Significant improvement in catalytic activity and enantioselectivity of a Phaseolus vulgaris epoxide hydrolase, PvEH3, towards ortho-cresyl glycidyl ether based on the semi-rational design

The investigation of substrate spectrum towards five racemic (rac-) aryl glycidyl ethers (1a–5a) indicated that E. coli/pveh3, an E. coli BL21(DE3) transformant harboring a PvEH3-encoding gene pveh3, showed the highest EH activity and enantiomeric ratio (E) towards rac-3a. For efficiently catalyzing the kinetic resolution of rac-3a, the activity and E value of PvEH3 were further improved by site-directed mutagenesis of selected residues. Based on the semi-rational design of an NC-loop in PvEH3, four single-site variants of pveh3 were amplified by PCR, and intracellularly expressed in E. coli BL21(DE3), respectively. E. coli/pveh3E134K and /pveh3T137P had the enhanced EH activities of 15.3 ± 0.4 and 16.1 ± 0.5 U/g wet cell as well as E values of 21.7 ± 1.0 and 21.2 ± 1.1 towards rac-3a. Subsequently, E. coli/pveh3E134K/T137P harboring a double-site variant gene was also constructed, having the highest EH activity of 22.4 ± 0.6 U/g wet cell and E value of 24.1 ± 1.2. The specific activity of the purified PvEH3E134K/T137P (14.5 ± 0.5 U/mg protein) towards rac-3a and its catalytic efficiency (kcat/Km of 5.67 mM−1 s−1) for (S)-3a were 1.7- and 3.54-fold those (8.4 ± 0.3 U/mg and 1.60 mM−1 s−1) of PvEH3. The gram-scale kinetic resolution of rac-3a using whole wet cells of E. coli/pveh3E134K/T137P was performed at 20 °C for 7.0 h, producing (R)-3a with 99.4% ees and 38.5 ± 1.2% yield. Additionally, the mechanism of PvEH3E134K/T137P with remarkably improved E value was analyzed by molecular docking simulation.

and being cofactor-independent catalysts, can enantio-and/or regio-selectively catalyze the opening of an active three-membered oxirane ring of racemic (rac-) epoxides, retaining single epoxide enantiomers and/or producing chiral vicinal diols 8 . The hydrolysis of epoxides by EHs mainly proceeds in two steps. A nucleophilic side-chain oxygen atom of Asp residue, one of the catalytic triad (Asp-His-Asp/Glu), in EH regioselectively attacks on the αand β-carbon atoms (i.e., C α and C β ) in the oxirane ring of (R)-or (S)-epoxide, forming an EH−hydroxyalkyl intermediate. Then, a water molecule, activated by His residue located in EH's catalytic triad, interacts with the intermediate, releasing a vicinal diol product 9 . Based on the catalytic mechanisms of the given EH−epoxide pairs, the asymmetric ring-opening hydrolysis of rac-epoxides was classified into two major pathways: both the kinetic resolution and enantioconvergent hydrolysis 10 .
The kinetic resolution of rac-epoxides by highly enantioselective EHs provides an environment-friendly bioprocess for the preparation of (R)-or (S)-epoxides with an intrinsic limitation of 50% yield 11 . However, the majority of wild-type (WT) EHs displayed unsatisfactory catalytic activity and/or E value, making them unable to be effectively used in the kinetic resolution of rac-epoxides 12 . With the development of protein engineering, the laboratory-directed evolution or modification of enzymatic structures based on the rational or semi-rational design has been applied to improving the catalytic characteristics of existing EHs and/or conferring the desired properties upon them 13 . Reportedly, EH222 with remarkably enhanced enantioselectivity, one mutant of a WT EH from Aspergillus niger (abbreviated to AnEH), was obtained through four rounds of iterative saturation mutagenesis (ISM). The E value of EH222 towards rac-phenyl glycidyl ether (2a) was much higher than that of a WT AnEH 14 . For another example, an EH from Agrobacterium radiobacter (ArEH) was also subjected to ISM. Through high-throughput screening, the best mutant of ArEH, T247K/I108L/D131S, was selected. Its catalytic efficiency (k cat /K m ) and E value towards rac-epichlorohydrin were 4.5-and 2.1-fold higher than those of ArEH 12 . Theoretically, the structural modification of EHs based on the rational or semi-rational design contributes to a better understanding on the molecular catalytic mechanism of the kinetic resolution of rac-epoxides.
