Introduction

Enzymatic biocatalysis is a process through which several valuable compounds and bulk chemicals are synthesized. Because this process is associated with lower levels of environmental pollution than typical synthetic processes, it has become an attractive alternative (Dalal et al. 2018). Several enzymatic reactions require flavin adenine dinucleotide and dihydroflavine adenine dinucleotide (FAD/FADH2), which function as electron carriers to supply or take up reducing or oxidizing equivalents. Flavoenzymes include single-component flavoenzymes that tightly bind the flavin cofactor and two-component flavoenzymes (predominantly monooxygenases, FPMOs) that use a flavin reductase and a diffusible flavin for catalysis (Paul et al. 2021). Most reported flavin-dependent halogenases are two-component FDHs, which require a flavin reductase as the reaction partner to function (Song et al. 2019). FADH2-dependent halogenase and FAD-dependent l-amino acid deaminase used in our study are a few two-component FPMO applied as biocatalysts. However, the high cost resulting from the addition of FAD/FADH2 has greatly hampered the application of FAD/FADH2-dependent enzymatic reactions, emphasizing the need for cost-effective regeneration systems (Hou et al. 2017).

The available nonenzymatic methods for FAD/FADH2 regeneration include electrochemical and chemical regeneration. In electrochemical regeneration, an artificial electrode is used, and H2O2 is produced as a by-product, resulting in the deactivation of the FAD/FADH2-dependent enzyme at the electrode surface, low selectivity, and poor productivity (Amongre and Gassner 2021). In chemical regeneration, an artificial organometallic complex [Cp*Rh(bpy)(H2O)]2+ is widely used for FADH2-dependent epoxidation and halogenation (Deng et al. 2020). Artificial nicotinamide adenine dinucleotide (NADH) mimics obtained from inexpensive starting materials in two synthetic steps were successfully used for the regeneration of FADH2 (replacement of flavin reductase) to perform the chlorination of l-tryptophan (Trp) (Ismail et al. 2019). However, the coupling efficiency of Trp halogenases was low, and large amounts of NADH mimic were consumed due to the competing formation of H2O2, which limited the applicability of the preparative scale reaction. Moreover, in halogenation reactions wherein an NADH analog acts as the FAD cofactor regeneration system, the generated FADH2 reacts with chloride to produce hypochlorous acid for substrate attack. Depending on the enzyme and the reaction conditions, there is a competitive reaction, in- or outside the enzyme where FADH2 reacts with oxygen to produce hydrogen peroxide because of futile decomposition of the peroxyflavin. The reactions compete with each other, resulting in the consumption of a large amount of the NADH analog. Higher concentrations of FAD also promote the aerobic reoxidation of FADH2 to form H2O2, and only lower concentrations of FAD (20–50 μm) can be used. The accumulation of H2O2 can destabilize the activities of mimics and halogenases, although catalase can be added to the reaction to prevent H2O2 accumulation (Ismail et al. 2019).

The construction of enzymatic systems for FAD/FADH2 regeneration requires the combination of a flavin reductase with an alcohol/formate dehydrogenase, which together regenerate NADH from NAD+ by catalyzing the oxidation of alcohol/formate to CO2, resulting in a trienzymatic cascade (Minges and Sewald 2020). It has been proven that the total turnover and reaction rates of the trienzymatic cascade are higher than those seen in nonenzymatic regeneration coupling of biocatalytic asymmetric epoxidation with NADH regeneration in organic-aqueous emulsions (Hofstetter et al. 2004). Therefore, many effective enzymatic regeneration systems have been constructed to regenerate FAD/FADH2. For FADH2-dependent epoxidation in a biphasic system, formate dehydrogenase and formate have been used for NADH regeneration. Effective epoxidation requires the co-expression of two separate enzymes (StyA and StyB) in Escherichia coli. StyB transfers the reduction equivalent from NADH to FAD to generate FADH2. FADH2 and oxygen are used as co-substrates of StyA epoxidized olefins to generate FAD-OOH. The final epoxidation process of styrene substrate is completed at the active site of StyA to realize the regeneration of oxidized FAD (Corrado et al. 2018). For FADH2-dependent halogenation, the cross-linking of precipitated Trp-7-halogenase RebH, flavin reductase PrnF, and an alcohol dehydrogenase leads to the production of multifunctional and recyclable cross-linked enzyme aggregates, which can be applied for the regioselective halogenation of Trp to 7-bromo-tryptophan on the gram scale (Frese and Sewald 2015). For FAD-dependent oxidative deamination, we previously introduced formate dehydrogenase and NADH oxidase to accelerate FAD regeneration from FADH2, resulting in increased whole-cell catalytic ability (Hou et al. 2017). l-amino acid deaminase (l-AAD) is a protein containing flavin adenine dinucleotide (FAD), which has some similar properties with PmaLAAD. For example, they need to tight binding of the FAD to play an active and transmembrane role α- helix interacts with the membrane. l-AAD can also transfer electrons to cytochrome electron acceptor. However, trienzymatic cascade catalysis may have the characteristics of low catalytic efficiency and difficult to regulate due to the larger number of enzymes needed than the two enzymes catalysis, which limits the application of the trienzymatic cascade catalysis to some extent. Accordingly, a more efficient and simpler FAD/FADH2 regeneration approach is needed.

