In the current work, to investigate amino acid substitutions that might enhance the catalytic efficiency of SMO, a rationaldesign approach was undertaken, which relied on molecular docking assisted by the AutoDock program and focused on those aminoacid residues that might interact with the phenyl ring of styrene.The SMO from Pseudomonas sp. LQ26 (designated as StyAB2) wasused as the parental enzyme since it has been well studied in ourlaboratory as a highly selective biocatalyst (Lin et al., 2010, 2011b,c;Qaed et al., 2011), and its high homology with other SMOs from thegenus of Pseudomonas would facilitate the docking study. This strategy led to several SMO mutants with increased enzymatic activitiestoward styrene, and one mutant displayed reversed enantioselectivity toward the substrate 1-phenylcyclohexene.
2. Materials and methods
2.1. Chemicals
The substrates styrene, trans- -methyl styrene, 2-vinylpyridineand 1-phenylcyclohexene were purchased from Alfa Aesar (Tianjin, China). Racemic styrene oxide (1S, 2S)-1-phenylpropyleneoxide and (1R, 2R)-1-phenylpropylene oxide were purchased fromSigma–Aldrich (St. Louis, MO, USA), and used as standard products.Other standard products including racemic 1-phenylcyclohexeneoxide and 2-(oxiran-2-yl)pyridine were synthesized from the corresponding alkenes according to the literatures (Fieser and Fieser,1967; Hanzlik et al., 1976). Other reagents were purchased from general suppliers and were used without further purification
2.2. Docking studies
The X-ray crystal structure of the oxygenase subunit of SMO(SMOA) from P. putida S12 was available from the PDB database (PDB ID: 3IHM). The amino acid sequence of this subunit shares
an 89% identity with the same subunit of the SMO from Pseudomonas sp. LQ26. To prepare the structure for docking, chain Bof the StyA homodimer and all water molecules of SMOA were removed, and charges and non-polar hydrogen atoms were added using MGLTools 1.5.4. AutoDock 4.0 was used for docking, and the docking parameters were kept to their default values in general (Morris et al., 1998) except that the grid spacing was changed to 0.275, the number of AutoDock 4 GA runs was increased from 10 to 50, and the docking grids were set as 22 × 32 × 22˚ A for styrene and 28 × 32 × 28˚ A for 1-phenylcyclohexene. The 50 independent runs from AD4 were analyzed in MGLTools 1.5.4 and the results were visualized using the program Pymol
2.3. Construction of point mutations
Site-directed mutagenesis was performed according to the QuikChange® site-directed mutagenesis protocol (Stratagene, La Jolla, CA) using the plasmid pETAB (Lin et al., 2010) encoding the wild-type StyAB2 (GenBank ID: GU593979) as the template. The sequences of mutagenic oligonucleotide primers (Table 1) were synthesized by Shanghai Invitrogen Life Technologies. The PCR product was treated with 20 U of Dpn I at 37 ◦ C for 2 h and transformed into DH5˛ competent cells. The successful introduction of the desired mutations was confirmed by sequencing at Shanghai Invitrogen Life Technologies.
2.4. Expression of the wild-type and mutant StyAB2 in E. coli BL21
E. coli strain BL21(DE3) containing the constructed plasmids was used to produce the wild-type and mutant StyAB2. Single colonies were picked up and grown overnight at 37 ◦ C in LuriaBertani broth containing 50 g kanamycin/ml. For each mutant and the wild-type, two single colonies were picked and cultivated to make two independent heterologous expressions. The overnight culture (2 ml) was inoculated into Terrific Broth (200 ml) containing 50 g kanamycin/ml in a 500 ml flask and incubated at 37 ◦ C for 3 h followed by 18 h incubation at 20 ◦ C with gyratory shaking at 220 rpm. The cells were harvested by centrifugation, washed twice with potassium phosphate buffer (0.1 M, pH 7) and stored at 4◦C.
