Efficient biosynthesis of ethyl (R)-4-chloro-3-hydroxybutyrate using a stereoselective carbonyl reductase from Burkholderia gladioli.

BACKGROUND: Ethyl (R)-4-chloro-3-hydroxybutyrate ((R)-CHBE) is a versatile chiral precursor for many pharmaceuticals. Although several biosynthesis strategies have been documented to convert ethyl 4-chloro-3-oxobutanoate (COBE) to (R)-CHBE, the catalytic efficiency and stereoselectivity are still too low to be scaled up for industrial applications. Due to the increasing demand of (R)-CHBE, it is essential to explore more robust biocatalyst capable of preparing (R)-CHBE efficiently.
RESULTS: A stereoselective carbonyl reductase toolbox was constructed and employed into the asymmetric reduction of COBE to (R)-CHBE. A robust enzyme designed as BgADH3 from Burkholderia gladioli CCTCC M 2012379 exhibited excellent activity and enantioselectivity, and was further characterized and investigated in the asymmetric synthesis of (R)-CHBE. An economical and satisfactory enzyme-coupled cofactor recycling system was created using recombinant Escherichia coli cells co-expressing BgADH3 and glucose dehydrogenase genes to regenerate NADPH in situ. In an aqueous/octanol biphasic system, as much as 1200 mmol COBE was completely converted by using substrate fed-batch strategy to afford (R)-CHBE with 99.9 % ee at a space-time yield per gram of biomass of 4.47 mmol∙L-1∙h-1∙g DCW-1.
CONCLUSIONS: These data demonstrate the promising of BgADH3 in practical synthesis of (R)-CHBE as a valuable chiral synthon. This study allows for the further application of BgADH3 in the biosynthesis of chiral alcohols, and establishes a preparative scale process for producing (R)-CHBE with excellent enantiopurity.

Background

Stereoselective carbonyl reductases (E.C. 1.1.1.x; SCRs) are nicotinamide cofactor-dependent enzymes capable of catalyzing the reversible redox reaction between alcohols and aldehydes/ketones. During the past decade, SCRs have been considerably applied to the synthesis of chiral pharmaceutical intermediates, including anticholesterol drugs [1, 2], β-lactams antibiotics [3], anticancer drugs [4], and other important drugs [5]. However, the scale-up of SCR-catalyzed reactions were restricted due to the limited commercially available SCRs, narrow substrate specificity, expensive cofactor dependency, and substrate insolubility.

Ethyl (R)-4-chloro-3-hydroxybutyrate ((R)-CHBE) is a versatile precursor for pharmacologically valuable products, such as L-carnitine [6], (R)-4-amino-3-hydroxybutyric acid (GABOB) [7], and (R)-4-hydroxy-pyrrolidone [8]. Several synthetic strategies for optically active CHBE were developed, wherein the enzymatic asymmetric synthesis is the most promising way. Although various biocatalysts have been found to give (S)-CHBE [1, 2, 9–13], (R)-isomer is in great demand yet less attainable. Since then, several microorganisms and enzymes capable of converting ethyl 4-chloro-3-oxobutanoate (COBE) to (R)-CHBE have been documented, including gox2036 from Gluconobacter oxydans [14], AKRs from Sporobolomyces salmonicolor and Lodderomyces elongisporus [15-17], and a reductase from Bacillus sp. ECU0013 [18]. However, all of them suffer from impediment such as low substrate concentration, unsatisfactory stereoselectivity, or high substrate/catalyst (S/C) ratio. These shortcomings hindered their applications in the industrial synthesis of (R)-CHBE. Exploring more robust SCRs with the ability to prepare enantioenriched (R)-CHBE efficiently is thus of great interest.

Herein, we designed and implemented two strategies for identifying novel SCRs, and constructed an enzyme toolbox to screen a robust SCR that can biotransform COBE to (R)-CHBE. As the promising SCR, BgADH3 from Burkholderia gladioli CCTCC M 2012379 was selected for further study. The substrate spectrum of BgADH3 was evaluated toward varied aryl ketones and ketoesters. Furthermore, the practical applicability of BgADH3 was investigated in the asymmetric synthesis of (R)-CHBE using Escherichia coli cells co-expressing BgADH3 and a glucose dehydrogenase (GDH). Since the substrate was poorly soluble and unstable in aqueous environments, biphasic system was established using substrate fed-batch strategy to solve this issue. To our knowledge, this is the first report of SCR from B. gladioli subjected to the asymmetric synthesis of enantioenriched (R)-CHBE in aqueous/octanol biphasic system.

