Decoding Enzymatic Dechlorination with Multiscale Modeling: Mechanistic Insights into Native Haloalkane Dehalogenase from Xanthobacter autotrophicus and Its Designed Variants

Abstract

Chlorinated hydrocarbons are widely used as solvents and synthetic intermediates, but their chemical persistence can cause hazardous environmental accumulation. Haloalkane dehalogenase from Xanthobacter autotrophicus (DhlA) is a bacterial enzyme that naturally converts toxic chloroalkanes into less harmful alcohols. Using a multiscale approach based on the empirical valence bond method, we investigate the catalytic mechanism of 1,2-dichloroethane dehalogenation within DhlA and its mutants. The reaction proceeds through two chemical steps: a bimolecular nucleophilic substitution followed by hydrolysis to form the alcohol. Our simulations accurately reproduce experimentally observed activation barriers for both steps and reveal how specific amino acids influence catalytic efficiency. While the catalytic D124-H289-D260 triad is well established, our results show that secondary active-site residues affect the reaction rates by shaping an electrostatic network that controls a trade-off between the two chemical steps. This interplay means that improving one step may compromise the other, highlighting the complexity of enzyme optimization. Guided by extensive experimental data alongside generative AI predictions, we propose a multiple mutant with the potential for enhanced overall biocatalytic performance. These findings deepen the mechanistic understanding of DhlA and provide a predictive framework for the rational design of improved dehalogenases, with promising applications in biocatalytic degradation of environmental pollutants.

Graphical Abstract

INTRODUCTION

Microorganisms utilize enzymes to convert hazardous compounds into harmless substances, often transforming them into benign byproducts. Haloalkane dehalogenases are bacterial enzymes that have been employed in bioengineering applications to assist in the recycling and bioremediation of halogenated aliphatic compounds, such as 1,2-dichloroethane (DCE), which are widespread environmental pollutants.¹ DCE is a volatile organic compound² commonly used as an industrial solvent and as a precursor in the production of vinyl chloride, a key material for polyvinyl chloride plastics. Haloalkane dehalogenase from Xanthobacter autotrophicus GJ10 (DhlA) is a promising enzyme that catalyzes the conversion of DCE into a less harmful product, an alcohol. Moreover, DhlA has demonstrated its practical applicability in full-scale bioremediation efforts, including groundwater purification plants targeting DCE contamination.^3,4 In the present study, we aim to computationally explore the origin of the catalytic power of this enzyme to identify specific residues in the protein environment, the modification of which could enhance the catalysis.

The catalytic mechanism of DhlA has been extensively investigated using various methods, including crystallography,^5,6 site-directed mutagenesis,^7–9 and kinetic studies.^10,11 As a member of the α/β-hydrolase superfamily, DhlA shares structural features with other hydrolases. Its catalytic core comprises an oxyanion hole and a catalytic triad, while the substrate binding site lies between the cap and catalytic domains. The DhlA-catalyzed degradation of DCE is proposed to proceed via the formation of the Michaelis complex, followed by two distinct chemical steps, as shown in Figure 1. In the first step, S_N2 displacement occurs when the bond between the carbon and chlorine atoms of the substrate molecule is cleaved by the nucleophilic attack of D124 in the active site. This results in the formation of an ester covalently bound to DhlA through the Oδ2 of D124. The ester intermediate is subsequently hydrolyzed. A tandem of catalytic acid D260, H-bonded to catalytic base H289, activates a nucleophilic water molecule, enabling it to attack the ester carbonyl.¹² The resulting tetrahedral intermediate then breaks down, ultimately yielding the alcohol.

Figure 1.

(A): Crystallographic structure of DhlA with the substrate molecule DCE (PDB entry 2DHC). (B): S_N2 displacement of the chlorine atom from DCE, forming a covalent bond between DCE and the carboxylate group of D124, resulting in the formation of an ester. (C): Hydrolysis of the ester, yielding 2-chloroethanol. (D): Active site of DhlA with the substrate-binding molecule (PDB entry 2DHC). (E): The substrate covalently bound to DhlA (PDB entry 2DHD). The protein is represented as a blue ribbon, while the substrate and key residues facilitating the chemical conversion are shown as a ball-and-stick model with the color code for atoms as follows: carbon in beige, nitrogen in blue, oxygen in red, and chlorine in green.