In our present work, the substrate spectrum investigation exhibited that E. coli/pveh3, an E. coli BL21(DE3) transformant intracellularly expressing PvEH3 mediated by pET-28a(+), had the highest EH activity (6.03 ± 0.24 U/g wet cell) and E value (13.3 ± 0.7) towards rac-3a among all tested rac-1a-5a (Table 1 and Fig. 1). Through reviewing the experimental and analytical results reported previously 15,16 , the NC-loop, linking the N-terminal catalytic region with the cap domain, of PvEH3 (GenBank accession no. ATG22745), was identified as the research object. To further improve the catalytic activity and E value of PvEH3 towards rac-3a, four single-and one double-site variants of pveh3 were amplified by two-stage whole-plasmid PCR 17 based on the semi-rational design, and intracellularly expressed in E. coli BL21(DE3), respectively. Then, the EH activities and E values of all E. coli transformants as well as the specific activities of the purified PvEH3, PvEH3 E134K , PvEH3 T137P and PvEH3 E134K/T137P towards rac-3a were determined, respectively. The kinetic parameters of PvEH3 E134K/ T137P , such as Michaelis constant (K m ) and turnover number (k cat ), for (S)-and (R)-3a were also determined, EH activity and protein assays. The activities of PvEH3 and its mutants towards rac-3a were determined as described previously 19 , with slight modification. 950 μL cell suspension of 4.2 mg wet cells/mL or purified EH solution of 6.3 μg protein/mL, diluted with 100 mM Na 2 HPO 4 −NaH 2 PO 4 buffer (pH 7.0), was well mixed with 50 μL 200 mM rac-3a dissolved in methanol (at a final concentration of 10 mM). The hydrolytic reaction was carried out at 20 °C for 10 min (for assaying the initial velocities of PvEH3 and its mutants by controlling the conversion ratio of rac-3a within 10%) and terminated by the addition of 4 mL methanol. All the reaction samples were analyzed, respectively, by high-performance liquid chromatography (HPLC), using a Waters e2695 apparatus (Waters, Milford, MA) equipped with an XBridge BEH C18 column. Herein, the column temperature was set at 30 °C. The mobile phase of methanol/H 2 O (7:3, v/v) was used at 0.8 mL/min, and monitored by a Waters 2489 UV−Vis detector at 220 nm. One activity unit (U) of EH was defined as the amount of whole wet cells of E. coli transformant or purified EH hydrolyzing 1 μmol rac-3a per minute under the given assay conditions. Analogously, the EH activities of E. coli/pveh3 were determined using 200 mg wet cells/mL towards 10 mM rac-1a, while 15 mg wet cells/mL towards 10 mM rac-2a, 4a and 5a, respectively. SDS-PAGE was carried out according to the method of Laemmli 20 on a 12.5% agarose gel, and the isolated proteins were visualized by staining with Coomassie Brilliant Blue R-250 (Sigma-Aldrich, St. Louis, MO). The apparent molecular weights of the expressed PvEH3 and its mutants were estimated by comparison with those of standard proteins using a Quantity One software (https://www.bio-rad.com/). The enzyme protein concentration was determined with the BCA-200 protein assay kit (Pierce, Rockford, IL).
Substrate spectrum assay of PvEH3. The hydrolytic reactions of rac-epoxides, in aliquots of 2 mL 100 mM phosphate buffer (pH 7.0) system consisting of 20 mM rac-1a-5a and 30 mg wet cells/mL of E. coli/pveh3, were carried out at 20 °C, respectively. During the hydrolytic process, aliquots of 100 μL reaction sample were drawn out periodically, and extracted with 900 μL ethyl acetate. The extracted samples were analyzed by chiral HPLC equipped with a Chiralcel OD-H column (Daicel, Osaka, Japan) under the same operating conditions as stated above, except for using isopropanol/n-hexane (2:8, v/v) as the mobile phase. The absolute configurations of the single enantiomers of rac-1a−2a and 4a−5a were confirmed, respectively, by comparing their retention times with those reported previously 10,18 . The conversion ratio (c value) of rac-substrate was defined as the percentage of its hydrolyzed amount to initial amount, while the yield of a single epoxide enantiomer, such as (R)-3a, was referred as the percentage of its retained amount to initial amount of rac-epoxide. The ee s of retained single aryl glycidyl ether was calculated based on the equation: ee s = [(R − S)/(R + S)] × 100%, where R and S represent the instantaneous concentrations of (R)-and (S)-1a-5a, respectively.