The current central problem in the cascade catalytic system is the imbalance caused by different intermediate specific activities or expression levels of the multiple heterologous enzymes, resulting in substrate accumulation and/or decreased production (Muschiol et al. 2015; Yi et al. 2021). Currently, the expression levels are controlled by adjusting the strength of promoters and ribosome-binding sites (RBSs) (Jiang and Fang 2016), altering the number of gene copies (Yuan et al. 2018), and using compatible plasmids with different copies (Hou et al. 2017). In our previous studies, a combination of these approaches was successfully used to improve the catalytic abilities of an E. coli whole-cell catalyst with metabolic engineering of FAD/FADH2 supply and its regeneration for the production of α-keto acids. It was also successfully applied in an E. coli whole-cell catalyst with multi-enzyme expression for the production of phenyllactic acid (Hou et al. 2017, 2019). Therefore, combination of these approaches appears to be an efficient strategy for coordinating the expression levels of multiple enzymes and improving overall efficiency.

In the present study, we constructed an FAD/FADH2 regeneration system using a dual-enzyme cascade called CombiAADHa, which included a FADH2-dependent halogenase and an FAD-dependent l-amino acid deaminase (l-AAD). Halogenase catalyzes the halogenation reaction of Trp to 7-chloro-tryptophan (7-Cl-Trp), which is widely used in the pharmaceutical and agricultural industries. l-AAD catalyzes the stereospecific oxidative deamination of l-amino acids to their corresponding α-keto acids, which are widely used in the pharmaceutical, food, and chemical industries. The electron transfer occurs between the reduced FADH2 and the oxidized FAD, thus providing coenzyme FADH2 and FAD for halogenation reaction and oxidative deamination reaction (Fig. 1). First, a cell-free biotransformation system was constructed and optimized to ensure the feasibility of CombiAADHa for FAD/FADH2 regeneration. Then, a whole-cell CombiAADHa system was constructed, and the activity ratio was optimized using gene duplication and RBS regulation. Finally, ultrasound treatment was applied to improve the membrane permeability and adjust the activity ratio.

Fig. 1
figure 1

Schematic representation of the constructed FAD/FADH2 regeneration system using a dual-enzyme cascade called CombiAADHa which included a FADH2-dependent halogenase and a FAD-dependent l-AAD

Full size image

Materials and methods

Bacterial strains, chemicals, and culture conditions

Wild-type l-AAD (Genbank: U35383.1) was used in this study (Hou et al. 2015). Wild-type halogenase RebH from Lechevalieria aerocolonigenes (PDB ID: 2OAM) (Genbank: AJ414559) and the triple mutant gene 3-LSR (S130L/N166S/Q494R) were synthesized by Shanghai Sangon Biological Engineering Technology and Services Co. Ltd. (Shanghai, China). E. coli BL21(DE3), pRSFDuet-1, and pETDuet-1 were purchased from Novagen (Madison, WI, USA). The enzymes (PrimeSTAR HS DNA polymerase, BamHI, HindIII, NdeI, and XhoI) and kits (DNA purification kit, plasmid isolation, DNA ligation kit, and competent cell preparation kit) were supplied by Takara (Dalian, China). All chemical reagents were purchased from Shanghai Sangon Biological Engineering Technology and Services Co. Ltd. (Shanghai, China), except FAD, 7-Cl-Trp, and indole pyruvic acid (IPA), which were purchased from Sigma-Aldrich (Shanghai, China).

Purification of halogenase and isolation of membrane fractions

A synthetic plasmid containing the halogenase genes was digested with NdeI and XhoI and ligated to the pRSFDuet-1 vector. The resulting constructs, pRSFDuet-RebH and pRSFDuet-lsr, were transformed into E. coli BL21(DE3) cells. The insertion of the halogenase genes into the transformants (pRSFDuet-RebH and pRSFDuet-lsr) was confirmed using polymerase chain reaction (PCR).

Terrific Broth medium was prepared as follows (per liter): 12 g tryptone, 24 g yeast extract, 12.54 g K2HPO4, 2.3 g KH2PO4, 5 g glycerol, and 100 mg kanamycin. The recombinant strains E. coli BL21(pRSFDuet-RebH) and E. coli BL21(pRSFDuet-lsr) were inoculated in 500 mL of Terrific Broth containing kanamycin (100 μg/mL). The induction conditions were as follows: pH 8, isopropyl-β-d-thiogalactopyranoside (IPTG) 0.04 mM, optical density (OD) of 0.6 at 600 nm, and temperature of 20 °C for 24 h. All purification steps were performed at 4 °C. The cells were obtained via centrifugation (Beckman Instruments, San Jose, CA, USA) at 8000 × g for 20 min and washed twice with 20 mM phosphate buffer (39 mL of 20 mM NaH2PO4·H2O + 61 mL of 20 mM Na2HPO4·7H2O; pH 7.0). A 100-mL portion of harvested cells was suspended in the same volume of binding buffer (20 mM PB, 500 mM NaCl, and 5 mM imidazole) and ultrasonicated for 20 min. Whole cells and cell debris were removed by centrifugation at 8,000 × g for 5 min, and His Mag Sepharose™ Ni affinity beads (50 µL; BEAVER, Jiangsu, China) were then added to the supernatant. The bead–protein mixture was incubated for 30 min and then washed twice with wash buffer (20 mM PB, 500 mM NaCl, and 50 mM imidazole). Halogenase was eluted using elution buffer (20 mM PB, 500 mM NaCl, and 500 mM imidazole), and then desalted and concentrated using an Ultra-4 Centrifugal Filter Device (Amicon, Pineville, NC, USA). The collected fractions were analyzed via sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), and the protein concentration was measured using a BCA protein assay kit (TianGen, Beijing, China). The membrane fraction was isolated by ultracentrifugating the cell lysate (100,000 × g, 1 h, 4 °C) of E. coli BL21(pRSFDuet-laad) cells and washed twice with 20 mM PB.