Results

Identification and screening of SCRs

Strain B. gladioli CCTCC M 2012379 isolated from soil samples exhibited activity for catalyzing COBE to (R)-CHBE (Additional file 1: Table S1). Genome hunting and data mining strategies were selected to discovering robust SCRs from CCTCC M 2012379. Based on bioinformatics analysis of sequence-similarity with gox2036 [14], which is a known NADH-dependent SCR giving enantiopure (R)-CHBE. 35 candidates were cloned or synthesized, and heterologously overexpressed in E. coli BL21 (DE3), wherein BgADH3 displayed high activity toward COBE and afforded (R)-CHBE (Additional file 2: Table S2). Sequence analysis indicated that the BgADH3 gene contained an open reading frame with 1011 bp encoding a 336 amino-acid protein, in which the conserved NADP-binding motiff T97G98XXXG102XG104 and key catalytic residues N202S228Y241K245 were found (Additional file 3: Figure S1). The recombinant BgADH3 with His6-tag mainly presented in the soluble fraction was purified through nickel chelating affinity chromatography [19]. As shown in Fig. 1, the BgADH3 was not homogeneous, and the estimated molecular mass was around 37 kDa, in accordance with its theoretical value (37.3 kDa). The molecular mass of native BgADH3 determined on the Discovery BIO GFC 150 column suggested that the quaternary structure of BgADH3 was dimer.

Fig. 1

SDS-PAGE analysis of BgADH3. Lane 1, molecular mass standard; Lane 2, E. coli BL21 (DE3)/pET28b-BgADH3 without IPTG induction; Lane 3, crude extract; Lane 4, soluble fraction; Lane 5, purified BgADH3

Biochemical characterization of BgADH3

To investigate the effect of pH on the activity and stability of BgADH3, different pH values ranging from 3.5 to 10.5 were tested. As a result, a bell-shaped pH profile was observed and the maximum activity was found at pH 6.5 (Fig. 2a). Remarkably, the recombinant BgADH3 is highly stable between pH 5.0 and 9.0 (Fig. 2b). The effect of temperature on BgADH3 activity was assessed by measuring activity at 25–65 °C. The optimum temperature occurred at 40 °C (Fig. 2c). Thermostability of BgADH3 was evaluated at temperatures of 25, 35, 45, 55, and 65 °C, respectively. Half-life time of 47 h was found at 45 °C. However, the enzyme was labile at higher temperatures (55 and 65 °C) (Fig. 2d). The presence of Ca2+, Co2+, Cu2+, Fe3+, Zn2+, and Mg2+ slightly enhanced the activity of BgADH3, whereas Ag+ inhibited the activity of the enzyme with a loss of 75 % (Additional file 4: Table S3). When the metal-chelating agent EDTA-Na2 was added, the residual activity of BgADH3 was maintained intact (99 %). Cofactor preference of BgADH3 was evaluated using COBE as substrate and the result was given in Table 1. The purified BgADH3 displayed specific activities for the asymmetric reduction of COBE using both NADH and NADPH as cofactors. Although the K mNADPH value (0.081 mM) of BgADH3 was slightly greater than that toward NADH (0.058 mM), the overall catalytic efficiency (k cat/K m) for NADPH was approximately 2-fold enhancement compared to that against NADH.

Fig. 2

Effects of pH and temperature on the activity and stability of BgADH3. a Optimal pH. b pH stability. c Optimal temperature. d Thermostability. Reaction conditions: COBE (50 mM), NADPH (0.5 mM), and purified BgADH3 (0.1 mg mL−1). The activities were determined after the addition of NADPH for 2 min. All reactions were performed in triplicate

Table 1

Apparent kinetic parameters of BgADH3a

Substrate	NADH	NADPH
K m (mM)	0.058	0.081
V max (μmol min−1 mg−1)	3.4	9.1
k cat (s−1)	3.97	10.6
k cat/K m (mM−1s−1)	68.3	131

aReaction conditions: substrate COBE (50 mM), NADH or NADPH (0.01–0.5 mM), purified BgADH3 (0.1 mg mL−1), pH 6.5, and 40 °C

Effects of organic solvents

A series of hydrophobic solvents were selected to assess their effects on BgADH3 activity and histograms representing the residual activities were arrayed in accordance with the increasing log P values (Fig. 3). BgADH3 activity was roughly correlated with the log P values of the corresponding solvents, except for dichloromethane, which strongly inhibited the enzyme activity. High activity was observed in the present of hydrophobic solvents with high log P value, such as octanol (91 %), cyclohexane (94 %), n-hexane (95 %), n-heptane (102 %), and iso-octane (99 %). Similarly, the effect of the hydrophobicity of the solvents on the retention of BgADH3 stability was investigated. Satisfactory results were obtained in the present of methyl tert-butyl ether (MTBE), butyl acetate, octanol, iso-octane, and n-heptane, wherein the highest stability was observed in the presence of MTBE (100 %).