The enantioselectivity of EH towards a given rac-epoxide, being quantitatively described by its enantiomeric ratio (i.e., E value), was used to estimate the degree of enantiopreferential hydrolysis of one epoxide enantiomer over its antipode 21 . Based on the hydrolytic parameters of rac-epoxides (i.e., c and ee s values) as defined and calculated above, E values of E. coli/pveh3 towards rac-1a-5a (the substrate spectrum investigation), and those of the five E. coli transformants expressing PvEH3's mutants towards rac-3a (the screening of mutants) were calculated, respectively, using the equation: Semi-rational design of the NC-loop in PvEH3 for its site-directed mutagenesis. Using the known crystal structure of a Solanum tuberosum EH (StEH, PDB: 2CJP) at 1.95 Å resolution as a template, with which PvEH3 shares 59.0% primary structure similarity, the 3-D structures of PvEH3 and its best mutant, PvEH3 E134K/T137P , were homologically modeled using the MODELLER 9.21 program (https://salilab.org/modeller/) 23 , and then subjected to molecular mechanics optimization by CHARMM27 force field in the GROMACS 4.5 package (https://www.gromacs.org/) 24 . Meanwhile, the 3-D structures of substrates, (S)-and (R)-3a, were disposed in minimized energy using the ChemBioOffice 2010 package (https://www.cambridgesoft.com/) 25 .
The substrate spectrum investigation indicated that E. coli/pveh3 showed the highest EH activity and E value towards rac-3a (Table 1), and preferentially hydrolyzed (S)-3a over its antipode. To further improve the catalytic activity and E value towards rac-3a, i.e., the enantiopreference for (S)-3a, PvEH3 was subjected to site-directed mutagenesis of selected residues based on the semi-rational design. Firstly, the interaction between the modeled 3-D structures of PvEH3 and (S)-3a was predicted by molecular docking (MD) simulation using the AutoDock vina program (https://autodock.scripps.edu/) 26 to locate the most appropriate binding sites and steric orientation (having the lowest binding free energy). Then, the 3-D conformation of the docked enzyme−substrate complex, PvEH3−(S)-3a, was optimized using the GROMACS 4.5 package, and visualized using a PyMol software (http:// pymol.org/) 27 to identify the amino acid residues in PvEH3 in proximity to (S)-3a within 10 Å. Secondly, other four plant-derived EHs, VrEH1, PvEH1, PvEH2 and NbEH 3,28,29 , having superior catalytic activities and/or E values towards rac-3a and sharing over 65% identity with PvEH3, were searched by a BLAST server in the Swiss-Prot Protein database (http://www.ebi.ac.uk/swissprot/). Then, the multiple sequence alignment among five plant EHs was performed using the Clustal Omega program (https://www.ebi.ac.uk/Tools/msa/clustalo/) 30 . Finally, considering the sites of non-conserved residues among five NC-loops, the several specific residues in an NC-loop of PvEH3, in proximity to (S)-3a within 10 Å, were selected to be separately substituted with the corresponding and frequently emerging ones among other four EHs (identities equal to 75 and 100%).
Construction of E. coli transformants harboring variant genes of pveh3. The site-directed mutagenesis of selected residues in PvEH3's NC-loop was carried out by two-stage whole-plasmid PCR. The PCR primers for the single and double site-directed mutagenesis of pveh3 were designed according to pveh3 sequence and the codons coding for mutation residues, and synthesized by Sangon (Shanghai, China) (Supplementary Table 1). Using a recombinant plasmid, pET-28a-pveh3, as the template, the first round of PCR was conducted with an upstream primer (such as E134K-U or T137P-U) and a downstream primer (pET-28a-D) as following conditions: an initial denaturation at 95 °C for 4 min, 30 cycles of at 98 °C for 10 s, 55 °C for 15 s and 72 °C for 3.5 min, and an extra elongation at 72 °C for 10 min. Then, the second round of PCR was continued using the first-round PCR product (a double-strand DNA) as megaprimers: 30 cycles of at 98 °C for 10 s, 55 °C for 15 s and 72 °C for 3 min. The amplified target PCR products, such as pET-28a-pveh3 E134K or -pveh3 T137P , were digested with Dpn I endonuclease to decompose the methylated template, and then transformed into E. coli BL21(DE3), respectively. Thereafter, four resulting E. coli tranformants harboring single-site variants of pveh3, such as E. coli/pve-h3 E134K , were sent to Sangon company for DNA sequencing. Analogously, one recombinant plasmid connecting a double-site pveh3 variant, pET-28a-pveh3 E134K/T137P , was also amplified by PCR from pET-28a-pveh3 E134K using a pair of primers (T137P-U and pET-28a-D) and transformed into E. coli BL21(DE3), followed by DNA sequencing. One E. coli transformant containing a correct double-site variant gene was designated as E. coli/pveh3 E134K/ T137P .