Enzyme activity assay

An l-AAD assay solution (50 μL) containing 20 mM Trp, 0.1 mM FAD, and 0.6 g/L cell (dry cell weight [DCW], w/v) in 50 mM PB buffer (pH 7.0) was kept at 40 °C for 0.5 h. The levels of α-keto acids (phenylpyruvic acid, α-ketoisocaproate, and indole pyruvic acid) were measured using high-performance liquid chromatography (HPLC), as described previously (Hou et al. 2017). The HPLC system (Agilent 1200, Palo Alto, CA, USA) was equipped with an ultraviolet detector (210 nm) and an Aminex HPX-87H Column (300 × 7.8 mm; Bio-Rad Laboratories Inc., Hercules, CA, USA) and was kept at a column temperature of 40 °C. The mobile phase was 5 mM sulfuric acid, and the flow rate was 0.6 mL·min−1. The specific activity was defined as the amount of enzyme required to catalyze the conversion of 1 μM of l-amino acid to an α-keto acid at 25 °C in 1 min.

The halogenase assay solution (50 μL) contained 1 mM Trp, 0.1 mM FAD, 10 mM NaCl, and 0.6 g/L recombinant cell (DCW, w/v) in 50 mM PB buffer (pH 7.0) and was kept at 28 °C for 1 h. The samples were analyzed using HPLC, based on a previously described protocol study (Veldmann et al. 2019). The specific activity was defined as the amount of enzyme required to catalyze the conversion of 1 μM of Trp to 7-Cl-Trp at 25 °C in 1 min.

Preparation of the biocatalyst

Luria–Bertani medium was prepared as follows (per liter): 10 g tryptone, 5 g yeast extract, 10 g sodium chloride, and 100 mg kanamycin. Modified MS medium was prepared as follows (per liter): 20.0 g glucose, 3.8 g Na2HPO4, 1.5 g KH2PO4, 1.0 g (NH4)2SO4, 0.2 g MgSO4, 5.0 g yeast extract, 2% (v/v) trace element solution (KI, H3BO3, MnSO4·4H2O, ZnSO4·7H2O, Na2MoO4·2H2O, CuSO4·5H2O, and CoCl2·6H2O), and 100 mg kanamycin. Recombinant E. coli were inoculated into 20 mL Luria–Bertani medium and grown overnight in a rotary shaker at 37 °C. A seed culture (1%, v/v) was inoculated into 50 mL of the modified MS medium for expression. The induction conditions were as follows: pH 8.0, IPTG 0.04 mM, OD600 0.6, and 20 °C for 12 h. The cells were isolated via centrifugation at 8,000 × g and 4C for 20 min and washed twice with 20 mM PB (pH 7.4). Then, they were resuspended in 20 mM PB (pH 7.4). Biomass concentration was evaluated spectrophotometrically (UV-2450 PC; Shimadzu Co., Kyoto, Japan) at a wavelength of 600 nm.

Cell-free biotransformation

The initial rate was determined by measuring the production of 7-Cl-Trp under standard reaction conditions. A mono-amino acid reaction system (pH 7.4) containing 2 g/L Trp, 1 U of l-AAD, 1 mM FAD, and halogenases with different activities was prepared in a total volume of 2 mL. The mixture was incubated in a shaker at 35 °C for 6 h. The dual-amino acid reaction system contained an extra 2 g/L of Phe or Leu. The optimal activity ratio of l-AAD:halogenase was determined over a range of values from 1:10 to 1:150. The concentration of H2O2 was determined by spectrophotometry at 415 nm using an H2O2 detection kit (Shanghai Sangon Biological Engineering Technology and Services Co. Ltd., Shanghai, China).

Construction of recombinant strains

To clone the l-AAD gene, the recombinant plasmid pET20b-laad constructed in a previous study was used as the template (Hou et al. 2016), and BamHI and HindIII restriction sites were added using PCR with the primers laad-F and laad-R (Table 1). The PCR product was digested with BamHI and HindIII and ligated to the pRSFDuet-lsr vector. The resulting construct, pRSFDuet-lsr-laad, was transformed into E. coli BL21(DE3) cells for expression, and the insertion of laad into the transformants was confirmed with PCR. Then, two or three lsr segments were ligated by fusion PCR, and the PCR product was digested with BamHI and HindIII and ligated to pRSFDuet-1 to construct pRSFDuet-2lsr and pRSFDuet-3lsr (lsr: halogenase LSR gene). RBSs with different strengths were obtained using the Sails Lab platform (https://salislab.net/software/doLogin). Genes involved in the FAD biosynthesis pathway—ribH, ribC, and ribF—were amplified from the E. coli PHCF8 constructed in a previous study using PCR with the primers listed in Table 1.

Table 1 Primers used in this study
Full size table

Whole-cell biotransformation

The reaction system (pH 7.4) contained 4.2 g/L (DCW, w/v) cells and 2.0 g/L Trp in a total volume of 5 mL, and the mixture was incubated in a shaker at 35 °C for 12 h. The intracellular FAD concentration was measured using an FAD assay kit (Sigma-Aldrich, Shanghai, China). The recombinant E. coli were subjected to ultrasonication in an ultrasonic cell disruptor (JY88-IIN, Xinzhi Biotechnology Co., LTD, Ningbo, China). For this, 50-mL centrifuge tubes containing 10 mL of recombinant E. coli cells were immersed in the ultrasonic chamber and maintained at a height of 3 cm from the bottom of the bath throughout all experiments. The temperature was maintained at 4 °C owing to recirculation of a water coolant. The ultrasonic treatment conditions were optimized to increase 7-Cl-Trp and IPA production. The effects of ultrasonic power (100, 200, and 300 W), duty cycle (40%, 50%, 60%, and 70%), and irradiation time (1, 2, 3, and 4 min) on enzyme activity and production were determined. During the ultrasonic enhancement of fermentation, the duty cycle can enhance biological processes. The duty cycle can range from 10% (i.e., ultrasound on for 1 s and off for 9 s) to 100% (i.e., on for 10 s) over a 10-s cycle. The duty cycle can be selected according to the microorganisms and enzymes of interest, and it can directly affect cell morphology and metabolic activities during fermentation. Therefore, we verified the effect of different duty cycle ratios on enzyme activities and products (Pawar and Rathod 2020).

