Design of a modularized multi-enzymatic cascade for converting (−)-limonene into (−)-perillamine

In the cascade process, (−)-limonene was initially hydroxylated by a cytochrome P450 enzyme to produce (−)-perillyl alcohol. The (−)-perillyl alcohol was further oxidized by an alcohol dehydrogenase (ADH) to produce (−)-perillyl aldehyde and then aminated by an w-transaminase (w-TA) to give (−)-perillamine simultaneously. We tried to recycle the effective nicotine cofactor by introducing a formate dehydrogenase (FDH) and an NADH oxidase (NOX) for regenerating the desired NADH (for C–H monooxygenation) and NAD+ (for alcohol dehydrogenation), respectively. Considering the instability of P450 enzymes and the interference between the two coenzyme recycling systems which might result in a thermodynamic balance, we divided this cascade into two independent modules (Module 1 and Module 2). Module 1 was composed of a P450 enzyme and an FDH for effectively hydroxylating (−)-limonene to (−)-perillyl alcohol, while Module 2 comprises an ADH, an NOX and an w-TA to transform the (−)-perillyl alcohol into (−)-perillamine (Scheme 1).

Scheme 1
scheme 1

A multi-enzyme cascade designed for the regiospecific bioamination of (−)-limonene into (−)-perillamine. 2-PTAM 2-pentanamine, 2-PTON 2-pentanone

Recruiting enzymes for constructing Module 1

Terpenoid hydroxylases are mainly cytochrome P450 enzymes, which have significant substrate promiscuity. According to the type of reaction catalyzed by the enzyme and the structure of substrate, we selected several terpene hydroxylases for the first hydroxylation step. Through functional characterization, we found that CYP153A7 could catalyze the regiospecific hydroxylation of (−)-limonene into (−)-perillyl alcohol. CYP153A7, originated from Sphingomonas sp. HXN-200, is a class I P450 enzyme with a broad substrate scope and applications (Chang et al. 2000, 2002a, b; Dong et al. 2022; Li et al. 2001). It requires ferredoxin (Fdx) and ferredoxin reductase (FdR) for electron transfer and can be used for the regio- and stereo-selective hydroxylation of non-activated carbon atoms (Pham et al. 2012; Yang and Li 2015). Therefore, we utilized an engineered E. coli BL21(DE3) strain (A7F), constructed previously in our laboratory, which co-expresses the CYP153A7 gene with its redox partners (Fdx & FdR) and a formate dehydrogenase from Candida boidinii (CbFDH) (Hummel 1999; Slusarczyk et al. 2000; Tang et al. 2021) for the effective cofactor regeneration.

Recruiting enzymes for constructing Module 2

Considering the high similarity of the chemical structures between (−)-perillyl alcohol and p-mentha-1,8-dien-3-ol, we tested the toolbox of p-mentha-1,8-dien-3-ol dehydrogenases preserved in our laboratory and found that a double mutant of alcohol dehydrogenase, LkADHT91F/I195V from Lactobacillus kefiri (LkADH), could catalyze the oxidation of (−)-perillyl alcohol into (−)-perillyl aldehyde. In order to regenerate the consumed cofactor NADH, we introduced a variant of the highly efficient NADH oxidase, SmNOXV193R/V194H from Streptococcus mutans (SmNOX) (Jiao et al. 2016). Since many w-TAs were reported for the transamination of aldehydes, we first screened a panel of TAs preserved in our laboratory for the enzymatic amination of (−)-perillyl aldehyde. ATA-117 from Arthrobacter sp. was found to be the most efficient enzyme, exhibiting a high catalytic activity toward (−)-perillyl aldehyde, approximately 2.8 μmol min−1 mg−1 lyophilized enzyme powders. Therefore, ATA-117 was finally chosen as the transaminase for our work, which was previously reported to be highly productive for the biocatalytic synthesis of sitagliptin after 11 rounds of directed evolution (Gomm et al. 2016; Guan et al. 2015; Savile et al. 2010).