Expression of PvEH3 and its mutants in E. coli BL21(DE3). A single colony of E. coli transformant,
such as E. coli/pveh3 E134K , was inoculated into LB medium (1.0% tryptone, 0.5% yeast extract and 1.0% NaCl, pH 7.2) supplemented with 100 μg/mL kanamycin sulfate, and cultured at 37 °C overnight as the seed culture. Then, the same fresh medium was inoculated with 2% (v/v) seed culture, and cultured until OD 600 value reached 0.6−0.8, followed by the addition of 0.05 mM IPTG to induce the expression of PvEH3 or its mutant at 20 °C for 8 h. The E. coli transformant cells were collected by centrifugation (8000 rpm) at 4 °C, and resuspended in 100 mM Na 2 HPO 4 −NaH 2 PO 4 buffer (pH 7.0) to 200 mg wet cells/mL used as the whole-cell biocatalyst unless stated otherwise. In this work, E. coli/pveh3 was used as the positive control, while E. coli BL21(DE3) transformed with pET-28a(+), designated E. coli/pET-28a, as the negative control.
Purification of the expressed PvEH3 and its mutants. The collected E. coli/pveh3, /pveh3 E134K , /pveh3 T137P or /pveh3 E134K/T137P cells, intracellularly expressing EH with a 6 × His tag at its N-terminus, were resuspended in buffer A (20 mM Tris-HCl, 500 mM NaCl and 20 mM imidazole, pH 7.5), and broken by ultrasonication in an ice-water bath. Then, the resulting supernatant was loaded onto a nickel-nitrilotriacetic acid (Ni-NTA) column (Tiandz, Beijing, China) preequilibrated with buffer A, followed by elution at a flow rate of 0.4 mL/min with buffer B as the same as buffer A except for 200 mM imidazole. Aliquots of 1.0 mL eluent only containing the target enzyme protein, PvEH3, PvEH3 E134K , PvEH3 T137P or PvEH3 E134K/T137P , analyzed by SDS-PAGE were pooled, dialyzed against 20 mM phosphate buffer (pH 7.0), and concentrated using a 10 kDa cut-off ultrafilter membrane (Millipore, Billerica, MA).
Kinetic parameter assays of the purified PvEH3 and PvEH3 E134K/T137P . The initial hydrolytic rates of (S)-and (R)-3a (μmol/min/mg protein) using 6.3 μg protein/mL purified PvEH3 (or PvEH3 E134K/T137P ) were determined, respectively, under the standard EH activity assay conditions, except for the concentrations of (S)and (R)-3a ranging from 1.5 to 15 mM. The hydrolytic rate versus (S)-or (R)-3a concentration was plotted to verify whether the hydrolytic mode of PvEH3 or PvEH3 E134K/T137P conformed to Michaelis−Menten equation. Both the K m and V max values of PvEH3 (or PvEH3 E134K/T137P ) for (S)-and (R)-3a were calculated by non-linear regression analysis using an Origin 9.0 software (http://www.orignlab.com/). The k cat value was deduced from the V max and apparent molecular weight of EH, while the catalytic efficiency (k cat /K m ) value was defined as the ratio of k cat to K m . All the experimental data from three independent replicates were expressed as the mean ± standard deviation (SD).
Effects of pH and temperature on the activity and stability of PvEH3 E134K/T137P . The pH optimum of the purified PvEH3 E134K/T137P towards rac-3a was examined under the standard EH activity assay conditions, except for using different buffers (Na 2 HPO 4 −citric acid: pH 5.0-7.0 and Tris−HCl: pH 7.5-9.0). For estimating the pH stability, aliquots of PvEH3 E134K/T137P solution were incubated at 20 °C for 1 h, in the absence of substrate, at a pH range of 5.0-9.0. The residual enzyme activity was determined under the standard EH activity assay conditions. Herein, the pH stability was defined as a pH range, over which the residual enzyme activity retained over 85% of its original activity.