Product purification

  • Four 5-L fermenters (300 series, Beijing DSA Instruments Co., Ltd, China) were used to obtain 10 L culture broth of E. coli under the same conditions. The whole cell catalyst was obtained by centrifugation and washing. Five-liter whole cell catalyst was sonicated (ultrasonic intensity 200 W power, duty cycle 50% for 2 min), and another 5-L culture fluid was not sonicated (conditions: the optimal strain was selected for a scaled up culture in a 5-L fermenter. A well-grown single colony was picked and inoculated into a 50-ml shake flask and incubated at 37 °C, 220 rpm for 10 h, at 1% inoculum volume in a 5-L fermenter containing 2.5-L MS medium. The ventilation was 3 mL/min, and the controlled fermentation temperature was 37 °C, 400 rpm for the previous period after inoculation. After incubation for 3.5 h, the incubation temperature was held at 20 °C until OD600 was 6, and the stirring speed was associated with dissolved oxygen. Setting the dissolved oxygen to 30%, 0.04 mM IPTG was added to induce the culture for 12 h at pH 8.)

  • The above two kinds of culture broth were respectively centrifuged at 4 °C, 8000 × g, for 15 min (Sorvall lynx 6000, Thermo Scientific, USA) at the same time, and then, the supernatant was separately collected in two sterile glass vessels.

  • The above two kinds of supernatant were adjusted to acidic (pH: 2 ~ 3), respectively, and subsequently, ethyl acetate (EA) was added separately for extraction, the EA phase was collected after extraction three times, respectively, and then, rotary evaporation to remove EA (re100 pro, scilogex, USA), the crude products were respectively recovered in two rotary evaporation bottles.

  • Separation and purification between the products are performed by column chromatography, the adsorbent is selected as silica gel, the loading mode is dry loading, and the mobile phase is petroleum ether: ethyl acetate = 2:1.

  • Identification of the product was performed by thin layer chromatography (TLC), color was developed under UV, and the same components as the standard (7-Cl-Trp and IPA) ratio shift value (RF) were separately recovered, and rotary evaporation.

  • Examine the purity and content of the two products separately (7-Cl-Trp and IPA) by HPLC (HPLC conditions: Agilent 1200, Palo Alto, CA, USA) equilibrated with an ultraviolet detector at 210 nm and an Aminex HPX-87H Column (300 × 7.8 mm; Bio-Rad Laboratories Inc., Hercules, CA, USA) at a column temperature of 40 °C. The mobile phase was 5 mM sulfuric acid, and the flow rate was 0.6 mL/min.).

Results

Construction and optimization of the cell-free biotransformation process to ensure the feasibility of CombiAADHa for FAD/FADH2 regeneration

To test the feasibility of CombiAADHa for cofactor regeneration and optimize the activity ratio, cell-free biotransformation needs to be conducted using purified enzyme preparations. However, the purification of membrane-bound l-AAD requires the addition of a detergent, which may be unfavorable for halogenation and cofactor regeneration. Moreover, the activity and stability of purified l-AAD are very different from those of membrane-bound l-AAD (Hou et al. 2015), which could decrease the value of cell-free biotransformation as a reference for whole-cell biotransformation. Therefore, as described in our previous study, membrane fractions were isolated from E. coli BL21(pRSFDuet-AAD) using ultracentrifugation to retain the activity and stability of l-AAD (Hou et al. 2015). Wild-type l-AAD was used instead of engineered l-AADs with higher catalytic activities because the catalytic efficiency of wild-type l-AAD itself is much higher than that of halogenase (Hou et al. 2016). Mutant halogenase 3-LSR (S130L/N166S/Q494R) has been found to show improved Trp conversion compared to its wild-type counterpart (Poor et al. 2014). Wild-type halogenase and 3-LSR were purified from recombinant E. coli, and their catalytic activities toward Trp were compared. 3-LSR showed a higher specific activity for Trp than the wild-type halogenase under the same conditions. Thus, 3-LSR was purified from E. coli BL21(pRSFDuet-lsr) and used in the subsequent experiments (Figs. 2a and b).l-AAD has broad substrate specificity and catalyzes the oxidation of aliphatic and aromatic l-amino acids to their corresponding keto acids with different catalytic rates (Nshimiyimana et al. 2019). Differences in oxidation deamination rates may lead to different halogenation rates. Thus, the feasibility of CombiAADHa was first tested for the biotransformation of Trp to 7-Cl-Trp in a cell-free system using membrane-bound l-AAD fractions with the addition of halogenase and different l-amino acids. Comparative studies of different cell-free biotransformation processes were performed, and the activity ratio (membrane L-AAD:halogenase), 7-Cl-Trp production, and H2O2 contents were examined.

Fig. 2
figure 2

Construction and optimization of cell-free biotransformation to ensure the feasibility of CombiAADHa for FAD/FADH2 regeneration. a Purification of halogenase 3-LSR. b Comparison of the specific activity of wild-type halogenase and 3-LSR for Trp (100%: 203 mU/mg). c Effect of activity ratio (membrane l-AAD:halogenase) on 7-Cl-Trp and H2O2 contents for dual-amino acid (Trp-Phe, Trp-Leu) and mono-amino acid (Trp-Trp) systems was determined. ****P < 0.0001. ***P < 0.001

Full size image

Figure 2c shows that there was no significant 7-Cl-Trp detection when the reaction system contained halogenase and Trp without FAD. However, a certain amount of 7-Cl-Trp was synthesized when the CombiAADHa regeneration system was introduced. This result suggested that the oxidative deamination catalyzed by l-AAD could efficiently accelerate the regeneration rate of FAD/FADH2 and enable halogenation. However, the reaction produced H2O2, which has a negative effect on the stability and activity of halogenase.