To determine the expression levels of the wild-type and mutant enzymes, crude cell extracts were analyzed by SDS-PAGE, then the oxygenase subunit was purified using Profinity IMAC Ni-Charged resin (Bio-Rad, Hercules, CA, USA) as described previously (Lin et al.,2010). Then, the purified proteins were analyzed by SDS-PAGE. The protein content was determined with a commercial BCA Protein Assay kit using bovine serum albumin as a standard (Beyotime, Beijing, China)
2.5. Biotransformation with whole cells and product analysis
The harvested recombinant E. coli BL21 cells expressing the wild-type and mutants with a cell dry weight (CDW) of 0.1 g were resuspended in a biphasic system (Lin et al., 2010; Panke et al., 1999) of 10 ml potassium phosphate buffer (100 mM, pH 6.5) containing 10% (v/v) bis-(2-ethylhexyl) phthalate (BEHP) with the addition of 10 mg styrene or 1-phenylcyclohexene. For 2-vinyl pyridine, a monophasic system without BEHP resulted in better conversion and thus was applied instead of the biphasic system. The reaction was carried out at 30 ◦ C for 4 h with shaking at 230 rpm and terminated by extraction with ether. The organic phases were combined, dried with anhydrous sodium sulfate, concentrated under vacuum, and subjected to GC and chiral HPLC analysis. Specific epoxidation activities were measured using whole cells following the literatures (Bae et al., 2008; Park et al., 2006). Briefly, recombinant E. coli BL21 cells with 0.5 g CDW/L were resuspended in 4 ml potassium phosphate buffer (100 mM, pH 7.0) containing glucose (5 g/L), and incubated at 30 ◦ C for 10 min before the addition of 1.5 mM substrate (30 mM stock solution of styrene or trans- -methyl styrene in ethanol). The reaction was continued for 5 min. The mixture was extracted with ether containing 0.1 mM dodecane as an internal standard and analyzed with gas chromatography (GC). One unit (U) is defined as the activity that produces 1 mol of oxide per min.
GC analysis was performed on a Fuli 9790 II system connected to a flame ionization detector using column BP5 (30 m × 0.22 mm ID × 0.25 m film thickness, SGE Analytical Science, Australia) to determine the conversion of each substrate (styrene, trans- -methyl styrene, 2-vinylpyridine or 1-phenylcyclohexene) to the corresponding epoxide. Enantiomeric excesses were determined using chiral HPLC on a Shimadzu LC 20-AD (Shimadzu, Japan) with a PDA detector using Daicel Chiralpak AS-H column for styrene oxide (hexane:2-propanol = 90:10, 0.5 ml/min, tR(R) 10.27 min, tR(S) 10.67 min), Chiralcel AD-H column for 2-methyl-3-phenyloxirane (hexane:2-propanol = 90:10, 1 ml/min, tR(R) 4.14 min, tR(S) 4.78 min), or Chiralcel OD-H column for 1-phenylcyclohexene oxide (hexane:2-propane = 99:1, 0.5 ml/min, tR(R,R) 12.53 min, tR(S,S) 13.67 min) and 2-(oxiran-2-yl)pyridine (hexane:2-propane = 95:5, 0.5 ml/min, tR(R) 15.05 min, tR(S) 15.83 min)
3. Results
The SMOA structure from P. putida S12 has been released without substrate and FMN. Based on the putative substrate-binding center of SMOA as well as our previous work, which has shown the residues 43–46 being close to the -substitute of ˛-ethylstyrene, the native substrate styrene was docked into SMOA using AutoDock 4.0. The returned 50 results were analyzed in the MGL tools 1.5.4, and the highest scoring conformer with the vinyl group of styrene adjacent to the residues 43–46 was shown in Fig. 2
The amino acid residues Tyr73, His76 and Ser96, which form the bottom of the substrate-binding pocket, were found to be adjacent to the benzene ring of styrene with their side chains facing the substrate with distances of 7.15, 11.93 and 9.48˚ A, respectively (Fig. 2). It is well recognized that residues adjacent to the substrate play a critical role in catalytic activity, and thus commonly act as the target sites for rational design. In addition, unlike the residues 43–46, residues Tyr73, His76 and Ser96 are away from the putative FAD binding channel, which would avoid negative impacts on catalytic