A Quantum Mechanical Approach to The Mechanism of Asymmetric Synthesis of Chiral Amine by Imine Reductase from Stackebrandtia Nassauensis

Abstract

The asymmetric synthesis of tetrahydroisoquinolines (THIQs) has gained importance in recent years due to their significant potential in drug development studies. In this study, the conversion of 1‐methyl‐3,4‐dihydroisoquinoline substrate to a chiral amine, 1‐methyl‐1,2,3,4‐tetrahydroisoquinoline, under the catalysis of the stereoselective imine reductase enzyme from Stackebrandtia nassauensis (SnIR) was investigated in detail to elucidate the mechanism and explain the experimental enantioselectivity. The results were found to be in agreement with the experimental data. To elucidate the reaction mechanism, quantum mechanical calculations were performed by considering a large cluster of the active site of the enzyme. In this regard, possible reaction pathways leading to both R‐ and S‐products with the corresponding intermediates and the transition states for the hydride transfer from the cofactor to the substrate were considered by density functional theory (DFT) calculations, and the factors contributing to the observed stereoselectivity were sought. The calculations supported a stepwise mechanism rather than the concerted protonation and the hydride transfer steps. The stereoselectivity in the hydride transfer was found to be due not only to the stability of the enzyme‐subtrate complex but also to the corresponding reaction barriers. The calculations were performed at the wB97XD/6‐311+G(2df,2p)//B3LYP/6‐31G(d,p) level of theory using the PCM approach.

In this study, in which the asymmetric synthesis of 1‐Me THIQ was investigated computationally using SnIR from Stackebrandtia nassauensis, a willingness to form S product was observed, consistent with the experimental selectivity in the literature. Computational examination of the mechanism of asymmetric chiral amine synthesis, which has an important place in drug studies, constitutes the backbone of the study.

Introduction

The increasing importance of the role of chiral amines, especially in drug studies, has rendered them as important and challenging candidates in organic synthesis in recent years. They are extremely prominent structures for the synthesis of many pharmaceuticals ^[1] against diseases, including hyperparathyroidism, tuberculosis, depression, Parkinson's, Alzheimer's, prostate and as painkillers.[ ² , ³ , ⁴ , ⁵ ] Tetrahydroisoquinolines (THIQs) are found in the structures of some drugs such as cancer, gout, Alzheimer's, pain, Parkinson's, muscle relaxants[ ⁶ , ⁷ , ⁸ , ⁹ ] and drug development studies continue by diversifying the tetrahydroisoquinoline structure with different substituents.

Diastereoselective hydrogenation of Schiff bases, diastereoisomeric crystallization, C−H insertion, nucleophilic addition and enantioselective reduction of imines or enamines are among the several methods used in chiral amine synthesis.[ ⁴ , ¹⁰ , ¹¹ , ¹² ] Although transition metal‐catalyzed asymmetric hydrogenation has come to the fore in recent years,[ ¹¹ , ¹³ , ¹⁴ ] the use of enzymes in asymmetric chiral amine synthesis enables high enantioselectivity along with more environmentally friendly technologies.[ ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ ] It is known that enzymes such as transaminases, amine dehydrogenases, amine oxidases, lipases and imine reductases are used for the enantioselective synthesis of amines.[ ⁴ , ¹⁶ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ , ²⁴ ] Imine reductases are NADPH‐dependent enzymes and play an important role in the stereoselective synthesis of amines from corresponding imines. ^[4] In the study by Mitsukura et al. in 2010; Streptomyces sp. GF3587 and GF3546 enzymes with imine reducing properties, which show high R‐ and S‐selectivity by using 2‐MPN, were reported. ^[22] This was followed by the purification and analysis of Streptomyces sp. GF3587, an R‐imine NADPH‐dependent reductase ^[25] and then another group reported Q1EQE0, another NADPH‐dependent oxidoreductase from Streptomyces kanamyceticus. It was found that (R)‐2‐methylpyrrolidine was synthesized with high selectivity from 2‐methyl‐1‐pyrroline. ^[26] Leipold et al. carried out studies with an (S)‐imine reductase from Streptomyces sp. GF3546. They reported that five‐, six‐ and seven‐membered imines were reduced with high enantioselectivity. It was also stated that the mentioned enzyme also reduces iminium ions. ^[24] An interesting situation is encountered in the experimental study conducted with AoIRED. It was observed that the stereoselectivity of the enzyme acting on 1‐methyl‐3,4‐dihydroisoquinoline changed depending on the storage time such that the selectivity changed from S to R depending on the storage time. ^[27]