Furthermore, the activity ratio (membrane l-AAD:halogenase) was optimized, and three amino acids (Phe, Leu, and Trp) were individually used for oxidative deamination. This resulted in the construction of dual-amino acid (Trp-Phe, Trp-Leu) and mono-amino acid (Trp-Trp) systems. For the Trp-Phe system, as the activity ratio decreased, H2O2 synthesis decreased. A minimum of 2.0 mg/L H2O2 was obtained at a ratio of 1:100, which was 64.5% and 80.0% that of the ratio obtained at 1:10 and 1:50, respectively. Meanwhile, 7-Cl-Trp synthesis increased, and a maximum of 81 mg/L was obtained at a ratio of 1:100, which was 2.7 and 1.4 times that observed at a ratio of 1:10 and 1:50, respectively. A further decrease in the ratio resulted in increased H2O2 synthesis and decreased 7-Cl-Trp synthesis. The possible reasons were as follows: when there is excessive activity of l-AAD in the system (membrane l-AAD:halogenase = 1:10), the FADH2 generation catalyzed by l-AAD becomes far higher than the consumption catalyzed by halogenase, resulting in FADH2 accumulation, H2O2 generation, and decreased 7-Cl-Trp production. In contrast, excessive halogenase activity in the system (membrane l-AAD:halogenase = 1:150) causes the FADH2 generation rate to be lower than the FADH2 consumption rate, resulting in decreased 7-Cl-Trp production.

For the Trp-Leu system, the peak amount of 7-Cl-Trp synthesized was 182 mg/L, which was 2.2 and 1.1 times that of the maximum amount obtained in the Trp-Phe and Trp-Trp systems, respectively. Comparisons of yield revealed that the yield in the Trp-Leu system was 2.2 times higher than that in the Trp-Phe system, indicating that the Trp-Leu system had advantages over Trp-Phe system. Although the Trp-Leu system showed a slightly higher yield than the Trp-Trp system (difference of 10%), a mono-amino acid system has certain advantages over dual-amino acid systems, like a reduced cost, simpler configuration, and more convenient product purification. Moreover, a certain amount of IPA (193 mg/L) was synthesized in the mono-amino acid system. IPA is an important intermediate and a building block for various high-value products in the pharmaceutical and food industries (Zhu et al. 2017). Zhu et al. (2017) obtained a fivefold higher IPA yield than that in wild-type E. coli using the codon-optimized recombinant expression of tdiD (tdiDco), blocking the branching pathway and optimizing tdiDco expression by increasing substrate utilization. Kerbs et al. (2022) produced halogenated tryptophan derivatives in Corynebacterium glutamicum with a high expression of tryptophanase and decarboxylase genes, and produced 16 mg/L 7-Cl-indole, 23 mg/L 7-Br-indole, and 0.36 g/L 7-Br-tryptamine during fed-batch fermentation. The above experiments all used shaking flask fermentation, which is simple and can be used for coenzyme regeneration through cell metabolism. In contrast, except for 7-Cl-Trp, the dual-amino acid system offers no additional high-value products. Therefore, the mono-amino acid (Trp-Trp) system was selected as the optimal configuration for CombiAADHa construction for the co-production of 7-Cl-Trp and IPA.

Furthermore, these abovementioned results proved that the ratio of l-AAD to halogenase has a significant effect on cofactor regeneration and the synthesis of 7-Cl-Trp, IPA, and H2O2. Given the selected mono-amino acid (Trp-Trp) system, the optimal activity ratio of l-AAD to halogenase was set from 1:50 to 1:60, in order to achieve the synthesis of 170 mg/L 7-Cl-Trp, 193 mg/L IPA, and 0.3 mg/L H2O2. Taken together, these findings suggested that CombiAADHa could effectively regenerate FAD/FADH2 and increase 7-Cl-Trp and IPA synthesis upon the application of a certain enzyme activity ratio.

Combination of expression fine-tuning and strengthening of FAD/FADH2 synthesis for the whole-cell CombiAADHa system

Due to the laborious separation procedure for membrane fractions, high cost of purification, and requirement of external FAD addition, it was essential to develop a whole-cell CombiAADHa system with the recombinant E. coli strain for the co-production of 7-Cl-Trp and IPA.

Based on the optimal activity ratio of halogenase and l-AAD in the cell-free biotransformation system, whole-cell biotransformation was optimized using gene duplication and RBS regulation. Gene duplication is a method in which gene copies are increased through repeated tandem expression of the same gene, which increases the expression of rate-limiting enzymes (Hepworth et al. 2017). Since the catalytic efficiency of l-AAD was much higher than that of halogenase, the expression level of halogenase was increased by the repeated expression of lsr 1–4 times, resulting in the construction of E. coli 1–4 (Fig. 3a). The enzyme activities and 7-Cl-Trp and IPA synthesis were evaluated (Fig. 3b and c). As shown in Fig. 3, as lsr repetitions increased from 1 (E. coli 1) to 3 (E. coli 3), halogenase activity increased from 0.5 to 7.1 U/g, and the ratio (l-AAD:halogenase) changed from 102:1 to 4.6:1. As a result, compared with the yield from a single copy of lsr (E. coli 1), the 7-Cl-Trp and IPA titers increased to 1.4- and 1.3-fold, respectively. Further increase in lsr repetitions (E. coli 4) resulted in a slight decrease in halogenase activity and 7-Cl-Trp and IPA synthesis, probably due to the increased plasmid burden.