Fig. 3

Effect of organic solvents on the activity and stability of recombinant BgADH3. The activities were measured using COBE as substrate under the standard assay protocol. Stability was determined by measuring the residual activities after incubation with organic solvents at 30 °C for 30 min. The activity in the absence of organic solvent was taken as control. THF, tetrahydrofuran; MTBE, methyl tert-butyl ether; CH2Cl2, dichloromethane

Substrate specificity and stereoselectivity

Specific activity was determined rapidly by using spectrophotometric standard assay described above and the stereoselectivity was evaluated in aqueous phosphate buffer by adding cosubstrate. As shown in Table 2, BgADH3 were active on all the tested aryl ketones and ketoesters, and exhibited dramatically high activity toward 4’-fluoroacetophenone 6, 3’,5’-bis(trifluoromethyl) acetophenone 7, and COBE 12. In contrast, substrate 4, 5, 8, 15, 16, and 17 significantly diminished the activity of BgADH3. Notably, excellent stereoselectivity of BgADH3 was also observed for most of the tested substrates, resulting in the corresponding chiral alcohols with >99 % ee (Additional file 5: Table S4). BgADH3 generally exhibited S preference, but the corresponding bioproducts with R configuration were obtained toward substrates 12–15 and 17.

Table 2

Substrate specificity and stereoselectivity of the BgADH3

aThe unit of specific activity was U mg−1 protein. bThe ee or de values of the corresponding products. cDetermined by chiral GC analysis. dDetermined by chiral HPLC analysis

Co-expression of BgADH3 and GDH genes

The BgADH3 gene and GDH gene (786 bp) encoding 261-amino acids were introduced into the MCSI and MCSII sites of pCDFDuet-1 vector (Fig. 4a). After induced by IPTG, the recombinant BgADH3 and GDH protein were successfully expressed in the same E. coli cell. The molecular weights of 37 kDa for BgADH3 and 28 kDa for GDH was observed in the sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) (Fig. 4b), which indicated that the expressed products are consistent with the predicted proteins.

Fig. 4

Schematic presentation of the co-expression plasmid contained BgADH3 and GDH genes and SDS-PAGE analysis. a Structure of the co-expression plasmid pCDFDuet-1-BgADH3-GDH. b SDS-PAGE analysis of recombinant co-expressed protein. Lane 1, protein molecular weight marker; Lane 2, soluble fraction of E. coli/pET28b-BgADH3; Lane 3, precipitate of E. coli/pET28b-BgADH3; Lane 4, soluble fraction of E. coli/pET28b-GDH; Lane 5, precipitate of E. coli/pET28b-GDH; Lane 6, soluble fraction of E. coli/pCDFDuet-1-BgADH3-GDH; Lane 7, precipitate of E. coli/pCDFDuet-1-BgADH3-GDH; Lane 8, E. coli/pCDFDuet-1-BgADH3-GDH without induction

Asymmetric synthesis of (R)-CHBE in aqueous-organic solvent system

Screening of organic phase

To screen an appropriate organic phase in the biphasic system for (R)-CHBE production by using co-expression strain E. coli/pCDFDuet-1-BgADH3-GDH, n-butanol, tert-amylalcohol, MTBE, butyl acetate, toluene, and octanol were selected to evaluate the effects on product yield and ee. The result was presented in Table 3. All yields of aqueous-organic solvents systems were below 50 % or less, except for octanol, in which a yield of 86.6 % was obtained. It was noteworthy that reduction of COBE was strongly inhibited in butyl acetate, affording too low yield (12 %). Interestingly, the enantioselectivity was not impaired and maintained >99 % ee in the presence of all the tested organic solvents.