The origin of the catalytic activity of the stereoselective enzymes has been sought in many studies. To this end, many studies have been conducted to investigate the effects of critical residues on catalytic activity and stereoselectivity in imine reductase enzymes.[ ¹ , ²³ , ²⁶ , ²⁷ , ²⁸ , ²⁹ ] In addition to experiments, in silico studies are also of great importance for understanding the mechanisms. Examples include molecular dynamics simulations, docking and DFT calculations that have been performed to understand the changes in stereoselectivity and mutation effects in IREDs.[ ³⁰ , ³¹ , ³² ] In a recent example, the enzyme reaction mechanism of IRED from Amycolatopsis orientalis was investigated by DFT calculations and the possible transition states for the hydride transfer were modelled. In their study, a quantum chemical cluster approach was used where the residues in the active site of the enzyme were utilized in the B3LYP−D3(BJ)/6–31G(d,p) calculations. ^[30]

The experimental studies have shown that the selectivity and the yield obtained from the stereoselective enzymes depend on both the enzyme and the substrate. Therefore, in order to study the enantioselectivity of stereoselective enzymes in depth, each case must be handled separately until the principles can be generalised. As stated by Wu et al. the field of enantioselective catalysis via enzymes is still in its early stages, with much yet to be discovered. ^[33] Although numerous new and groundbreaking studies have been published in recent years, the structure‐activity relationship and mechanistic knowledge required to engineer these enzymes in the desired way are not yet sufficiently established. ^[33] In this regard, it is envisaged that molecular dynamics or quantum mechanical calculations may be useful, especially for mechanistic studies, which could lead to a more rational design of tailor‐made enzymes. ^[33] As part of these efforts, the enantioselectivity of an IRED enzyme will be thoroughly investigated in this study. The experimental study on SnIR from Stackebrandtia nassauensis (Accession No: WP 013019548.1) has demonstrated that it has a high conversion rate of 3,4‐DHIQ derivatives. Although the enantiomeric excess ratios differ depending on the substrate, the enzyme was found to be highly S‐selective for 1‐methyl‐3,4‐dihydroisoquinoline (1‐Me DHIQ). ^[34] In this study, the mechanism of the conversion of 1‐methyl‐3,4‐dihydroisoquinoline substrate to a chiral amine, 1‐methyl‐1,2,3,4‐tetrahydroisoquinoline (1‐Me THIQ) under the catalysis of the imine reductase from Stackebrandtia nassauensis (SnIR) enzyme is investigated in detail, based on the experimental study (Scheme 1). This compound is of particular significance as it forms the backbone of numerous pharmaceutical structures employed in the treatment of neurodegenerative diseases. ^[34] In order to elucidate the reaction mechanism, quantum mechanical calculations were employed by considering a large cluster of the enzymes's active site. The quantum cluster approach is a frequently employed method in enantioselectivity studies on enzymes and its successful results are also available in the literature.[ ³⁵ , ³⁶ ] In this context, potential reaction pathways with corresponding intermediates and transition states for the hydride transfer from the cofactor NADPH to the substrate were considered using by density functional theory calculations (DFT), with the objective of elucidating the factors that govern the observed stereoselectivity.

Scheme 1.

Chiral amine (1‐Me THIQ) synthesis reaction with imine reductase from Stackebrandtia nassauensis (SnIR).

Results and Discussion

The synthesis of chiral amines starting from imines involves two main steps: firstly, the formation of the iminium ion from the imine and secondly, the subsequent reduction of the iminium to the chiral amine (Scheme 2). In fact, it is known from the literature that the initial step in the reduction of the imine structure to a chiral amine is the formation of an iminium ion by proton transfer, which is followed by a hydride transfer that takes from NADPH. ^[26] It is postulated that a protic residue in the vicinity of the active site of the enzyme may be involved in the iminium formation step. In R‐type IREDs, this proton donor residue is usually Asp, whereas in S‐type IREDs it is usually Tyr.[ ¹ , ²⁶ , ²⁸ ] While some studies ^[28] have found that mutations in these critical residues only alter the catalytic activity of IREDs without affecting stereoselectivity, others ^[27] have reported changes in stereoselectivity as well.

Scheme 2.

The stepwise mechanism of the reduction of 1‐Me DHIQ (1) by imine reductase to chiral amine 3.

One of the points to consider here is that it has been reported that “in the absence of a protic residue, the chiral amine formation step is successfully performed by imine reductase from a preformed iminium cation”. ^[33] Furthermore, in an imine reduction study conducted with dihydrofolate reductase, it has been stated that protonation of the imine substrate occurs even with a water molecule.[ ¹ , ²⁶ ] In a separate study conducted with imine reductase from Paenibacillus lactis (PISIR), it was observed that especially when imines and their corresponding prochiral iminium salts were compared in terms of kinetic data, both k_cat and k_cat /K_m values of iminium salts were found to be considerably higher than those of imines. This finding indicates that iminium salts exhibit faster reactivity than imines. ^[37]

While some studies posit that the mechanism commences with the catalytic protonation, ^[28] others emphasise that the enzyme synthesizes the appropriate chiral amine by utilising the preformed iminium ion, thanks to its catalytic activity. ^[33] However, the two processes share a common feature: the catalysis of the second step, in which the chiral amine is formed by hydride transfer to a prochiral iminium structure, is catalyzed by the enzyme.

