QM/MM calculations and MD simulations of acetolactate decarboxylase to reveal substrate R/S-acetolactate binding mode and stereoselective catalytic mechansim

Abstract

Acetolactate decarboxylase [EC 4.1.1.5] (ALDC) catalyzes both decarboxylation/protonation of the natural substrate and the rearrangement of the non-natural substrate, which is an interesting phenomenon in biological systems. It is a metalloenzyme that converts natural substrate (S)-acetolactate to (R)-acetoin by decarboxylation and subsequent protonation. For the enantiomer (R)-acetolactate, ALDC does not directly initiate decarboxylation to make (R)-acetoin; it has to carry out a previous rearrangement reaction to produce (S)-acetolactate. Formerly, due to the lack of detailed information on the enzyme structure, we could not predict what are the binding modes of the two enantiomers and how amino acids and zinc ion affect the rearrangement and decarboxylation. The recent crystal structures of the ALDC, developed computer technology and quantum mechanics help us to speculate the reaction mechanism. In this study, a hybrid quantum mechanical/molecular mechanical (QM/MM) method at the ONIOM level was used to elicit the details of the catalytic mechanism and energy profiles along the reaction at the atomic level. In the rearrangement process, the hydrogen bond between Glu253 and carboxyl of (R)-acetolactate is crucial when the carboxyl rearranges to adjacent carbonyl carbon to form (S)-acetolactate. In the decarboxylation process, ALDC catalyzes (S)-acetolactate by a two-step reaction. The first step is the carboxyl departure with the formation of an enediol intermediate and carbon dioxide. The next step is the protonation reaction on the face opposite to the enediol intermediate. The calculations show that the water-assisted protonation reaction is more favorable than that without water. Moreover, the protonation reaction is the rate-limiting step of the decarboxylation process. This study may help us to understand ALDC catalyzing R/S-acetolactate and similar stereoselective catalytic mechanism in biological systems.

---

Introduction

Acetolactate decarboxylase [EC 4.1.1.5] (ALDC) is a crucial enzyme during beer maturation,¹ which catalyzes α-acetolactate (2-hydroxy-2-methyl-3-oxobutanoate) to (R)-acetoin (3-hydroxy-2-butanone) and carbon dioxide via decarboxylation.^2,3 ALDC is known to exist in numerous bacteria.^2–4

During beer fermentation, α-acetolactate produced by yeast is a biological precursor of amino acids, i.e. valine and leucine, which are essential for yeast growth.⁵ Acetolactate is successively converted into diacetyl and acetoin during the metabolism of yeast.⁵ Acetoin does not affect the beer flavor; however, diacetyl gives the beer a bad flavor with a slow chemical reaction.^1,6 In the beer brewing industry, enzyme ALDC is used to overcome the diacetyl formation. It directly converts α-acetolactate to acetoin without diacetyl. The use of ALDC can shorten beer fermentation.^6,7 Moreover, ALDC has been proved to synthesize enantiomerically pure diols.⁸

Decarboxylation is a fundamental and important reaction in biological systems. In general, most of carbon dioxide is produced by decarboxylation of organic acids in fermentation and respiration.⁹ Decarboxylation usually utilizes either an organic cofactor, such as flavin or pyridoxal 5′-phosphate, or an inorganic metal ion cofactor such as iron or zinc ion.¹⁰ There also are some decarboxylase that conduct decarboxylation without any cofactors.¹⁰ In decarboxylase, several factors may affect decarboxylation such as the hydrophobic environment of active site,¹¹ hydrogen bonding network created by nearby residues,¹² electrostatic repulsion between amino acids and carboxylate group¹³ and cofactor stabilizing the carbanion upon the elimination of carbon dioxide.^14,15

Experiments¹⁶ show that ALDC catalyzes the natural substrate (S)-acetolactate to produce (R)-acetoin through decarboxylation and protonation. C-Labeling and circular dichroism experiments have proved that (S)-acetolactate would give intermediate enediol after decarboxylation and then the intermediate is protonated on the opposite site of carbon atom from which carboxyl left.^17,18 Interestingly, the non-natural substrate (R)-acetolactate also produces (R)-acetoin catalyzed by ALDC¹⁶ (Scheme 1). Experiments show that (R)-acetolactate is catalyzed by ALDC at lower rate than the (S)-isomer.¹⁶ C-Labeling experiment shows that the process of ALDC catalyzing (R)-acetolactate to produce (R)-acetoin need a prior migration of the carboxylate group to the contiguous β-C to produce (S)-acetolactate.¹⁹ ALDC is likely to catalyze both decarboxylation/protonation of the natural substrate and the rearrangement of the non-natural substrate. This conversion appears to be an unusual example of the enzyme conversion of racemic substrate into a single enantiomer of the product. The special feature of ALDC reflects the enzyme specificity in biology.