The temperature optimum of PvEH3 E134K/T137P towards rac-3a was measured, at pH optimum, at temperatures ranging from 15 to 45 °C. To evaluate the thermostability, aliquots of PvEH3 E134K/T137P solution were incubated at the same temperature range for 1 h. The thermostability here was defined as a temperature, at or below which the residual enzyme activity was more than 85% of its original activity. Additionally, to ascertain a suitable reaction (2020) 10:1680 | https://doi.org/10.1038/s41598-020-58693-1 www.nature.com/scientificreports www.nature.com/scientificreports/ temperature for the kinetic resolution of rac-3a, the effect of temperature on the E value of PvEH3 E134K/T137P was also investigated at 10-40 °C.
Kinetic resolution of rac-3a using whole cells of E. coli/pveh3 E134K/T137P . The gram-scale kinetic resolution of rac-3a at elevated concentrations was carried out using the whole wet cells of E. coli/pveh3 E134K/T137P . The hydrolytic reactions, in aliquots of 20 mL 100 mM phosphate buffer (pH 7.0) system consisting of rac-3a at elevated concentrations ranging from 200 to 1000 mM and 200 mg wet cells/mL of E. coli/pveh3 E134K/T137P , were conducted at 20 °C. During the hydrolytic process, aliquots of 50 μL reaction sample were periodically drawn out, extracted with 950 μL ethyl acetate, and analyzed by chiral HPLC for calculating the c value of rac-3a as well as the ee s , yield and space-time yield (STY) of retained (R)-3a. In this work, the STY (g/L/h) was defined as the amount of retained (R)-3a in per unit volume and time, and calculated based on the equation: STY = (M × C 0 × Y)/t, in which C 0 and t represent the initial concentration and hydrolytic time of rac-3a, respectively, while M and Y are the molar molecular weight and yield of retained (R)-3a.

Molecular docking simulation of EH with (S)-or (R)-3a. The interaction between 3-D structures of
PvEH3 (or PvEH3 E134K/T137P ) and (S)-or (R)-3a was predicted by MD simulation using the AutoDock vina program. On the basis of the 3-D conformations of the simulatedly docked enzyme−substrate complexes, such as PvEH3− and PvEH3 E134K/T137P −(S)-3a, optimized using the GROMACS 4.5 package, the through-space distance (d β ) and hydrogen bond length (l 1 or l 2 ) were identified by a PyMol software. Herein, the d β was defined as the distance between a nucleophilic side-chain oxygen atom of Asp 101 residue in PvEH3 (or PvEH3 E134K/T137P ) and a C β (the less hindered terminal carbon atom in an oxirane ring) of (S)-or (R)-3a, while the l 1 or l 2 was the hydrogen bond length from a hydroxyl group of proton donor (Tyr 150 or Tyr 232 ) to an oxygen atom in the oxirane ring of epoxide. The binding free energy of PvEH3 (or PvEH3 E134K/T137P ) simulatedly docking with (S)-or (R)-3a, which is closely related to the affinity between them, was calculated using the molecular mechanics Poisson−Boltzmann surface area (MM-PBSA) method 22 .

