Development of artificial enzyme cascade for the conversion of styrene to (R)-mandelic acid
Previously, a two-enzyme cascade comprising styrene monooxygenase (SMO, consisting of StyA and StyB) from Pseudomonas sp. VLB120 and epoxide hydrolase (StEH) from Solanum tuberosum was developed for converting styrene to (R)-PED. Here, we sought to extend this enzyme cascade to produce (R)-MA. Alditol oxidase (AldO) from Streptomyces coelicolor A3(2) was chosen for the conversion of (R)-PED to (R)-MA (Dominic P. H. M. Heuts 2007; van Hellemond et al. 2009). AldO gene was cloned in pRSFDuet-1 plasmid and transformed into E. coli T7 strain to give E. coli (AldO). Biotransformation of (R)-PED to (R)-MA was investigated with E. coli (AldO) cells (10–15 g cdw/L) in KP buffer system (200 mM, pH 8.0) at 30 °C for 24 h, and 15 g cdw/L of cells gave the best results. The time course is shown in Fig. 2a: (R)-MA was produced at a linear rate within the first 8 h with the fast consumption of (R)-PED. After 8 h of biotransformation, 1.9 g/L of (R)-MA was produced, giving a space–time yield of 237.5 mg/L/h. These results clearly demonstrated AldO catalyzed the conversion of (R)-PED to (R)-MA. As the reaction continued, the reaction rate gradually dropped down and 2.43 g/L (16 mM) of (R)-MA was produced at 24 h in 80% yield, while 600 mg/L (4 mM) of (R)-PED remain unreacted. This is probably because of high Km (101 mM) toward (R)-PED (van Hellemond et al. 2009) and relatively low activity of AldO. Enzyme evolution might be needed for increasing its affinity with (R)-PED as well as the activity. Although H2O2 could be produced during the reaction with 20 mM substrate, the low level of the produced H2O2 in cells did not influence the enzymatic reactions. This is probably because the E. coli cells produced endogenous catalase which could decompose H2O2. The co-expression of the catalase from E. coli with AldO was also tried; however, the product yield was not increased. pRSF-AldO were co-transformed together with pCDF-SMO-StEH to give E. coli (S2A) co-expressing SMO, StEH, and AldO for the direct conversion of styrene to (R)-MA.
Bioproduction of (R)-mandelic acid from styrene via whole cell biotransformation in n-hexadecane-aqueous two-phase system
Cascade biotransformation of styrene to (R)-MA was conducted using E. coli (S2A) cells (15 g cdw/L) in a mixture of KP buffer (200 mM, pH 8.0) containing 0.5% glucose and n-hexadecane (1:1; v/v) at 30 °C for 24 h. The organic phase functions as a reservoir for toxic substrate styrene and intermediate styrene oxide. The time course on the production of (R)-MA from styrene is shown in Fig. 2b: as time went on, styrene concentration rapidly decreased and the concentration of (R)-MA and (R)-PED increased gradually. 1.07 g/L of (R)-MA was successfully obtained within the first 10 h, which corresponds to a space–time yield of 107 mg/L/h. After 24 h reaction, 1.52 g/L (10 mM) of (R)-MA was produced from 2.08 g/L (20 mM) of styrene, together with the remaining (R)-PED (800 mg/L; 6 mM) and styrene (300 mg/L; 3 mM). A small amount of mass imbalance (about 1 mM) was observed, which is probably due to styrene evaporation.
Development of artificial enzyme cascade for the conversion of L-phenylalanine to (R)-mandelic acid
To develop an enzyme cascade for producing (R)-MA from L-Phe, deamination reaction catalyzed by phenylalanine ammonia lyase (PAL) from Arabidopsis thaliana and decarboxylation reaction catalyzed by phenylacrylic acid decarboxylase [PAD, consisting of ferulic acid decarboxylase (FDC1) and phenylacrylic acid decarboxylase (PAD1)] from Aspergillus niger were introduced into above mentioned epoxidation-hydrolysis-double oxidation artificial enzyme cascade, yielding a new five-enzyme cascade.
