Efficient Production of (R)-2-Hydroxy-4-Phenylbutyric Acid by Using a Coupled Reconstructed d-Lactate Dehydrogenase and Formate Dehydrogenase System

Background (R)-2-Hydroxy-4-phenylbutyric acid [(R)-HPBA] is a key precursor for the production of angiotensin-converting enzyme inhibitors. However, the product yield and concentration of reported (R)-HPBA synthetic processes remain unsatisfactory. Methodology/Principal Findings The Y52L/F299Y mutant of NAD-dependent d-lactate dehydrogenase (d-nLDH) in Lactobacillus bulgaricus ATCC 11842 was found to have high bio-reduction activity toward 2-oxo-4-phenylbutyric acid (OPBA). The mutant d-nLDHY52L/F299Y was then coexpressed with formate dehydrogenase in Escherichia coli BL21 (DE3) to construct a novel biocatalyst E. coli DF. Thus, a novel bio-reduction process utilizing whole cells of E. coli DF as the biocatalyst and formate as the co-substrate for cofactor regeneration was developed for the production of (R)-HPBA from OPBA. The biocatalysis conditions were then optimized. Conclusions/Significance Under the optimum conditions, 73.4 mM OPBA was reduced to 71.8 mM (R)-HPBA in 90 min. Given its high product enantiomeric excess (>99%) and productivity (47.9 mM h−1), the constructed coupling biocatalysis system is a promising alternative for (R)-HPBA production.

In previous studies, enzymatic resolution and asymmetric reduction were used in the biological production of (R)-HPBA. Compared with enzymatic resolution catalyzed by hydrolases, especially lipases [4,9,13], asymmetric bio-reduction of 2-oxo-4phenylbutyric acid (OPBA) by dehydrogenases is more desirable because of its excellent stereoselectivity and high theoretical yield up to 100% [1,14]. For practical production of (R)-HPBA from OPBA through bio-reduction, highly efficient reductases and cofactor regeneration systems are needed.
In contrast to the (R)-HPBE preparation processes, which often utilize a specific carbonyl reductase, the production of (R)-HPBA from OPBA is catalyzed by 2-ketoacid reductases, especially NADdependent D-lactate dehydrogenase (D-nLDH) [12,15]. However, as an unnatural substrate of D-nLDH, OPBA could not be efficiently catalyzed by the biocatalyst because of its large aromatic group at C-4.On the other hand, cofactor regeneration systems that utilize glucose as a co-substrate in (R)-HPBE production may not be the proper choice in the (R)-HPBA production. The addition of glucose to the reaction system may result in the production of organic acids (such as gluconic acid and lactic acid) as byproducts and increase the complexity of the (R)-HPBA separation process [16,17]. In a previous study, a partially purified D-nLDH was used to transform OPBA to (R)-HPBA. The cofactor NADH was regenerated by formate dehydrogenase (FDH) present in whole cells of Candida boidinii ATCC 32195. Although this NADH regeneration system produced CO 2 as the only byproduct, which facilitated the isolation of (R)-HPBA, the whole cells of C. boidinii should be pre-treated with toluene to make them permeable [12].
In our previous studies, the D-nLDH in Lactobacillus bulgaricus ATCC 11842 was rationally re-designed and then used for the bioreduction of substrates with large aliphatic or aromatic groups at C-3 [14]. In this study, the activities of different D-nLDH mutants toward OPBA (2-oxo carboxylic acids with an aromatic group at C-4) were assayed. The most active reconstructed D-nLDH was co-expressed with FDH from C. boidinii NCYC 1513 in E. coli BL21 (DE3). Then, a novel process utilizing whole cells of recombinant E. coli was developed for efficient production of (R)-HPBA from OPBA (Fig. 1).

Microorganisms and growth conditions
The bacterial strains, plasmids, and oligonucleotide primers used in this study are listed in Table 1. E. coli DH5a and BL21 (DE3) were used for general cloning and expression procedures, respectively. E. coli WD, E. coli D1, and E. coli D2 were used to express wild D-nLDH, D-nLDH F299Y , and D-nLDH Y52L/F299Y , respectively [14]. E. coli PD containing the vector pETDuet-1 was used as a control. The plasmid pETDuet-ldhD Y52L/F299Y -fdh was constructed as follows: the ldhD Y52L/F299Y gene was amplified using primers D.f and D.r with plasmid pETDuet-ldhD Y52L/F299Y as a template. The fdh gene was amplified using primers F.f and F.r with genomic DNA of C. boidinii NCYC 1513 as a template. The resulting PCR products ldhD Y52L/F299Y and fdh were digested with NcoI-BamHI and NdeI-XhoI, respectively, and cloned into MCS1 and MCS2 of pETDuet-1 successively to construct pETDuet-ldhD Y52L/F299Y -fdh. The plasmid pETDuet-ldhD Y52L/ F299Y -fdh was then transformed into E. coli BL21 (DE3) to construct E. coli DF. All of the E. coli strains were grown in Luria-Bertani (LB) medium, and ampicillin was added at a concentration of 100 mg m1 21 if necessary.  Table 1. Strains, plasmids, and oligonucleotide primers used in this study.