Scheme 1 Chemical reaction of R/S-acetolactate catalyzed by ALDC.

Although these experimental studies have clarified the intermediates along the catalytic process, several questions still remain unclear, partly owing to the lack of the detail structure of ALDC active center. Recently, Fulop et al.,²⁰ reported the crystal structures of ALDC (PDB code: 4BT2, 4BT3, 4BT4, 4BT5, 4BT6, 4BT7), which enable us to comprehend the active center structure in an atomic level and speculate the reaction process on the theoretical level. With the development of quantum mechanics and computer technology, a combined quantum mechanical/molecular mechanical (QM/MM) has been a valuable tool to study enzyme-catalyzed processes.^21,22 In this study, we employed molecular dynamics (MD) simulations and quantum mechanical/molecular mechanical (QM/MM) calculations to examine the R/S-acetolactate binding to ALDC in the enzyme environment and its mechanisms of enzyme catalysis. For the mechanisms, we investigated the ALDC catalyzed (R)-acetolactate rearrangement reaction and (S)-acetolactate decarboxylation/protonation reactions.

Computational methods

Molecular docking and molecular dynamics studies of R/S-acetolactate and ALDC complexes

In the ALDC crystal structure, there are 236 amino acids and a zinc ion. The protonation states of residues were calculated using H++ server^23–25 at pH = 6.0. In the reaction center, considering hydrogen bonding network and zinc ion, His194, His196, His207 and Glu253 were manually protonated. His194 and His196 have a proton on the Nδ position, His207 has a proton on the Nε position, and Glu253 is defined as protonated. With the aid of AutoDockVina package,²⁶ the two complexes of R/S-acetolactate and ALDC were obtained based on the crystal structures 4BT5, 4BT4 and 4BT3.

To obtain stable structures, MD simulations were run to relax each complex in explicit water solvent box. Energy minimization of the complex was performed to remove bad contacts and relax models. 20 ps heating simulation was performed under NVT ensemble with temperature gradually increasing from 0 to 300 K. Then, 10 ns MD simulation was performed under NPT at 300 K. Then, root-mean-square deviation (RMSD) calculation was employed to see the structure equilibrium and cluster analysis of trajectory was used to obtain the representative structures. All the MD simulation and analysis were used with the Amber 11 package.²⁷ The TIP3P²⁸ model was employed for the water molecules. AMBER ff99SB force field^29–31 was used to describe the protein. The acetolactate force field parameters were obtained from the GAFF force field,³⁰ and the RESP³² partial charges were assigned. Some atoms in quantum mechanical (QM) region were fixed to remove system collapse, as the force field parameters do not suitably describe the metal–ligand interaction.^33–35 Zinc ion and R/S-acetolactate were employed force constant of 10 kcal (mol Ų)⁻¹ and 50 kcal (mol Ų)⁻¹, respectively. For S-acetolactate, the hydrogen bond between Glu253 and acetolactate was described with a weak force constant 5 kcal (mol Ų)⁻¹.