Table 3

Screening of organic solvents for biosynthesis of (R)-CHBE

Organic solvent	Log P a	Partition coefficientb		Yield (%)	ee (%)
		COBE	CHBE
n-Butanol	0.839	2.6c	1.8c	37.7	>99
tert-Amylalcohol	1.094	3.3c	2.9c	39.4	>99
MTBE	1.296	8.9c	4.8c	48.1	>99
Butyl acetate	1.804	10.6	15.8	12.8	>99
Toluene	2.27	21.3	2.8	44.1	>99
Octanol	2.876	5.3	3.8	86.6	>99

aCalculated using Advanced Chemistry Development (ACD/Labs) Software V11.02

bPartition coefficient is the molar ratio of each compound found in the organic phase to the same compound in the aqueous phase (data from ref. [18, 46, 47])

cCOBE (73.9 μmol) and CHBE (71.2 μmol) in 0.7 mL phosphate buffer (100 mM, 6.5) were shaken with organic solvents (0.7 mL) at 30 °C in an eppendorf tube. The concentrations of COBE and CHBE in each phase were determined by GC analysis

Optimization of reaction conditions

To improve the asymmetric reduction of COBE to (R)-CHBE by using co-expression E. coli, different reaction conditions were further investigated. As shown in Fig. 5a, it was found that the yield of (R)-CHBE was significantly improved when the glucose/COBE (mmol/mmol) ratio was increased to 3, and then was clearly diminished when the ratio was over 8. To effectively synthesize (R)-CHBE, the amount of NADP+ was also optimized. The highest yield was obtained using a NADP+/COBE (μmol/mmol) ratio of 1.0 and a huge excess of NADP+ was not necessary (Fig. 5b). The maximum level of yield was found at 30 °C and pH 6.5 (Fig. 5c and d). An increased production of (R)-CHBE was observed as the cell dosage was raised from 0.02 to 0.04 g DCW. When the cell dosage was excess 0.04 g DCW, the yield of (R)-CHBE was not increased significantly (Fig. 5e). Keeping constant the S/C (mmol/g DCW) ratio of 150, different amount of COBE were further estimated. The yield of (R)-CHBE was significantly decreased when more than 24 mmol of the substrate was added (Fig. 5f). Based on the above results, the optimum reaction conditions in 20 mL system were obtained: COBE 24 mmol, glucose 72 mmol, NADP+ 24 μmol, cell dosage 0.16 g DCW, pH 6.5, and 30 °C.

Fig. 5

Effects of glucose concentrations (a), NADP+ concentrations (b), temperature (c), pH (d), cell dosage (e), and substrate loading (f) on the biosynthesis of (R)-CHBE using co-expression E. coli cells

Substrate fed-batch strategy and preparative scale

The reaction was performed in a water/octanol biphasic system with a total of 1200 mmol COBE addition that was fed in three steps, during which the resulting gluconic acid was neutralized with 2 M NaOH to maintain an initial pH of 6.5. As shown in Fig. 6, at 20 h, 1101.6 mmol of (R)-CHBE was produced with a satisfactory enantiomeric purity of 99.9 % ee and the remaining COBE substrate was not detectable. After normal downstream processing, the desired chiral alcohol product was obtained with a yield of 91.8 %. To further confirm the structure and configuration of the final product, NMR, MS, and optical rotation were performed. 1H NMR (400 MHz, CDCl3): δ 4.3–4.28 (m, 1H), 4.18 (q, J = 7.2 Hz, 2H), 3.60–3.61 (m, 2H), 3.01 (s, 1H), 2.57–2.66 (m, 2H), and 1.24–1.29 (m, 3H) (Additional file 6: Figure S2); 13C NMR (126 MHz, CDCl3): δ 171.76, 67.93, 60.98, 48.12, 38.51, 14.07 (Additional file 6: Figure S3); MS calculated for C6H12ClO3: 167.0, found: 167.2 [M + 1]; [α]25 D = +15.1° (c = 20 mg/mL, CHCl3) ([α]23 D = + 14° (neat), Sigma-Aldrich).

Fig. 6

Production of (R)-CHEB using co-expression E. coli cells in a water/octanol biphasic system with a substrate fed-strategy. Reduction of COBE using sonicated co-expressed E. coli cells (8 g DCW) was conducted at 30 °C in 0.5 L phosphate buffer (pH 6.5) containing glucose (1800 mmol) and COBE (600 mmol) and 0.5 L octanol. COBE (300 mmol) together with 3 equiv. glucose were added to the system at 6 h and 12 h, respectively

Discussion

During the past decade, the efficiency of asymmetrically biocatalytic synthesis catalyzed by SCRs has emerged as an environmentally sustainably alternative to traditional organo- and metallocatalysis due to the inherent high stereoselectivity of the biocatalysts and the mild reaction conditions. However, there are still impeded in their large-scale application due to the limited commercially robust biocatalysts, the narrow substrate specificity and the dependency of expensive cofactor as well as the water-insoluble substrates [20]. With the development of genomics, proteomics, and bioinformatics, genome hunting and data mining have become powerful tools for exploiting novel enzymes [21, 22]. In the present study, on the basis of screening results of microorganisms, and the combination of the BLASTp searching using the reported carbonyl reductase (GenBank accession no. AAW61772.1), a toolbox contained 35 candidates predicted as carbonyl reductases was successfully constructed, and further subjected to the asymmetric synthesis of enantioenriched (R)-CHBE.