Optimizing the cascade reactions of Module 1

Subsequently, we tried to improve the reaction efficiency of Module 1. As shown in Table 1, the effects of a few parameters such as temperature, pH, substrate concentration, cell dosage and whether to add surfactants were investigated. The titer of (−)-perillyl alcohol was a little bit higher at pH 7.5, and on this basis, we lowered the reaction temperature to 25 ℃. To our surprise, the concentration of formed (−)-perillyl alcohol was doubled. Strangely, an increase in the cell dosage from 5 to 20 gcdw/L resulted in a slight decrease in the analytic yield of the final product. This might be caused by the increase of the viscosity of the reaction system, which may affect the mass transfer rate. To address this problem, we tried to add 1 mg/mL of Triton X-100 to the reaction system. Finally, under the conditions of 15 gcdw/L cell dosage, 10 mM substrate and 1 mg/mL Triton X-100, the concentration of product (−)-perillyl alcohol reached 6.8 mM. Partial volatilization of (−)-limonene led to the loss of substrate (Cornelissen et al. 2013), as confirmed in a system without cells, while the remaining (−)-limonene (ca. 68%) was completely oxidized into (−)-perillyl alcohol.

Table 1 Optimization of reaction conditions of Module 1

Optimizing the cascade reactions of Module 2

In order to optimize the reaction efficiency of Module 2, the effects of a few parameters were examined, including temperature, pH, amine donor and cofactor doses were examined. Based on actual effects on the reaction, the optimal pH and temperature for the reaction were pH 7.5 and 35 ℃ (Fig. 1). In a one-pot reaction, there are multiple enzymes coexisting in the same space. In order to improve the overall efficiency of cascaded reactions, it is important to coordinate the ratio of each element added into the system. We have investigated the ratio of the three enzyme doses in Module 2, suggesting that the optimal dose ratio of [LkADH]/[SmNOX]/[ATA-117] should be 1:2:10 (Additional file 1: Fig. S1). After determining the proportion of the enzyme added, we examined the effect of enzyme doses. Keeping a constant ratio of the three enzymes, the product titer increased with the increase in the enzyme doses. When the amount of LkADH was 0.2 U/mL, the product titer was the highest, so the optimum dose of LkADH was fixed to be 0.2 U/mL (Additional file 1: Fig. S2).

Fig. 1
figure 1

Optimization of the temperature (a) and pH (b) of Module 2. Reaction conditions (0.5 mL): (−)-Perillyl alcohol 10 mM, LkADH 0.2 U/mL, [LkADH]/[NOX]/[ATA-117] = 1/5/10, 0.2 mM NAD+, 0.1 mM PLP, 50 mM 2-pentanamine, potassium phosphate buffer (pH 7.5, 100 mM) or Tris–HCl buffer (pH 8–9, 100 mM) or Gly-NaOH buffer (pH 9.5–10, 100 mM), 30℃, 35 ℃ or 40 ℃, 800 rpm, 12 h. Temp.: temperature; Concn.: concentration

Transaminases have the significant advantages that do not require redox cofactors, however the unfavorable thermodynamic equilibrium of enzymes limits its application (Kelefiotis-Stratidakis et al. 2019). The amino donor may be an important factor to overcome the shortcomings of reaction equilibrium. We optimized the type and amounts of amino donors. Among several commonly used amine donors, 2-pentanamine is the most effective one (Fig. 2A). In previous studies, researchers usually added excessive amounts of amino donor to shift the equilibrium [14], but excessive amino donor will also impair the activity of enzyme. The product titer was the highest when 80 mM 2-pentanamine was added to the reaction system (Fig. 2B). In addition, the doses of cofactors NAD+ and pyridoxal-5-phosphate (PLP) were also optimized, indicating that 0.2 mM NAD+ and 0.2 mM PLP were sufficient for the cascaded reaction (Additional file 1: Fig. S3). Through the optimization of Module 2, the titer of (−)-perillamine was increased up to nearly 7 mM.