Fig. 3
figure 3

Combination of expression fine-tuning and strengthening FAD/FADH2 synthesis for whole-cell CombiAADHa system. a Schematic representation of the constructed plasmids. b l-AAD and halogenase activities at the initial of whole-cell biotransformation. c IPA and 7-Cl-Trp production of whole-cell biotransformation by recombinant E. coli harboring different plasmid corresponding to Fig. 3a. d Time courses for the production of FAD by the recombinant strain

Full size image

Then, to further adjust the activity ratio, laad was equipped with RBSs of different strengths (Table 2). The RBS of the plasmid pRSFDuet-1 was considered the highest. The other RBSs used successively weakened expression in the following order: RBS1, RBS2, and RBS3. The RBS of laad was replaced with the weaker RBS in combination with three lsr repeats, resulting in the construction of E. coli 5–7 (Fig. 3a). After the RBS of laad was replaced with RBS2 (E. coli 6), the halogenase activity increased from 7.1 to 21.2 U/g and the ratio (l-AAD:halogenase) changed from 4.6:1 to 1:4.2. The titers of 7-Cl-Trp and IPA both became 1.3 times higher than those observed with the original RBS (Fig. 3b and c).

Table 2 RBS used in this study
Full size table

The activity ratio remained beyond optimal despite expression fine-tuning via various strategies. Site-directed mutagenesis was performed to decrease the l-AAD activity and leave halogenase activity unchanged, yielding the optimal ratio. In our previous study (Hou et al. 2015), during the process of l-AAD modification to improve l-AAD activity, many mutants with varying degrees of reduced l-AAD activity were unexpectedly obtained. For example, mutation at position 160 caused a 12% decrease in l-AAD activity, and mutation at position 240 caused a 21% decrease in l-AAD activity. Using these l-AAD mutants, obtain reaction systems with different activity ratios of l-AAD and halogenase.

Furthermore, the time courses of FAD concentration changes were compared between E. coli 1 and E. coli 6 (FAD concentration was determined using an FAD kit) (Fig. 3d). At 3 h, the FAD concentration in E. coli 1 decreased dramatically by 80% and remained relatively low between 3 and 12 h. This was because that the catalytic rate of l-AAD was higher than that of halogenase, owing to which FAD was dramatically consumed and remained at a low concentration. In contrast, the FAD concentration in E. coli 6 remained stable and merely decreased by 50% at 9 h. This was because that the activity ratio was regulated, and the halogenase activity was higher. As a result, the FAD/FADH2 regeneration rate increased and the FAD concentration remained relatively stable. These results suggested that through the combination of gene duplication and RBS regulation, the expression levels of halogenase and l-AAD could be regulated effectively, and the FAD/FADH2 regeneration rate could be increased, leading to improved whole-cell catalytic abilities. We hypothesized that 7-Cl-Trp and IPA production would further increase if the FAD/FADH2 biosynthesis pathway is strengthened.

Our previous studies have shown that the FAD/FADH2 supply pathway can be strengthened via the overexpression of lumazine synthase (RibH, EC: 2.5.1.9), riboflavin synthase (RibC, EC: 2.5.1.9), and riboflavin kinase/FMN adenylyltransferase (RibF, EC: 2.7.7.2/2.7.1.26), which is favorable for improving the catalytic efficiency of FAD-dependent whole-cell biotransformation (Hou et al. 2017). Therefore, the key genes (ribH, ribC, and ribF) for FAD/FADH2 supply were simultaneously overexpressed with lsr and laad in E. coli using two compatible plasmids to strengthen the FAD/FADH2 supply pathway (Fig. 3a, E. coli 8). After strengthening FAD/FADH2 synthesis in E. coli 6, the initial intracellular FAD concentration increased by 50% (from 0.5 to 1.0 mM, Fig. 3d), and the enzyme activity ratio changed to 1:4.5 (Fig. 3b). As a result, 7-Cl-Trp and IPA synthesis was further enhanced by 15% (from 96 to 110 mg/L) and 12% (from 115 to 129 mg/L), respectively (Fig. 3c).

To characterize the effect of FAD/FADH2 supply pathway strengthening on whole-cell biotransformation, we constructed E. coli 9, in which the FAD/FADH2 supply pathway was strengthened and the expression of halogenase and l-AAD was not fine-tuned (see Fig. 3a). As shown in Fig. 3d, the initial FAD concentration in E. coli 9 was 2.4 times that of E. coli 1. However, at 3 h, the FAD concentration in E. coli 9 dramatically decreased by 75% and remained relatively low at 3–12 h. Hence, no obvious increase in 7-Cl-Trp and IPA synthesis was observed in E. coli 9.

Ultrasound-assisted whole-cell biotransformation in recombinant E. coli

The expression levels of halogenase and l-AAD were effectively controlled via a combination of gene duplication and RBS regulation, leading to a changed activity ratio and increased whole-cell catalytic activity. However, given the optimal ratio observed in cell-free biotransformation (1:50–1:60), other strategies needed to be developed to further fine-tune the ratio for whole-cell biotransformation.

For whole-cell CombiAADHa biotransformation, first, Trp and FAD were catalytically converted to IPA and FADH2 outside the cell by membrane-bound l-AAD. Then, Trp and FADH2 were transported into the cell and catalytically converted to 7-Cl-Trp and FAD by the intracellular halogenase (Fig. 1). It was clear that the biotransformation rate was limited by the mass transfer of Trp, FAD, FADH2, and 7-Cl-Trp across the cell membrane. To address this issue, ultrasound treatment was applied with the goal of improving membrane permeability and the efficiency of the whole-cell biotransformation of Trp to 7-Cl-Trp and IPA. Furthermore, since l-AAD is a membrane-bound protein, ultrasound treatment could very likely decrease the l-AAD activity and improve the activity ratio.