QM/MM studies of R/S-acetolactate and ALDC complexes

After MD simulation and cluster analysis,³⁶ the water molecules of the representative structures within 5 Å of protein were kept for QM/MM calculations. All other waters were deleted to reduce computing costs. Two-layers were defined within the enzyme, according to ONIOM scheme:^21,37–39 the QM model system and the MM real system. The zinc ion, His194, His196, His207, Glu65, Arg145 and Glu253 were chosen as the core residues. In the ONIOM calculations, all residues within 6 Å of the core residues including all QM atoms and part MM part atoms were allowed to move, whereas all others were frozen. The initial QM model system contained the zinc ion, entire acetolactate and part of His194, His196, His207, Glu65, Arg145, and Glu253 (total 66 atoms without H-link atoms). The QM part calculation was carried out by density functional theory B3LYP/6-31G(d) as the B3LYP^40,41 and 6-31G(d)⁴² were proved adequate to describe the zinc ion and similar structure.^43–46 The MM part of the system was described by AMBER parm96 force field parameters³¹ for amino acid residues. A mechanical embedding scheme was used for geometry optimization of minimum-energy states and transition states. Frequency calculations were done to verify the absence of imaginary frequencies in minimum-energy state and presence of only one imaginary frequency in transition state. After geometry optimization, single point energies at a higher level of theory were calculated using ONIOM (B3LYP/6-311 + G(2d,2p):AMBER) level.

For qualitative analysis of the variation of charges in reactions, the optimized QM part was extracted and calculated. In this part, the RESP charges were assigned.

Gaussian09 software⁴⁷ was used for QM/MM calculations. Gaussview⁴⁸ and VMD⁴⁹ were used as visualization tools and TAO package⁵⁰ was used for assistance in handling the input and output files.

Results and discussion

Molecular docking and molecular dynamics studies

By means of the molecular docking method, we obtained ten lowest energy conformers of R/S-acetolactate and ALDC. The high score conformers were considered for further MD simulations. 10 ns classical MD simulation was performed with R/S-acetolactate and ALDC complexes obtained from the docking results, and the RMSD value of protein backbone was measured for each simulation to ensure the stability of protein structures (Fig. S1 in ESI†). Structures of R/S-acetolactate and ALDC complexes were clustered into 5 groups, and the most representative structures were chosen as the initial QM/MM models. The complex structure of (R)-acetolactate and ALDC was defined as MODEL R, and the complex structure of (S)-acetolactate and ALDC was defined as MODEL S (displayed in Fig. 1).

Fig. 1 The representative active center structure of MODEL R and MODEL S. The dashed lines between residues represent hydrogen bonds.

In MODEL R, zinc ion and nearby three histidine residues (His194, His196 and His207) form three coordinate bonds, which fix zinc ion and is in favor of substrate binding. Moreover, amino acid residues Glu253, Arg145 and Glu65 form the hydrogen bonding network. For the substrate, oxygen atoms on carboxyl and hydroxyl of (R)-acetolactate form coordinate bonds with the zinc ion. Carboxyl and hydroxyl also form hydrogen bonds with nearby amino acids Glu253 and Glu65, respectively. Similar to MODEL R, zinc ion and nearby three histidine residues (His194, His196 and His207) form three coordinate bonds, and residues Glu253, Arg145 and Glu65 form hydrogen bonding network in the MODEL S active center. The substrate, (S)-acetolactate, is anchored to oxygen in hydroxyl group and carbonyl group by coordinate bonds. The amino acids, Glu65 and Glu253, in active center also form hydrogen bonds with (S)-acetolactate.

As mentioned above, zinc ion fixes the substrate and hydrogen bonds determine the binding direction. However, for different enantiomers, they have different binding modes, i.e. (R)-acetolactate and (S)-acetolactate should have different reaction mechanisms.

QM/MM studies

In the following section, we present and discuss the (R)-acetolactate rearrangement reaction and (S)-acetolactate decarboxylation/protonation reactions in ALDC by the two models MODEL R and MODEL S, respectively.

Model R: rearrangement mechanism

According to QM/MM calculations, relevant structures of rearrangement reaction were suggested and shown in Fig. 2. In this reaction, carboxyl of (R)-acetolactate rearranges to the adjacent β-carbon; moreover, a proton transfers from the hydroxyl group of substrate to amino acid Glu65.

Fig. 2 The structures of (R)-acetolactate rearrangement contained reactant (R-R), transition structure (R-TS) and the product (R-P). All the structures were the results of ONIOM (B3LYP/6-31G(d):Amber) calculations. The unit of energy is kcal mol⁻¹ for the rearrangement energy profile.

