今天推送的文章是2023年发表在Chinese Journal of Catalysis上的 “Precise regulation of the substrate selectivity of Baeyer-Villiger monooxygenase to minimize overoxidation of prazole sulfoxides”，通讯作者是许建和教授、郁惠蕾教授和王斌举教授。
Bayer-Veliger monooxygenase (BVMO) catalyzes the asymmetric oxidation of thioethers to valuable chiral sulfoxides, but the peroxidation of sulfoxides to generate unwanted sulfones limits the synthesis of BVMO. This peroxidation is due to insufficient substrate selectivity of BVMO, in which case the desired product, sulfoxide, can be further oxidized. In this study, a mathematical model was established to quantitatively determine the selectivity of the substrate based on the ratio of the specificity constant (kcat/km) between thioether and sulfoxide. A structure-oriented substrate tunneling approach was used to successfully minimize the sulfoxide peroxidation reaction and precisely adjust the substrate selectivity. Molecular dynamics simulations and quantum mechanical/molecular mechanics studies have shown that the altered H-bond network around flavin hydrogen peroxide (FADH-OOH) can modulate the mechanism and activity of sulfoxide oxidation. In addition, the redesigned AcPSMO mutant has been successfully applied to the controlled synthesis of other chiral sulfoxides.
A quantitative model was established to characterize the substrate selectivity of BVMO between thioether and sulfoxide
Previous studies have demonstrated the substrate selectivity of BVMOs. Since the content of sulfoxide is not only determined by the enzymatic oxidation efficiency of sulfoxide, but also by the changing concentration of sulfoxide during the reaction, the substrate selectivity of BVMOs for thioether oxidation is not well defined at a certain reaction time. In this paper, quantitative parameters were established to characterize the selectivity of AcPSMO for pthionine and sulfoxide by enantiomer ratio (E) in kinetic resolution
In the AcPMSO-catalyzed continuous oxidation reaction (Scheme 1), the concentrations of the first substrate (thioether, S1) and the second substrate (sulfoxide, also a product of the oxidation of the first substrate, P1) are interrelated variations. Firstly, a kinetic model of the continuous oxidation of thioether and sulfoxide was established. The oxidation reaction rate of thioether and sulfoxide was determined using steady-state kinetics (equation). S1-S3 (SI)。 The change in sulfoxide concentration can be calculated by subtracting the consumption of further oxidized sulfoxide from the sulfoxide produced by the oxidation of the thioether. The ratio of the oxidation reaction rate of sulfoxide to thionate (ν2/ν1) can be written as equation (1) and further simplified to define the kcat/km ratio of sulfoxide to thionate as R (Eq. (1) in SI.
The quantitative model (Equation (2)) is obtained by integrating Eq. (1), and the R-value can be calculated at any point in the reaction when the sulfide conversion (c) and product purity (pp) are known. Since these assumptions are not always satisfied during the reaction, the calculated R-value is referred to as the apparent R-value (Rapp), which is used to represent the substrate selectivity of AcPMSO during subsequent directed evolution.
Structure-oriented engineering of AcPSMO substrate selectivity
The crystal structure of AcPSMO complexed with FAD and NADP+ was determined by co-crystallization. In contrast to the known BVMO catalytic cycle, the identified structure was found to be the "close" form of the corresponding CHMO, which is consistent with the structure in the non-catalytic state. However, the complexes of AcPSMO-FAD-NADP+ to substrate were not obtained, either by immersion or co-crystallization experiments. Using the AcPSMO structure and the crystal structure of another CHMO as input templates, the complex was constructed by multi-template homology modeling, and the crystal structure of the CHMO was determined to be a catalytic open state. Since the location of amino acid residues adjacent to the substrate tunnel has been reported to transfer the substrate preference from cyclohexanone to the bulky substrate prazole sulfoxide, it is speculated that residues in the substrate tunnel may also affect recognition or selectivity between similar substrates omeprazole thionether and (S)-omeprazole.
The chemoselective quantification of substrates catalyzed by AcPSMO catalyzed sulfide oxidation relies on high-performance liquid chromatography and is not suitable for the screening of a large number of mutants. In order to improve the success rate of identifying and controlling chemoselective hot spots in AcPSMO, the amino acid valine with low hydrophobicity was selected to replace the 6 residues in the 4 Å range around the substrate tunnel in combination with the characteristics of large steric hindrance and strong hydrophobicity of omeprazole sulfide substrate as an optimization of the classical alanine scanning strategy (Fig. 1). The two residues, F277 and L432, located at the bottleneck of the substrate channel, have a clear effect on the substrate selectivity between sulfide and sulfoxide. Compared to AcPSMO (Rapp = 12.6), the substrate selectivity of the variant F277V was significantly improved (Rapp = 170) and the substrate selectivity of the variant L432V was greatly reduced (Rapp = 1.44), while the other variants in this small library had less difference in substrate selectivity. Therefore, saturated mutation libraries at sites 277 and 432 were constructed for further study. The results showed that all variants with hydrophobic residues were superior to wild-type (WT) except phenylalanine at site 277, with the variant F277L performing well in both reaction conversion (91.8%) and substrate selectivity (Rapp = 430). At the same time, all variants of L432X exhibit similar or worse substrate selectivity, and even different enantiomeric selectivity.