Ultrasonic treatment conditions were optimized to increase 7-Cl-Trp and IPA production. The effect of ultrasonic power on enzyme activities and 7-Cl-Trp and IPA production was determined over the power range of 100 to 400 W, with ultrasound applied for 1 min on a 60% duty cycle (Fig. 4). As the power increased, the production of 7-Cl-Trp and IPA increased, and maximum concentrations of 145 mg/L and 156 mg/L (Fig. 4a), respectively, were obtained at 200 W; the ratio was changed to 1:8 (Fig. 4b). With further increase in the power to 300 W, the ratio changed to 1:7.5, with the production of 7-Cl-Trp and IPA decreasing.

Fig. 4
figure 4

Effect of ultrasonic treatment conditions on 7-Cl-Trp and IPA production (a) and l-AAD and halogenase activities at the initial of whole-cell biotransformation (b)

Full size image

Then, once the optimal power had been determined, the effect of duty cycle on 7-Cl-Trp and IPA production was studied by varying the ON–OFF times of ultrasonic irradiation at 200 W for 1 min. 7-Cl-Trp and IPA production peaked at 155 mg/L and 167 mg/L (Fig. 4a), respectively, with a 50% (5 s ON and 5 s OFF) duty cycle. Moreover, the ratio changed to 1:20 (Fig. 4b).

Finally, the effect of irradiation time on 7-Cl-Trp and IPA production was studied over the range of 1–4 min at 200 W power and a 50% duty cycle. When the irradiation time was set to 2 min and 3 min, the activity ratios changed to 1:50 and 1:56, respectively, which were among the optimal ratios observed during cell-free biotransformation (Fig. 4a). Nonetheless, the 7-Cl-Trp and IPA production at 2 min was only 1.1 times that at 3 min (Fig. 4b). A further increase in the irradiation time to 4 min caused a dramatic decrease in 7-Cl-Trp and IPA production, although the activity ratio remained in the optimal range. This was likely because excessively strong ultrasound treatment had a destructive effect on halogenase activity. Therefore, the enzyme activity ratio could not be the only indicator of whole-cell catalytic ability.

In summary, the optimal ultrasonic intensity was 200 W with a 50% duty cycle for 2 min. Treatment at these parameters resulted in the maximum production of 172 mg/L 7-Cl-Trp and 181 mg/L IPA, respectively—1.6 and 1.4 times higher than that observed with whole-cell biotransformation without ultrasonic treatment. A further increase in the ultrasonic intensity imparts a more intense mechanical shear and severe shock to whole-cell catalysts, possibly resulting in cellular lysis and the expulsion of cellular content, thus decreasing production.

Product purification

A 5-L fermenter was used to prepare 10 L of fermentation broth, of which 5 L was treated with ultrasound, and the other 5 L was not. The purity and content of 7-Cl-L-Trp and IPA were detected using HPLC in both groups. The purity and yield of 7-Cl-L-Trp purified without ultrasonic treatment were 97.9% and 73.3%, respectively, while these values were 98.2% and 71.6%, respectively, for IPA (Data not shown). After ultrasonic treatment, the purity of 7-Cl-L-Trp reached 98.1%, and the yield was 72.8%. The purity and yield of IPA became 97.8% and 70.9%, respectively (Data not shown). Thus, ultrasonic treatment appeared to have little effect on the purity and yield of purified products.

Discussion

FAD/FADH2-dependent reactions are limited by cofactor regeneration rate due to the complexity of the existing trienzymatic cascade regeneration (Ismail et al. 2019). Irrespective of whether the FAD/FADH2 regeneration system is an electrochemical, chemical, or three-enzyme regeneration system, there has been rare research on cell-free biotransformation and whole-cell transformation in these systems. In this study, we developed an efficient dual-enzyme cascade called CombiAADHa for FAD/FADH2 regeneration based on a FADH2-dependent halogenase and a FAD-dependent l-AAD and the constructed cell-free and whole-cell biotransformation systems were used to co-produce 7-Cl-Trp and IPA from one substrate-Trp. Compared with trienzymatic cascade using two kinds of substrates, CombiAADHa is easily to be controlled and optimized to coordinate the reactions and obtain the maximal reaction rate, moreover, the usage of one substrate is favorable to reduce cost and simplify download purification. This was favorable for reducing the cost and increasing the commercial value of the reaction.

For the synthesis of α-keto acids from l-amino acids, the FAD-dependent membrane-bound l-AAD shows a better performance than l-amino acid oxidases (l-AAO, EC 1.4.3.2), like the broad substrate specificity, no formation of H2O2, and the membrane-bound nature, which are favorable for the construction of whole-cell biocatalysts (Motta et al. 2019). Although the membrane-bound protein used in this experiment had a transmembrane structure, the catalytic properties of l-AAD have been well-verified. Therefore, the FAD dependent l-AAD was selected to construct our FAD/FADH2 regeneration system.

First, a cell-free biotransformation system was constructed and optimized, proving the feasibility of CombiAADHa for FAD/FADH2 regeneration. The generation of H2O2, which is unfavorable for enzyme stability and activity, is a common problem in FAD/FADH2-dependent reactions. This is because the FADH2 reacts with oxygen to produce flavin hydroperoxide, leading to the generation of hypohalous acid for halogenation or H2O2 (Schroeder et al. 2018). In this study, 7-Cl-Trp and H2O2 synthesis varied with different activity ratios, proving that H2O2 generation is unfavorable for whole-cell biotransformation. In some cases, this H2O2 can be partially removed by the addition of catalase, but this not only increases production costs but also makes large-scale production more challenging (Ismail et al. 2019). Therefore, the addition of halogenase and l-AAD was controlled to obtain an optimal activity ratio, leading to decreased H2O2 levels and increased 7-Cl-Trp synthesis.