In reactant (R-R), the substrate, (R)-acetolactate, is part of the hydrogen bonding network, involving Glu253, Arg145 and Glu65, which stabilizes (R)-acetolactate conformation in the active center. Zinc ion forms a three coordinated geometry with nearby amino acids (His194, His196, and His207). It also coordinates with carboxyl and hydroxyl of R-acetolactate to fix the substrate on the active center. Geometry optimization shows that the substrate binds to active center of ALDC by pentacoordinated structure. The transition structure (R-TS) shows that oxygen atom of carbonyl of (R)-acetolactate is close to zinc ion, forming a six-coordinated structure. The distance of the hydrogen bond between carboxyl of (R)-acetolactate and Glu253 changed only 4.4% (R-R: 1.59 Å and R-TS: 1.66 Å). In the product (R-P), zinc ion also forms pentacoordinated complex with carboxyl and hydroxyl of acetolactate.

Intrinsic coordinate reaction (IRC) analysis shows that the proton of hydroxyl transfers to the carboxyl of Glu65, when carboxyl group rearranges to β-carbon. The proton transfer process is essential for the formation of hydroxyl and carbonyl of product. In the rearrangement process, we also observed the enlarged active pocket. It implies that six-coordinated structure needs more space in active pocket, which may weaken the hydrogen bonding network and reduce the structural stability. During the rearrangement reaction, the carboxyl of α-C of (R)-acetolactate rearranges to adjacent β-C forming (S)-acetolactate, which is consistent with the experimental observation.¹⁹ The energy barrier for this rearrangement is 23.3 kcal mol⁻¹. The energy profile for the arrangement reaction is displayed in Fig. 2.

We also considered other possible binding models as speculated by Fulop et al.,²⁰ wherein hydroxyl and carbonyl form hydrogen bond with Glu65 and Glu253, respectively, and carboxyl forms hydrogen bond with Arg145. Fulop et al.,²⁰ proposed that carboxyl performs the rearrangement reaction with the aid of Arg145. The main difference between this model and MODEL R is the hydrogen bond between carboxyl of (R)-acetolactate and nearby amino acid (Arg145 or Glu253). However, the binding model speculated by Fulop et al.,²⁰ could not be obtained by molecular docking. We also manually built (R)-acetolactate into active center according to the proposal of Fulop et al.²⁰ Compared to the crystal structure with similar substrate structure, the MD result seems not reasonable. Thus, we suggest MODEL R, in which the amino acid Glu253 induces the rearrangement reaction, is a more preferred model for the rearrangement reaction.

To explore the possibility of the direct decarboxylation reaction of this style of structure (MODEL R), we processed flexible scanning of the distance between carboxyl group and α/β-carbon. The transition state and energy barrier for the direct decarboxylation can be seen in the ESI Fig. S2.† The results show that Glu253 is vital to prevent direct decarboxylation, due to the presence of the hydrogen bond. This is unlike the decarboxylation of methylmalonyl CoA decarboxylase (MMCD), in which the conserved polar residue Tyr also forms a hydrogen bond; however, the residue destabilizes the ground state of the substrate and facilitates decarboxylation.⁵¹

MODEL S: decarboxylation reaction and protonation reaction

The process of ALDC catalyzing (S)-acetolactate include decarboxylation and protonation reactions, and the related structures of catalytic reactions are displayed in Fig. 3.

Fig. 3 The structures of (S)-acetolactate decarboxylation and protonation optimized at ONIOM (B3LYP/6-31G(d):Amber) level. The (a) row represents the relevant optimized structures of decarboxylation reaction. The (b) row represents the relevant optimized structures of protonation reaction without water aiding. The (c) row represents the relevant optimized structures of protonation reaction with water aiding.

Decarboxylation reaction

Similar to MODEL R, in MODEL S, hydroxyl forms hydrogen bond with Glu65. Oxygen atoms of carboxyl and hydroxyl of (S)-acetolactate form coordinate bonds with zinc ion, which fixed the substrate in the active centre. However, carboxyl does not form hydrogen bond with adjacent amino acids, and it is towards outside of the hydrogen bonding network, which has enough space to perform decarboxylation reaction. Moreover, in MODEL S, there is a hydrophobic residue Phe93 near the carboxyl group, which may create a hydrophobic environment to facilitate decarboxylation. In addition, carbonyl acts as a hydrogen bond acceptor and carboxyl in Glu253 acts as a hydrogen bond donor with the bond length 1.65 Å, which is different from MODEL R wherein Glu253 forms hydrogen bond with carboxyl of acetolactate. In other decarboxylases such as orotidine 5′-monophosphate decarboxylase (OMPDC),¹¹ the hydrophobic pocket has been proved to facilitate decarboxylation.

