A Novel Whole-Cell Biocatalyst with NAD+ Regeneration for Production of Chiral Chemicals

Background The high costs of pyridine nucleotide cofactors have limited the applications of NAD(P)-dependent oxidoreductases on an industrial scale. Although NAD(P)H regeneration systems have been widely studied, NAD(P)+ regeneration, which is required in reactions where the oxidized form of the cofactor is used, has been less well explored, particularly in whole-cell biocatalytic processes. Methodology/Principal Findings Simultaneous overexpression of an NAD+ dependent enzyme and an NAD+ regenerating enzyme (H2O producing NADH oxidase from Lactobacillus brevis) in a whole-cell biocatalyst was studied for application in the NAD+-dependent oxidation system. The whole-cell biocatalyst with (2R,3R)-2,3-butanediol dehydrogenase as the catalyzing enzyme was used to produce (3R)-acetoin, (3S)-acetoin and (2S,3S)-2,3-butanediol. Conclusions/Significance A recombinant strain, in which an NAD+ regeneration enzyme was coexpressed, displayed significantly higher biocatalytic efficiency in terms of the production of chiral acetoin and (2S,3S)-2,3-butanediol. The application of this coexpression system to the production of other chiral chemicals could be extended by using different NAD(P)-dependent dehydrogenases that require NAD(P)+ for catalysis.


Introduction
Many approaches have been developed to produce chiral chemicals for the manufacture of a wide range of intermediates in the pharmaceutical, agrochemical, fine chemical and food industries [1]. In recent years, biocatalysis has emerged as a powerful strategy for the production of enantiomerically pure building blocks that cannot be obtained through chemical and fermentation processes [2]. NAD(P)-dependent oxidation systems have been used for the kinetic resolution of chiral alcohols and amines from a mixture of stereoisomers and for the production of ketones that are difficult to synthesize chemically [3][4]. However, due to the high costs of pyridine cofactors, an efficient cofactor regeneration system is a prerequisite for the commercial viability of a process [5][6].
Whole cells have some NAD + and NADP + reserves that provide a continuous source of cofactors [7]. Therefore, whole cells are used in many applications of dehydrogenases. Simultaneous overexpression of target enzymes and NAD(P)H regeneration enzymes in whole-cell biocatalysts has been carried out in many asymmetric reduction systems [8][9]. In this study, NAD + regeneration by whole-cell biocatalysis was used to extend the applications of NAD(P) + -dependent oxidoreductases. Chiral acetoin (AC) is widely used to synthesize novel optically active a-hydroxyketone derivatives and liquid crystal composites [10]. Chemical syntheses or fermentations generally lead to the production of a mixture of both isomers [11]. Therefore, it is essential to find an effective biocatalytic process to produce enantiomerically pure AC. In the presence of NAD + , NADdependent (2R,3R)-2,3-butanediol dehydrogenase (BDH) can theoretically catalyze the stereospecific oxidation of (2R,3R)-2,3butanediol (BD) and meso-2,3-BD to (3R)-AC and (3S)-AC, as shown in Scheme S1, respectively [12][13][14].
Similar to other dehydrogenases, BDHs require NAD + and NADH as cofactors [12][13][14]. During the catalytic process, NAD + is reduced to NADH, and 2,3-BD is oxidized to AC. Thus, during the course of one cycle, NAD + is depleted while NADH and AC are accumulated. A cofactor regeneration process is necessary to obtain high AC productivity. In contrast to the more common NADH oxidase (NOX) that converts O 2 to H 2 O 2 , the unique NOX from Lactobacillus brevis regenerates NAD + from NADH by reducing O 2 to H 2 O [15][16][17][18]. Since proteins are deactivated upon exposure to H 2 O 2 , the novel characteristics of the NOX from L. brevis make it a promising alternative for NAD + regeneration. In this study, we developed a coexpression system in which acetoin reductase/2,3-BDH encoded by the Bacillus subtilis ydjL (bdhA) gene was the producing enzyme and NOX was the cofactorregenerating enzyme. Our intention was to develop a novel biocatalytic process for the efficient production of (3R)-AC and (3S)-AC with high enantiomeric excess (ee).
In comparison with free enzymes, whole-cell biocatalysts are much more convenient to use and less expensive to prepare [19].