For whole-cell biotransformation, through the combination of different regulation strategies, the expression of halogenase was increased, accelerating halogenation and the regeneration of FAD. This offered more FAD for oxidative deamination, resulting in increased IPA synthesis. Meanwhile, l-AAD effectively regenerated FADH2, promoted the halogenation reaction, and enhanced 7-Cl-Trp synthesis. These results further proved that the FAD/FADH2 regeneration rate was a limiting factor for the reaction, and strengthening FAD/FADH2 regeneration could effectively improve whole-cell catalytic capacity. It was also indicated that an increased expression of one enzyme would result in the decreased expression of the other enzyme in the whole-cell biocatalyst, consistent with findings from the construction of an E. coli whole-cell catalyst with multi-enzyme expression for phenyllactic acid production (Hou et al. 2019).

Then, intracellular FAD levels were increased by the strengthening of FAD biosynthesis during cell growth. The initial FAD concentration in E. coli 9 was higher than that in E. coli 8, but the 7-Cl-Trp and IPA production was lower, likely owing to a dramatic decrease in FAD concentration caused by inefficient FAD/FADH2 regeneration (Fig. 3d). In constract, the biocatalytic capability of the E. coli whole-cell system was improved via engineering of FAD/FADH2 supply pathways and the construction of an FAD/FADH2 regeneration system simultaneously. Compared with the control strain, the engineered system yielded a higher intracellular FAD concentration and the 7-Cl-Trp and IPA production was 108% and 95% of the original, respectively. These findings suggest that engineering of the FAD biosynthesis pathway is a good approach for improving the initial FAD concentration but is not effective in maintaining adequate FAD levels during whole-cell biotransformation, and cofactor biosynthesis and regeneration have a synergistic effect on whole-cell catalytic capability, which are consistent with our previous study in which the whole-cell catalytic ability of l-amino acid for the corresponding α-keto acid was increased by the combination of FAD biosynthesis and regeneration (Hou et al. 2017).

Since halogenation reaction occurs inside the cell and oxidative deamination occurs outside, production can be effectively increased by improving membrane permeability and accelerating the transport of substrate and products through ultrasound. Ultrasound was previously used to increase D-tartaric acid production in recombinant E. coli for whole-cell catalysis (Dong et al. 2018). Meanwhile, we found that ultrasonic conditions could affect l-AAD activity and the enzyme activity ratio owing to the destruction of cell membrane structure. Under high ultrasonic intensity, although the enzyme activity ratio was in the optimal range, synthesis decreased due to an excessive decrease in enzyme activity. The findings thus indicated that production is affected by a combination of the activity ratio, enzyme activity, and ultrasonic intensity.

In addition, the ultrasound conditions need to be controlled with the aim of not affecting the downstream purification (Kwon et al. 2019). Moharkar et al. (2022) used ultrasound-mediated sugaring out extraction (UMSOE) to extract erythromycin from the actual fermentation broth, and ultrasound had no significant effect on cell morphology and the release of intracellular substances. Kielkopf et al. (2021) found that ultrasonic treatment before protein purification could destroy the cell membrane of E. coli and improve the content in the soluble protein fraction, thus improving the efficiency of protein extraction and separation. In the present study, the ultrasonic duration was only 2 min, less than one-tenth of the fermentation incubation time. Ultrasonic treatment had no significantly effect on product purification in this case.

In this study, enzymatic transformation showed a shorter reaction time, and the productivities of 7-Cl-Trp and IPA (28.3 mg/L/h and 32.2 mg/L/h) were 2.0 and 2.1 times higher than that of the whole-cell reaction (14.3 mg/L/h and 15.1 mg/L/h), respectively. This shows that the isolated l-AAD and halogenase showed certain advantages over whole-cell catalysts, since there was no side-reaction during enzymatic transformation and the substrate was more easily available for the enzyme, without any hindrance. Given the laborious and costly process of enzyme purification, system stability, requirement of external FAD addition, and recycling, whole-cell transformation is an attractive alternative to a cell-free system. However, in terms of productivity, the free enzyme systems show better performance (Chen et al. 2020; Rohman and Dijkstra 2021). Therefore, both biotransformation methods are promising and need improvement in the future.

In conclusion, we constructed an ecofriendly and efficient whole-cell FAD/FADH2 regeneration system by co-expressing l-AAD and halogenase, and we tested the feasibility of CombiAADHa for the production of 7-Cl-Trp and IPA from Trp. The catalytic abilities of this system were then improved through a combination of expression fine-tuning, strengthening of FAD/FADH2 synthesis, and whole-cell ultrasonication. Overexpression of the YadH Flux pump in Escherichia coli increased the secretion of heterologous polykeone-6-deoxyerythromycin Endotone B (6 dEB, erythromycin precursor) by 15%, thus improving whole-cell catalytic capacity (Lim et al. 2015). Based on the good effect of ultrasonic treatment on the whole cell production of 7-Cl-Trp and IPA, we plan to use the metabolic engineering strategy to improve the purification ability of recombinant cell membrane in the next step. Furthermore, the FAD/FADH2 regeneration mechanism of the dual-enzyme regeneration system, the synthesis of a single product (such as 7-Cl-Trp), and the externally supplied of halide salts (Veldmann et al. 2019) will be clarified in future studies.