2.59 Å in IM3, as shown in Fig. 4. The free energy barrier of hydride transfer is calculated to be 11.4 kcal mol⁻¹, suggesting this elementary reaction is easily to occur. It is noteworthy that the hydride transfer is spontaneously followed by a proton transfer from Tyr151 to C4-keto group, thereby forming a low-energy enol. Compared with reactant and other intermediates, this enol is quite stable in dehydratase active pocket. Based on our QM/MM calculation results, it can be postulated that the enol is an enzymatic product (P′), which is in agreement with the previously theoretical calculation results.³⁴ Interestingly, although the formation of product has been observed by using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry,⁶¹ there is no convincing evidence to prove that the enzymatic product exists in its keto form. Considering the enol and keto show same molecular weight, it is difficult to distinguish them by mass spectrometry alone.

It is also interesting to explore whether the enol rearranges to give the final keto sugar in the dehydratase active pocket or in the solution after the enol diffuses from active site. Fig. 7b shows the optimized active site structure of the enol product, in which the C4-hydroxyl forms two hydrogen bonds with the side chains of Thr127 and Tyr151 with distances of 1.84 and 1.49 Å, respectively. Besides, the water molecule lies within hydrogen bonding distance to the side chains of Asp128 and Glu129 (1.48 and 1.51 Å, respectively). Therefore, we firstly explored the possibility that the proton transfer from Glu129 to glycosyl C5 to give the final product (P′′). As mentioned above, although the side chain of Glu129 is located in an acceptable position, the Glu129 positions its proton toward an unsuitable orientation to acidify the glycosyl C5 owing to the hydrogen bonding interaction. The free energy barrier of this proton transfer process is calculated to be 38.9 kcal mol⁻¹, as shown in Fig. 8, which suggests that the proton transfer from Glu129 to C6 is very difficult.

Fig. 8 The calculated energy profile and corresponding stable states and transition state for proton transfer from Glu129 to C5. Distances are given in Å. ^aRelative single-point energies. ^bRelative Gibbs free energies.

As the previously generated water in the dehydration step is not free to move, and the glycosyl binding site has no enough space to accommodate other solvated water molecules, we exclude the other possibility of water-assisted enol–keto tautomerism in the dehydratase active site. Based on the structural analysis of enol product, we boldly speculate that the enol may rearrange to give dTDP-4-keto-6-dexoyglucose product (P) in the solution after it dissociates from the active site. Interestingly, after the enol product is released from the dehydratase active site, Tyr151 is deprotonated, and Asp128 and Glu129 will recover their original ionization states via water-assisted proton exchange between the Asp128 and Glu129.

Here, we employed four simple cluster models to explore the enol–keto tautomerism of sugar ring (Scheme 3). All calculations were performed by using the DFT method with B3LYP function implemented in Gaussian09 program package.⁶² Geometry optimizations were performed using the 6-31G(d,p) basis set. Based on the optimized geometries, single-point calculations with the larger basis set 6-311++G(2d,2p) were performed to obtain more accurate energies. The polarizable continuum model (PCM)^63,64 was employed to calculate the solvent effects, in which the dielectric constant (ε) of 80 was used to simulate aqueous solvent environments. To simplify the calculations, we assume that the tautomerisation takes place in the neutral aqueous solution.

Scheme 3 Enol–keto tautomerization of sugar ring.

The energetic data for the four different models are given in Table 1. In the case where no water is involved in enol–keto tautomerism, the energy barrier of direct proton transfer from C4-hydroxyl to C5 is calculated to be 61.7 kcal mol⁻¹ in gas phase and 64.2 kcal mol⁻¹ in water solvent. Compared with the direct tautomerism, the water-assisted proton transfer is proved to be more favorable. For example, when two molecules are involved in the tautomeric reaction, the calculated barrier drastically decreases to 26.8 kcal mol⁻¹ in gas phase and 27.9 kcal mol⁻¹ in water solvent. The last model contains eight water molecules, in which two of them were employed to directly participate in the tautomeric reaction and the others were set to stabilize the transition state. The calculated barrier is 20.5 kcal mol⁻¹ in gas phase and 22.2 kcal mol⁻¹ in water solvent. The optimized structures of reactant, transition state and product for the enol–keto tautomerism are shown in Fig. S3.†

Table 1 Summary of calculated barriers and reaction energies for the enol–keto tautomerization in water solvent using DFT method

