Multi-Enzymatic Cascade for Efficient Deracemization of dl-Pantolactone into d-Pantolactone

d-pantolactone is an intermediate in the synthesis of d-pantothenic acid, which is known as vitamin B5. The commercial synthesis of d-pantolactone is carried out through the selective resolution of dl-pantolactone catalyzed by lactone hydrolase. In contrast to a kinetic resolution approach, the deracemization of dl-pantolactone is a simpler, greener, and more sustainable way to obtain d-pantolactone with high optical purity. Herein, an efficient three-enzyme cascade was developed for the deracemization of dl-pantolactone, using l-pantolactone dehydrogenase from Amycolatopsis methanolica (AmeLPLDH), conjugated polyketone reductase from Zygosaccharomyces parabailii (ZpaCPR), and glucose dehydrogenase from Bacillus subtilis (BsGDH). The AmeLPLDH was used to catalyze the dehydrogenated l-pantolactone into ketopantolactone; the ZpaCPR was used to further catalyze the ketopantolactone into d-pantolactone; and glucose dehydrogenase together with glucose fulfilled the function of coenzyme regeneration. All three enzymes were co-expressed in E. coli strain BL21(DE3), which served as the whole-cell biocatalyst. Under optimized conditions, 36 h deracemization of 1.25 M dl-pantolactone d-pantolactone led to an e.e.p value of 98.6%, corresponding to productivity of 107.7 g/(l·d).


Introduction
As a chiral alcohol, D-pantolactone serves as an important intermediate in the synthesis of D-pantothenic acid (vitamin B 5 ), which is widely used as a pharmaceutical and cosmetic ingredient as well as a food and feed additive [1,2]. In commercial production, Dpantothenic acid is synthesized through the condensation reaction between D-pantolactone and β-alanine, and D-pantolactone is mainly prepared through the lactonase-catalyzed resolution of racemic DL-pantolactone [3][4][5][6]. In the process of enzymatic resolution, the generated enantiopure D-pantothenic acid is separated from the remaining L-pantolactone and further converted into D-pantolactone by a chemical lactonization step, requiring the consumption of energy, solvents, and acids/alkalis [7]. In efforts to develop a simpler and greener process, the oxidoreductase-based asymmetric synthesis of D-pantolactone has received increasing attention in recent years [7][8][9][10][11][12][13]. These efforts have mainly focused on the enzymatic reduction of ketopantolactone to D-pantolactone with the assistance of a glucose dehydrogenase/glucose coenzyme regeneration system. For instance, an E. coli strain co-expressing conjugated polyketone reductase from Saccharomyces cerevisiae (SceCPR1) and glucose dehydrogenase from Exiguobacterium sibiricum (EsGDH) was applied for the asymmetric reduction of ketopantolactone, providing 458 mM D-pantolactone with an e.e. p value of 99.9% and a yield of 91.6% after 6 h [8]. Ketopantolactone is prone to spontaneous Scheme 1. One-pot deracemization of DL-pantolactone to its D-enantiomer through a biocatalysis cascade using E. coli cells co-expressing L-pantolactone dehydrogenase (LPLDH), conjugated polyketone reductase (CPR), and glucose dehydrogenase (GDH).

Three-Enzyme Cascade for Deracemization of DL-Pantolactone
A whole-cell biocatalyst co-expressing SceCPR1 and EsGDH in E. coli was previously constructed to catalyze the asymmetric reduction of ketopantolactone to D-pantolactone. To upgrade the asymmetric reduction to the deracemization process, it is necessary to use Scheme 1. One-pot deracemization of DL-pantolactone to its D-enantiomer through a biocatalysis cascade using E. coli cells co-expressing L-pantolactone dehydrogenase (LPLDH), conjugated polyketone reductase (CPR), and glucose dehydrogenase (GDH).