Fig. 2
figure 2

Optimization of the kind (a) and equivalents (b) of amino donor in Module 2. Reaction conditions (0.5 mL): shaken at 35 °C, 800 rpm for 12 h, KPB buffer (pH 7.5, 100 mM). The reaction mixture (0.5 mL) was composed of 10 mM (−)-perillyl alcohol (with 2% DMSO), 0.2 U/mL LkADH, 0.4 U/mL SmNOX, 2 U/mL ATA-117, 0.2 mM PLP, 0.2 mM NAD+, in addition to: a 50 mM DL-Ala/IPA/2-pentanamine; b 10−100 mM 2-pentanamine. IPA isopropanyl alcohol, Concn. Concentration, 2-PTAM 2-pentanamine, equiv. equivalent

One-pot two-step cascade reactions

Initially, we tried one-pot one-step and one-pot two-step modes to perform this cascade reaction, respectively. When we performed the cascaded reaction in the one-pot one-step mode, only a trace amount of the final product (−)-perillamine was detected. The efficiency of the cascade reaction was relatively low, implying that the one-pot one-step mode might be not conducive to the catalytic reaction of unstable P450 enzyme in the complicated reaction system. Therefore, the cascade reactions were alternatively performed in a one-pot two-step mode. When the cascaded reactions of Module 1 were complete, the reaction mixture was heated in a 65 °C water bath to inactivate formate dehydrogenase to avoid any interference with the subsequent coenzyme recycling. All the elements desired for Module 2 were supplemented into the reaction system after being cooled to room temperature and the incubation was continued for 12 h. In the one-pot two-step mode, the titer of (−)-perillamine had a slight increase to approximately 2 mM, suggesting that there was another bottleneck existing in our cascade reaction system. Through step-by-step investigation, we found that only a trace amount of (−)-perillyl aldehyde could be detected in the dehydrogenation step catalyzed by LkADH. By measuring the activity of SmNOX, it turned out that the surfactant added in Module 1 had negative effects on the enzyme activity of SmNOX (Additional file 1: Table S1) and the recycling efficiency of the cofactor. Therefore, we had to increase the SmNOX dose from 0.2 to 1 U/mL for a compensation, which resulted in a significant increase of the product titer.

100-mL preparative-scale cascade reaction

Under the optimized conditions, we carried out a 100 mL preparative-scale cascade reaction, where the time-course analysis revealed that the formation of (−)-perillamine reached a maximum 5.4 mM at approximately 12 h (Fig. 3). Since monoterpenoids are extremely volatile, it was difficult to accurately determine their content. We only monitor the concentrations of the intermediate product (−)-perillyl alcohol, and the final product (−)-perillamine, which are relatively not easy to be volatilized. The reaction of Module 1 quickly reached the end point in 2 h, and there was no significant increase in product titer when the reaction time was prolonged. After the reaction mixture of Module 2 was supplemented, the resulting (−)-perillyl alcohol in Module 1 was rapidly converted into (−)-perillyl aldehyde, accompanied by the formation of (−)-perillamine as the final product. The product was isolated from the reaction system and purified by silica gel column chromatography, giving 76 mg (−)-perillamine with a space–time yield of 1.5 g L−1 d−1. 1H-NMR (400 MHz, CDCl3) δ/ppm (Additional file 1: Fig. S4): 5.58 (brs, 1H), 4.71 (brs, 2H), 3.17 (brs, 2H), 2.19–1.85 (m, 7H), 1.73 (s, 3H), 1.53–1.40 (m, 1H). 13C NMR (101 MHz, CDCl3) δ/ppm: 149.97, 138.50, 120.22, 108.56, 47.92, 41.28, 30.50, 27.64, 27.12, 20.81.

Fig. 3
figure 3

Time course of the 100-mL preparative-scale multi-enzyme cascade reaction. Step I: 10 mM (−)-limonene (with 2% DMSO), 15 gcdw L−1 A7F resting cells, 0.2 mM NAD+, 100 mM sodium formate, 1 mg/mL Triton X-100, KPB buffer (100 mM, pH 7.5), 25 °C, 200 rpm. Step II: supplement of 0.2 U mL−1 LkADH, 1 U mL−1 SmNOX, 2 U mL−1 ATA-117, 80 mM 2-pentanamine, 0.2 mM NAD+ and 0.2 mM PLP, incubated at 35 °C and shaken at 200 rpm. Concn.: concentration

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit


This article is autogenerated using RSS feeds and has not been created or edited by OA JF.

Click here for Source link (