In light of the aforementioned details, our investigation on the mechanism in this study will start from the second step, since the second step, the hydride transfer to iminium, is the step that decides the stereoselectivity. Subsequently, further consideration will be given to the first step, which is the formation step of the iminium ion due to the presence of a protic residue in SnIR, will be considered further.

The initial stage of the study involved a preliminary model to approximately determine the reaction path for the hydride transfer step (Figure S1). The model study was attempted directly with and without water assistance. It was observed that a significantly higher barrier (71.42 kcal) was required in the water assisted case than in the non‐assisted case (30.76 kcal). This high barrier is in accordance with the fact that the negative charge of the hydride to be transferred, is not favored by an electron‐rich atom such as oxygen. Subsequently, the studies presented herein focused on direct hydride transfer without the assistance of water, utilising the formed quantum cluster model.

Detailed calculations on the mechanism using the cluster model were first performed to determine whether the reaction takes place in a single step, as in Scheme 3, or in a stepwise manner as depicted in Scheme 2. In the concerted mechanism, the transfer of the proton from the Tyr to the substrate and the hydride transfer from the cofactor were performed in a single step. It was observed that the concerted barrier for R product formation was notably high (36.04 kcal) when started with the substrate in its neutral form. This high energy barrier was incompatible with the enzyme kinetics.

Scheme 3.

Concerted mechanism.

Then, the stepwise mechanism was investigated in detail with the cluster model (Scheme 2). In the first step of the two‐step reaction, the proton transfer from Tyr171 resulted in the formation of an iminium structure.

As previously stated, the enantioselectivity in the reaction was determined by the second step, in which hydride transfer occurs from NADPH to the iminium substrate (2). Therefore, the aforementioned transfer was subjected to comprehensive analysis through computational modelling.

As detailed in the Experimental Section, the possible conformers of the transition states (TS) for the hydride transfer step were considered (TS2 in Scheme 2), and their corresponding enzyme‐substrate complexes were identified from the IRC calculations as the “reactive conformers” for catalytically facile enzyme‐substrate geometries. In the cluster approach, Sheng et al states that the substrate binding and product release are not considered explicitly and that the substrate binding is assumed to be reversible. Accordingly, in our study, this binding was evaluated on the assumption that it is reversible. ^[36] The barriers and the relative energies of the modelled enzyme‐substrate complexes (Figure 1) were considered in conjunction with one another and, two types of reactant‐product pairs were chosen such that one pair has the lowest barrier and the other pair is from the lowest enzyme‐substrate complex (Table S1). The pro‐S and pro‐R enzyme‐substrate structures have permitted the classification of the structures with respect to the proximity of the substrates to Tyr171. In the hydride transfer, to form the amine structure, the proximity or the distance of the nitrogen environment of the substrate to the Tyr amino acid was observed to affect the calculated barrier as well. It is also important to recall at this point that in earlier studies, the tyrosine residue was identified as a potential source of the proton required for the formation of the iminium ion. ^[28] For this reason, the selected structures for the hydride transfer were investigated in two groups for the S and R forming structures: If the substrate's nitrogen and Tyr amino acid are in close proximity to one another, the structures belonging to these pathways will be denoted as R₁ and S₁ ; if distant, as R₂ and S₂ . Throughout the discussion that follows, the prochirality of the structures will be represented by R/S following the number of the structure (intermediate or transition state, as shown in Scheme 2), printed in bold.

Figure 1.

Calculated energy profiles for R and S paths.