To engineer E. coli strains expressing the five-enzyme artificial cascade, the previously constructed pET-PAL-PAD (Zhou et al. 2016) were transformed into the E. coli (S2A) cells to give E. coli (P2S2A). The enzyme expression in E. coli (P2S2A) was analysed by SDS-PAGE (Additional file 1: Figure S19). The corresponding protein bands of PAL, PAD, SMO, StEH, and AldO were clearly observed, indicating these enzymes were well-expressed.
Bioproduction of (R)-mandelic acid from L-phenylalanine via whole cell biotransformation in n-hexadecane-aqueous two-phase system
Cascade biotransformation of L-Phe to (R)-MA was conducted at 30 °C and 250 rpm using E. coli (P2S2A) cells (15 g cdw/L) in a mixture of KP buffer (200 mM, pH 8.0) containing 0.5% glucose and 2.48 g/L (15 mM) L-Phe and n-hexadecane (1:1; v/v). The organic phase functions as a reservoir for toxic intermediates such as styrene and styrene oxide. The time course of reaction is shown in Fig. 2c. With the rapid reduction of L-Phe, styrene, and (R)-PED and (R)-MA gradually increased. After six hours of the biotransformation, 650 mg/L of (R)-MA was produced, giving a space–time yield of 108.3 mg/L/h. 913 mg/L (6 mM) of (R)-MA was obtained after 24 h biotransformation of 2.48 g/L (15 mM) L-Phe. These results indicated the successful production of (R)-MA from L-Phe by the newly developed five-enzyme cascades. Considering the remaining L-Phe (3 mM), styrene (1 mM), and (R)-PED (4 mM), a small amount of mass imbalance (about 1 mM) was observed, which is also possibly due to styrene evaporation. The (R)-MA titer is much higher than those previously reported (236 mg/L in yeast cells and 680 mg/L in E. coli cells) (Reifenrath and Boles 2018; Sun et al. 2011). The biotransformation was also performed using 5 mM or 10 mM L-Phe under the same reaction condition, 3 mM and 6 mM of (R)-MA was produced with 60% conversion in both cases, which is higher than the conversion obtained from 15 mM L-Phe. Although the product titer is much higher than the previously reported, it is yet insufficient for commercial application. The expression of AldO was in a relatively low level when many different enzymes were co-expressed in E. coli cells (Fig. 1d). In addition, AldO showed the insufficient activity. These were probably the reasons for the accumulation of (R)-PED. Further optimizations, such as the expression of different enzymes in high level and at the good ratio, as well as engineering of much more efficient enzyme, are needed to increase the conversion, enhance the product titer, and reduce the accumulation of intermediates.
Engineering of E. coli strains for the production of (R)-mandelic acid from glycerol or glucose
Direct fermentation of glucose or glycerol provides a prosperous approach for the green and sustainable biosynthesis of pharmaceuticals and high-value chemicals. For synthesis of (R)-MA from glycerol or glucose, the E. coli T7 host was changed into E. coli NST74 strain. The latter is a L-phenylalanine-overproducing strain, producing L-Phe from glycerol or glucose (Choi and Tribe 1982). The pRSF-AldO, pET-PAL-PAD, and pCDF-SMO-StEH were transformed into the previously engineered E. coli NST74(DE3) strain (Zhou et al. 2016) to give the E. coli NST74 (P2S2A) strain (Fig. 1c). The protein expression in E. coli NST74 (P2S2A) was analyzed by SDS-PAGE (Fig. 1d). Similar to the results from E. coli (P2S2A), all enzymes involved in the artificial enzyme cascade were well-expressed in E. coli NST74 (P2S2A) strain.
Fermentative production of (R)-mandelic acid from glycerol or glucose using a L-phenylalanine-overproducing E. coli strain expressing the five-enzyme artificial cascade
Fermentative production of (R)-MA from glycerol or glucose in single E. coli strain was first investigated in aqueous system. E. coli NST74 (P2S2A) strain was grown in M9 (with glucose as the carbon source) or TB medium (with glycerol as the carbon source) containing respective antibiotics (50 µg/mL kanamycin, 50 µg/mL streptomycin, and 100 µg/mL ampicillin) at 37 °C and 250 rpm. When OD600nm reached to ~ 0.6, 0.1 mM IPTG was added and then the cells were continued to grow at 25 °C for another 24 h. Samples were taken at different time points for monitoring the concentration of L-Phe, styrene and (R)-MA in medium. Fermentative biotransformation of glycerol (Fig. 3a) and glucose (Fig. 3b) successfully gave (R)-MA concentration of 75 and 48 mg/L, respectively. The fermentation of glycerol gave higher (R)-MA titer in comparison with the one performed with glucose as carbon source, suggesting that the cells tend to consume the glucose more for their growth than for the (R)-MA production. Notably, high accumulation of styrene in the aqueous phase was observed during the fermentation, which caused the toxicity to the cells.