The F277 mutant reduces sulfoxide peroxidation
The enzymatic oxidation of omeprazole thioether was catalyzed by AcPSMOs (WT, L432V, and F277L) with different substrate selectivity, and the variable accumulation of product (S)-omeprazole and unwanted sulfone by-products during the reaction was compared (Figure 2). For AcPSMO and the negative variant L432V (with low substrate selectivity), the sulfone content in the biological sulfoxide reaction after 24 h was more than 65%. It is mainly determined by the biotransformation rate of (S)-omeprazole (Figure 2, black line). For the positive variant F277L, which has the highest substrate selectivity, the accumulation of by-products (sulfone) is much less, and the amount of sulfone is consistently less than 1% throughout the reaction. In this case, the maximum content of the accumulated (S)-omeprazole increases to 97% at 24 h, which greatly improves the yield of chiral sulfoxide and greatly simplifies the purification of (S)-omeprazole from reaction mixtures containing by-products with similar physicochemical properties.
Kinetic study of oxidation of omeprazole thioether and (S)-omeprazole by AcPSMO and its variants
The apparent kinetics of omeprazole thioether and (S)-omeprazole by AcPSMO and two canonical variants (F277L and L244V) were further investigated (Table 2). The catalytic specificity constant (kcat/km) with omeprazole sulfide as substrate varies slightly with different mutants. The mutation had no significant effect on the affinity of AcPSMO for omeprazole thioether, but enhanced the catalytic conversion rate (kcat), as both mutations converted large phenylamphetamine to smaller valine, which may lead to better adaptation to the entry of omeprazole thioether. Interestingly, for (S)-omeprazole oxidation, the catalytic capacity of F277 is almost eliminated, with a catalytic constant (kcat/Km) of only 6.7% compared to the AcPSMO wild-type enzyme. In contrast, the single mutant L432V exhibits significantly improved kcat and has a higher binding affinity for (S)-omeprazole. The results showed that the increase in Rapp value of F277L was mainly due to the reduction of the catalytic efficiency of (S)-omeprazole, especially the slowdown of the peroxidation rate of the desired product (S)-omeprazole.
ELUCIDATE SUBSTRATE-SPECIFIC MECHANISMS BY MOLECULAR DYNAMICS SIMULATIONS AND QM/MM CALCULATIONS
The mechanisms by which CHMO catalyzes the Baeyer-Villiger oxidation reaction include NADPH reduction of FAD at the active site, reduced E-FADH2 reacts with oxygen to form c4a-oxyflavin, flavin acts as a nucleophile to attack the electrophilic substrate ketone, product and NADP+ release. The NADP+ release process of the Baeyer-Villiger oxidation reaction was reported to be rate-limited, with a rate constant of ~2 s−1. When the substrate becomes unnatural methylphenylthionether and prazole sulfoxide, the total reaction rate is greatly reduced and the rate-limiting step becomes a different step of C4A-oxyflavin sulfide oxide.
To determine the structural basis of the intrinsic selectivity of AcPSMOs between two unnatural substrates (thioether and sulfoxide), the crystal structures of L432V and F277L were determined. Depending on the crystal structure, thioether or sulfoxide is docked into the AcPSMO WT, L432V, and F277L variants. Then, combined with molecular dynamics (MD) simulations and QM/MM calculations, the mechanism of anti-thiide and sulfoxide oxidation of the WT, L432V, and F277L AcPSMO variants was clarified. For thioether oxidation, calculations show that the reaction is carried out by a stepwise mechanism in which the distal OH of flavin C4a-OOH attacks thioether, which combines with the heterocleavage of the O-O bonds, resulting in a well-stabilized hydrogen bonding network around the S-OH+ intermediate and C-O-by. The calculated sulfoxide barriers are WT 14.9 kcal/mol, L432V 11.5 kcal/mol, and F277L 11.7 kcal/mol, respectively, indicating that thionylation is kinetically very favorable in all of these variants.