As a start, the energies of the prochiral structures, corresponding to the initial geometries prior to the hydride transfer to the iminium, namely, 2_R₁ , 2_R₂ , 2_S₁ and 2_S₂ , were compared and 2_R₁ was found as the most stable one (Figure S2). However, the free energy barrier for the hydride transfer from 2_R₁ is 30.77 kcal/mol, which is a high barrier for an enzyme reaction. If the substrate is in the pro‐chiral S structure with a close distance to Tyr as in 2_R₁ (2_S₁), then this geometry is 11.36 kcal/mol higher in energy than 2_R₁ but the transition state is at 23.53 kcal/mol of relative energy with respect to 2_R₁ . In 2_R₁ , the phenolate oxygen atom on Tyr interacts with both the substrate and the NADPH fragments (Figure 2). However, in 2_S₁ , a similar interaction is not present between substrate and Tyr interacts with only NADPH. On the other hand, in 2_S₁ , the substrate forms a favourable H‐bonding (2.11 Å) with the NADPH, between the NADPH C=O and the H of sp³ C next to the substrate N, which are in the vicinity of the hydride donor‐acceptor carbons (Figure 2). However, in 2_R₁ such a H‐bonding is not present and the reacting C centers are at 4.74 Å (the farthest among the E−S complexes) while in 2_S₁ they are at 3.57 Å, which is the shortest reacting distance among the enzyme‐substrate complexes. Therefore, the aforementioned H‐bonding enables a closer distance for the reacting carbons in their pro‐chiral states. The TS2_R₁ requires a high‐energy because the strong interaction of Tyr inhibits the lability of the substrate to undergo reaction. Additionally, in this geometry, the methyl group of the substrate is directed under the pyridine ring in the nicotinamide moiety of NADPH. However, in TS2_S₁ , neither methyl group is exposed to this steric interaction, nor is there any overlap between the substrate isoquinoline and the NADPH's pyridine rings (Figure 3). The substrate is more labile and the favourable H‐bonding between the oxygen of NADPH and the substrate H is retained in both 2_S₁ and TS2_S₁ geometries. It is important to emphasise that the current 2_S₁ geometry precludes the possibility of both a hydride transfer and the Tyr‐substrate interaction occurring simultaneously, as observed in 2_R₁ (Figure 2). Furthermore, TS2_R₁ is a late transition state (with respect to TS2_S₁ ) with C_donor‐H_hydride, and H_hydride‐C_acceptor distances of 1.42 Å and 1.36 Å, respectively, where these distances were 1.30 Å and 1.45 Å, respectively in TS2_S₁ . (Figure 2).

Figure 2.

Critical distances for 2_R₁ ,TS2_R _1, 2_S₁ and TS2_S _1.

Figure 3.

 A schematic representation of the overlap of the isoquinoline ring of substrate (gray colored) and the nicotinamide ring of NADPH (green colored) in the transition states.

In the case of the reacting sites being situated away from Tyr171, as in 2_S₂ and 2_R₂ , the interaction of the Tyr with NADPH is conserved in both complexes, while Tyr interaction with the substrate is lost due to directionality of the N atom. The relative energies of 2_R₂ and 2_S₂ are 11.96 kcal/mol and 5.06 kcal/mol higher than that of 2_R₁ , respectively. However, these geometries of 2_R₂ and 2_S₂ lead to favourable H‐bondings built between the C=O of NADPH and the proton of substrate N. These strong H‐bondings are retained in their corresponding transition state structures as well.

In TS2_R₂ , there is a significant steric overlap between the nicotinamide ring and the substrate's isoquinoline ring, but TS2_S₂ does not suffer from this steric interaction since these rings are almost gauche to each other and do not overlap (Figure 3). Both the relative stabilities of the enzyme‐substrate complexes and their corresponding barriers suggest that the transfer from 2_S₂ appears as a more favourable path, which is in line with the experimental stereoselectivity (Figure 1). ^[34] It is important to note the position of the substrate in the crystal structure at this stage. In the crystal structure of the enzyme (PDB ID: 6JIT), the substrate, which is a derivative of imine isoquinoline, is positioned away from the Tyr171 residue, in a manner similar to that observed in 2_S₂ .

As previously stated, while Tyr interacts with only NADPH at 2_R₂ , 2_S₁ and 2_S₂ structures, it interacts with both NADPH and the substrate at 2_R₁ (Figure 2 and Figure 4). This interaction network at 2_R₁ renders the structure more stable, however, it actually increases the barrier by widening the distance between the carbons where hydride transfer will occur and creates a kind of restrain on the lability of the substrate.

Figure 4.

Critical distances for 2_R₂ and TS2_R_2, 2_S₂ and TS2_S_2.