Three-Enzyme Cascade for Deracemization of DL-Pantolactone
A whole-cell biocatalyst co-expressing SceCPR1 and EsGDH in E. coli was previously constructed to catalyze the asymmetric reduction of ketopantolactone to D-pantolactone. To upgrade the asymmetric reduction to the deracemization process, it is necessary to use Lpantolactone dehydrogenase with strict L-enantioselectivity to initiate the dehydrogenation of L-pantolactone and keep the D-pantolactone unreacted. L-pantolactone dehydrogenase from Nocardia asteroides (NasLPLDH) was purified from the wild-type strain and was able to catalyze the dehydrogenation of L-pantolactone into ketopantolactone [30]. The amino acid sequence encoding NasLPLDH was chosen to perform a homology search in the NCBI database. Four putative L-pantolactone dehydrogenases were selected from Nocardia cyriacigeorgica GUH-2 (91% sequence identity), Nocardia farcinica IFM 10152 (88% sequence identity), Amycolatopsis methanolica 239 (76% sequence identity), and Cnuibacter physcomitrellae strain XA(T) (71% sequence identity) and named NcyLPLDH, NfaLPLDH, AmeLPLDH, and CphLPLDH, respectively. All of the genes were synthesized onto the pET28a plasmid and overexpressed in E. coli BL21(DE3) ( Figure S1). When the substrate specificity was investigated, all five LPLDHs showed the capacity to catalyze the dehydrogenation of L-pantolactone but not D-pantolactone, indicating their strict L-enantioselectivity.

Purification and Characterization of Key Enzymes AmeLPLDH and ZpaCPR
As newly mined enzymes, AmeLPLDH and ZpaCPR were purified and then characterized to gain insights into their enzymatic and catalytic properties. The purification was conducted by affinity chromatography ( Figure S5). Various factors, including temperature, pH, metal ions, and organic solvents, were investigated. The influence of temperature was determined over a range of 20-50 °C, and the highest activity for AmeLPLDH and ZpaCPR was observed at 30 and 45 °C, respectively (Figure 3a). To determine the effect of pH on activity, an enzyme assay was carried out at pH ranging from 5.0 to 10.0. The optimal pH for ZpaCPR was 5.5, and an increase in pH from 5.5 to 10.0 decreased the activity. In contrast, AmeLPLDH adapted to a much wider pH range (Figure 3b). Various metal ions were investigated for their effect on enzyme activity, and only Ca 2+ slightly enhanced the enzyme activity of AmeLPLDH. None of the tested metal ions had a beneficial effect on the enzyme activity of ZpaCPR (Figure 4a). The use of organic solvents may have a profound effect on the activity of AmeLPLDH and ZpaCPR; thus, various readily available and cost-effective organic solvents (chloroform, ethyl acetate, isopropanol, acetone, octanol, and DMSO) were tested in this regard. Except for acetone, the tested organic solvents inhibited enzyme activity; octanol was the most harmful ( Figure 4b). The kinetic parameters were determined, and the kcat/Km values of AmeLPLDH and ZpaCPR were 63.85 and 11.60 mM −1 s −1 , respectively (Table 1). Overall, the difference in enzymatic prop-

Purification and Characterization of Key Enzymes AmeLPLDH and ZpaCPR
As newly mined enzymes, AmeLPLDH and ZpaCPR were purified and then characterized to gain insights into their enzymatic and catalytic properties. The purification was conducted by affinity chromatography ( Figure S5). Various factors, including temperature, pH, metal ions, and organic solvents, were investigated. The influence of temperature was determined over a range of 20-50 • C, and the highest activity for AmeLPLDH and ZpaCPR was observed at 30 and 45 • C, respectively ( Figure 3a). To determine the effect of pH on activity, an enzyme assay was carried out at pH ranging from 5.0 to 10.0. The optimal pH for ZpaCPR was 5.5, and an increase in pH from 5.5 to 10.0 decreased the activity. In contrast, AmeLPLDH adapted to a much wider pH range (Figure 3b). Various metal ions were investigated for their effect on enzyme activity, and only Ca 2+ slightly enhanced the enzyme activity of AmeLPLDH. None of the tested metal ions had a beneficial effect on the enzyme activity of ZpaCPR ( Figure 4a). The use of organic solvents may have a profound effect on the activity of AmeLPLDH and ZpaCPR; thus, various readily available and cost-effective organic solvents (chloroform, ethyl acetate, isopropanol, acetone, octanol, and DMSO) were tested in this regard. Except for acetone, the tested organic solvents inhibited enzyme activity; octanol was the most harmful ( Figure 4b). The kinetic parameters were determined, and the k cat /K m values of AmeLPLDH and ZpaCPR were 63.85 and 11.60 mM −1 s −1 , respectively (Table 1). Overall, the difference in enzymatic properties between AmeLPLDH and ZpaCPR indicated that it was necessary to subsequently orchestrate their catalytic performance with BsGDH.