activity caused by reduced FAD binding (Feenstra et al., 2006; Ukaegbu et al., 2010). Therefore, residues Tyr73, His76 and Ser96 were modified using site-directed mutagenesis to investigate their effects on the enzymatic activity and enantioselectivity. Three amino acid substitutions were designed for each site in a way to reflect typical changes in size and hydrophobicity of the side chain of the residue. Tyr73 was replaced with Phe, Val and Ser. Compared with Tyr, Phe lacks the hydroxyl group while retaining the aromatic structure; Ser lacks the aromatic structure while retaining the hydroxyl group; and Val lacks both the hydroxyl group and the aromatic structure, but retains part of the steric hindrance. His76 was replaced with Asn, Val and Ala, all of which lack the electronically charged side chain, but contain a polar but uncharged residue (for Asn), or representative hydrophobic residues (for Val and Ala). Ser96 was replaced with Ala, Leu and Thr. Compared with Ser, Thr retains the hydroxyl group with an additional methyl group; and Leu and Ala lack the hydroxyl group and are representative hydrophobic residues with varied side chain sizes
All the constructed mutants listed in Fig. 3 were functionally expressed in E. coli at a level similar to the wild-type. Their activities were examined in the epoxidation of styrene in the biphasic reaction for 4 h (Lin et al., 2010; Panke et al., 1999). All active mutants retained excellent enantioselectivity, yielding the product (S)-styrene oxide with >99% ee. Mutants Y73F, Y73V and S96A exhibited higher activities than the wild-type, while other mutants showed lower activities or even major deleterious effect (Fig. 3). The best mutant S96A displayed a relative activity of 180% (Fig. 3)
Specific epoxidation activities were then measured for the most active mutants when the assays were carried out in an aqueous system for 5 min (Panke et al., 1998; Park et al., 2006). The results confirmed the increased enzymatic activity of mutants Y73F, Y73V and S96A for the epoxidation of styrene, as well as for trans- -methyl styrene (Fig. 4) without any negative impact on their enantioselectivities. The whole cell specific epoxidation activity of the wild-type StyAB2 was 66.5 U/g CDW, which was comparable to that of the other SMOs measured under similar conditions, such as that from Pseudomonas sp. VLB120 (79 ± 5 U/g CDW) and P. putida SN1 (55 ± 5 U/g CDW) (Panke et al., 1998; Park et al., 2006). The specific epoxidation activities of the most active mutant S96A toward styrene and trans- -methyl styrene were 2.6 and 2.3-fold of the wild-type, respectively (Fig. 4).
For substrates 2-vinyl pyridine and 1-phenylcyclohexene, the changes in activities were varied for the mutants Y73F, Y73V and S96A. Only S96A and Y73V displayed slight increases in the 2-vinyl pyridine and 1-phenylcyclohexene conversions, respectively (Table 2). Interestingly, the asymmetric epoxidation of 1-phenylcyclohexene catalyzed with the Y73V mutant displayed
reversed enantioselectivity compared to the wild-type, resulting in the (R,R)-enantiomer with 60% enantiomeric excess (Table 2), while the epoxidation of styrene, 2-vinylprydine and trans- -methyl styrene catalyzed with the same mutant yielded (S)-enantiomers with >99% ee, the same as the wild-type.
reversed enantioselectivity compared to the wild-type, resulting in the (R,R)-enantiomer with 60% enantiomeric excess (Table 2), while the epoxidation of styrene, 2-vinylprydine and trans- -methyl styrene catalyzed with the same mutant yielded (S)-enantiomers with >99% ee, the same as the wild-type.
4. Discussion
Crystal structure-based rational design of proteins focuses on a small number of variants and directly tests the substrates of interest to avoid the screening of a huge number of mutants using substrate analogues. This method has proven to be efficient for the improvement of a variety of enzymatic properties (Bornscheuer and Pohl, 2001; Schmidt et al., 2009; Voigt et al., 2001).