Effects of pH and Substrate Concentration on Chiral AC Biosynthesis
To achieve high product concentrations, the effect of pH on the yield of AC from 2,3-BD was investigated. As shown in Fig. 3A, reactions with 43.0 g L 21 of meso-2,3-BD as the substrate and 5.0 g dry cell weight (DCW) L 21 of E. coli BL21(DE3) (pETDuet-ydjLnox) as the biocatalyst were carried out under different conditions. After 6 h of reaction with E. coli BL21(DE3) (pETDuet-ydjLnox), the reaction rates were measured by monitor-  ing the yield of (3S)-AC. The results showed that the reaction rate was the highest at pH 8.0. It was reported that the optimum pH value for the conversion of 2,3-BD to AC by (2R,3R)-2,3-BDH was 9.0 and 5.5,7.0 for NOX [13,18]. When the two enzymes were coexpressed and functioned together, the optimum pH value was 8.0. Substrate concentration is another important parameter in the catalytic process, and it requires careful investigation because high substrate concentrations may lead to substrate inhibition. The effect of the substrate concentration on the conversion rate was tested (Fig. 3B). The results showed that in the range 8,61 g L 21 , the conversion rate of meso-2,3-BD decreased sharply from 100% to 47.5% as the concentration increased. Simultaneously, the concentration of (3S)-AC increased linearly with the increase in the substrate concentration from 8 g L 21 to 43 g L 21 , but when the substrate concentration increased beyond 43 g L 21 , the concentration of (3S)-AC decreased rapidly. It is known that AC can be easily separated from the reaction solution due to its low boiling point and that residual 2,3-BD can be reused. Therefore, 43 g L 21 of meso-2,3-BD was selected as a suitable substrate concentration based on the final yield of (3S)-AC.
Since (3S)-AC and (3R)-AC standards could be separated by GC, the ee of the products was calculated by the amount of each isomer. The two isomers of AC were produced in 96.0% ee. The results from GC analyses of substrates and products of the catalytic reaction are shown in Fig. 5A, B.
The production of chiral AC by biocatalysis has been reported earlier [20][21]. Yamada-Onodera et al. reported that about 9.68 g L 21 of both (3R)-AC from (2R,3R)-2,3-BD and (3S)-AC from meso-2,3-BD were obtained after 24 h of incubation with recombinant E. coli expressing only glycerol dehydrogenase (GDH) [20]. In this study, a recombinant strain that coexpressed (2R,3R)-2,3-BDH and NOX was constructed, and the yield of both (3S)-AC and (3R)-AC was found to be much higher than that in previous reports. The results also showed that the recombinant strain that coexpressed (2R,3R)-2,3-BDH and NOX was significantly more efficient than the recombinant strain that only expressed (2R,3R)-2,3-BDH in terms of AC production. As shown in Fig. 4, the yield of (3S)-AC from the two-enzyme system was 42.8% higher than that from the one-enzyme system (similar result was found for (3R)-AC production, data not shown).
Although the yield of (2S,3S)-2,3-BD was not very high due to the low content of (2S,3S)-2,3-BD in the mixture of 2,3-BD (only 12.4%), the biocatalytic efficiency of the (2R,3R)-2,3-BDH and NOX coexpressing recombinant strain was much higher than that of recombinant E. coli expressing GDH only (1.4 g L 21 ) [20]. This shows that the incorporation of a cofactor regeneration system in the E. coli whole-cell biocatalyst is a powerful strategy for enhancing the catalytic efficiency of oxidoreductases.
Many dehydrogenases that are useful in industrial production are cofactor-dependent. By using a suitable enantioselective dehydrogenase, important enantiomerically pure compounds can be prepared by kinetic resolution. For example, NAD + dependent (S)-ADH can be used for the production of (R)-phenylethanol from a racemate with 100% ee and 100% yield. Regeneration of the cofactor is important in large-scale applications of coenzymedependent reactions [24]. NAD(P)H regeneration has been widely studied, whereas NAD(P) + regeneration, which is required for reactions involving the oxidized forms of the pyridine nucleotide cofactors, is less well developed [15]. An NAD-dependent alcohol dehydrogenase from L. brevis coupled with the NOX from Lactobacillus sanfranciscensis was successfully used to produce enantiomerically pure alcohol from a racemic mixture [15]. The two enzymes were first purified, and then coupled in a reaction in which NOX was used to regenerate the oxidized cofactor. Although this method did not generate any byproducts, the complex purification processes and the oxygen sensitivity of NOX precluded its applications on a preparative scale [3].
In comparison with isolated enzymes, utilization of whole-cell biocatalysts circumvents laborious protein purification steps, which simplifies the reactions in many cases. NAD(P) + regeneration in whole-cell biocatalysts would help in extending the applications of NAD(P)-dependent oxidoreductases. In this study, we constructed a novel whole-cell biocatalyst from recombinant E. coli in which (2R,3R)-2,3-BDH and NOX were coexpressed to catalyze the production of (3R)-AC, (3S)-AC, and (2S,3S)-2,3-BD with high ee. The present study clearly demonstrates the importance of providing sufficient amounts of the oxidized form of the pyridine nucleotide cofactors to NAD-dependent dehydrogenases. The value of the NAD + regeneration technique presented in this paper could extend beyond chiral AC and (2S,3S)-2,3-BD production to other chiral chemicals. Coupling NOX from L. brevis in the biocatalytic system provided a suitable method for NAD + regeneration, and this should open up the possibilities for constructing other whole-cell biocatalysts with different NADdependent dehydrogenases.
Constructions of E. coli BL21(DE3) (pETDuet-ydjL), E. coli BL21(DE3) (pETDuet-nox) and E. coli BL21(DE3) (pETDuet-ydjLnox) The bacterial strains and plasmids used in this study are listed in Table 2. B. subtilis 168 genomic DNA was extracted with the Wizard Genomic DNA Purification Kit (Promega, Madison, WI, USA). The ydjL gene was amplified by PCR using forward primer py1 with an NcoI restriction site insertion and reverse primer py2 with a SalI restriction site insertion. The PCR product was firstly ligated to the pEasy-Blunt vector, and the resulting plasmid was designated pEasy-Blunt-ydjL. The pEasy-Blunt-ydjL was then sequenced (Sangon, Shanghai, China) to verify that no mutations were introduced by PCR. Next, to construct the recombinant plasmid pETDuet-ydjL under the control of the T7 promoter, pEasy-Blunt-ydjL was digested with NcoI and SalI, and the gel- purified ydjL fragment was ligated to the pETDuet-1 vector that had been digested with the same restriction enzymes. Using the same process that described above, the nox gene fragment was obtained from the genome of L. brevis using primers pn1 (with the BglII restriction site) and pn2 (with the XhoI restriction site). To construct a coexpression system carrying ydjL and nox, pEasy-Blunt-nox was digested with BglII and XhoI, and the gel-purified nox fragment was ligated to pETDuet-ydjL digested with the same restriction enzymes. E. coli DH5a was used for general cloning, and E. coli BL21(DE3) was used for protein expression. Luria-Bertani (LB) medium was used for both E. coli and B. subtilis culture. DeMan-Rogosa Sharpe (MRS) medium was used for L. brevis culture.

