Skip to main content
NCBI home page
Nature Communications logoLink to Nature Communications
. 2025 Feb 23;16:1904. doi: 10.1038/s41467-025-57280-0

A concise enzyme cascade enables the manufacture of natural and halogenated protoberberine alkaloids

Fei Li 1,#, Zhenbo Yuan 1,#, Yue Gao 1,#, Zhiwei Deng 1, Yan Zhang 2, Zhengshan Luo 1, Yijian Rao 1,✉
PMCID: PMC11847921  PMID: 39988594

Abstract

The application and drug development of plant-derived natural products are often limited by their low abundance in medicinal plants and the lack of structural complexity and diversity. Herein, we design a concise enzyme cascade to efficiently produce natural and unnatural protoberberine alkaloids from cost-effective, readily available substrates. Through enzyme discovery and engineering, along with systematic optimization of the berberine bridge enzyme to address remaining manufacturing challenges in protoberberine alkaloid biosynthesis, the high production of drug Rotundine is achieved at an impressive gram-scale titer, demonstrating its industrial potential. More importantly, this cascade also enables the efficient biosynthesis of various unnatural halogenated protoberberine alkaloids. Thus, this work not only unlocks the potential of enzyme cascades in overcoming longstanding challenges in the efficient biosynthesis of plant-derived alkaloids, but also opens avenues to introduce structural complexity and diversity into alkaloids through synthetic biology, offering significant potential for drug development.

Subject terms: Applied microbiology, Synthetic biology, Metabolic engineering, Natural product synthesis

Plant-derived alkaloids are an important class of natural products due to their diverse pharmacological properties. Here, the authors report a concise enzyme cascade for the biosynthesis of natural and halogenated protoberberine alkaloids at higher productivity levels.

Introduction

Plant-derived alkaloids are an important class of natural products with various pharmacological properties1–4, including Rotundine (L-tetrahydropalmatine), berberine, morphine, colchicine, galanthamine and hyoscyamine (Fig. 1a). Many of them have been used as traditional medicines in China, Native America, India and the Islamic region. For instance, Rotundine was first isolated from Corydalis5, a plant that has been used as traditional Chinese herbal medicine for over a thousand years, known for its analgesic, anti-inflammatory, neuroprotective, anti-addictive, and antitumor activities6–8. Today, it also serves as an alternative to anxiolytic and sedative drugs from the addictive benzodiazepine group, as well as analgesics9. However, similar to many plant-derived natural products10,11, the commercial use of plant-derived alkaloids still mainly relies on extraction from medicinal plants with low abundance12–15, which is further affected by climate change, cultivation methods and location. Moreover, due to the lack of appropriate functional groups, derivatization of naturally occurring alkaloids to increase structural complexity and diversity through chemical methods remains challenging, restricting further drug development. Although chemical synthesis methods have been developed to overcome these issues, they often involve harsh conditions and heavy-metal catalysts16,17. In addition, the structural complexity of alkaloids, with their chiral centers and regioselective modifications, often results in low yields.

Fig. 1. Design of an artificially concise multi-enzyme cascade for the biosynthesis of natural and halogenated protoberberine alkaloids.

Fig. 1

Open in a new tab

a Representative plant-derived alkaloids. b Natural biosynthetic pathway of representative Rotundine in plants. PPP: Pentose phosphate pathway; EMP: Embedn-Meyerhof-Parnas pathway (glycolysis pathway); E4P: erythrose-4-phosphate; PEP: phosphoenolpyruvate; DHS: 3-deoxy-d-arabino-heptulosonate-7-phosphate synthase; EPSPS: 5-enolpyruvylshikimate-3-phosphate synthase; CM: chorismate mutase; PDH: prephenate-specific TyrA dehydrogenase; TyrAT: Tyr aminotransferase; TyrH: tyrosine hydroxylase; DODC: DOPA decarboxylase; PPDC: phenylpyruvate decarboxylase; 4HPAAs: 4-HPAA synthase; 4HPP: 4-hydroxyphenylpyruvate; 4HPAA: 4-hydroxyphenylacetaldehyde. c Design of an artificial multi-enzyme cascade for biosynthsizing various protoberberine alkaloids with the BIA, MT, BBE and S9OMT modules. CAR carboxylic acid reductase, Sfp phosphopantetheine transferase, NCS norcoclaurine synthase, 6OMT 6-O-methyl transferase, CNMT coclaurine-N-methyl transferase, BBE berberine bridge enzyme, S9OMT scoulerine 9-O-methyl transferase.