Computational chemistry methods have provided valuable insight into the catalytic mechanism of DhlA, particularly with regard to the S_N2 displacement step. Computational work has revealed how the unique hydrophobic environment and precise positioning of active-site residues in DhlA facilitate efficient S_N2 displacement and subsequent hydrolysis.¹³ Several studies have characterized the geometry of substrate alignment, showing the importance of in-plane positioning of DCE relative to the carboxylate group of D124.^13,14 Molecular dynamics (MD) simulations have also been used to probe the active site and binding channel architecture, identifying potential water-access pathways.¹⁵ These results highlighted the compact, buried nature of the active site and its preferential fit for small ligands such as DCE. The role of the protein environment has been addressed in several computational studies using cluster models or semiempirical QM/MM approaches. For example, W125 and W175 have been identified as critical for stabilizing the leaving Cl⁻ ion and lowering the activation barrier in the first reaction step.^16,17 In addition, mutational analyses supported by computational data have identified other residues—F172, P223, and V226—that also contribute to leaving group stabilization.¹⁶ Semiempirical QM/MM modeling has further shown that the enzyme preferentially stabilizes the transition state via electrostatic effects, significantly lowering the reaction barrier compared to aqueous media.^18,19 The full two-step enzymatic mechanism has been previously modeled but using a reduced system comprising 13 active-site residues.^20,21 These studies established the catalytic roles of D124, H289, and D260 and provided estimates of the energy barriers for both chemical steps.

At this stage, we found it useful to explore the enzyme in silico to gain a deeper understanding of the role of the protein environment in the full catalytic process. Identifying the residues that facilitate or hinder the catalytic function of the enzyme is essential for pinpointing targets for potential enhancements. Optimizing these key residues allows for the rational design of more efficient modifications to the protein sequence with improved catalytic properties. The entire chemical process in a full protein environment can be modeled using multiscale approaches, which allow for the simultaneous treatment of a large molecular system at different levels of theory.²² The empirical valence bond (EVB) method is particularly well suited for this purpose, as it enables free energy simulations that track the evolution of the substrate–enzyme system along the reaction pathway in a biologically relevant environment.^23,24 This method provides an effective way to evaluate catalytic effects and the influence of the protein medium on the enzymatic reaction (e.g., refs 25–28). Previous EVB studies have successfully reproduced experimental activation barriers for wild-type DhlA and various mutants,^29–31 and have also proposed mutations predicted to enhance the catalytic efficiency of the S_N2 step.²⁹

While previous simulations have focused primarily on the S_N2 reaction step, this study aims to provide a comprehensive model of the entire chemical conversion process, with particular emphasis on the hydrolysis step. Our goal is to fully account for the effects of the protein environment and to explore the impact of specific amino acid replacements on enzyme activity. To achieve this, we begin by computing the free energy profile of the reference reaction in aqueous solution, allowing us to identify the rate-limiting step and to derive the parameters required for EVB calibration. By applying these calibrated diabatic potentials to a more complex biological medium, we model the same reaction within the protein environment, with particular attention to the hydrolysis step. A detailed analysis of structural, energetic, and electrostatic factors in the wild-type enzyme allows us to evaluate the effects of single- and double-point mutations in newly designed, optimized variants of DhlA. Based on these findings, we propose a new DhlA variant that could accelerate both chemical steps, thereby improving the enzyme’s overall catalytic efficiency.

METHODS

Quantum Chemistry Calculations.

Quantum chemistry calculations were employed to estimate the energetics of dehalogenation in an aqueous environment. The chemical reactions were modeled at the ωB97X-D/6–311+G(d,p)/SMD level.^32,33 The choice of a range-separated hybrid functional was guided by previous computational studies on nucleophilic addition and hydrolysis reactions.^34,35 Second-order energy derivatives were computed to confirm the local minima for the reactant and product states as well as the local maxima for the transition states. A relaxed scan through the intermolecular hydrogen bond was performed to identify the energetics of the proton transfer. All calculations were carried out using Gaussian16 software.³⁶ The atomic charges and coordinates of the calculated molecules, along with technical details, are provided in Section S1 of the Supporting Information.

Empirical Valence Bond Calculations.

MD simulations were performed to elucidate the structural, energetic, and electrostatic effects of the protein environment on the chemical reaction. The EVB method, implemented in the Molaris-XG program,³⁷ was used for this purpose. Atomic coordinates from PDB entry 2DHC were employed as the starting configuration. This structure, derived from X. autotrophicus GJ10, was determined by X-ray diffraction at a resolution of 2.30 Å.³⁸ It contains a water molecule in the enzyme’s active site, and thus a model incorporating one explicit water molecule in the active site was considered. The experimental structure was modified using the Chimera program³⁹ to introduce specific amino acid substitutions. For each substitution, side-chain conformations were selected from the Dunbrack rotamer library.⁴⁰ The enzymes were then protonated and relaxed for 1 ns by using the Molaris-XG program.