Biocatalyst preparation
The recombinant strains of E. coli PD, E. coli WD, E. coli D1, E. coli D2, and E. coli DF were all cultured in LB medium (100 mg ml 21 ampicillin) at 37uC to an optical density of 0.6 at 600 nm. IPTG (1 mM) was then added to induce protein expression, and cultures were grown at 16uC for a further 12 h. Cells were harvested by centrifugation at 6,000 rpm for 10 min, washed twice with 67 mM phosphate buffer solution (pH 7.4), and then subjected to successive biotransformation.

Optimization of biocatalysis conditions
To optimize the biotransformation conditions, 5-ml reaction mixtures were incubated at 37uC and 120 rpm in a 25-ml flask. The pH was adjusted from 5.5 to 8.5. The concentrations of OPBA and formate were 25-175 mM. The concentration of the whole cells was 1-8 g dry cell weight (DCW) l 21 . Samples (0.2 ml) were collected periodically and centrifuged at 12,000 rpm. The concentrations of OPBA and (R)-HPBA in the supernatant were analyzed by a high-performance liquid chromatography (HPLC) system (Agilent 1100 series, Hewlett-Packard, USA).

Analytical procedures
Cells of E. coli PD, E. coli WD, E. coli D1, E. coli D2, and E. coli DF were harvested, suspended in 67 mM phosphate buffer solution (pH 7.4) containing 1 mM PMSF, and then disrupted by sonication (Sonics 500 W; 20 KHz) for 5 min in an ice bath. Thereafter, intact cells and cell debris were removed by centrifugation, and the resultant crude extracts were subjected to successive D-nLDH activity assays. The reduction activities of D-nLDH wild-type and mutants toward OPBA were assayed at 37uC in 1 ml of 50 mM Tris-HCl buffer (pH 7.5) containing 0.2 mM NADH, 10 mM OPBA, and the crude extracts of E. coli PD, E. coli WD, E. coli D1, E. coli D2, and E. coli DF. The rate of NADH decrease was determined by measuring the absorbance change at 340 nm [14,18,19]. One unit of D-nLDH activity was defined as the amount that catalyzed the oxidation of 1 mmol of NADH per minute. The protein concentration was determined by the Lowry procedure by using bovine serum albumin as the standard [20].
OPBA and (R)-HPBA were measured by HPLC (Agilent 1100 series) equipped with an Agilent Zorbax SB-C18 column (15064.6 mm, 5 mm) and a variable-wavelength detector at 210 nm. The mobile phase consisted of 1 mM H 2 SO 4 and acetonitrile with a ratio of 85:15 (v/v) at a flow rate of 0.7 ml min 21 at 30uC. Stereoselective assays for (R)-HPBA and

Activity of D-nLDH wild-type and mutants toward OPBA
To evaluate the possibility of transforming OPBA into (R)-HPBA by D-nLDH, the wild type D-nLDH from L. bulgaricus ATCC 11842 and its mutants were overexpressed in E. coli BL21 (DE3). Crude extracts of E. coli PD, E. coli WD, and E. coli D1 exhibited rather low OPBA reduction activity (Fig. 2A). The Y52L/F299Y mutant of D-nLDH caused the specific activity of the crude extract of E. coli D2 to be 233.2-312.3 fold higher than that in extracts of E. coli PD, E. coli WD, and E. coli D1. These results suggest that the mutant D-nLDH Y52L/F299Y is rather active toward OPBA and may have the potential to efficiently produce (R)-HPBA from OPBA.
Feasibility of (R)-HPBA production through the cofactor regeneration system Asymmetric reduction of OPBA by whole cells of E. coli PD, E. coli WD, E. coli D1, E. coli D2, and E. coli DF was investigated to further explore the potential by using D-nLDH in the synthesis of (R)-HPBA. OPBA at 50 mM was used as the substrate. Whole cells of E. coli PD, E. coli WD, E. coli D1, and E. coli D2 at a concentration of 8 g DCW l 21 were added to the reaction broth. The reaction was conducted at 37uC for 2 h. Here, NADH was regenerated through the direct addition of 50 mM glucose in the reaction system. Whole cells of E. coli D2 exhibited higher (R)-HPBA producing capability than did cells of E. coli PD, E. coli WD, and E. coli D1 (Fig. 2B). However, the (R)-HPBA productivity (3.7 mM h 21 ) was still rather low because of the low efficiency of the NADH regeneration system. Additionally, organic acids, including pyruvic acid, lactic acid, and acetic acid, accumulated in the reaction broth (Fig. S1).
FDH is a good choice for NADH regeneration in a biocatalysis system because its substrate, formate, has a low cost and its product, carbon dioxide, is easily separated [21][22][23][24][25]. In this work, FDH was coexpressed with D-nLDH Y52L/F299Y in E. coli DF and the (R)-HPBA production capability of the novel biocatalyst was investigated. Formate (50 mM) was added to the reaction broth for the regeneration of NADH. Although the activity of D-nLDH Y52L/ F299Y in the crude extract of E. coli DF was lower than in the extract of E. coli D2, whole cells of E. coli DF exhibited much higher (R)-HPBA producing capability than other biocatalysts ( Fig. 2A and Fig. 2B). (R)-HPBA at 49.0 mM was obtained from 50 mM OPBA. The productivity of (R)-HPBA was 24.5 mM h 21 . Thus, whole cells of E. coli DF were selected as biocatalysts for (R)-HPBA production in the subsequent experiments.