To explore the decarboxylation reaction, in reactant (S-R), the distance between carboxyl and α-C of (S)-acetolactate was taken as a reaction coordinate (ESI Fig. S3†). The transition structure (S-TS1) and the intermediate (S-IM1-0) were obtained by flexible scanning. The energy barrier for the decarboxylation reaction is 5.3 kcal mol⁻¹. The relevant differences between the S-R, S-TS and S-IM1-0 show that the bond between carboxyl and α-C stretched from 1.57 Å to 1.98 Å at a transition structure and to 3.26 Å at intermediate. At the same time, the angle of carboxyl (∠O–C–O) gradually increases, resulting in the formation of carbon dioxide. Thus, the intermediate structure enediol was obtained, which supports the experimental observation of the intermediate.¹⁶

In addition to the hydrophobic residues that facilitate decarboxylation, the zinc ion and the intermediate structure also affect the reaction. The zinc ion and the double bond of intermediate serve as an electron sink to stabilize the negative charge carbanion. In the decarboxylation, α-carbon of acetolactate gradually changes configuration from pyramidal sp³ to planar sp², and the distance between α-carbon and β-carbon changes from 1.54 Å to 1.36 Å. By analysing the charge distribution of decarboxylation, zinc ion charge changes from 0.66 to 0.49 (ESI Table S1†), which shows that the zinc ion can be considered to stabilize the intermediate. Zinc ion also helps to polarize this group and facilitate the decarboxylation reaction. The divalent metal ion Zn²⁺ plays an important role in both catalysis and structure stability. The hydrogen bonds between the intermediate and residues, Glu65 and Glu253, stabilize the anionic intermediate.

Protonation reaction

Considering the fact that carbon dioxide produced by the previous decarboxylation reaction does not take part in the next reaction, we removed carbon dioxide.

In the intermediate, two oxygen atoms of enediol form hydrogen bonds with nearby amino acids, Glu65 and Glu253, and coordinate bonds with zinc ion. In order to illustrate the reaction more clearly, the methyl carbon close to Glu65 in the enediol structure was renamed to C1. The carbon linked to C1 was named C2, and so on (Fig. 3, S-IM1-1). Protonated Arg145 can transfer a proton to enediol to form acetoin. In this reaction, the position of HH21 and C2 is close to each other with NH2–HH21 (Arg145) distances being 1.01, 1.36, and 2.59 Å, wherein the three numbers in parentheses represent the NH2–HH21 distance for intermediate (S-IM1-1), transition structure (S-TS2-1) and product (S-P-1), respectively. In the protonation reaction, C2 of enediol gradually changes its configuration from planar sp² to pyramidal sp³. The energy barrier for this step is 21.6 kcal mol⁻¹ (Fig. 4), which is the rate determining step in whole ALDC catalyzing (S)-acetolactate. The energy is higher than the experimental value of 13.3 kcal mol⁻¹.²⁰

Fig. 4 Relative energy profiles (kcal mol⁻¹) for (S)-acetolactate decarboxylation reaction (a), with (b), without (c) water-assisted protonation reactions in ALDC. In the entire decarboxylation reaction, a–c don’t use the same structure to calculate the relative energy, because carbon dioxide leaving and addition of a water molecule.