With the elucidation of the biosynthetic pathways of alkaloids and advancements in synthetic biology18–27, many efforts have been made to biosynthesize natural and unnatural alkaloids in microorganisms, including Saccharomyces cerevisiae and Escherichia coli28–35 (Fig. 1b). However, challenges such as the complexity of their biosynthetic pathways, the difficulties in expressing plant-derived P450 enzyme36–38 and berberine bridge enzyme (BBE)29,34,39,40, and the cytotoxicity from the accumulation of alkaloids or its intermediates34,41 always results in low production titers28,29,34, such as 16.9 mg L-1 production in berberine and 68.6 mg L-1 production in Rotundine in engineered yeasts, which still lack commercial viability. In fact, this remains a common manufacturing challenge for the heterologous biosynthesis of many plant-derived alkaloids in microorganisms.

Recently, it was reported that a designed nine-enzyme catalytic cascade enabled the efficient biosynthesis of the HIV drug islatravir42, and therapeutic oligonucleotides could be produced through an enzyme cascade in a single operation43. These seminal examples suggest that the designed enzyme cascades will revolutionize drug synthesis and development. Furthermore, specific enzymes can control the stereo- and chemoselectivity of chiral compounds44,45. Importantly, the use of modular “plug-and-play” strategy allows the easy incorporation or removal of enzymes to tailor the cascade for synthesizing different target compounds46,47, thereby introducing structural complexity and diversity. As for plant-derived natural products, steps catalyzed by enzymes that are difficult to express in engineered cells or that are still not identified can be bypassed through the careful selection of substrates46, making the process more efficient or feasible. Therefore, we envisaged that an artificial enzyme cascade could be constructed to address current production challenges associated with plant-derived alkaloids, facilitating potential industrial preparation. Given that halogenated derivatives often provide enhanced therapeutic effects48,49, we anticipated that these derivatives could be produced using this cascade, thereby addressing the challenges of derivatization through chemical methods.

Herein, a designed enzyme cascade is developed for the efficient synthesis of various protoberberine alkaloids from readily available substrates (Fig. 1c). This cascade is shorter than the original biosynthetic pathway and avoids the difficult plant-derived P450 modification step (Fig. 1b). After enzyme discovery and engineering, a high production titer of various protoberberine alkaloids is realized. The presentative drug Rotundine reaches 2.44 g L-1 through one-pot whole-cell (E. coli) catalytic system, demonstrating industrial potential. More importantly, this cascade also facilitates the efficient biosynthesis of unnatural halogenated protoberberine alkaloids. Notably, we address a common manufacturing problem in the biosynthesis of protoberberine alkaloids, the heterologous expression of the enzyme BBE, through a multi-faceted approach that includes molecular chaperone optimization, promoter engineering, directed evolution, and cofactor enhancement. This enables the efficient production of the critical intermediate (S)-scoulerine at impressive concentrations of 3.19 g L-¹, achieving a 28-fold increase over previously reported results29,30. This streamlines the further biosynthesis of other protoberberine alkaloids. Thus, this study provides a method for the biomanufacturing of natural and unnatural plant-derived alkaloids.

Results

Design of a concise enzyme cascade for the biosynthesis of protoberberine alkaloids

To develop a proof-of-concept enzyme cascade for the efficient biosynthesis of protoberberine alkaloids, their reported natural biosynthetic pathways, such as coptisine, epiberberine and jatrorrhizine23, were comprehensively analyzed. It indicates that all of them could be produced on the basis of the dedicated intermediate (S)-reticuline with the skeleton of benzylisoquinoline alkaloids (BIAs) (Fig. 1c)50,51. Here, we used drug Roundine as a model compound. It suggests that its core BIA skeleton of (S)-reticuline could be generated by the norcoclaurine synthase (NCS)-catalyzed coupling reaction of dopamine and 2-(3,4-dihydroxyphenyl) acetaldehyde (2a) (Figs. 1c, 2)52. Considering that aldehyde is unstable in microorganisms25,46, a carboxylic acid reductase (CAR) was employed to reduce the readily available 2-(3-hydroxy-4-methoxyphenyl) acetic acid (1a)53, thereby synthesizing 2a in situ. Subsequent O-methylation and N-methylation could be achieved by norcoclaurine 6-O-methyltransferase (6OMT) and coclaurine-N-methyltransferase (CNMT) to produce (S)-reticuline23. The above reactions were designed as the BIA module and the MT module (Fig. 1c, Fig. 2), respectively. By introducing substrate 1a, which is already installed with the hydroxy and methoxy group at C3′ and C4′ (Fig. 2a), respectively, the need for difficult hydroxylation reaction catalyzed by P450 N-methylcoclaurine-3′-hydroxylase (NMCH)40 and then the methylation reaction catalyzed by 3′-hydroxy-N-methylcoclaurine 4′-O-methyltransferase (4′-OMT)38 is avoided, thereby shortening the biosynthetic pathway. Ultimately, a characteristic modification was constructed, comprising the BBE module for the oxidative cyclization and the S9OMT module for successive two-staged methylation (Fig. 2). To this end, an artificially concise six-enzyme cascade was designed to potentially biosynthesize Rotundine or its derivatives, streamlining the complex de novo biosynthetic pathway and mitigating cytotoxicity caused by product or intermediate accumulation in microorganisms.