For both the relaxation and the subsequent production runs, the system was divided into three regions. Region I consisted of the reaction center; Region II included all protein atoms within a 20 Å sphere from the reaction center; and Region III comprised the rest of the protein, which was kept frozen throughout the simulations. Electrostatic potential (ESP) charges were applied to the atoms in Region I, and the ENZYMIX force field⁴¹ was used for all other atoms. The protein and substrate were immersed in a 20 Å water sphere, and the surface constrained all-atom solvent (SCAAS) model⁴² was employed. Long-range interactions were modeled using the local reaction field (LRF) method.⁴³

After relaxation, five structures were selected for the production run to model the S_N2 reaction. The production run, which employed free energy perturbation/umbrella sampling (FEP/US)⁴⁴ modeling, was divided into 31 frames, each lasting 30 ps with a time step of 1 fs. The same settings (31 frames, 1 fs time step, and 30 ps per frame) were applied to simulate the second chemical step, hydrolysis, using the structures extracted from the final frame of the S_N2 step as starting coordinates. All computations were conducted at 303 K (30 °C). Calibration was performed based on experimental^45,46 and/or calculated activation energies and free energies for related reactions in water. Further details on the model system setup and dynamics are provided in Section S2 of the Supporting Information.

RESULTS AND DISCUSSION

Mechanism and Energetics of Dechlorination in Water.

The dechlorination of DCE in an aqueous environment was first examined by considering the reaction mechanism in the DhlA protein, as demonstrated in ref 6 and shown in Figure 1. This process consists of two steps: acylation of the haloalkane (Reaction I, Figure 2A), followed by deacylation to yield the primary alcohol (Reaction II, Figure 2B). Quantum chemical calculations were performed to estimate the energetics of these reactions, specifically, the free energies (ΔG) and activation energies (ΔG^⧧). The energetics of both steps in water are summarized in Figure 2.

Figure 2.

(A,B) show chemical schemes, while (C,D) display energetic diagrams of the S_N2 and hydrolysis reactions, respectively. The energies of the states shown in green are given relative to the RS of each reaction separately, with units in kcal·mol⁻¹. Important interatomic distances are shown in black, with units in Ångströms. Atom color coding: hydrogen in white, carbon in beige, nitrogen in blue, oxygen in red, and chlorine in green.

In Reaction I, the model system included DCE and the acetate anion as a simplified nucleophile (RS in Figure 2C), with the chemical conversion resulting in a chlorine anion and 2-chloroethyl acetate (PS in Figure 2C). The energetic profile of this reaction is shown in Figure 2C. According to the QM simulations, the reaction is exothermic, with a computed barrier of 23.9 kcal·mol⁻¹, which closely matches the 23.6 kcal·mol⁻¹ and 24.6 kcal·mol⁻¹ values estimated in earlier works.^45,46 Reaction II (Figure 2B) involves base-catalyzed hydrolysis, where imidazole accelerates the nucleophilic attack of water on 2-chloroethyl acetate. The calculated reaction profile is shown in Figure 2D, along with the relative computed molecular structures. The process begins with a proton transfer from the water molecule to the nitrogen atom of the imidazole ring, after which the resulting hydroxyl anion attacks the carboxylic carbon atom of the ester. This nucleophilic addition leads to the formation of the tetrahedral intermediate (TI), as shown in Figure 2D. The imaginary mode of the transition state (TS1) involves the motion of the carboxylic carbon and the water molecule, as shown in Figure S1 of the Supporting Information. No barrier was detected for the proton transfer step (Figures S2 and S3 of the Supporting Information), suggesting that the nucleophilic addition is concerted.

The TI is predicted to be relatively unstable, undergoing further transformation via deacylation to yield acetic acid and the 2-chloroethylate anion, which then accepts a proton from imidazole to form 2-chloroethanol. An early computational study on the dehalogenase LinB⁴⁷ suggested the existence of an additional intermediate, TI’. The intermediates TI and TI’ should differ in the coordination of the protonated imidazole to the hydroxy or ester oxygen of the ester molecule, as shown in Figure 3. These states were found to be energetically similar to an energy difference of 0.04 kcal·mol⁻¹, which is within computational error. This small difference suggests a rapid transformation of TI to TS2.