Results and discussion
Catalytic activities and E values of PvEH3 towards rac-1a-5a. In our previous studies, pveh3 was cloned and expressed. To efficiently conduct the enantioconvergent hydrolysis, the catalytic activity of PvEH3 towards rac-para-chlorostyrene oxide (pCSO) and its regioselectivity for (S)-pCSO were improved by site-saturation and mutiple site-directed mutagenesis 31,32 . The regioselectivities, quantitatively described by selectivity coefficients, β R and α S , were adopted to evaluate the probabilities attacking on the C β of (R)-epoxide and C α of (S)-epoxide, respectively. It should be stressed that one specific EH is not likely to optimally function on all rac-epoxides and/or on two kinds of hydrolytic pathways towards a given rac-substrate. In this work, to expand the industrial application of PvEH3 in the kinetic resolution of rac-aryl glycidyl ethers, the hydrolytic processes of rac-1a-5a using whole wet cells of E. coli/pveh3 (Supplementary Figs. 1-5) and the substrate spectrum of PvEH3 towards rac-1a-5a were investigated. As shown in Table 1, E. coli/pveh3 enantiopreferentially catalyzed the hydrolysis of (R)-1a but (S)-2a-5a, and possessed the highest EH activity of 6.03 ± 0.24 U/g wet cell and E value of 13.3 ± 0.7 towards rac-3a, indicating that different aryl groups in aryl glycidyl ethers brought remarkable effects on the catalytic activity and E value of PvEH3. Unfortunately, the catalytic properties, especially the E value, of PvEH3 were still unsatisfactory even towards rac-3a. Therefore, it is highly desirable to further improve its activity and E value by site-directed mutagenesis of selected residues for efficiently producing (R)-3a via the kinetic resolution of rac-3a. To the best of our knowledge, several EHs have been applied to the kinetic resolution of rac-aryl/alkyl glycidyl ethers, but they exhibited unsatisfactory catalytic properties towards rac-3a. For examples, the EH activity of E. coli/vreh3, an E. coli transformant harboring a VrEH3-encoding gene vreh3, was only 4.8 U/g wet cell, while no activity of a Tsukamurella paurometabola EH (TpEH1) was detected towards rac-3a 22,33 . Selection of the specific residues in the NC-loop of PvEH3. The majority of characterized EHs, belonging to an α/β-hydrolase fold superfamily, were structurally divided into two main regions: an α/β domain, i.e., a β-sheet surrounded by a cluster of α-helices, and a cap domain. Both α/β and cap domains are connected by a variable NC-loop in amino acid composition and peptide chain length. Some experimental and analytical results have confirmed that the NC-loops of EHs are closely related to their catalytic characteristics 34,35 . For example, the laboratory-directed evolution study on the A. niger M200 EH demonstrated that Leu 215 and Ala 217 , which are located in the NC-loop, remarkably affected its catalytic activity and E value towards rac-SO 15 . For another example, S-B17C*10 (a multiple-site mutant of StEH), whose E value towards trans-2-methylstyrene oxide was 1.93-fold than that of S-B17 (an StEH mutant), was obtained through two rounds of ISM. The structural analysis indicated that the mutation site, Leu 145 , in the NC-loop might remarkably influence the enantioselectivity (i.e., E value) of StEH 16 . In view of the above reported results, the NC-loop of PvEH3 in this work was identified as the research object. Therefore, the 3-D structure of PvEH3 was homologically modeled using the MODELLER 9.21 program (Fig. 2a), from which the topology diagram of secondary structures was deduced (Fig. 2b). As shown in Supplementary Fig. 6, PvEH3 contains an NC-loop of 24 amino acid residues located at the sites from Ser 125 to Asp 148 , which harbors a short α-helix.
A total of 87 residues in PvEH3 in proximity to (S)-3a within 10 Å were identified using a PyMol software based on the 3-D comformation of a simulatedly docked complex PvEH3−(S)-3a. Among them, 11 residues were found to be located in the PvEH3's NC-loop (Fig. 3a). Furthermore, according to the result of multiple sequence alignment of the NC-loop of PvEH3 with those of four selected plant EHs (Fig. 3b), five absolutely conserved residues in the NC-loops (Val 126 , Pro 127 , Met 141 , Asp 147 and Asp 148 , sharing 100% identity with the corresponding residues in other four EHs) were eliminated from consideration. After further considering the remaining six non-conserved residues, two residues were also excluded from consideration because the two highest identity Scientific RepoRtS | (2020) 10:1680 | https://doi.org/10.1038/s41598-020-58693-1 www.nature.com/scientificreports www.nature.com/scientificreports/ residues, Val 138 and Tyr 145 , are the same as those of PvEH3. Consequently, four specific residues, Leu 128 , Leu 129 , Glu 134 and Thr 137 , in the NC-loop of PvEH3 were selected to be separately substituted with the corresponding and frequently emerging residues, Phe, Met, Lys and Pro, among other four plant EHs (identities equal to 75 and 100%).

Construction and screening of E. coli transformants harboring variants. Based on the above
semi-rational design, four recombinant expression plasmids harboring single-site variant genes, such as pET-28a-pveh3 L128F , were amplified by whole-plasmid PCR, and then transformed into E. coli BL21(DE3), respectively, resulting in the corresponding E. coli transformants, designated E. coli/pveh3 L128F , /pveh3 L129M , /pveh3 E134K and /pveh3 T137P . SDS-PAGE analysis showed that PvEH3 and its four mutants were successfully expressed, respectively, while no target band was detected in E. coli/pET-28a (Supplementary Fig. 7). Furthermore, the EH activities and E values of these transformants towards rac-3a were determined, respectively. Both E. coli/pveh3 E134K and /pveh3 T137P showed the improved EH activities of 15.3 ± 0.4 and 16.1 ± 0.5 U/g wet cell as well as E values of 21.7 ± 1.0 and 21.2 ± 1.1. Comparatively, both the EH activity and E value of E. coli/pveh3 were 6.0 ± 0.2 U/g wet cell and 13.3 ± 0.7, respectively, under the same expression conditions (Table 2).