To tackle the styrene toxicity issue, n-hexadecane-aqueous two-phase system was used for the fermentation. E. coli NST74 (P2S2A) cells were grown in M9 or TB medium under the above-mentioned conditions. After the addition of IPTG, n-hexadecane was added into the aqueous medium at a ratio of 1:2. The reaction was then conducted at 25 °C and 250 rpm for 24 h. Time courses of the fermentation are shown in Fig. 3c, d. The concentration of (R)-PED and (S)-styrene oxide were very low, as shown in Fig. 3, possibly caused by the low concentration of L-Phe produced from glycerol or glucose. Compared to previous single aqueous system; however, the two-phase system still significantly increased the product concentration. While the L-phenylalanine, (R)-PED, and (R)-MA were formed in the aqueous phase, the intermediates styrene and (S)-styrene oxide remained in the organic phase to avoid the toxicity issue. As the best results, 228 and 152 mg/L of (R)-MA was obtained from the fermentative biotransformation of glycerol and glucose in two-phase system, respectively, with the significant decrement of styrene accumulation.
Production of (R)-mandelic acid from glycerol or glucose by coupling of two E. coli strains expressing the artificial enzyme cascade and L-phenylalanine biosynthesis pathway, respectively
Production of (R)-MA from glycerol or glucose was examined by coupling E. coli NST74-Phe, which over-expressed five limiting enzymes to convert glycerol or glucose to L-Phe via Shikimate pathway (Sekar et al. 2019), and E. coli (P2S2A) expressing the novel artificial cascade to convert L-Phe to (R)-MA (Fig. 2c). E. coli NST74-Phe was first grown in modified NH4-media containing 10 g/L glycerol or glucose for 28 h to produce 12.87 g/L (Fig. 4a) and 11 g/L of L-Phe from glycerol and glucose, respectively. Freshly made E. coli (P2S2A) cells, concentrated KP buffer (pH 8.0), and glucose solution were directly added into fermentation medium, resulting in a reaction mixture containing 1.65 g/L of L-Phe, 200 mM KP buffer (pH 8.0), 1% glucose, and 15 g cdw/L of E. coli (P2S2A) cells. n-Hexadecane was added to give a ratio of 1:1 (v/v) for aqueous and organic phases. The biotransformation was performed at 30 °C and 250 rpm for 24 h. Nevertheless, 760 and 455 mg/L of (R)-MA was obtained from glycerol-fermented media and glucose-fermented media, respectively (Fig. 4b and Additional file 1: Fig. S18). While the optimal temperature for expression and production of L-phe from glycerol or glucose is 37 °C (Sekar et al. 2019), the optimal temperature for the expression of artificial enzyme cascade is 22 °C and for cascade biotransformation of L-Phe to (R)-MA is 30 °C. Therefore, the single-cell fermentation at 25 °C gave lower product concentration. In contrary, the two-cells system could be operated at their respective optimum temperatures to give higher productivity. The results with the use of glycerol as feedstock showed higher (R)-MA titer than those previously reported (236 mg/L in yeast cells and 680 mg/L in E. coli cells) (Reifenrath and Boles 2018; Sun et al. 2011).
Preparative Biotransformation of L-Phe to (R)-MA in aqueous-n-hexadecane two-phase system
Preparative biotransformation of L-Phe to (R)-MA was performed in two-phase system with E. coli (P2S2A) cells, yielding 6 mM (R)-MA in 24 h with 60% conversion. After purification, pure product was obtained with an isolated yield of 46%. 1H and 13C NMR analysis confirmed the chemical structure of (R)-MA products (Additional file 1: Figure S17). Chiral HPLC analysis showed > 99% ee of the produced (R)-MA (Additional file 1: Figure S16).