Figure 3.

Recoordination of the tetrahedral intermediate (TI, left) to form the transition state (TS2, right) via an intermediate state (TI’, middle), where imidazole is recoordinated from the oxygen of the hydroxy group to the oxygen of the 2-chloroethyl moiety. Energies are relative to TI. Important interatomic distances are shown in black, with units in Ångstroms. Atom color coding: hydrogen in white, carbon in beige, nitrogen in blue, oxygen in red, and chlorine in green.

In summary, the chemical transformation of the haloalkane begins with an S_N2 reaction with the carboxylate anion, forming the ester. This exothermic reaction has an energy barrier of about 24 kcal·mol⁻¹. The chlorine displacement step is followed by base-catalyzed nucleophilic attack by a water molecule, which produces an unstable intermediate state. Cleavage of the C–O bond in the intermediate leads to the final product, the primary alcohol. The rate-limiting step of hydrolysis is the formation of the tetrahedral intermediate with a ΔG^⧧ of 27 kcal·mol⁻¹, while its conversion to the alcohol is nearly barrierless.

Dechlorination in the Protein Environment.

The chemical transformation of DCE in the protein follows the same steps as in water: the acetate group of D124 plays the role of the nucleophile in the S_N2 displacement, and the imidazole ring of H289 serves as the base catalyst in the hydrolysis step.¹² The rest of the protein environment affects the chemical process through spatial restrictions and electrostatic interactions. The active site of DhlA consists mainly of hydrophobic amino acids: four phenylalanines (F128, F164, F172, and F222), two tryptophans (W125 and W175), two leucines (L262 and L263), a valine (V226), and a proline (P223). All of these residues influence the enzymatic processes via stabilization or destabilization of RS or TS. A few polar residues of the active site—E56, D124, D260, and H289—play essential roles in the chemical transformation of DCE within the enzyme’s cavity.^6,48

To clarify the detailed mechanism of dechlorination catalyzed by DhlA, a computational study was conducted on both chemical steps using the EVB method. First, the reactions were modeled in water, and the diabatic potentials were calibrated using the data obtained from the QM calculations as discussed above. The calibration of the EVB outputs, along with other technical details, is explained elsewhere.⁴⁹ The two chemical steps were then simulated in the protein environment, and the energetic profiles in aqueous and enzyme media are depicted in Figure 4. A significant decrease in the activation energy barrier is observed in all cases when the enzyme facilitates the reaction. However, the free energies of both the S_N2 and the hydrolysis reactions do not differ between water and the protein. The TI in the hydrolysis step is almost equally unstable in both media relative to TS1, which agrees with the earlier MD calculations for the dehalogenase LinB.⁴⁷

Figure 4.

Free energy profiles obtained with EVB for (A): the S_N2 displacement, (B): formation of the tetrahedral intermediate, and (C): the final product 2-chloroethanol. The reference profile in water is shown in black, while the DhlA-catalyzed profile is displayed in blue.

The representative molecular structures of the RS, TS, and PS in the nucleophilic substitution reaction, corresponding to the energetic diagram in Figure 4A, are shown in Figure 5. The key changes in the distances between C_DCE–O_D124 and C_DCE–Cl_DCE are from 2.74 to 1.53 Å and from 1.89 to 3.22 Å, respectively. Compared to the reaction in water, the distances at the TS are nearly the same; however, DCE and the acetate are coordinated much closer in the protein at both the RS and PS states. The leaving group of the substrate, the chloride anion Cl⁻, directly interacts with two tryptophan residues, W125 and W175. While residues F172, P223, and V226 do not directly interact with Cl⁻, they still influence the kinetics. In terms of distances, the two tryptophans coordinate the anion at 2.30–2.40 Å, while the hydrogens of the side chains of F172, P223, and V226 are farther, at 3.42, 3.26, and 3.43 Å, respectively. E56, one of the few polar residues near Cl⁻, forms H–Cl contacts at distances of 2.82 (Hγ2) and 3.18 Å (Hβ1). The central hydrogen of the phenyl ring of F128 is in close proximity to the other chlorine atom, which does not undergo a chemical change. This may help orient the substrate properly in the active site and stabilize the TS via intermolecular Cl–H contacts of 2.91 Å. The other residues closest to this chlorine are F164 (3.32 Å) and L262 (3.37 Å). The final product of the reaction, ester, is chemically bound to the protein through D124. The ester then undergoes further transformation in the hydrolysis reaction.