Deracemization of DL-Pantolactone at High Concentrations
The enzymes AmeLPLDH, ZpaCPR, and BsGDH have different catalytic properties, and the deracemization process must be optimized to achieve adequate productivity and product concentrations [28]. AmeLPLDH, ZpaCPR, and BsGDH require FMN or NADPH as a coenzyme. However, neither FMN nor NADPH was exogenously supplied to the reaction mixture, so that the developed deracemization process would be cost-effective.
To achieve synergistic effects, factors including temperature, pH, agitation, glucose concentration, and substrate loading were investigated. The substrate concentration was set as 1000 mM and the pH was kept constant through titration with 1 M NaOH. The optimal temperature was 30 • C, and the e.e. p value remained >30% between 20 and 40 • C ( Figure S7a). The effect of pH on catalytic efficiency was determined within the range of 5.0-10.0 ( Figure S7b). The optimal pH was 6.0, and the e.e. p value decreased as the pH rose to 10.0, indicating that the enzymatic stability was more adapted to a weak acidic environment. Agitation was optimized to 400 rpm, although the e.e. p value remained >80% between 300 and 600 rpm ( Figure S7c). As a co-substrate for coenzyme regeneration, glucose in various concentrations was used to test the effect on catalytic efficiency. When no glucose was added, the e.e. p value was 24.2%, suggesting the presence of an intracellular proton supply. The e.e. p value rose when the glucose concentration increased from 0 to 2000 mM, and concentrations higher than 2000 mM caused no further improvement. Thus, the ratio of glucose to substrate was set as 2:1. Under optimized conditions, substrate loading was increased stepwise from 500 to 1750 mM in order to investigate the effect on catalytic efficiency ( Figure 6). When the substrate concentration was not greater than 1000 mM, the e.e. p value remained at >98%. When the substrate concentration was 1250 mM, the yield of both L-pantolactone and ketopantolactone began to increase, resulting in a lower e.e. p value (81.4%). When the substrate concentration was 1750 mM, the yield of L-pantolactone and ketopantolactone was 14.4 and 16.3%, respectively, indicating that the catalytic efficiency of the enzymes was insufficient after 24 h reaction. To identify the rate-limiting enzyme, E. coli cells expressing AmeLPLDH, ZpaCPR, or BsGDH were added to the reaction mixture (Figure 7). The addition of AmeLPLDH or ZpaCPR resulted in similar catalytic performance to the control, whereas supplementation with BsGDH dramatically increased the e.e. p value to 98.6%. A high proportion of D-pantolactone simplified the subsequent product separation procedure, and crude product was obtained through solvent extraction and solvent evaporation. Finally, the crude product was verified to be D-pantolactone through a combination of chiral GC, GC-MS, 1 H NMR, and 13 C NMR analysis.