Reportedly, there was a cooperative impact on the improvement in catalytic activity and E value of EHs via combinatorial or multiple-site mutagenesis 36,37 . Hence, to further improve the catalytic characteristics of PvEH3 for the kinetic resolution of rac-3a at the elevated concentration, it was subjected to combinatorial substitution of both E134K and T137P based on the results of single site-directed mutagenesis. The recombinant plasmid, pET-28a-pveh3 E134K/T137P , connecting a double-site variant pveh3 was also amplified by whole-plasmid PCR as designed theoretically, transformed into E. coli BL21(DE3), and expressed successfully. As expected, E. coli/pveh3 E134K/T137P had the highest EH activity of 22.4 ± 0.6 U/g wet cell and E value of 24.1 ± 1.2, which were 3.7-and 1.8-fold those of E. coli/pveh3 ( Table 2). The EH activity determination along with SDS-PAGE analysis indicated that, compared with that of E. coli/pveh3, the remarkably improved EH activities of E. coli/pveh3 T137P and /pveh3 E134K/ T137P were partially attributed to the expression level increases of target proteins in them. It was also reported that the expression level of Bacillus pumilus laccase in E. coli BL21(DE3) was increased through the site-directed mutagenesis of it 38 . www.nature.com/scientificreports www.nature.com/scientificreports/

Specific activities and kinetic parameters of the purified PvEH3 and its mutants. The expressed
PvEH3, PvEH3 E134K , PvEH3 T137P and PvEH3 E134K/T137P were purified to homogeneity, respectively, by one-step Ni-NTA affinity column chromatography, displaying single target protein bands on SDS-PAGE with the same apparent molecular weight of about 36.1 kDa (Supplementary Fig. 7), which was close to their respective theoretical ones (36,023, 36,022, 36,019 and 36,018 Da) predicted by ProtParam program (https://web.expasy.org/ protparam/). The specific activities of the purified PvEH3 E134K , PvEH3 T137P and PvEH3 E134K/T137P towards rac-3a under the standard EH activity assay conditions were determined to be 12.8 ± 0.4, 10.6 ± 0.3 and 14.5 ± 0.5 U/ mg protein, which were about 1.5-, 1.3-and 1.7-fold that (8.4 ± 0.3 U/mg protein) of a WT PvEH3.
The kinetic parameters of the purified PvEH3 (or PvEH3 E134K/T137P ) for (R)-and (S)-3a were measured (Table 3)    www.nature.com/scientificreports www.nature.com/scientificreports/ These changes in kinetic parameters suggested that the combinatorial substitution, E134K and T137P, in the NC-loop of PvEH3 had a significantly positive effect on its enantiopreference for (S)-3a.
pH and temperature characteristics of the purified PvEH3 E134K/T137P . PvEH3 E134K/T137P displayed higher catalytic activity towards rac-3a at a pH range of 6.5-7.5, over which the highest catalytic activity (i.e., the pH optimum) was at pH 7.0. When the pH value was at 5.0 or 9.0, the catalytic activity of PvEH3 E134K/T137P was only about 20 or 50% of that at pH 7.0. After incubation at 20 °C for 1 h at pH values of 5.0-9.0, PvEH3 E134K/T137P exhibited higher stability (>85%) at a pH range of 6.5-7.5 (Fig. 4a).