Figure 5.

First chemical step of DCE conversion in the DhlA environment: the transformation of DCE into the ester bonded to the enzyme. The most important interatomic distances are shown in black, with units in Ångströms. The data were averaged over a 30 ps frame across five replicas. Atom color coding: hydrogen in white, carbon in beige, nitrogen in blue, oxygen in red, and chlorine in green.

The few polar amino acids in DhlA’s active site, E56, D124, and D260—form an oxyanion hole. In the first step of the ester hydrolysis displayed in Figure 4B, H289 in tandem with the catalytic acid D260 activates the nucleophilic water molecule via deprotonation, facilitating nucleophilic attack at the ester’s carbonyl group. The molecular structures of RS, TS1, and TS are shown in Figure 6. The catalytic residue D260 is approximately 8 Å away from the catalytic water molecule and is hydrogen-bonded to the catalytic base H289, with intermolecular distances of Nδ1–Oδ1 at 2.89 Å and Nδ1–Oδ2 at 2.10 Å. To examine the contribution of this residue to catalysis, two types of simulations were performed: one where the residue was kept ionized and another where it was nonionized. A similar check was previously applied to the E56 residue in ref 29. The results indicate an effect of about 5 kcal· mol⁻¹: the activation energy related to TS1 was found to be 16.8 kcal·mol⁻¹ when D260 was kept ionized vs 22.4 kcal·mol⁻¹ when it was un-ionized. Atomic coordinates of the initial states were identical in both types of calculations, and the results were averaged over five replicas. Among the surrounding residues, F128 is the one that contacts the leaving 2-chloroethylate group: the Hζ–Cl distance changes from 2.81 Å at the RS to 3.10 Å at the TI. Q123 is near the catalytic water, with the Hε21–Ow distance changing from 2.39 Å at RS to 2.87 Å at TI. F290 is close to H289 at 2.91 Å.

Figure 6.

Second chemical step of DCE conversion in the DhlA environment includes the transformation of the ester into a tetrahedral intermediate. The most important interatomic distances are shown in black, with units in Ångströms. The data were averaged over a 30 ps frame for five replicas. Atom color coding: hydrogen in white, carbon in beige, nitrogen in blue, oxygen in red, and chlorine in green.

The protein environment influences the reactions by creating an electrostatic network through nearby amino acid residues. Both chemical steps occur in the active site pocket, which can be roughly divided into two regions, as shown in Figure 7A, where residues close to the leaving chloride anion are colored royal blue, while those near the water molecule are in coral. The distance between the water and chloride is approximately 7.5 Å, and residues within 4 Å of either the water oxygen or chloride anion are considered, as shown in Figure 7A. Residues E56, W125, F222, P223, V226, F172, and W175 surround Cl⁻, and these are thought to be important for Cl⁻ stabilization, especially W125 and W175, as highlighted in ref 50. The second chemical step has not received as much attention, and there are limited data on the overall mechanism and the contribution of nearby residues. As shown in the figure, residues G55, T58, Q123, D124, H289, and F290 are closest to the catalytic water molecule. Information about the electrostatic effects of some active-site residues is provided in Figure 7B, which demonstrates the residues’ electrostatic interactions with D124, the key residue in both chemical steps. The most important residues involved are E56, Q123, W125, G127, F128, and N148. Notably, while the same residues contribute, their roles in stabilizing the RS and TS alter between the S_N2 displacement and hydrolysis. A more detailed understanding of how exactly active-site residues function in the enzymatic process can be gained through mutational analysis using both experimental and computational techniques as discussed below.

Figure 7.

(A): Relaxed molecular structure of the reactant state in the hydrolysis reaction: the oxygen of the catalytic water is shown as a red sphere, and the chloride anion is shown as a blue sphere. Residues with atoms within 4 Å of this oxygen or chloride are depicted in coral and royal blue colors, respectively. (B): Electrostatic interactions of the D124 residue with all other residues within a 10 Å radius in the S_N2 (top panel) and hydrolysis (bottom panel) reactions, in kcal·mol⁻¹. The results were averaged over five replicas.

Unraveling and Improving the Catalysis via Active Site modification.