Optically active (R)-CHBE is a useful chiral building block for L-carnitine, a biomolecule capable of modulating the serum oxidation stress and lipid profile [23], and is synthesized through the asymmetrical reduction of COBE. A SCR from G. oxydans 621H was reported to provide for accessing (R)-CHBE, but with low substrate concentration (6 mM) [14]. Using an AKR from S. salmonicolor, as much as 268 g L−1 of COBE was converted to (R)-CHBE yet with moderate enantioselectivity (only 91.7 % ee) [15]. A reductase from Bacillus sp. ECU0013 was capable of producing (R)-CHBE with high substrate loading (215 g L−1) and enantioselectivity (99.6 % ee), however, it required high S/C and only small-scale was accomplished [18]. In this study, a robust SCR designed as BgADH3 (GenBank Accession No. AEA63541) from B. gladioli CCTCC M 2012379 demonstrated high catalytic efficiency and excellent enantioselectivity thus was considered as a promising SCR for the asymmetric synthesis of optically active (R)-CHBE, accomplishing an economic and environmental balance.

Considering the majority of pairwise similarities between SCR members are in the range of 15–30 %, and the presence of conserved NADP-binding motiff TGXXXGXG as well as N-S-Y-K catalytic terads [24], BgADH3 was defined as the ‘classical’ short-chain dehydrogenases/reductases (SDR). Although the well-known Lactobacillus brevis SDR exhibited strong Mg2+ dependency [25], SDRs generally does not require metal ions. In this case, the metal chelator EDTA did not impact the activity, and most of the tested metal ions did not significantly improved BgADH3 activity, revealing that BgADH3 is typical of non-metal SDR. Several SCRs have been described capable of oxidizing both NADH and NADPH in the asymmetric reduction [26, 27]. In this study, cofactor preference of BgADH3 was evaluated using COBE as substrate, indicating that BgADH3 was not completely specific to NADH or NADPH as well. Due to the higher catalytic efficiency, NADPH was selected as the cofactor in the reduction of COBE using BgADH3. Notably, the catalytic ability of BgADH3 was approximately 5-fold improvement compared to that found on the literature for gox2036, an NADH-dependent SDR provided access to (R)-CHBE [14].

Enzymes generally require aqueous media in which the hydrophobic substrates are poorly soluble and even unstable. A straightforward solution to this issue might be the application of an aqueous organic solvent system [28, 29]. There are several documented cases in which organic solvents can affect the enzymatic activity [30-33]. Compared to hydrophobic solvents, hydrophilic ones are usually easier impact the activity due to the greater tendency of stripping tightly bound water in the enzyme molecules [34]. Herein, batteries of water-immiscible solvents were tested for the influence on the activity of BgADH3. High activity observed in the presence of hydrophobic solvents with high log P value was consistent with the recent study of organic tolerance of alcohol dehydrogenases [35]. In contrast, the stability of BgADH3 showed no dependence on the hydrophobicity of solvents. This finding is in agreement with the assumption that solvent functionality is also significance for enzyme deactivation [36]. Thus, it was established that log P was not an adequate parameter to describe the trend of enzyme tolerance to organic solvents. Although BgADH3 exhibited high activity and stability in the presence of cyclohexane, n-hexane, n-heptane, and iso-octane, COBE and (R)-CHBE exhibited low solubility due to the poor partition coefficient thus were not suitable for the application in aqueous organic solvent system. Butyl acetate has been frequently introduced as the non-aqueous phase in the reduction of COBE [2, 9], however, BgADH3 was strongly inactivated in this solvent. In this case, octanol was eventually chosen as the organic phase in the biphasic system.

Different position and size of the substituents in carbonyl compounds play an critical role in the enzyme activity [37]. With regard to acetophenone derivatives, the ortho-substituted group (4) or multi-substitution (5 and 8) reduced the BgADH3 activity due to the steric hindrance effects on the hydrogen attack from NADPH to the carbonyl group. Likewise, the bulky substituents in ketoesters adjacent to the carbonyl group also have negative effects. BgADH3 generally obeys Prelog’s rule with S preference by the hydride transfer from the cofactor to the Re-face of the carbonyl group during the enzymatic reductions, similar to the majority of known SCRs [14]. The inverted stereochemical assignment observed toward 12–14 and 17 was explained by a higher Cahn-Ingold-Prelog priority in the smaller side of substrate than the large one, thus still followed the Prelog’s rule of hydride delivery. However, the reduction of 15 catalyzed by BgADH3 also afforded the (R)-enantiomer, suggesting that bulky substituent might affect the configuration of the corresponding alcohol.