Deracemization of DL-Pantolactone at High Concentrations
The enzymes AmeLPLDH, ZpaCPR, and BsGDH have different catalytic properties, and the deracemization process must be optimized to achieve adequate productivity and product concentrations [28]. AmeLPLDH, ZpaCPR, and BsGDH require FMN or NADPH as a coenzyme. However, neither FMN nor NADPH was exogenously supplied to the reaction mixture, so that the developed deracemization process would be cost-effective.
To achieve synergistic effects, factors including temperature, pH, agitation, glucose concentration, and substrate loading were investigated. The substrate concentration was set as 1000 mM and the pH was kept constant through titration with 1 M NaOH. The optimal temperature was 30 °C, and the e.e.p value remained >30% between 20 and 40 °C ( Figure S7a). The effect of pH on catalytic efficiency was determined within the range of 5.0-10.0 ( Figure S7b). The optimal pH was 6.0, and the e.e.p value decreased as the pH rose to 10.0, indicating that the enzymatic stability was more adapted to a weak acidic environment. Agitation was optimized to 400 rpm, although the e.e.p value remained >80% between 300 and 600 rpm ( Figure S7c). As a co-substrate for coenzyme regeneration, glucose in various concentrations was used to test the effect on catalytic efficiency. When no glucose was added, the e.e.p value was 24.2%, suggesting the presence of an intracellular proton supply. The e.e.p value rose when the glucose concentration increased from 0 to 2000 mM, and concentrations higher than 2000 mM caused no further improvement. Thus, the ratio of glucose to substrate was set as 2:1. Under optimized conditions, substrate loading was increased stepwise from 500 to 1750 mM in order to investigate the effect on catalytic efficiency ( Figure 6). When the substrate concentration was not greater than 1000 mM, the e.e.p value remained at >98%. When the substrate concentration was 1250 mM, the yield of both L-pantolactone and ketopantolactone began to increase, resulting in a lower e.e.p value (81.4%). When the substrate concentration was 1750 mM, the yield of Lpantolactone and ketopantolactone was 14.4 and 16.3%, respectively, indicating that the catalytic efficiency of the enzymes was insufficient after 24 h reaction. To identify the ratelimiting enzyme, E. coli cells expressing AmeLPLDH, ZpaCPR, or BsGDH were added to the reaction mixture (Figure 7). The addition of AmeLPLDH or ZpaCPR resulted in similar catalytic performance to the control, whereas supplementation with BsGDH dramatically increased the e.e.p value to 98.6%. A high proportion of D-pantolactone simplified the subsequent product separation procedure, and crude product was obtained through solvent extraction and solvent evaporation. Finally, the crude product was verified to be Dpantolactone through a combination of chiral GC, GC-MS, 1 H NMR, and 13 C NMR analysis.

Chemicals, Genes, Plasmids, and Organisms
DL-pantolactone, ketopantolactone, and other chemicals and reagents were purchased from Shanghai Macklin Biochemical Co., Ltd. (Shanghai, China). Kits and enzymes for genetic manipulation were purchased from Takara Biomedical Technology Co., Ltd. (Beijing, China). Chromatography columns for protein purification were purchased from GE Healthcare Life Sciences (Shanghai, China). The genes encoding L-pantolactone dehydrogenase, ketopantolactone reductase and glucose dehydrogenase were codon-optimized and then synthesized by Qingke Biotechnology Co., Ltd. (Hangzhou, China). The pET28a and pACYCDuet-1 vectors were used for co-expression of L-pantolactone dehydrogenase, ketopantolactone reductase, and glucose dehydrogenase, and E. coli strain BL21(DE3) was used as the host. E. coli cells were grown routinely in LB medium at 37 °C for 12 h.

Chemicals, Genes, Plasmids, and Organisms
DL-pantolactone, ketopantolactone, and other chemicals and reagents were purchased from Shanghai Macklin Biochemical Co., Ltd. (Shanghai, China). Kits and enzymes for genetic manipulation were purchased from Takara Biomedical Technology Co., Ltd. (Beijing, China). Chromatography columns for protein purification were purchased from GE Healthcare Life Sciences (Shanghai, China). The genes encoding L-pantolactone dehydrogenase, ketopantolactone reductase and glucose dehydrogenase were codon-optimized and then synthesized by Qingke Biotechnology Co., Ltd. (Hangzhou, China). The pET28a and pACYCDuet-1 vectors were used for co-expression of L-pantolactone dehydrogenase, ketopantolactone reductase, and glucose dehydrogenase, and E. coli strain BL21(DE3) was used as the host. E. coli cells were grown routinely in LB medium at 37 • C for 12 h.