The mutational analysis of DhlA performed with site-directed mutagenesis has revealed that residues within the active site play crucial roles in the enzyme’s catalytic efficiency, with each mutation affecting the enzymatic process in a distinct way. Starting with D124, which is critical for both chemical steps, mutations like D124G,⁷ D124A,⁷ D124E,⁷ and D124N⁹ resulted in a loss of activity, confirming its importance in the formation of the ester intermediate. The E56Q mutation,²⁹ which alters the oxyanion hole, shows an improved activity with DCE and its bromine-substituted analogue 1,2-dibromoethane (DBE). The mutation W175Y/E56N²⁹ exhibited a similar to the wild type catalytic activity and it showed an improvement over the W175Y mutant.

W125 and W175, which are involved in stabilizing the leaving chloride ion, were also targeted for mutation. Variants like W125G,⁵⁰ W125F,⁵⁰ W125R,⁵⁰ W175Q,⁵⁰ W175F,⁵¹ and W175Y^51,52 all slowed down the catalysis, emphasizing the significance of these residues in stabilizing the transition state. Other mutations, such as F172Y,⁵³ F172W,⁵³ and F164A,²⁹ demonstrated how changes in the aromatic nature of residues surrounding the active site can influence the reaction. For example, the F172Y⁵³ and F172W⁵³ mutations affected the enzyme’s ability to carry out the S_N2 reaction without altering the hydrolysis step, suggesting that aromatic residues in the active site provide structural support to the transition state and leaving group in the S_N2 reaction. In terms of V226, variants like V226G, V226A, and V226L illustrated the effects of hydrophobic residue alterations.⁵⁴ V226 has van der Waals contacts with W125 and F172, seems to contribute to leaving group stabilization via W125, and can influence halide release. The V226A and V226L mutations both increased the reaction rate in the case of DBE, while the V226A variant slowed down the S_N2 step for DCE, suggesting that steric changes near the active site can reduce reaction efficiency. The double mutant W125F/V226Q²⁹ did not show any activity.

The D260N mutation, along with D260N/N148D and D260N/N148E,⁵⁵ revealed the importance of the catalytic acid in hydrolysis. These mutations disrupt the acid–base catalysis mechanism, further confirming the role of D260 in activating the water molecule for nucleophilic attack on the ester. The D260N/N148E mutant showed a significant decrease in k_cat for the debromination of DBE due to slower carbon–bromine bond cleavage and alkyl-enzyme hydrolysis.⁵⁵ The inherent role of H289 was proven with the H289Q mutant that demonstrated no catalytical activity with haloalkanes.¹²

Although several variants were obtained, kinetic measurements for individual chemical steps are available for only a few of these mutants. Table 1 presents the previously determined experimental rate constants for both chemical steps in the wild-type enzyme and four of its mutants alongside the results of the EVB simulations. The EVB simulations were carried out at 30 °C, in line with earlier experimental measurements for the native DhlA and the F172W and V226A mutants. In contrast, the L262V and F128L/L262V variants were tested at 15 °C. For consistency, the activation free energies in Table 1 were converted to 30 °C by using transition state theory, enabling a direct comparison between experiment and simulation for all protein variants.

Table 1.

Calculated Activation Free Energy (calc. ΔG^‡), Experimentally Estimated Rate Constants (exp. ΔG^‡), and Experimental Activation Free Energy Derived from Them Using the Eyring Equation at 303 K (exp. k₂ and exp. k₃)^e

	SN2 reaction			hydrolysis reaction
protein variant	calc. ΔG‡	exp. ΔG‡	exp. k2	calc. ΔG‡	exp. ΔG‡	exp. k3	temp., pH
wild type	15.4	15.4	50 ± 10a	16.8	16.2	14 ± 3b	30 °C, 8.248
F172W	18.8	16.8	4.5 ± 1	16.4	16.4	9.5 ± 1	30 °C, 8.253
V226A	15.0	16.2	14 ± 1	14.9	16.4	9 ± 2	30 °C, 8.254
L262V	15.7	16.2c	14.1 ± 0.2c	16.4	18.6c	0.24 ± 0.01c	15 °C, 8.431
F128L/L262V	18.1	19.0c	0.134 ± 0.001c	15.2	16.3c	10.6 ± 0.4c,d	15 °C, 8.431

^aThe experimental rate constant k₂ for the wild type at 15 °C is 7.6 ± 0.8 s⁻¹.

^bThe experimental rate constant k₃ for the wild type at 15 °C is 0.19 ± 0.02 s⁻¹.