Biocatalyst Preparation
Recombinant E. coli cells were grown at 37uC on a rotary shaker (180 rpm) in LB medium containing ampicillin (100 mg mL 21 ) to an OD 620nm value of 0.6. Expression of the recombinant gene was induced by adding 1 mM IPTG at 16uC to avoid the formation of inactive inclusion bodies. After induction, the cells were harvested by centrifugation at 6,0006g for 5 min at 4uC and then washed twice with 1/15 M phosphate buffer (PB) (pH 7.4). The cell pellet was resuspended in 1/15 M PB (pH 7.4), and maintained at 4uC for further studies.

Assay of Whole-Cell Biocatalytic Activity
The whole-cell biocatalytic activity was assayed by measuring the increase of AC in the reaction solution. The reaction solution consisting of 10.0 g L 21 2,3-BD and the whole-cell biocatalyst (the final concentration was 1.5 g DCW L 21 ) in 200 mM PB (pH 8.0) was incubated at 37uC on a rotary shaker for 20 min. The mixture was centrifuged at 6,0006g for 5 min to stop the reaction, and the AC concentration in the supernatant was measured. One unit of whole-cell biocatalytic activity was defined as the amount of cells that catalyzed the formation of 1.0 mmol of AC per minute at 37uC.

Product Analysis
Samples (0.15 mL) were taken periodically and centrifuged at 15,0006g. The concentrations of 2,3-BD and AC in the supernatant were analyzed by GC. Prior to GC analysis, the supernatant was extracted with an equal volume of ethyl acetate after the addition of isoamyl alcohol as the internal standard. The GC (Agilent GC6820) system consisted of a flame ionization detector and a fused silica capillary column (Supelco Beta DEX TM 120, inside diameter, 0.25 mm; length, 30 m). The operating conditions were as follows. Nitrogen was used as the carrier gas. The injector temperature and detector temperature were both 280uC. The column oven was maintained at 40uC for 3 min and then programmed to increase to 80uC at a rate of 1.5uC min 21 . The temperature was then raised to 86uC at a rate of 0.5uC min 21 and finally to 200uC at a rate of 30uC min 21 . The injection volume was 3 mL. In this study, the ee values of (3S)-AC and (3R)-AC were calculated as  . GC analyses of substrates and products of the catalytic reaction (* Isoamyl alcohol was used as the internal standard). A. Conversion of meso-2,3-BD to (3S)-AC: A1-before the reaction; A2-after the reaction; B. Conversion of (2R,3R)-2,3-BD to (3R)-AC: B1-before the reaction; B2-after the reaction; C. The starting material and products of kinetic resolution from a mixture of 2,3-BD: C1before the reaction; C2-after the reaction. doi:10.1371/journal.pone.0008860.g005