Purification of AmeLPLDH and ZpaCPR
Cells expressing AmeLPLDH and ZpaCPR were resuspended in 50 mM Tris-HCl buffer (pH 8.0) at a concentration of 50 g/L, then sonicated for cell disruption. Brij 35 was added to the AmeLPLDH cell homogenate to give a final concentration of 1%. After gentle stirring for 2.5 h, the cell debris was removed by centrifugation. The cell-free extracts of AmeLPLDH and ZpaCPR were then analyzed by Ni-NTA chelating affinity chromatography. Unbound proteins were washed off by applying a binding buffer (5 mM imidazole and 300 mM NaCl dissolved in 50 mM Tris-HCl, pH 8.0). AmeLPLDH was eluted out by applying an elution buffer (300 mM imidazole and 300 mM NaCl dissolved in 50 mM Tris-HCl, pH 8.0), and ZpaCPR was eluted out by applying an elution buffer (200 mM imidazole and 300 mM NaCl dissolved in 50 mM Tris-HCl, pH 8.0). The purity of the purified AmeLPLDH and ZpaCPR was verified using SDS-PAGE. The purified AmeLPLDH and ZpaCPR enzymes were desalted with 50 mM Tris-HCl buffer (pH 8.0) and stored at −20 • C for further investigation.

Activity Assay of AmeLPLDH and ZpaCPR
The activity of AmeLPLDH was measured at 30 • C by monitoring changes in absorbance at 450 nm. The 1 mL assay mixture of AmeLPLDH consisted of 1 µg purified enzyme, 10 mM L-pantolactone, 0.15 mM FMN, and 200 mM PBS buffer (pH 8.0). The enzyme assays began with the addition of the coenzyme FMN. One unit of activity represents the reduction of 1 µmol FMN per minute. The pure enzyme activity of ZpaCPR was measured at 45 • C by monitoring changes in absorbance at 340 nm. The 1 mL assay mixture of ZpaCPR consisted of 1 µg purified enzyme, 10 mM ketopantolactone, 0.15 mM NADPH, and 200 mM PBS buffer (pH 5.5). The enzyme assay began with the addition of the coenzyme NADPH. One unit of activity represented the oxidation of 1 µmol NADPH per minute. The BCA method was used to determine the protein concentration of all samples, with bovine serum albumin used as the standard protein. All enzyme assays were performed in triplicate [34].
In the characterization of AmeLPLDH and ZpaCPR, the effects of temperature and pH on the activity were investigated at 20-50 • C and pH 5.0-10.0. The metal ions (Cu 2+ , Ni 2+ , Li + , Co 2+ , Na + , Ag + , Al 3+ , Ca 2+ , Mn 2+ , Zn 2+ ) and organic solvents (chloroform, ethyl acetate, isopropanol, acetone, octanol, DMSO) were individually added to the reaction mixture to determine their effects on the activity. In the determination of kinetic parameters, the tested substrate concentrations included 1, 2,4,6,8,10,20,30, and 40 mM. According to Michaelis-Menten kinetics, the parameters K m and V max were calculated through curve fitting using Origin Pro software (version 8.5).