^cThe rate constants were measured at 15 °C, and the corresponding ΔG^‡ values were derived using the Eyring equation and reported at 30 °C for a comparison purpose.

^dFor the F128L/L262V mutant, the hydrolysis and product release steps were not distinguished experimentally.

^eThe computational results were averaged over five replicas. The important experimental conditions—temperature and pH—are noted in the far-right column.

Two single-point mutations, F172W and V226A, reveal that the F172 and V226 residues indirectly influence the first chemical step, slowing the S_N2 reaction without significantly affecting hydrolysis. The more recent L262V and F128L/L262V variants were designed to enhance catalysis, particularly in the S_N2 step. Amino acid selection for these replacements was guided by generative artificial intelligence (AI) predictions⁵⁶ and further explored both experimentally and computationally.³¹ EVB calculations agree with the experimentally observed trends and provide additional insight into the effects of amino acid substitutions. Figure 8A displays the active site of the WT and L262V and F128L/L262V mutants at the RS of the hydrolysis reaction. As noted above, the side chain of F128 forms intermolecular Cl–H contacts with the DCE molecule, likely stabilizing the TS in the S_N2 reaction. This interaction is lost with the F128L substitution, which also eliminates the π-stacking interaction with W125, potentially affecting the C–Cl bond cleavage step.

Figure 8.

(A): The representative molecular structures of the active site of the wild-type, L262V, and F128L/L262V DhlA’s variants at the RS in the hydrolysis reaction; (B): D260 and the amino residues within 3 Å from its carboxy group in the F128L/L262V variant; (C): electrostatic interactions of the D260 residue with all other residues within a 10 Å radius at the RS in the hydrolysis reaction, in kcal·mol⁻¹. The results were averaged over five replicas. The color code for atoms: hydrogen in white, carbon in beige (turquoise for D260), nitrogen in blue, oxygen in red, and chlorine in green.

During the hydrolysis reaction, residues surrounding H289 play a role. As previous studies suggest, D260 is the most influential residue among those not undergoing chemical transformation, and substituting nearby amino acids reduces catalytic activity.⁵⁵ The structure of D260 and its surrounding residues is shown in Figure 8B, and the contribution of adjacent residues to electrostatic interactions with D260 is illustrated in Figure 8C. The most prominent neighboring residues influencing D260 are N148, K259, K261, L262, and L263. The L262V substitution occurs in the proximity of D260, with a distance of approximately 3.3 Å between the carboxylate group of D260 and the side chain of L262 or V262 (Figure 8A). Given the importance of D260 in the catalysis, even subtle changes in its local environment could impact the enzyme’s activity. Since valine has a slightly smaller and less flexible hydrophobic side chain compared to leucine, this substitution may affect local electrostatic and steric interactions. Accordingly, the intermolecular distance between the Oδ1 of D260 and Hδ12 of L262 increases from 2.91 Å in the wild type to 3.33 and 3.25 Å between the Oδ1 of D260 and Hγ22 of V262 in the mutants; see Figure 8A.

By optimization of active site residues that directly influence substrate binding, transition state stabilization, and the catalytic environment, it is possible to create enzyme variants that accelerate the overall reaction rate. A promising strategy for achieving this goal involves targeting mutations that indirectly enhance catalysis in both the S_N2 displacement and hydrolysis steps. The main challenge lies in selecting the appropriate residues, or combinations of residues, for substitution, a task that can be addressed using AI-based predictions informed by sequence analysis. Previous successful predictions of modified dehalogenases have focused on S_N2 reaction modeling, and here we aim to extend these efforts to the hydrolysis step. In ref 56, a large pool of variants was generated using an AI tool known as the Maximum Entropy method (MaxEnt). A subset of these sequences was subsequently investigated experimentally in ref 31. Two notable mutants are L262V, which promotes faster nucleophilic attack, and F128L and L262V, which improve the efficiency of the hydrolysis step. Additional MaxEnt predictions identified E56, along with F128 and L262, as promising targets for mutation (see Supporting Information of ref 56).