In this work, three potentially critical residues in the SMO from Pseudomonas sp. LQ26 were proposed according to structure-based molecular modeling, and three mutants, Y73F, Y73V and S96A, were identified to exhibit higher activity in styrene epoxidation compared to the wild-type enzyme. Amino acid residues at positions 73 and 96 are the same for all SMOs originating from the genus Pseudomonas, as well as the one from the metagenome (van Hellemond et al., 2007), i.e. Tyr and Ser, respectively. However, it is noteworthy that self-sufficient one-component SMOs have Ile and Ala at the corresponding positions, respectively, which includes the StyA2B from Rhodococcus opacus 1CP (Tischler et al., 2009) and two putative SMOs from Nocardia farcinica IFM10152 and Arthrobacter aurescens
TC1 (Fig. 5). Furthermore, for several putative SMOs, both the amino acid substitutions Y73V and S96A exist naturally (Fig. 5), indicating that these substitutions have already been explored by natural evolution, although whether they would indicate higher activity in native proteins is correctly unknown. Experimental studies on those putative enzymes might provide more information on the structure–functional relationship of SMOs in the future
TC1 (Fig. 5). Furthermore, for several putative SMOs, both the amino acid substitutions Y73V and S96A exist naturally (Fig. 5), indicating that these substitutions have already been explored by natural evolution, although whether they would indicate higher activity in native proteins is correctly unknown. Experimental studies on those putative enzymes might provide more information on the structure–functional relationship of SMOs in the future
Hydrophobic interaction appeared to be critical at position 73. Replacement of the Tyr residue with Phe or Val increased the activity, while loss of the hydrophobic side chain led to significantly impaired activity for the Y73S mutant. In fact, the putative substrate binding site of SMO is completely buried within the protein core, surrounded by more than ten hydrophobic residues and only four hydrophilic residues. The high hydrophobicity is regarded as being consistent with the hydrophobic nature of the substrate styrene (Ukaegbu et al., 2010). On the other hand, the size of the side chain of the residue at position 96 dramatically affects the enzymatic activity of SMO. The mutant S96T only added an additional methyl group on the side chain compared with the wild-type, but lost most of the enzymatic activity (Fig. 3). The substitution of Ser with Ala was well accepted and resulted in increased activity, while larger residue such as Leu led to a complete loss of activity (Fig. 3). Surprisingly, when the mutants S96T, S96A and S96L were modeled using the SWISS-MODEL version 8.05 (Kiefer et al., 2009) and applied in the automatic docking of styrene, no difference was found in terms of the binding energy, intermolecular energy, internal energy or torsional energy for all three resulting docking complexes compared with the wild-type. Based on the mechanism of this biocatalytic epoxidation reaction, the reactive cavity should accommodate not only styrene, but also the reduced FAD and oxygen, and the binding of substrate is indeed affected by the presence of FAD (Ukaegbu et al., 2010). The docking model where only styrene occupies the cavity could only provide limited information and may not reflect subtle changes in protein structure. It could be hypothesized that the replacement of Ser with larger residues such as Leu might affect FAD binding indirectly or weaken the interaction of reduced FAD with the substrate through the subtle movement of the substrate in the active pocket pushed by the steric hindrance of the side chain. In addition, residues Tyr73 and Ser96 are located at the same domain of the oxygenase subunit of the SMO on the 3- and 4-sheet, respectively, and are very close to each other with a distance of 3.41˚ A. Therefore, the amino acid substitution with large side chain at position 73 might cause distortion of the local structure within the limited space, and thus result in major deleterious effects on the catalytic reaction. On the contrary, smaller side chains at either position 73 or 96 may benefit the intermolecular interaction causing higher activity in the mutants Y73V and S96A. However, the double mutant combining the beneficial substitutions of Y73V and S96A did not show cumulative effect, and its enzymatic activity toward styrene remained similar to that of the mutant S96A (data not shown). The result indicates that the S96A mutation might have created enough space for the proper orientation of the -sheets. Therefore, further reduction of the size of the side chain at position 73 would have little beneficial effect.
In the majority of cases in this study, the mutations did not affect the enantioselectivity of the enzyme. Therefore, we assumed that the putative active cavity of SMO should be strictly shaped by surrounding residues and the mutations could hardly change the orientation of the substrate. The reversal of enantioselectivity for the mutant Y73V during the epoxidation of 1-phenylcyclohexene was unexpected because all other experimentally assigned SMOs display the same enantioselectivity (Bernasconi et al., 2000; Di Gennaro et al., 1999; Lin et al., 2010; Panke et al., 1998; Park et al., 2005; Tischler et al., 2009; van Hellemond et al., 2005), and the overall structure of the active cavity of SMOs should not be flexible enough to generate complementary enantiomers. Moreover, the mutants from directed evolution were also reported to produce (S)-enantiomers (Gursky et al., 2010).