The temperature optimum of PvEH3 E134K/T137P towards rac-3a, at the pH optimum of 7.0, was determined to be 30 °C. After incubation at various temperatures (15-45 °C) for 1 h, PvEH3 E134K/T137P retained over 85% of its original activity at 35 °C or below (Fig. 4b), which was similar to that of A. mediolanus EH 39 . Additionally, the effect of reaction temperature for the kinetic resolution on the E value of PvEH3 E134K/T137P towards rac-3a was investigated at temperatures of 10-40 °C (Fig. 4b). The catalytic activity of PvEH3 E134K/T137P was increased as the temperature was elevated up to 30 °C, while its E value was decreased all through as the temperature rising. As a result, a suitable temperature of 20 °C was selected for the kinetic resolution, at which the catalytic activity of PvEH3 E134K/T137P was 85% of that at 30 °C while its E value was 24.1 ± 1.2. It is recognized that the lower the temperature, the higher the E value is. For examples, the E value of A. usamii EH towards rac-SO was elevated from 11.5 to 25.5 when the temperature was decreased from 35 to 0 °C, while the similar result was also observed in the case of the kinetic resolution of rac-SO by Rhodotrorula glutinis EH 40,41 . Preparative-scale production of (R)-3a by whole cells of E. coli/pveh3 E134K/T137P . To realize the gram-scale preparation of (R)-3a, the enantioselective hydrolytic reactions of rac-3a at concentrations of 200, 400, 600, 800 and 1000 mM, in aliquots of 20 mL 100 mM phosphate buffer (pH 7.0) system, were investigated, respectively, at 20 °C using 200 mg wet cells/mL of E. coli/pveh3 E134K/T137P . The reaction process was monitored by chiral HPLC at given intervals to confirm a suitable time for preventing the substrate from excessive hydrolysis. To our knowledge, using the whole cell instead of purified EH as the biocatalyst was because that the former was easily accessible and possessed higher stability 42 .
As shown in Table 4, rac-3a was efficiently and enantioselectively hydrolyzed until its concentration up to 800 mM, producing (R)-3a with ee s over 99.0% and yield from 38.5 ± 1.2 to 40.2 ± 1.5% (a theoretical yield of 50%). However, when rac-3a concentration was raised to 1000 mM, the ee s of (R)-3a was only 74.8 ± 2.1% until 14.0 h. The reason may be that there was a severe inhibition or denaturation effect on EHs when the substrate reached or exceeded a limit concentration 7,22 . Therefore, the maximum allowable concentration of rac-3a was 800 mM. The hydrolytic process of 800 mM rac-3a was shown in Fig. 5. After rac-3a was incubated at 20 °C for 7.0 h at 61.4 ± 1.6% c, (S)-3a was almost completely hydrolyzed, while (R)-3a was retained with 99.4% ee s and 38.5 ± 1.2% yield. The STY of (R)-3a was calculated to be 7.2 g/L/h, being 14.7-and 30.0-fold those by the whole cells of Trichosporon loubierii and A. niger expressing TlEH and AnEH 43,44 . As the hydrolytic reaction of rac-3a   pocket (SBP) of EHs, harboring a cluster of amino acid residues such as a catalytic triad and two proton donors, was located between the α/β and cap domains which are connected by an NC-loop 22 . It was reported that the spatial position and orientation of substrate-binding residues, influencing the catalytic activity and substrate preference, can be changed by specific residue substitutions in the NC-loop and even more distant regions 45, 46 . In the case of PvEH3 (or PvEH3 E134K/T137P ), the substrate-binding residues are identified as one catalytic nucleophile (Asp 101 ) and two proton donors (Tyr 150 and Tyr 232 ). The 3-D structure alignment between PvEH3 and PvEH3 E134K/ T137P revealed that there were obvious distinctions in the steric position and orientation of Asp 101 , Tyr 150 and Tyr 232 ( Supplementary Fig. 8), which may improved the affinity of PvEH3 E134K/T137P −(S)-3a.

Conclusions
Using an NC-loop in PvEH3 as the research object, its single and double site-directed mutagenesis was carried out based on the semi-rational design. The specific residue substitution, E134K or T137P in the NC-loop, had a positive effect on the catalytic activity and E value of PvEH3. Furthermore, E. coli/pveh3 E134K/T137P harboring a double-site variant of pveh3 possessed the highest EH activity and E value towards rac-3a among all the tested E. coli transformants, suggesting that a combinatorial substitution, E134K and T137P, had a synergistic impact. Compared with PvEH3, the purified PvEH3 E134K/T137P displayed the obviously improved specific activity towards rac-3a and the enhanced catalytic efficiency (k cat S /K m S ) for (S)-3a, The gram-scale kinetic resolution of 800 mM rac-3a was conducted using whole wet cells of E. coli/pveh3 E134K/T137P , producing (R)-3a with high ee s , yield and STY. Furthermore, the molecular mechanism of PvEH3 E134K/T137P with markedly improved enantiopreference for (S)-3a was explained by MD simulation. Our present work provided an efficient technical strategy to customize the desired EHs for preparing target chiral epoxides.