In the protonation reaction, Fulop et al.,²⁰ suggested that a water molecule takes part in the proton transfer process. Moreover, experiment¹⁸ shows that ALDClase in D₂O induces a D/H exchange in (R)-acetoin to give a deuterated product. Taking into account the size of the activity center (R_(HH21–C2) = 2.98 Å), there is enough space to accommodate a water molecule. We could presume that a water molecule might assist the reaction in the protonation. In order to verify whether water molecule might favor the reaction, we built a water molecule included complex wherein water is between C2 of enediol and Arg145, which is based on the speculation by Fulop et al.²⁰ The optimization calculations show that one of the hydrogen of water forms hydrogen bond with Glu65 (1.76 Å) and oxygen forms hydrogen bond with Arg145 (1.64 Å). Frequency calculation and IRC analysis show that the proton transfer reaction is a concerted reaction, wherein the proton of water moved to C2 and HH21 moved to water at the same time. In the intermediate (S-IM1-2), transition structure (S-TS2-2) and product (S-P-2), the distance of NH2–HH21 is 1.05, 1.58, 1.73 Å, respectively, and HW–C2 is 2.14, 1.42, 1.10 Å, respectively. The energy barrier for this proton transfer reaction is 16.3 kcal mol⁻¹, which is less than the previous model without a water assisted proton transfer energy barrier. The decarboxylation reaction and protonation reaction are in agreement with the previous speculation on ALDC by Fulop et al.²⁰ From the perspective of structural change, in the intermediate complex, the distance between proton and C2 of enediol is 2.98 Å (R_(HH21–C2) = 2.98 Å) without a water molecule assisted reaction. While in the water assisted reaction, the distance between the proton and C2 enediol is 2.14 Å. These imply that proton needs to transfer to a relatively longer distance to obtain the product without a water-assisted protonation reaction. The relatively longer distance proton transfer may require relatively large reaction barrier.

Compared to the energy barriers of ALDC catalyzing R/S-acetolactate, ALDC catalyzes (S)-acetolactate (16.3 kcal mol⁻¹) easier than (R)-acetolactate (23.3 kcal mol⁻¹). It suggests that ALDC catalysis of (R)-acetolactate decarboxylation needs a prior higher energy barrier of rearrangement to obtain (S)-acetolactate, which is the rate determining step in ALDC catalyzing (R)-acetolactate to produce (R)-acetoin. From the view of chemical kinetics, reaction rate of ALDC catalyzing (R)-acetolactate to obtain (R)-acetoin is lower than that of (S)-acetolactate. Experimental observation also suggests that (R)-acetolactate is decarboxylated by ALDC at lower rate than that of the (S)-isomer.^3,19 The conclusion from our calculation is consistent with the observations.

To validate the energy calculations, we also employed some other DFT functionals to map the free energy profiles of ALDC catalyzing R/S-acetolactate. The DFT functionals and the related energy are listed in Table S3 of the ESI.† The discussion can also be seen in the ESI on page S7.†

Conclusions

In this study, we explored the catalytic mechanism of (R)-acetolactate rearrangement and (S)-acetolactate decarboxylation/protonation of acetolactate decarboxylase (ALDC) at atomic level by QM/MM calculations. The results suggest that amino acids and zinc ion in active center play a crucial role to stabilize and orient acetolactate, which is important for enzymatic reaction. ALDC catalyzing (R)-acetolactate decarboxylation needs a prior rearrangement reaction to afford (S)-acetolactate. The rearrangement mechanism shows that carboxyl of α-C rearranged to adjacent β-C with the help of Glu253. ALDC catalyzing (S)-acetolactate includes two steps. The first decarboxylation step is characterized by the formation of an intermediate enediol and carbon dioxide. In this step, hydrophobic residue, zinc ion and hydrogen bonds facilitate the decarboxylation reaction. The second step is the protonation reaction. The amino acid Arg145 and a water molecule transfer a proton to enediol intermediate through a collaborative reaction. The energy pathways illustrate that the second reaction step is rate-limiting. We hope that the catalytic mechanism will help understand ALDC catalyzing R/S-acetolactate and the stereochemistry of biological processes.

Acknowledgements

This study is supported by the Natural Science Foundation of China (Grant No. 21273095).