Based on the fact that no reversal of stereoselectivity was observed for the Y73V mutant with the substrates styrene, 2-vinylprydine or trans- -methyl styrene, we hypothesized that this
reversal of stereoselectivity might be due to the structural flexibility of this particular substrate, 1-phenylcyclohexene, which contains a cyclohexenyl group that adopts a half-chair conformation with C2symmetry. The representative binding modes for 1-phenylcyclohexene into the putative active site of SMO resulting from automated docking are shown (Fig. 6A and B). Unlike styrene or styrene derivatives which displayed distinct preference to one single orientation in automated docking, and often produced enantiopure oxide with >99% ee (Lin et al., 2011b), the symmetric cyclohexenyl group in 1-phenylcyclohexene apparently resulted in much reduced stereoselectivity as the two highest scoring conformers in Fig. 6 were estimated by the program with very similar energy and docking scores. Since the putative substrate binding cavity is located at the bottom of the FAD binding site (Ukaegbu et al., 2010), oxygen and reduced FAD would come from the same direction for both conformers (Fig. 6A and B). Therefore, the (S,S) and (R,R)-enantiomers of the oxide product would be achieved from the two conformers shown in Fig. 6A and B, respectively.
reversal of stereoselectivity might be due to the structural flexibility of this particular substrate, 1-phenylcyclohexene, which contains a cyclohexenyl group that adopts a half-chair conformation with C2symmetry. The representative binding modes for 1-phenylcyclohexene into the putative active site of SMO resulting from automated docking are shown (Fig. 6A and B). Unlike styrene or styrene derivatives which displayed distinct preference to one single orientation in automated docking, and often produced enantiopure oxide with >99% ee (Lin et al., 2011b), the symmetric cyclohexenyl group in 1-phenylcyclohexene apparently resulted in much reduced stereoselectivity as the two highest scoring conformers in Fig. 6 were estimated by the program with very similar energy and docking scores. Since the putative substrate binding cavity is located at the bottom of the FAD binding site (Ukaegbu et al., 2010), oxygen and reduced FAD would come from the same direction for both conformers (Fig. 6A and B). Therefore, the (S,S) and (R,R)-enantiomers of the oxide product would be achieved from the two conformers shown in Fig. 6A and B, respectively.
The biotransformation results indicated that the conformer in Fig. 6A should be slightly preferred by the wild-type enzyme, leading to the formation of (1S, 2S)-1-phenylcyclohexene oxide with medium enantioselectivity (71%ee). The replacement of tyrosine with valine may have weakened the strong – interactions between the substrate and enzyme by the loss of the phenyl group from the side chain at position 73 and increased the flexibility of the active pocket. In addition, the size of the pocket may have expanded due to the reduced volume of the side chain, which may also facilitate the reversal of enantiomeric preference. However, the active site of the Y73V mutant should not have changed significantly as its selectivity toward styrene, 2-vinylprydine and trans- -methyl styrene remained the same as the wild-type. Therefore, the stereoswitch of the Y73V mutant toward 1-phenylcyclohexene was apparently triggered by subtle changes in the protein structure, which only took effect for this particular substrate with a symmetric cyclohexenyl group and having two conformers with similar energy in the automatic docking study.
In conclusion, three amino acid substitutions, Y73F, Y73V, and S96A were identified via a rational design approach to enhance the catalytic efficiency of SMO. These residues are located in the putative active pocket of the enzyme and could possibly interact with the phenyl ring of the native substrate, styrene. One of the mutants, Y73V, displayed reversed enantiomeric preference toward the substrate 1-phenylcyclohexene while retaining the same enantioselectivity toward other substrates. The results extended the knowledge of the structural–sequence relationship of SMOs and demonstrated that structure-based rational design was an efficient approach to altering the characteristics of this enzyme.