In the other enzyme‐substrate complexes, the substrate is more labile and prefers or forced to make H‐bondings on the other side of the NADPH. Consequently, the distances between the reacting C−C atoms are the greatest in 2_R₁ (4.71 Å), while the shortest distance is observed with 2_S₁ (3.57 Å). Apart from bringing the substrate and NADPH together, H‐bonding also affects the charge of the substrate's nitrogen, which in turn renders the accepting carbon more positive (Figure S3). This effect can be traced from the charge of the N_substrate, which ranges from−0.58 au to −0.65 au in TS2_S₁ , TS2_S₂ and TS2_R₂ while in TS2_R₁ , where such H‐bonding from NADPH to substrate is not present, it is −0.27 au. Accordingly, C_acceptor charges are in the range of 0.48 au to 0.51 in TS2_S₁ , TS2_S₂ and TS2_R₂ , whereas in TS2_R₁ the charge is 0.32 au. The phenate oxygen of Tyr is a highly negatively charged site (−0.72 au in 2_R₁ and in −0.86‐ −0.79 au range in others) for Tyr‐substrate and Tyr‐NADPH interactions but it should be noted that this interaction is shared by two parties in the R₁ case. Furthermore, the charge on the carbonyl oxygen of NADPH, which is an available site for strong H bonding with the proton of N_substrate, is also a highly negatively charged site (in −0.66–−0.71 au range in all prochiral 2 structures), comparable to the phenate oxygen. This interaction pattern contributes to the stabilities of the H‐bonded structures in R₂ , S₁ and S₂ systems.

The cluster models of the enzyme‐substrate complexes and their transition states demonstrate that the H‐bondings, steric effects and their interplay between the two are the most important parameters that dictate the stereoselectivity. If the tyrosine is negatively charged and the substrate is positioned such that its nitrogen atom is directed towards it in the enzyme‐substrate complex, the attack from the Si face will allow strong interactions between the Tyr‐NADPH and the substrate (as inTS2_R₁ ). If the attack is taking place from the Re face of the iminium ion, the substrate's iminium proton is unable to interact simultaneously with the Tyr in a reactive conformer for a facile transition state geometry (as in TS2_S₁ ). In this case, a favourable H‐bonding is formed between the NADPH C=O and the substrate. If the substrate's iminium N proton is located away from the Tyr, then, an effective H‐bonding interaction between NADPH and the substrate imposes the rings of NADPH and isoquionline to overlap, creating steric hindrance (TS2‐R₂ ). However, the transition state geometry for a Re face attack is free from these destabilizing interactions (TS2_S₂ ). These S‐type geometries discussed in the text being energetically attainable indicate a favourable Re face attack, producing the experimentally observed S‐product. As illustrated by the model employed here, NADPH plays a role in maintaining the substrate in a favourable position for reaction through the key H‐bonding interaction between its C=O and the substrate's iminium proton. In the context of hydride transfer, the role of the Tyr residue is limited to maintaining the NADPH in a favorable geometry. This is in line with the literature reports where Fademrecht et al. state that “the main function of IREDs would be in positioning the protonated imine substrate in an optimal orientation and distance to the cofactor NADPH”. ^[1] Nevertheless, it is essential to consider the intrinsic limitations of the current cluster model, particularly with regard to protein dynamics and environmental influences.

In an another study by Prejanò et al on a similar IRED enzyme (PDB ID: 5FWN which is within the 30 % sequence identity group and has a very similar amino acid pattern of the active site), the reaction barrier was reported as 20.1 kcal/mol for the S enantiomer formation. This is in line with the findings of our study, in which the barrier from the most stable pro‐S structure (2_S₂ ) was found to be at the same value and the R‐product formation was unfavorable (Figure 1). ^[30] This study has also demonstrated the importance of evaluating the reaction barriers as well as the enzyme substrate stability.

The first step, protonation of the imine to form the iminium ion was also considered as a part of the mechanism in this study. The protonation of the imine was assumed to take place in the catalytic site. ^[28] As stated earlier, tyrosine residue (171) in the vicinity of the 1‐Me DHIQ was considered as the source of the proton to protonate 1‐Me DHIQ.[ ¹ , ²⁶ , ²⁸ ] The proton transfer was investigated using the quantum cluster with or without the water assistance of water (Figure S4). Both the water‐assisted and the non‐assisted proton transfers require very low barriers of 6.92 and 8.67 kcal/mol, respectively. The closeness of these barriers suggests that the first step can be carried out both with or without water assistance. A comparison of the low barrier of protonation with the second step shows that the reduction is the slower step in the mechanism.