	Barrier (kcal mol^−1)		Reaction energy (kcal mol^−1)
	Gas phase	ε = 80	Gas phase	ε = 80
0H2O	61.7	64.2	−11.2	−11.5
1H2O	31.2	33.7	−9.2	−14.1
2H2O	26.8	27.9	−7.7	−10.2
8H2O	20.5	22.2	−7.9	−13.2

---

These results indicate that the presence of explicit water molecules can greatly lower the energy barrier of enol–keto tautomerism, and the keto form is usually more favorable than the enol form in water solvent. Based on the calculation results, we speculate that once the enol sugar is released from the dehydratase active site and moves into the solution, it facilely rearranges to give a more stable keto sugar, dTDP-4-keto-6-dexoyglucose, the final product.

4. Conclusion

We report here an extensive QM/MM investigation of the detailed mechanism of dTDP-glucose 4,6-dehydratase from Streptomyces venezuelae. Based on our calculations, we revised the previously proposed three-step catalytic process. The computational results demonstrated that the whole catalytic reaction contains four enzymatic elementary steps and one non-enzymatic enol–keto tautomerisation step. In the oxidation step, the NAD⁺ abstracts the hydride from glycosyl C4 and a conserved tyrosine residue (Tyr151 in S. venezuelae) removes a proton from the glycosyl C4-hydroxyl group. This oxidation step proceeds through a concerted asynchronous mechanism with a calculated free energy barrier of 21.1 kcal mol⁻¹, in which the hydride transfer lags behind the proton transfer. Different from the proposal based on experiments that the dehydration follows a concerted mechanism, the QM/MM calculation results show that the elimination of water is stepwise, in which two highly conserved residues are involved. Glu129 firstly deprotonates the glycosyl C5 and subsequently Asp128 protonates the leaving C6-hydroxyl group to give a water molecule and dTDP-4-ketoglucose-5,6-ene intermediate. In the reduction step, NADH returns the hydride back to glycosyl C6 and the phenolic hydroxyl of Tyr151 spontaneously denotes its proton to C4-keto group, forming an enol. Compared with the reactant and other intermediates, this enol is more stable, which is speculated to be the enzymatic product. We employed four simple cluster models to explore the enol–keto tautomerism of sugar ring. The calculated results suggest that the enol sugar should be easily converted into keto sugar with the assistance of water molecules in solution. In summary, after dissociated from the dehydratase active site and diffused into the solution, the enol sugar will facilely rearrange to give the more favorable dTDP-4-keto-6-dexoyglucose product. Although residue Thr127 is not directly involved in the whole enzymatic reaction, it may be responsible for promoting catalysis by forming a hydrogen bond with substrate sugar ring. Deeper insight into the detailed reaction mechanism of dTDP-glucose 4,6-dehydratase may provide help for the researchers to better understand the biosynthesis of dexoy sugar. In an attempt to shed more light on the dehydration mechanisms, other sugar dehydratases are now being investigated in our laboratory.

Acknowledgements

This work was supported by the Natural Science Foundation of China (21373125, 21173129) and Project of Shandong Province Higher Educational Science and Technology Program (no. J12LD10).