Although the upscale application of SCRs was impeded by the requirement for expensive cofactors such as NAD(P)+/NAD(P)H, enzyme-coupled and substrate-coupled cofactor recycling strategies have been developed for surmounting this challenge [38-40]. Compared to the substrate-coupled cofactor recycling with reversibility and poor thermodynamic driving force, enzyme-coupled cofactor regeneration exhibited more efficient and inexpensive. In this study, GDH from Exiguobacterium sibiricum 255–15 (GenBank: ACB59697.1) and glucose was successfully coupled with BgADH3 in the biosynthesis of (R)-CHBE. Expressing SCR and GDH in same cell is the possibility to avoid using an additional expensive cofactor when performing the biotransformation with high substrate loading [41-43]. Herein, the BgADH3 and GDH genes were introduced into the pCDFDuet-1 vector to construct the co-expressed plasmid pCDFDuet-1-BgADH3-GDH and strain E. coli/pCDFDuet-1-BgADH3-GDH. The yield was enhanced approximately 17 % in the presence of 1.0 μmol NADP+/(mmol COBE) or more compared to that in the absence of external NADP+, likely because the amount of cofactor in vivo is insufficient to this reaction. Based on cost considerations, the ratio of NADP+/COBE (μmol/mmol) was limited to 1.0. Reaction pH and temperature play important roles in biotransformation, which can significantly affect the activity and stability of biocatalysts, the speed of substrate molecules motion, and the activation energy of the bioreduction. The optimum temperature of co-expressed E. coli was inconsistent with that of single BgADH3. This might be explained that the high temperature impact the GDH activity before affect the BgADH3 activity. The substrate loading was considerably improved with the increased of cell dosage at a fixed S/C (mmol/g DCW) of 150, while the yield was significantly reduced when the cell dosage was too high due to the increased viscosity in the reaction media which might be hinder the molecular transfer between substrate and enzyme [1]. To the best of our knowledge, this is the first report of a preparative-scale highly stereoselective reduction of COBE to (R)-CHBE by using a newly cloned BgADH3 co-expressed with GDH. From a practical viewpoint, this asymmetric synthesis of (R)-CHBE is promising since no example reported to produce (R)-CHBE with more than 1200 mmol of in 1-L reaction system yet (Table 4).

Table 4

Comparison of several reported biocatalysts for producing (R)-CHBE

Enzyme	SsAR	SsALR	CpSADH	BYueD	Gox2036	LEK	BgADH3
Family	AKR	AKR	MDR	-	SDR	AKR	SDR
Solvent system	water/n-butyl acetate	water	water	water/toluene	water	water	water/octanol
Cofactor enzyme	GDH	GDH	/	GDH	GDH	GDH	GDH
Cosubstrate	G	G	IPA	G	G	G	G
External cofactor	NADP+	NADP+	Without	NADP+	NAD(H)	Without	NADP+
Substrate loading (mmol)	91	0.61	1.525	13	0.06	111	1200
Biocatalyst (g)	2 (wet)	0.4 (wet)	/	0.5 (dry)	0.1 (wet)	17 (dry)	8 (dry)
Reaction volume (mL)	50	10	25	20	10	/	1000
Reaction time (h)	50	10	17	5	24	1	20
ee of (R)-CHBE	91.7	99	99	99.6	100	>99	99.9
References	[15]	[48]	[49]	[18]	[14]	[17]	This work

Conclusions

In summary, a SCR toolbox was constructed in this work through genome hunting and data mining approaches based on bioinformatics analysis of sequence-similarity with known stereoselective SCR, and further subjected to the asymmetric synthesis of (R)-CHBE. The newly cloned BgADH3 from B. gladioli displayed high activity and excellent enantioselectivity toward COBE, and served as a versatile SCR with a broad substrate spectrum converting varied aromatic ketones and ketoesters to the corresponding chiral alcohols with excellent enantiomeric purity (>99 % ee). Furthermore, asymmetric synthesis of (R)-CHBE was developed by using recombinant E. coli cells co-expressing BgADH3 and GDH. Using substrate fed-batch mode and aqueous/octanol biphasic system, the decomposition of COBE and its inhibition of the biocatalyst were significantly alleviated. Using the “designer cells”, (R)-CHBE was produced at a high space-time yield per gram of biomass of 4.47 mmol∙L−1∙h−1∙g DCW−1 and excellent enantiopurity (99.9 % ee) with the highest substrate total loading of 1200 mmol in 1-L reaction system reported so far. This study allows for the further application of BgADH3 in the biosynthesis of chiral alcohols, and establishes a preparative scale process for producing (R)-CHBE with excellent enantiopurity.