As mentioned earlier, the putative proton donor aspartic acid/tyrosine residues have been stated to play an important role in R/S selectivity of the enzyme. Some studies have indicated that the enzyme becomes inactive with the mutation of aspartic acid, while others have demonstrated a change in the conversion of the enzyme.[ ²³ , ²⁶ ] In a study on Q1EQE0 from Streptomyces kanamyceticus, the asymmetric reduction of 2‐methyl‐1‐pyrroline and the effect of mutation with this R‐selective enzyme was examined. It has been reported that the enzyme becomes inactive in the presence of D187 N and D187 A mutations of Asp187 in Q1EQE0. ^[26] Mutation studies on D172 A and D172 L mutants that were examined with 1‐Me DHIQ substrate showed that Asp172 had an effect on conversion [%] and ee [%] values in the catalytic mechanism in (R)‐IRED from Streptomyces kanamyceticus: conversion [%] decreased from 97 to 41 in both cases and the ee [%] changed from 47 to 81 and 68, respectively. ^[23] The mutation of tyrosine has also been demonstrated to be effective in these enzymes. ^[27] One such example is the AoIRED from Amycolatopsis orientalis where Y179 A, Y179F, N171 A, N171D and N241 A mutants have been shown to alter both the activity and the stereoselectivity. In addition to mutation, it was also observed that the stereoselectivity of the enzyme was dependent on time. ^[27] Based on the available data, the aim was to investigate whether the stereoselectivity in the hydride transfer step would be affected by a mutation in Tyrosine (Tyr171). The stereoselectivity behaviour of the enzyme was tested with calculations on the enzyme reaction in which Tyr was replaced with Asp. The mutation of Tyr171 to aspartic acid, which is considered to act as a proton donor, was examined in order to investigate the potential changes in the geometry and energies of the structures under consideration. The cluster employed in the calculations was modified as described above and the hydride transfers from the mutated prochiral 2_R₁ (M_2_R₁ , Figure 5) and 2_S₂ (M_2_S₂ , Figure 5) were modelled as representative structures. The hydride transfer barriers in the modified clusters were calculated as 29.24 kcal/mol from M_2_R₁ and 23.40 kcal/mol from M_2_S_{2 .} These values were obtained by the IRC calculations from the transition state structures to their corresponding enzyme‐substrate complexes. The two cluster structures, formed from the wild type and the mutated enzymes were aligned for the relative positions of the critical residues to gain insight into understanding the barriers and intermolecular forces in detail. The superposition of the prochiral iminium‐containing cluster structures with their corresponding mutated ones (M_2_R₁ with 2_R₁ and M_2_S₂ with 2_S₂ ) revealed no significant difference in the overall positions of the residues (Figure S5). The number and the effect of intermolecular forces and their changes upon mutation or upon hydride transfer reaction of wild‐type or mutant structures were also compared. The interactions between the Tyr‐substrate and Tyr‐NADPH, which constituted a restrained network in 2_R₁ , are absent in the mutated case, since Asp is unable to reach the substrate due to the long distance between them. This resulted in a more labile substrate, but conversely, a new H‐bonding interaction between the substrate and NADPH emerged in M_2_R₁ . This H‐bonding both created a stabilizing interaction in M_2_R₁ but it was not present in the transition state structure (M_TS2_R₁ ) rendering the hydride transfer still an unfavourable reaction at this geometry. The hydride donor‐ acceptor C atoms are even further apart from each other in M_2_R₁ (4.93 Å) as compared to 2_R₁ (4.71 Å). As a result of the interplay of these stabilizing (emerging H bonding) and destabilizing (more distant reacting centers) interactions, the barrier of hydride transfer decreases only slightly in the M_2_R₁ case. On the other hand, the NADPH‐substrate interactions in the 2_S₂ and M_2_S₂ structures are similar as well as the other surrounding interactions in the cluster. The critical H‐bonding between the NADPH and the substrate is conserved in the transition state (M_TS2_S₂ ), and the geometry of the transition state eliminates the steric interaction between the substrate isoquinoline and the NADPH's pyridine rings. The distance between the reacting C atoms in the mutated structure (4.76 Å) is 0.12 Å greater than in the wild type (4.64 Å) in the reactant, resulting with a 3.29 kcal/mol higher barrier for the hydride transfer.

Figure 5.

Comparison of interactions in M_2_R₁ , M_TS2_R₁ and M_2_S₂ and M_TS2_S₂ structures.

As there is no experimental mutation study, our evaluations are limited to computational studies. These studies propose that the selectivity may not be changed with the aforementioned mutation and the tendency to show S‐selective features continues.

Conclusions

In this study, we have modelled the reaction mechanism of the conversion of 1‐methyl‐3,4‐dihydroisoquinoline to a chiral amine, catalyzed by an imine reductase, namely SnIR.

The reaction mechanism involved protonation of the imine substrate to form the iminium ion and the following hydride transfer from NADPH to the substrate. A cluster model of the active site of the enzyme, which incorporates key residues in the vicinity of the substrate and the cofactor was employed in the calculations. The stereoselectivity in the hydride transfer was discussed by considering the enzyme‐subtrate complexes’ and the transition state geometries in addition to the corresponding reaction barriers. The calculations have supported a stepwise mechanism rather than the concerted protonation and hydride transfer steps. In the first step, the assistance of water did not result in a notable impact on the protonation barrier, suggesting that both pathways are facile. The calculations on the hydride transfer step, which is the step that determines the selectivity, have demonstrated a preference for the S‐product, in accordance with the literature results. ^[34] This conclusion is in line with the literature where Prejanò et al in their study on AoIRED proposed that all energies, including the transition state, should be considered in the evaluation of stereoselectivity rather than solely focusing on the energies of the enzyme‐substrate complex. ^[30] The calculations herein indicated that the stereoselective hydride transfer step is the slower step of the mechanism.