References

1. D. A. Johnson and H. Liu, Curr. Opin. Chem. Biol., 1998, 2, 642.
2. A. C. Weymouth-Wilson, Nat. Prod. Rep., 1997, 14, 99.
3. T. M. Hallis and H. Liu, Acc. Chem. Res., 1999, 32, 579.
4. A. Kirschning, A. F. W. Bechthold and J. Rohr, Top. Curr. Chem., 1997, 188, 1.
5. M. McNeil, M. Daffe and P. J. Brennan, J. Biol. Chem., 1990, 265, 18200.
6. S. D. Elliott and J. Y. Tai, J. Exp. Med., 1978, 148, 1699.
7. C. A. Schnaitman and J. D. Klena, Microbiol. Mol. Biol. Rev., 1993, 57, 655.
8. C. L. Marolda and M. A. Valvano, J. Bacteriol., 1995, 177, 5539.
9. Y. Ma, R. J. Stern, M. S. Scherman, V. D. Vissa, W. Yan, V. C. Jones, F. Zhang, S. G. Franzblau, W. H. Lewis and M. R. McNeil, Antimicrob. Agents Chemother., 2001, 45, 1407.
10. W. Blankenfeldt, M. Asuncion, J. S. Lam and J. H. Naismith, EMBO J., 2000, 15, 6652.
11. S. T. M. Allard, M. F. Giraud, C. Whitfield, M. Graninger, P. Messner and J. H. Naismith, J. Mol. Biol., 2001, 307, 283.
12. M. F. Giraud, G. A. Leonard, R. A. Field, C. Berlind and J. H. Naismith, Nat. Struct. Biol., 2000, 7, 398.
13. W. Blankenfeldt, I. D. Kerr, M. F. Giraud, H. J. McMiken, G. Leonard, C. Whitfield, P. Messner, M. Graninger and J. H. Naismith, Structure, 2002, 10, 773.
14. M. F. Giraud and J. H. Naismith, Curr. Opin. Struct. Biol., 2000, 10, 687.
15. S. T. M. Allard, K. Beis, M. F. Giraud, A. D. Hegeman, J. W. Gross, R. C. Wilmouth, C. Whitfield, M. Graninger, P. Messner, A. G. Allen, D. J. Maskell and J. H. Naismith, Structure, 2002, 10, 81.
16. J. H. Naismith, Biochem. Soc. Trans., 2004, 32, 647.
17. K. Madduri, J. Kennedy, G. Rivola, A. Inventi-Solari, S. Filippini, G. Zanuso, A. L. Colombo, K. M. Gewain, J. L. Occi, D. J. MacNeil and C. R. Hutchinson, Nat. Biotechnol., 1998, 16, 69.
18. P. J. Solenberg, P. Matsushima, D. R. Stack, S. C. Wilkie, R. C. Thompson and R. H. Baltz, Chem. Biol., 1997, 4, 195.
19. S. Omura, H. Takeshima, A. Nakagawa and J. Miyazawa, J. Antibiot., 1976, 29, 316.
20. J. Staunton and B. Wilkinson, Chem. Rev., 1997, 97, 2611.
21. F. Schlünzen, R. Zarivach, J. Harms, A. Bashan, A. Tocilj, R. Albrecht, A. Yonath and F. Franceschi, Nature, 2001, 413, 814.
22. J. L. Hansen, J. A. Ippolito, N. Ban, P. Nissen, P. B. Moore and T. A. Steitz, Mol. Cell, 2002, 10, 117.
23. Y. Q. Xue, L. S. Zhao, H. W. Liu and D. H. Sherman, Proc. Natl. Acad. Sci. U. S. A., 1998, 95, 12111.
24. S. Chen, J. B. Roberts, Y. Q. Xue, D. H. Sherman and K. A. Reynolds, Gene, 2001, 263, 255.
25. Y. Q. Xue and D. H. Sherman, Metab. Eng., 2001, 3, 15.
26. J. Thoden, A. D. Hegeman, P. A. Frey and H. M. Holden, PDB Data Bank, filename 1BXK 1999.
27. S. T. M. Allard, W. W. Cleland and H. M. Holden, J. Biol. Chem., 2004, 279, 2211.
28. Y. Kallberg, U. Oppermann, H. Jörnvall and B. Persson, Eur. J. Biochem., 2002, 269, 4409.
29. B. Persson, Y. Kallberg, U. Oppermann and H. Jörnvall, Chem.–Biol. Interact., 2003, 143–144, 271.
30. B. Gerratana, W. W. Cleland and P. A. Frey, Biochemistry, 2001, 40, 9187.
31. J. W. Gross, A. D. Hegeman, B. Gerratana and P. A. Frey, Biochemistry, 2001, 40, 12497.
32. J. A. Cerlt and P. G. Gassman, J. Am. Chem. Soc., 1992, 114, 5928.
33. A. D. Hegeman, J. W. Gross and P. A. Frey, Biochemistry, 2002, 41, 2797.
34. K. Beis, S. T. M. Allard, A. D. Hegeman, G. Murshudov, D. Philp and J. H. Naismith, J. Am. Chem. Soc., 2003, 125, 11872.
35. M. J. Field, P. A. Bash and M. Karplus, J. Comput. Chem., 1990, 11, 700.