In MD simulations of L432V, F277L, and WT, the typical binding conformation of sulfoxide is similar (Figure 3), with the benzimidazole ring of the substrate forming an π-π stacking interaction with the isoxazine ring of the FAD. It is worth noting that both F277 and R327 can form π-cationic interactions in the structures of WT and L432V. In addition, in both WT and L432V, there is a connection water (W1) between flavin hydrogen peroxide (FADH-OOH) and ASP57. However, during the MD simulation of F277L, water can diffuse out of the active site, resulting in a direct hydrogen bonding interaction between the distal H of FADH-OOH and ASP57. This is mainly because the variant F277L can cause significant spatial repulsion between L277 and R327, which extrudes W1 from the active site. In addition to the 277 residue, the 432 residue also affects the hydrogen bonding interaction between the benzimidazole ring N1-H and the FADH-OOH moiety. Clearly, both L432V and F277L mutations can significantly alter the hydrogen bonding network around FADH-OOH, which may have a fundamental impact on the mechanism and reactivity of sulfoxide oxidation.
Based on the representative structures of L432V (RC1), F277L (RC2) and WT (RC3), the oxidation mechanism of sulfoxide was further studied by QM/MM calculation. For L432V (Figure 4(a)), the electrophilic attack of sulfoxide by the distal o atom of FADH-OOH is accompanied by heterocleavage of O-O (TS1a in Figure 4(a)). Subsequently, water-assisted protons (Hd) were transferred from FADH-OOH to Oa (ASP) and protons were transferred from N1-H to Op, resulting in protonated ASP57 and n1-deprotonated sulfone products (Int1a). The calculated barrier for this step is 16.2 kcal/mol (RC1a→TS1a). From Int1a, the transfer of protons (Hn) to N1 in the substrate is combined with the transfer of protons from protonated ASP57 to Op, resulting in the formation of sulfone products in PC1. This step is thermodynamically and kinetically very favorable, with only a small barrier of 5.6 kcal/mol (Int1a→TS1b).
As a comparison, the oxidation mechanism of sulfoxide in F277L was also studied (Fig. 4b). Compared to L432V, the oxidation of sulfoxide in F277L follows a FADH-OOH-mediated oxidation mechanism, in which the electrophilic attack of sulfoxide by the distal o atom of FADH-OOH is associated with O-O heterocleavage and the transfer of protons (Hd) from distal Od to proximal Op to generate a sulfoxide product (PC2) in one step. This reaction requires a relatively high barrier of 23.8 kcal/mol (Figure 4b). Clearly, changes in the hydrogen bonding network fundamentally affect the oxidation mechanism of sulfoxide. QM/MM calculations showed that the formation of sulfone in WT had a medium barrier of 20 kcal/mol, between the mutant F277L (23.8 kcal/mol) and the mutant L432V (16.2 kcal/mol). The barrier order predicted by QM/mm is in good agreement with the experimental results. In particular, the important role of hydrogen bonding networks in regulating the reactivity of sulfone formation is elucidated.
From the analysis of the above classical MD simulations and mixed QM/MM calculations, D57 is essential for proton transfer in FADH-OOH-mediated sulfoxide oxidation. To further confirm the catalytic effect of D57, D57 was replaced with a non-polar amino acid, asparagine, or glutamine. In the F277L/D57Q mutant, the oxidation of sulfoxide to sulfone was largely inhibited, while its activity against omeprazole thioether remained unchanged. In addition to the experimental results, the mutant F277L/D57Q was also simulated. STARTING FROM THE REPRESENTATIVE STRUCTURE, QM/MM CALCULATION WAS CARRIED OUT TO STUDY THE OXIDATION MECHANISM OF SULFOXIDE. The electrophilic attack of sulfoxide by the distal o atom of FADH-OOH is accompanied by heterocleavage cleavage of O-O. At the same time, the proton transfer of FADH-OOH to Ow (W1) is coupled with the proton transfer of N1-H to Op. Since water is a rather weak base compared to D57, the reaction is found to be very unfavorable in terms of kinetics. This finding is consistent with the experimental results, indicating that the oxidation of sulfoxide to sulfone will be largely inhibited in the F277L/D57Q mutant. Therefore, D57 plays a vital role in the oxidation of sulfoxide to sulfone.
AcPSMO is used for the redesign of asymmetric oxidation of other omeprazole sulfide ethers
In the process of catalytic oxidation of omeprazole thioether by F277L/D57Q, the corresponding sulfone content was further reduced. AcPSMO and its variants, F277L and F277L/D57Q, were also oxidized for three other similar substrates. The mutants F277L and F277L/D57Q significantly reduced peroxidation of all tested substrates, especially the by-product sulfone, while substrate selectivity (Rapp) was increased for all substrates (Table 3). These results indicated that the mutants designed based on the substrate-selective regulatory mechanism of AcPSMO were successfully applied to the controllable synthesis of other chiral prazole sulfoxides.