The calculations presented in this study demonstrate that the hydrogen bonding (especially between the C=O oxygen of the amide group of NADPH and the substrate, which created a stabilization in the transition state structures) and steric effects (in particular, the overlap of the substrate isoquinoline and the NADPH's pyridine rings as a result of the hydride attack at Si vs Re faces of the substrate) and their complex interplay between the two are the most significant parameters that determine the stereoselectivity.

Inspired by the relationship of the selectivity with respect to the proton donor residue in the literature, the proton donor Tyr171 was converted to Asp in the cluster model. Changing the proton donor residue to Asp did not predict a change in enantioselectivity in our calculations.

The selectivity patterns exhibited by IREDs can vary depending on the substrate and the enzyme in question.[ ²⁷ , ³⁴ ] To fully comprehend the selectivity of such processes, it is essential to gain an understanding of both the chemical reactions involved and the interactions occurring within the active site of the catalyst, where the reaction takes place with the substrate and/or cofactor. Our research on this substrate, which is of particular importance for the pharmaceutical industry, has created a valuable opportunity to study the phenomenon of stereoselectivity in such a large system with an atomistic approach. ^[34] This study, in conjunction with other examples in the literature,[ ²³ , ²⁶ , ²⁸ , ²⁹ ] will assist in the in‐depth understanding of these systems and the development of approaches that will enable the tuning of selectivity in the desired direction. It is our hope that an understanding of the structure‐activity relationship in asymmetric synthesis with IREDs will provide a significant contribution to the engineering of such processes in the long term.

Experimental Section

Computational Study

Modelling enzyme reactions with quantum mechanical methods is extremely difficult since the size of the system restricts the use of accurate methods at atomic scale. For this reason, it is generally preferable to work with a model or an effective part of the system that will encompass the interactions and the reaction center. In the quantum chemical cluster approach, which has been successfully employed in numerous enzyme reactions, residues situated within the active site and engaged in substrate interaction are incorporated into the system under investigation. This is followed by the generation of a small enzyme model through the excision of the relevant residues from appropriate points. Then, calculations, predominantly DFT, are performed to gain insights into the reaction mechanism.[ ³⁰ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ ]

Although the cluster approach has already demonstrated its efficacy as a successful tool for studying the mechanism of enzymatic reactions, it is essential to be careful in its application, particularly in light of its inherent limitations.[ ³⁵ , ³⁶ ] The size of the cluster built to mimic the active site of the enzyme must be sufficiently large to include the residues that interact with the substrate and form a good representative of the active site. The issue is not a significant challenge for recent applications of the cluster approach, as increased computing power is more readily available. A further limitation of the quantum cluster approach is that the model cannot mimic an enzyme reaction when a distant but crucial residue for the reaction is excluded. Furthermore, the quantum cluster method is unable to account for allosteric effects. The cluster approach imposes constraints on the flexibility of the reacting parties, which limits the scope of DFT calculations to accounting for the conformational dynamics of the protein during a transformation. If such data is required, it would be advisable to implement alternative methods, such as molecular dynamics. In this example, as in other quantum cluster studies, the chemical step of the enzyme reaction is considered in the calculations, which is assumed to start with the substrate that has already bound to the active site, as stated by Sheng and Himo et al.[ ³⁵ , ³⁶ ] Thus, the reaction mechanism and the barriers, rather than the absolute binding energies of substrate to enzymes are considered in order to model the enantioeselectivity.

Initially, a model study was conducted to investigate the mechanism of the hydride transfer reaction possibilities and to predict the critical geometrical parameters in this study. B3LYP/6‐31G(d,p) level of theory was utilized in these model calculations. Subsequently, quantum mechanical optimizations of the clusters were conducted at the B3LYP/6‐31G(d,p) level[ ⁴⁶ , ⁴⁷ ] This level of theory has been frequently employed in geometry optimizations in quantum chemical cluster approach calculations of the enzymes.[ ⁴¹ , ⁴³ ] In some studies in the literature, B3LYP−D3(BJ), which incorporates a dispersion correction, has been utilized.[ ³⁰ , ⁴² ] In this study, energies were further refined with single‐point energy calculations with the 6–311+G(2df,2p) basis set using the wB97XD functional, ^[48] which also includes dispersion correction. A Polarizable Continuum Model (PCM) was employed in single point energies with the dielectric constant of 4.0. ^[49] All quantum mechanical calculations were performed by using Gaussian 16 Revision A.03 software. ^[50] The energies presented throughout the manuscript are free energies, calculated by adding, the free energy corrections from the frequency calculations in the gas phase to the high‐level single‐point electronic energies. All the 3‐dimensional figures in the manuscript were drawn by Discovery Studio Visualizer v21.1.0.20298. ^[51]