36. R. A. Friesner and V. Guallar, Annu. Rev. Phys. Chem., 2005, 56, 389.
37. D. Riccardi, P. Schaefer, Y. Yang, H. B. Yu, N. Ghosh, X. Prat-Resina, P. Konig, G. H. Li, D. G. Xu, H. Guo, M. Elstner and Q. Cui, J. Phys. Chem. B, 2006, 110, 6458.
38. H. M. Senn and W. Thiel, Angew. Chem., Int. Ed., 2009, 48, 1198.
39. C. S. Li and S. Shaik, RSC Adv., 2013, 3, 2995.
40. K. Gkionis, S. T. Mutter and J. A. Platts, RSC Adv., 2013, 3, 4066.
41. X. L. Pan, F. C. Cui, W. Liu and J. Y. Liu, J. Phys. Chem. B, 2012, 116, 5689.
42. M. Zheng and D. G. Xu, J. Phys. Chem. B, 2013, 117, 10161.
43. W. Humphrey, A. Dalke and K. Schulten, J. Mol. Graphics, 1996, 14, 33.
44. M. H. M. Olsson, C. R. Søndergard, M. Rostkowski and J. H. Jensen, J. Chem. Theory Comput., 2011, 7, 525.
45. B. R. Brooks, R. E. Bruccoleri, B. D. Olafson, D. J. States, S. Swaminathan and M. Karplus, J. Comput. Chem., 1983, 4, 187.
46. V. Zoete, M. A. Cuendet, A. Grosdidier and O. Michielin, J. Comput. Chem., 2011, 32, 2359.
47. A. D. MacKerell Jr, D. Bashford, M. Bellott, R. L. Dunbrack Jr, J. D. Evanseck, M. J. Field, S. Fischer, J. Gao, H. Guo, S. Ha, D. Joseph-McCarthy, L. Kuchnir, K. Kuczera, F. T. K. Lau, C. Mattos, S. Michnick, T. Ngo, D. T. Nguyen, B. Prodhom, W. E. Reiher III, B. Roux, M. Schlenkrich, J. C. Smith, R. Stote, J. Straub, M. Watanabe, J. Wiórkiewicz-Kuczera, D. Yin and M. Karplus, J. Phys. Chem. B, 1998, 102, 3586.
48. P. Sherwood, A. H. de Vries, M. F. Guest, G. Schreckenbach, C. R. A. Catlow, S. A. French, A. A. Sokol, S. T. Bromley, W. Thiel, A. J. Turner, S. Billeter, F. Terstegen, S. Thiel, J. Kendrick, S. C. Rogers, J. Casci, M. Watson, F. King, E. Karlsen, M. Sjøvoll, A. Fahmi, A. Schäfer and C. Lennartz, J. Mol. Struct.: THEOCHEM, 2003, 632, 1.
49. R. Ahlrichs, M. Bar, M. Haser, H. Horn and C. Kolmel, Chem. Phys. Lett., 1989, 162, 165.
50. W. Smith and T. R. Forester, J. Mol. Graphics, 1996, 14, 136.
51. D. Bakowies and W. Thiel, J. Phys. Chem., 1996, 100, 10580.
52. A. H. de Vries, P. Sherwood, S. J. Collins, A. M. Rigby, M. Rigutto and G. J. Kramer, J. Phys. Chem. B, 1999, 103, 6133.
53. S. R. Billeter, A. J. Turner and W. Thiel, Phys. Chem. Chem. Phys., 2000, 2, 2177.
54. J. Nocedal, Math. Comput., 1980, 35, 773.
55. D. C. Liu and J. Nocedal, Math. Programming, 1989, 45, 503.
56. A. Banerjee, N. Adams, J. Simons and R. Shepard, J. Phys. Chem., 1985, 89, 52.
57. I. F. Thorpe and C. L. Brooks III, J. Am. Chem. Soc., 2005, 127, 12997.
58. M. Klähn, S. Braun-Sand, E. Rosta and A. Warshel, J. Phys. Chem. B, 2005, 109, 15645.
59. P. Hu and Y. K. Zhang, J. Am. Chem. Soc., 2006, 128, 1272.
60. J. March, Advanced Organic Chemistry, John Wiley and Sons, New York, 3rd edn, 1985, pp. 218–236.
61. J. W. Gross, A. D. Hegeman, M. M. Vestling and P. A. Frey, Biochemistry, 2000, 39, 13633.
62. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery Jr, J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, N. J. Millam, M. Klene, J. E. Knox, J. B. Cross, Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels, O. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski and D. J. Fox, Gaussian09, Revision B.01, Gaussian, Inc., Wallingford, CT, 2009.
63. V. Barone, M. Cossi and J. Tomasi, J. Comput. Chem., 1998, 19, 404.
64. J. Tomasi and M. Persico, Chem. Rev., 1994, 94, 2027.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra04406a

---