Co-Expression of AmeLPLDH and Fusion Enzyme ZpaCPR-(GSG)-BsGDH
The fusion genes encoding ZpaCPR and BsGDH were constructed by multiple overlap extension PCR. To assemble the ZpaCPR-(linker)-BsGDH fusion gene, the stop codon of the ZpaCPR gene was removed and the linker GSG was introduced between the open reading frames of the genes encoding ZpaCPR and BsGDH via two rounds of PCR. The first round of PCR introduced the linker GSG into the ZpaCPR gene using one pair of primers (Table S2). Simultaneously, the complementary linker GSG was introduced into the BsGDH gene using the other pair of primers (Table S2). Each PCR product was purified and served as a template in the second round of PCR. The PCR program was as follows: 3 min at 98 • C, 30 cycles at 98 • C (10 s), 58 • C (15 s), and 72 • C (30 s), and a final 10 min extension at 72 • C. The PCR products of the ZpaCPR and BsGDH genes were joined by overlap extension PCR. The PCR program was as follows: 3 min at 98 • C, 30 cycles at 98 • C (10 s), 58 • C (15 s), and 72 • C (40 s), and a final 5 min extension at 72 • C. The purified PCR products were ligated into a pACYCDuet-1 vector. The fusion gene was confirmed by sequencing. Finally, the fusion gene was ligated into pACYCDuet-1 between EcoR I and Hind III sites, yielding pACYCDuet-1-ZpaCPR-(GSG)-BsGDH. Using the construction of pACYCDuet-1-ZpaCPR-(GSG)-BsGDH plasmid as an example, pET28a-ZpaCPR-(GSG)-BsGDH was also constructed ( Figure S8). Plasmids pACYCDuet-1-ZpaCPR-(GSG)-BsGDH and pET28a-ZpaCPR-(GSG)-BsGDH were transformed into competent cells with pET28a-AmeLPLDH and pACYCDuet-1-AmeLPLDH, respectively, resulting in E. coli strains BL21 (DE3)/pET28a-AmeLPLDH/pACYCDuet-1-ZpaCPR-(GSG)-BsGDH and BL21 (DE3)/pET28a-ZpaCPR-(GSG)-BsGDH/pACYCDuet-1-AmeLPLDH. Following the procedure in Section 3.4, the cells were induced and harvested; then, the catalytic performance in the deracemization of DL-pantolactone was determined.
The optimal conditions for the process were investigated using a single factor method. Temperature varied from 20 to 40 • C, and pH values ranged from 5.0 to 10.0. Agitation between 0 and 600 rpm and the concentration of glucose as co-substrate within the range of 0 to 2500 mm were tested. In addition, the effects of substrate concentration (500 to 1750 mM) on the deracemization of DL-pantolactone were also determined. After 24 h reaction, the samples were treated as described in Section 3.4, then subjected to GC analysis as described in Section 3.10.

Deracemization of 1.25 mM DL-Pantolactone through Supplementation with BsGDH
After optimization, the reaction mixture (10 mL) contained 1.25 M DL-pantolactone, 2.5 M glucose, 200 g/L wet E. coli BL21 (DE3)/pACYCDuet-1-AmeLPLDH/pET28a-ZpaCPR-(GSG)-BsGDH cells, and 50 mM PBS buffer (pH 6.0). After 24 h reaction at 30 • C and 400 rpm, 100 g/L wet E. coli BL21 (DE3)/pET28a-BsGDH cells were added to the reaction mixture. Then, the reaction proceeded at 30 • C and 400 rpm for another 12 h. For the whole reaction, pH was kept constant using an auto-titration system. When the reaction was terminated at 36 h, the samples were treated as described in Section 3.4, then subjected to GC analysis as described in Section 3.10.
Upon completion of the catalytic reaction, the reaction mixture was treated with ethyl acetate; then, the product in the resulting organic phase was collected through evaporation of the solvent. The resulting crude product was validated by GC-MS (Agilent 7890A/5975C, Agilent Technologies Inc., Santa Clara, CA, USA) using previously reported parameters ( Figure S10) [8]. The crude product was dissolved in CDCl 3 for NMR analysis (Avance NEO, Bruker, Switzerland) to further verify the remaining product, Dpantolactone. The NMR spectroscopy was operated at 600 MHz for 1 H and 151 MHz for 13 C detection ( Figure S11).

Conclusions
In summary, a toolbox consisting of five LPLDHs, three CPRs, and three GDHs was developed for carrying out multi-enzymatic deracemization of DL-pantolactone. By screening catalytic activity, the enzymes AmeLPLDH, ZpaCPR, and BsGDH were selected, and the newly mined enzymes AmeLPLDH and ZpaCPR were purified and then characterized. The k cat /K m values of AmeLPLDH and ZpaCPR were 63.85 and 11.60 mM −1 s −1 , respectively. Through genetic fusion and vector selection, E. coli strain BL21(DE3)/pACYCDuet-1-AmeLPLDH/pET28a-ZpaCPR-(GSG)-BsGDH was obtained and used as whole-cell catalyst, providing a competitive catalytic process to synthesize D-pantolactone with high optical purity (>98% e.e. p ). Under the optimized conditions, the process enabled efficient deracemization of 1 M DL-pantolactone at 24 h, and supplementation with BsGDH further yielded nearly complete deracemization of 1.25 M DL-pantolactone to D-pantolactone after 36 h.