E56 is particularly interesting, as it is located near the substrate, the departing Cl⁻, and the catalytic water molecule (see Figure 7A). This residue was targeted in several earlier studies and was highlighted by MaxEnt as a prospective site for enhancing catalysis. Based on this information, several combinations of amino acid substitutions were simulated using the EVB approach, and the resulting activation barriers are presented in Table 2. Modifications at position 56 have a notable impact on the S_N2 displacement, with several combinations, including E56N/F128L and E56N/F128L/L262V—leading to substantial reductions in the activation barrier. In particular, the E56N/F128L/L262V variant shows a 2.0 kcal·mol⁻¹ decrease in the S_N2 barrier compared to that of the wild type, which is generally considered significant in enzymatic catalysis. While the hydrolysis barrier for this mutant is reduced only slightly (by 0.2 kcal·mol⁻¹), it remains comparable to the wild-type value, suggesting that the variant avoids the common trade-off of improving one step at the expense of the other. Notably, other variants such as E56Q/L262V and E56N/L262V, although lowering the S_N2 barrier to some extent, result in a substantial increase in the hydrolysis barrier (>2 kcal·mol⁻¹). Therefore, E56N/F128L/L262V stands out as the most promising candidate, balancing favorable energetics for both steps and offering a plausible route toward an improved overall catalytic efficiency.

Table 2.

Calculated Activation Free Energy, ΔG^‡ (in kcal•mol⁻¹), for the S_N2 and Hydrolysis Reactions in the Proposed DhlA Variants

protein variant	SN2 reaction	hydrolysis
wild typea	15.4	16.8
E56Q/F128L	15.2	19.7
E56N/F128L	13.4	18.1
E56Q/L262V	18.4	18.6
E56N/L262V	17.8	20.2
E56Q/F128L/L262V	17.8	16.4
E56N/F128L/L262V	13.4	16.6

^aNote: based on experimental data, the ΔG^‡ of the wild type is 15.4 kcal·mol⁻¹ for the S_N2 reaction and 16.2 kcal·mol⁻¹ for the hydrolysis reaction.

CONCLUSIONS

Multiscale molecular modeling has been applied to uncover the molecular basis of DCE dechlorination in DhlA, identifying several factors that contribute to the decrease in the activation barrier within the protein environment. Our computational results, consistent with available experimental data, highlight the importance of residues E56, F128, F172, and V226 in the S_N2 step, while residues N148, D260, L262, and H289 contribute primarily to the electrostatic stabilization of the hydrolysis reaction catalyzed by DhlA. We reaffirm the essential role of D260 in the hydrolysis step, where it enhances the proton-accepting function of the nearby catalytic base H289 by modulating the basicity of its imidazole ring. This interaction helps to reduce the activation barrier of the hydrolysis step by approximately 5 kcal·mol⁻¹.

The overall mutational analysis of DhlA underscores the complexity of active site remodeling, as individual residues can have distinct and sometimes opposing effects on different chemical steps. This supports a functional categorization of active-site residues into those primarily facilitating the S_N2 step (e.g., F172 and V226) and those enhancing hydrolysis (e.g., D260). Additionally, depending on their role, certain residues—such as D124, W125, W175, H289, and D260—are central to the enzymatic mechanism, while others like V226 and F128 modulate catalytic efficiency by influencing substrate positioning and active-site geometry. Recent findings on the L262V and F128L/L262V mutants illustrate the importance of fine-tuning residue interactions to optimize catalytic performance. These results underscore the delicate balance required to maintain high enzymatic efficiency, as even minor alterations can lead to significant changes in activity. Based on our simulations, we propose the E56N/F128L/L262V mutant as a promising variant capable of significantly lowering the barrier in the S_N2 step while maintaining or exceeding wild-type efficiency in the hydrolysis step.

These insights contribute to a deeper understanding of the DhlA function and offer a rational framework for designing more efficient enzyme variants for industrial bioremediation applications. Although our findings are specific to DhlA, the general strategy may be applicable to other related dehalogenases, such as LinB from Sphingomonas paucimobilis UT26 and DhaA from Rhodococcus rhodochrous NCIMB 13064, which share conserved structural features and catalytic mechanisms. However, we recognize that enzymatic activity can be influenced by organism-specific physiological and environmental factors, and such extrapolations should be made with the appropriate caution. The development of enhanced dehalogenases can be further advanced by integrating QM/MM methodologies with state-of-the-art AI tools, enabling the identification of protein variants with superior catalytic performance.

Supplementary Material

ASSOCIATED CONTENT

Supporting Information

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acscatal.5c03557.

Details of the quantum chemical and EVB calculations, including Cartesian coordinates and absolute energies of stationary points, along with imaginary frequencies for transition states; overview and detailed description of the EVB methodology and simulations; electrostatic potential charges used in EVB simulations; and activation free energies from individual replicas of all DhlA variants studied (PDF)