The cluster that was utilized in the DFT calculations was formed using three principal elements, which were based on both empirical and computational evidence: 1. The crystal structure 2. Experimental data including kinetics and mutations of IREDs in general 3. Residue interaction patterns from docking. 6JIT (PDB ID) structure of the enzyme obtained from the Protein Data Bank included the cofactor and the substrate, namely 1‐(2‐phenylethyl)‐3,4‐dihydroisoquinoline. Firstly, the interactions between the cofactor and the substrate in the 6JIT structure were examined in order to select the residues that should be included in the cluster (Figure S6). Critical residues were examined from these interactions, and also residues that contribute to stereoselectivity were determined based on the findings of the literature. ^[27] Given that the crystal structure does not reflect the dynamic nature of enzyme‐substrate, docking was performed to assess whether any interacting residues other than those observed in the PDB code should be included in the cluster. The substrate in the X‐ray structure of the enzyme (6JIT) from Protein Data Bank was deleted and then 1‐Me DHIQ was docked back to the structure for our purpose (Figure S7) and the interactions were re‐examined and the critical residues were decided. The details of the docking are in the Supporting Information. After these processes, the residues to be included in the cluster were clarified.

Some of the residues used to construct the cluster for DFT calculations are included in their full form, while others are represented as backbones or side chains. The residues were generally cut from the Cα of their proceeding residue properly, ends capped with hydrogen atoms and the end atoms were frozen in their 3‐dimensional geometry in the crystal structure of the enzyme during geometry optimization. The details of the cutting scheme and the residues included in the calculations are submitted in the Table S2. The two‐dimensional shape of the cluster used in the mechanistic study is shown in Figure 6, and the atoms marked with asterisks are frozen during the geometry optimizations of the cluster.

Figure 6.

Two‐dimensional shape of the cluster used in the calculations.

The cluster formed was composed of 284 atoms, whose orientations with respect to each other were set according to the X‐ray structure of the enzyme (6JIT). After optimizing the cluster, its geometry was compared with the non‐optimized structure to determine if there were any significant changes in position particularly at the ends of the chain residues. (Figure S8). As shown in Figure S8, no significant difference was present in the positions of the critical residues when the two were superimposed.

In the mentioned cluster calculations, a validation process was carried out to understand whether the cluster was considered large enough, especially for the solution effects. For validation, single point energies of the corresponding structures (at the wB97XD/6‐311+G(2df,2p) level) were calculated for the hydride transfer barrier at different dielectric constants of 1.0, 4.0, 8.0, 16.0 and 80.0 (Table S3). The calculated barrier energies being close to each other have shown that the cluster would not be affected by the solution environment and that its size is good enough for a comparative study.

First, the cluster model was built as explained in detail. Then, conformer searches were performed for the S and R‐transition states (see Table S1 for details). In constructing the input geometries for transition states, the residues and NADPH were held stationary in their crystal structure and the substrate was positioned with respect to NADPH in a suitable geometry for a transition state structure for hydride transfer and the structure was optimized. Once a TS structure had been obtained, the N_NADPH‐C_donor‐C_acceptor‐ N_subs dihedral, dictating the position of the substrate with respect to NADPH was spanned for different geometries of all possible transition states, without ignoring the crystal structure geometry for the substrate. Intrinsic Reaction Coordinate (IRC) calculations were performed on the transition state structures to confirm the corresponding reactants and the products. Further optimizations of the IRC geometries have provided us with possible enzyme‐substrate/enzyme‐product complexes to consider as the reactants/products for the hydride transfer reactions. In evaluating the conformers, the original crystal structure and the excluded fragment of NADPH that occupied space in real enzyme structure were also taken into consideration. The conformers, that showed markedly divergent arrangements of their residue fragments, NADPH, and the substrate relative to the crystal structure, were not subjected to further evaluation and were excluded. All the structures optimized on the potential energy surface were verified to be minimum by all positive frequencies and the transition states were verified by single negative frequencies belonging to the reaction coordinate. In quantum clusters, frequently, small negative frequencies, which can safely be ignored are obtained as a result of restrained geometries, however, in these structures none was observed. ^[45]

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Kopar M., Senyurt Tuzun N., ChemPlusChem 2025, 90, e202400606. 10.1002/cplu.202400606

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.