Methods

Strains, plasmids and chemicals

The E. coli DH5α (Tiangen biotech Co., Ltd., Beijing, China) and E. coli BL21 (DE3) (Novagen, Darmstadt, Germany) were selected as hosts in this study. Plasmid pGEM-T (Promega, Beijing, China) and pET28a (Novagen, Darmstadt, Germany) were used for cloning and expression, respectively. Polymerase chain reaction (PCR) mix and restriction endonucleases were purchased from Vazyme (Nanjing, China). Ampicillin and kanamycin were obtained from solarbio (Beijing, China). Streptomycin and IPTG were purchased from Sangon Biotech (Shanghai, China). COBE and (R)-CHBE were provided by J&K Scientific Ltd. (Shanghai, China). All other chemicals were of analytical grade and commercially available.

Expression of SCRs in E. coli cells

Strain CCTCC M 2012379 was cultivated as described previously [44]. The genomic DNA of this strain was extracted using a FastDNA® Spin Kit for Soil (MPBio, Shanghai, China). The genes of SCRs were amplified by PCR using appropriate primers designed according to the putative proteins from B. gladioli BSR3 (Additional file 7: Table S5). Each amplified DNA fragment was inserted into pGEM-T to create pGEM-T-SCR, and digested with restriction endonucleases. The desired fragment was subcloned into linearied pET28a with the same restriction enzyme cutting sites and transformed in E. coli BL21 (DE3). Other SCRs genes were synthesized and transformated in E. coli BL21 (DE3).

Cell growth and enzyme purification

The recombinant E. coli strains with SCRs were cultivated, induced, and purified as described previously [45]. The protein expression level was verified by SDS-PAGE. The native molecular weight was measured using a Discovery BIO GFC 150 (300 × 7.8 mm, 3 μm) column (Sigma-Aldrich, Shanghai, China) equilibrated with phosphate buffer (150 mM) at 1.0 mL/min using cytochrome c (12.4 kDa), ovalbumin (43 kDa), bovine serum albumin (66 kDa), and yeast alcohol dehydrogenase (150 kDa) as standard proteins. Protein concentration was determined using BCA Protein Assay Kit (KeyGEN BioTECH, Nanjing, China) with bovine serum albumin (BSA) as a standard.

Enzyme assays

Specific activity was detected spectrophotometrically by monitoring the depletion in the absorption of NAD(P)H at 340 nm. One unit (U) of enzymatic activity was defined as 1 μmol of NAD(P)H consumed per minute under the assay conditions. Each assay mixture contained COBE (50 mM) and NAD(P)H (0.5 mM) in phosphate buffer (100 mM, pH 6.5). The activity was determined after the addition of NAD(P)H for 2 min. The same method was used in all reactions against other substrates.

Characterization of recombinant BgADH3

Specific activities were evaluated under standard assay conditions using COBE as substrate. The effect of pH on the activity of BgADH3 was assessed within a pH range of 3.5–10.5, using disodium hydrogen phosphate-citrate buffer (3.5–7.0), phosphate buffer (6.0–8.0), Tris–HCl buffer (7.5–9.0), and Gly-NaOH buffer (9.0-10.5). The pH stability was tested by pre-incubating purified BgADH3 in buffers with different pH values at 30 °C for 2 h, and then the residual activity was measured. The non-incubated enzyme was considered as a control. The effect of temperature on the activity was studied by assaying activities at temperatures ranging from 25 °C to 65 °C. The thermostability was estimated by incubating the purified BgADH3 in the phosphate buffer (100 mM, pH 6.5) at 25, 35, 45, 55, and 65 °C, respectively. Samples were withdrawn per hour and the residual activities were detected. The non-heated enzyme was taken as a control. The effect of metal ions on the activity was evaluated by pre-incubating recombinant BgADH3 in the presence of Fe2+, Ni2+, Fe3+, Ca2+, Ba2+, Cu2+, Mn2+, Zn2+, Co2+, Mg2+, Ag+, and EDTA (2 mM) at 30 °C for 30 min. Measurements of kinetic parameters were done by altering the concentration of NADH or NADPH (0.01–0.5 mM) at a constant COBE concentration (50 mM), and were calculated through nonlinear regression of the Michaelis-Menten equation. The influence of organic solvents on the activity was assessed by adding hydrophobic solvents (50 %, v/v), including tetrahydrofuran (THF), ethyl acetate, n-butanol, tert-amylalcohol, MTBE, dichloromethane, butyl acetate, toluene, octanol, cyclohexane, n-hexane, n-heptane, and iso-octane. Stability was determined by measuring the residual activity after incubation with organic solvents at 30 °C for 30 min. All assays were performed in triplicate.