References

1. S. Godtfredsen and M. Ottesen, Carlsberg Res. Commun., 1982, 47, 93–102.
2. M. I. Dolin and I. C. Gunsalus, J. Bacteriol., 1951, 62, 199–214.
3. E. Juni, J. Biol. Chem., 1952, 195, 715–726.
4. N. Goupil-Feuillerat, M. Cocaign-Bousquet, J. J. Godon, S. D. Ehrlich and P. Renault, J. Bacteriol., 1997, 179, 6285–6293.
5. H. Suomalainen and P. Ronkainen, Nature, 1968, 220, 792–793.
6. C. Dulieu, M. Moll, J. Boudrant and D. Poncelet, Biotechnol. Prog., 2000, 16, 958–965.
7. S. Dequin, Appl. Microbiol. Biotechnol., 2001, 56, 577–588.
8. Y. Yan, C.-C. Lee and J. C. Liao, Org. Biomol. Chem., 2009, 7, 3914–3917.
9. S. Ochoa, Physiol. Rev., 1951, 31, 56–106.
10. T. Li, L. Huo, C. Pulley and A. Liu, Bioorg. Chem., 2012, 43, 2–14.
11. C. A. Lewis and R. Wolfenden, Biochemistry, 2009, 48, 8738–8745.
12. J. J. Almrud, G. J. Poelarends, W. H. Johnson, H. Serrano, M. L. Hackert and C. P. Whitman, Biochemistry, 2005, 44, 14818–14827.
13. L. Cendron, R. Berni, C. Folli, I. Ramazzina, R. Percudani and G. Zanotti, J. Biol. Chem., 2007, 282, 18182–18189.
14. M. Blaesse, T. Kupke, R. Huber and S. Steinbacher, EMBO J., 2000, 19, 6299–6310.
15. G. L. Waldrop, B. F. Braxton, J. L. Urbauer, W. W. Cleland and D. M. Kiick, Biochemistry, 1994, 33, 5262–5267.
16. R. K. Hill, S. Sawada and S. M. Arfin, Bioorg. Chem., 1979, 8, 175–189.
17. D. H. G. Crout, J. Littlechild, M. B. Mitchell and S. M. Morrey, J. Chem. Soc., Perkin Trans. 1, 1984, 2271–2276,  10.1039/p19840002271.
18. A. F. Drake, G. Siligardi, D. H. G. Crout and D. L. Rathbone, J. Chem. Soc., Chem. Commun., 1987, 1834–1835,  10.1039/c39870001834.
19. D. H. G. Crout, J. Littlechild and S. M. Morrey, J. Chem. Soc., Perkin Trans. 1, 1986, 105–108,  10.1039/p19860000105.
20. V. A. Marlow, D. Rea, S. Najmudin, M. Wills and V. Fulop, ACS Chem. Biol., 2013, 8, 2339–2344.
21. L. W. Chung, W. M. C. Sameera, R. Ramozzi, A. J. Page, M. Hatanaka, G. P. Petrova, T. V. Harris, X. Li, Z. F. Ke, F. Y. Liu, H. B. Li, L. N. Ding and K. Morokuma, Chem. Rev., 2015, 115, 5678–5796.
22. L. W. Chung, H. Hirao, X. Li and K. Morokuma, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2012, 2, 327–350.
23. J. Myers, G. Grothaus, S. Narayanan and A. Onufriev, Proteins: Struct., Funct., Bioinf., 2006, 63, 928–938.
24. J. C. Gordon, J. B. Myers, T. Folta, V. Shoja, L. S. Heath and A. Onufriev, Nucleic Acids Res., 2005, 33, W368–W371.
25. R. Anandakrishnan, B. Aguilar and A. V. Onufriev, Nucleic Acids Res., 2012, 40, W537–W541.
26. O. Trott and A. J. Olson, J. Comput. Chem., 2010, 31, 455–461.
27. D. A. Case, T. E. Cheatham, T. Darden, H. Gohlke, R. Luo, K. M. Merz, A. Onufriev, C. Simmerling, B. Wang and R. J. Woods, J. Comput. Chem., 2005, 26, 1668–1688.
28. W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W. Impey and M. L. Klein, J. Chem. Phys., 1983, 79, 926–935.
29. J. Wang, P. Cieplak and P. A. Kollman, J. Comput. Chem., 2000, 21, 1049–1074.
30. J. Wang, R. M. Wolf, J. W. Caldwell, P. A. Kollman and D. A. Case, J. Comput. Chem., 2004, 25, 1157–1174.