Optimization of biocatalysis conditions
To achieve a higher product concentration, the biocatalytic conditions for (R)-HPBA production from OPBA by using whole cells of E. coli DF were optimized. The influence of the reaction pH was determined in reaction mixtures containing 13 g DCW l 21 whole cells of E. coli DF, 50 mM OPBA, 50 mM sodium formate, and 200 mM phosphate buffer (pH ranging from 5.5 to 8.5). After bioconversion at 37uC for 15 min, the highest (R)-HPBA production was detected at pH 6.5 (Fig. 3A).  To determine the effect of the OPBA concentration, reactions with eight different OPBA and sodium formate concentrations (25,50,75,100,125,150, and 175 mM) were conducted at pH 6.5 and 37uC for 30 min. The highest (R)-HPBA production was detected when 75 mM OPBA was used (Fig. 3B). The effect of the biocatalyst concentration was also investigated to determine the optimal range. The biotransformation was conducted with 75 mM OPBA, 75 mM sodium formate, 200 mM phosphate buffer (pH 6.5), and whole cells of E. coli DF at six different concentrations (1, 3, 5, 6, 7, and 8 g DCW l 21 ). When the reactions were conducted to approximate 80% theoretical yield, the highest specific productivity was observed at a biocatalyst concentration of 6 g DCW l 21 ( Table 2).

Production of (R)-HPBA under optimal conditions
On the basis of the results presented above, an optimal bioconversion system for production of optically pure (R)-HPBA from OPBA was developed. Biotransformation was conducted at 37uC in 200 mM phosphate buffer (pH 6.5) with 6 g DCW l 21 whole cells of E. coli DF as the biocatalyst. As shown in Fig. 4A, 71.8 mM (R)-HPBA with a high enantiomeric purity (ee .99%, Fig. S2) was obtained from 73.4 mM OPBA in 90 min. When whole cells of E. coli D2 only expressing D-nLDH Y52L/F299Y were used as the biocatalyst, and glucose was added for NADH regeneration, only 44.7 mM (R)-HPBA was produced with a yield of 60.9% after 360 min (Fig. 4B).
Many biocatalysts have been used in the enantioselective production of (R)-HPBE and (R)-HPBA through bio-reduction [1,12,[26][27][28]. Compared with (R)-HPBE production processes, the product concentrations of the reported (R)-HPBA synthesis processes were rather low (Table 3) [1,[10][11][12]15,29,30]. In the previous study, purified D-LDH from Staphylococcus epidermidis and FDH from Candida boidinii were applied for (R)-HPBA production. (R)-HPBA at a concentration of 182 mM was produced, which is the highest reported yield of (R)-HPBA to date [15]. However, problems concerning the application of the process, such as the complicated enzyme purification and costly cofactor addition, remain. In the present work, mutant D-nLDH and FDH were co-expressed in E. coli DF and used for (R)-HPBA production from OPBA. The productivity (47.9 mM h 21 ) and ee (.99%) of the product were rather high for (R)-HPBA production. Additionally, given the simple composition of the biocatalytic system, separation of (R)-HPBA from the biocatalytic system would be relatively inexpensive. Therefore, the novel process established in this study could also be used as a promising route for the production of highly optically pure (R)-HPBA.

Conclusions
In summary, whole cells of E. coli DF coexpressing D-nLDH Y52L/F299Y from L. bulgaricus ATCC 11842 and FDH from C. boidinii NCYC 1513 exhibited catalytic capability for (R)-HPBA production from OPBA. After optimization of the biotransformation conditions, 73.4 mM OPBA was reduced to 71.8 mM (R)-HPBA with a high productivity of 47.9 mM h 21 and an excellent ee (.99%). The constructed coupled biocatalysis Table 3. Comparison of recently reported processes for (R)-HPBA or (R)-HPBE production through bio-reduction. (R)-2-Hydroxy-4-Phenylbutyric Acid Production PLOS ONE | www.plosone.org system developed in this work may be a promising alternative for the production of the key medical intermediate (R)-HPBA. Figure S1 HPLC analysis of the product of the catalytic reaction by using whole cells of E. coli D2 (A) as the biocatalyst and glucose as the substrate for NADH regeneration or whole cells of E. coli DF (B) as the biocatalyst and sodium formate as the substrate for NADH regeneration.