Stereoselectivity

The stereoselectivity of BgADH3 was estimated by testing the asymmetric reduction of aryl ketones and ketoesters using glucose as the cosubstrate. Each reaction mixture was comprised of phosphate buffer (100 mM, pH 6.5), substrate in DMSO (50 mM, 5 % v/v), NADPH (0.05 mM), glucose (278 mM), GDH (0.1 mg mL−1), and purified BgADH3 (0.1 mg mL−1) in a total volume of 0.5 mL, and proceeded at 40 °C for 6 h. Reaction mixture without recombinant BgADH3 was used as the control. After complete of the reaction, each biotransformation solution was extracted with ethyl acetate. The ee or de values of products were determined by GC or HPLC analysis.

The pGEM-T-GDH was constructed as previously reported [9], and digested with Nde I and Xho I restriction enzymes. The fragment of interest was induced into the Nde I/Xho I site (MCSI) of plasmid pCDFDuet-1 to afford recombinant pCDFDuet-1-GDH vector. Subsequently, the BgADH3 gene cut from pGEM-T-BgADH3 with Nco I and Hind III was ligated to the Nco I/Hind III site (MCSII) of pCDFDuet-1-GDH, leading to the final recombinant plasmid pCDFDuet-1-BgADH3-GDH harboring BgADH3 and GDH genes. The resulting co-expressed plasmid was further transformed into E. coli BL21 (DE3) and grown in LB medium containing streptomycin (50 μg mL−1). The expression levels of this recombinant genes were verified by SDS-PAGE.

Efficient biosynthesis of (R)-CHBE in aqueous-organic solvent system

Screening of organic solvents

Different organic solvents were evaluated in the asymmetric synthesis of (R)-CHBE using co-expression E. coli/pCDFDuet-1-BgADH3-GDH at 40 °C for 6 h by mixing an equal volume of the tested solvents with 10 mL phosphate buffer (100 mM, pH 6.5) consisted of COBE (6 mmol), glucose (12 mmol), NADP+ (6 μmol), and sonicated co-expressed E. coli cells (0.04 g DCW). The pH of the reaction mixture was monitored at 6.5 with 2 M NaOH during reaction. The ee and quantity of COBE and (R)-CHBE were determined by GC [2].

Several reaction parameters (e.g., glucose concentration, NADP+ concentration, reaction pH, reaction temperature, cell dosage and substrate loading, etc.) affected the biotransformation were assessed by adding different concentrations of glucose (1–9 mmol glucose/(mmol COBE)), NADP+ (0–2 μmol/(mmol COBE)), cell dosage (0.02–0.1 g DCW), and COBE (6–48 mmol) in 10 mL phosphate buffer mixed with an equal volume of octanol. Unless otherwise stated, each assay was stirred at 40 °C for 6 h. The effects of reaction temperature and pH on the biosynthesis were performed at various temperatures (25–60 °C) and different pHs (4.0–9.0).

Substrate fed-batch was carried out in 1-L reaction system, which comprising phosphate buffer (100 mM, pH 6.5), COBE (600 mmol), glucose (1800 mmol), and sonicated E. coli/pCDFDuet-1-BgADH3-GDH (8 g DCW), in a final volume of 0.5 L, was mixed with an equal volume of octanol in a bioreactor with two electrodes for monitoring the temperature and pH. The resulting mixture was stirred at 200 rpm for 20 h. The temperature was maintained at 30 °C and the pH was adjusted to 6.5 with 2 M NaOH during the reaction. COBE (300 mmol) and glucose (900 mmol) were added to the reaction mixture at 6 and 12 h, respectively. After reaction, the two layers were separated and the aqueous layer was extracted twice with ethyl acetate. The extracted layers were combined with the original organic layer, washed with NaHCO3, and dried by anhydrous Na2SO4. The obtained solvent was evaporated under vacuum, offering the final product as oily liquid. High-resolution mass spectrum (MS) was obtained on an Agilent 6210 TOF LC/MS system (Palo Alto, CA, USA). Nuclear Magnetic Resonance (NMR) spectra of the final product were performed on Bruker AVANCE III (Bruker, Switzerland) spectrometer. Polarimetry were carried on Rudolph Autopol IV polarimeter (Rudolph, USA) using the sodium D line (589 nm) at 25 °C.