31. W. D. Cornell, P. Cieplak, C. I. Bayly, I. R. Gould, K. M. Merz, D. M. Ferguson, D. C. Spellmeyer, T. Fox, J. W. Caldwell and P. A. Kollman, J. Am. Chem. Soc., 1995, 117, 5179–5197.
32. C. I. Bayly, P. Cieplak, W. Cornell and P. A. Kollman, J. Phys. Chem., 1993, 97, 10269–10280.
33. C. Yan, Z. Xiu, X. Li, S. Li, C. Hao and H. Teng, Proteins: Struct., Funct., Bioinf., 2008, 73, 134–149.
34. R. Wu, S. Wang, N. Zhou, Z. Cao and Y. Zhang, J. Am. Chem. Soc., 2010, 132, 9471–9479.
35. G. V. Dhoke, M. D. Davari, U. Schwaneberg and M. Bocola, ACS Catal., 2015, 5, 3207–3215.
36. J. Shao, S. W. Tanner, N. Thompson and T. E. Cheatham, J. Chem. Theory Comput., 2007, 3, 2312–2334.
37. F. Maseras and K. Morokuma, J. Comput. Chem., 1995, 16, 1170–1179.
38. M. Svensson, S. Humbel, R. D. J. Froese, T. Matsubara, S. Sieber and K. Morokuma, J. Phys. Chem., 1996, 100, 19357–19363.
39. S. Dapprich, I. Komáromi, K. S. Byun, K. Morokuma and M. J. Frisch, J. Mol. Struct.: THEOCHEM, 1999, 461–462, 1–21.
40. A. D. Becke, J. Chem. Phys., 1993, 98, 5648–5652.
41. C. Lee, W. Yang and R. G. Parr, Phys. Rev. B: Condens. Matter Mater. Phys., 1988, 37, 785–789.
42. R. Ditchfield, W. J. Hehre and J. A. Pople, J. Chem. Phys., 1971, 54, 724–728.
43. C. H. Zhang, J. Y. Gao, Z. Q. Chen and Y. Xue, J. Mol. Graphics Modell., 2010, 28, 799–806.
44. Y. Xiong, H. T. Lu, Y. Li, G. F. Yang and C. G. Zhan, Biophys. J., 2006, 91, 1858–1867.
45. P. Tao, J. F. Fisher, Q. C. Shi, T. Vreven, S. Mobashery and H. B. Schlegel, Biochemistry, 2009, 48, 9839–9847.
46. S. Abdel-Azeim, X. Li, L. W. Chung and K. Morokuma, J. Comput. Chem., 2011, 32, 3154–3167.
47. M. T. Frisch, G. Schlegel, H. Scuseria, G. Robb, M. Cheeseman, J. Scalmani, G. Barone, V. Mennucci, B. Petersson, G. Nakatsuji, H. Caricato, M. Li, X. Hratchian, H. Izmaylov, A. Bloino, J. Zheng, G. Sonnenberg, J. Hada, M. Ehara, M. Toyota, K. Fukuda, R. Hasegawa, J. Ishida, M. Nakajima, T. Honda, Y. Kitao, O. Nakai, H. Vreven, T. Montgomery, J. A. Peralta Jr, J. Ogliaro, F. Bearpark, M. Heyd, J. Brothers, E. Kudin, K. Staroverov, V. Kobayashi, R. Normand, J. Raghavachari, K. Rendell, A. Burant, J. Iyengar, S. Tomasi, J. Cossi, M. Rega, N. Millam, J. Klene, M. Knox, J. Cross, J. Bakken, V. Adamo, C. Jaramillo, J. Gomperts, R. Stratmann, R. Yazyev, O. Austin, A. Cammi, R. Pomelli, C. Ochterski, J. Martin, R. Morokuma, K. Zakrzewski, V. Voth, G. Salvador, P. Dannenberg, J. Dapprich, S. Daniels, A. Farkas Foresman, J. Ortiz, J. Cioslowski and J. Fox, D. Gaussian 09 Revision D.01, Gaussian Inc., Wallingford, CT, 2009.
48. R. K. Dennington and T. Millam, J. GaussView, version 5Semichem Inc., Shawnee Mission, KS, 2009.
49. W. Humphrey, A. Dalke and K. Schulten, J. Mol. Graphics, 1996, 14, 33–38.
50. P. Tao and H. B. Schlegel, J. Comput. Chem., 2010, 31, 2363–2369.
51. M. M. Benning, T. Haller, J. A. Gerlt and H. M. Holden, Biochemistry, 2000, 39, 4630–4639.

---

Footnote

† Electronic supplementary information (ESI) available: RMSD values of MD simulations. Scanning curves of reaction coordinate of ONIOM calculations and so on. See DOI: 10.1039/c6ra19346c

---