Fig. 2. Establishment of a multi-enzyme cascade reaction to efficiently synthesize Rotundine in vitro.

Fig. 2

Open in a new tab

a The BIA module (Module I) containing CARSfp (activated by Sfp) and NCS. Mg2+, ATP, and NADPH were necessary for the activity of CAR. The catalytic activity of TpCAR and TfNCS was analyzed by high-performance liquid chromatography (HPLC). The intermediate (S)-3a in the BIA module was detected. b The MT module (Module II) containing 6OMT and NMT. SAM was required as the methyl donor in this module. The catalytic activity of Ps6OMT and CNMT was analyzed by HPLC. The intermediate (S)-4a and (S)-5a in the MT module were detected. c the BBE module (Module III) and the S9OMT module (Module IV). SAM was required as the methyl donor in the reaction catalyzed by S9OMTM. The catalytic activity of EcBBE and TfS9OMTM was analyzed by HPLC. The intermediate (S)-6a and (S)-7a were detected, and the final Rotundine was observed when enlarging the spectrum highlighted in red rectangle. Source data are provided as a Source Data file.

Validation of the designed cascade for the biosynthesis of Rotundine

To verify the designed enzyme cascade, the corresponding enzymes were identified, expressed, and purified from E. coli, and their ability to produce Rotundine through the cascade was tested in vitro. At first, TpCAR from Tsukamurella paurometabola53 and TfNCS from Thalictrum flavum52 for stereocontrol were selected based on their reported reactivity and high expression levels (Supplementary Fig. 1). The results showed that the effective synthesis of (S)-1-(3-hydroxy-4-methoxybenzyl)-1,2,3,4-tetrahydroisoquinoline-6,7-diol ((S)-3a) was realized without obvious accumulation of intermediate 2a (Fig. 2a, Supplementary Fig. 2), when 1a and dopamine were incubated with purified TpCAR and TfNCS.

Next, the appropriate OMTs and NMT for the MT module were examined using (S)-3a as the substrate. Since the gene responsible for the methylation reaction of the phenolic hydroxy group at 6-position in (S)-3a has not been characterized, RnCOMT from Rattus norvegicus46, GsOMT1 from G. superba20, Ps6OMT from Papaver somniferum54, Tf6OMT from Thalictrum flavum55 and Cj6OMT from Coptis japonica56 were selected. It showed that all of them successfully produced the desired product (S)-4a (Supplementary Fig. 3a). Considering the catalytic activity (Fig. 2b, Supplementary Fig. 3a) and expression level (Supplementary Fig. 3b), Ps6OMT was chosen for further investigation. When used in combination with CNMT from Coptis japonica56, it was found that (S)-3a was efficiently converted to (S)-5a with high catalytic efficiency without obvious accumulation of the intermediate (S)-4a (Fig. 2b), which was even not affected by the catalytic order of OMT and NMT (Supplementary Fig. 4), implying the robustness of the MT module.

Last, the BBE and S9OMT modules were investigated for Rotundine biosynthesis. It is worth noting that the functionally heterologous expression of plant-derived BBE, which catalyzes the critical regioselective oxidative cyclization, remains a major challenge in constructing the protoberberine alkaloids. This difficulty is due to its low expression level and weak catalytic activity29,34,39,40. Considering the reported crystal structure57, which will facilitate structure-guided evolution later, BBE from Eschscholzia californica (EcBBE) was selected. Despite the low expression level and very weak activity (Supplementary Fig. 5), the desired product (S)-6a was detected (Fig. 2c). For the final methylation step, (S)-6a could be efficiently converted to (S)-tetrahydrocolumbamine ((S)-7a) by TfS9OMT from Thalictrum flavum, consistent with previous report28. However, the reported mutant of TfS9OMT-M111L/N191D/F205S (TfS9OMTM) exhibited very weak catalytic activity in converting (S)-7a to Rotundine ((S)-8a) (Fig. 2c), with keeping the methylation activity towards (S)-6a to synthesize (S)-7a. These results indicate the need for further systematic engineering of EcBBE and TfS9OMTM to realize high production of Rotundine. Overall, a concise six-enzyme cascade was successfully constructed to synthesize Rotundine.

Construction and optimization of the BIA and MT modules in engineered strains

To enable large-scale production of Rotundine and to avoid complex protein purification and the high cost of cofactors associated with in vitro reconstitution, the designed enzyme cascade was integrated into a whole-cell catalytic system. However, considering that the optimal pH of EcBBE (pH 9.0) is not compatible with the other enzymes (Supplementary Fig. 6)58, and the low catalytic activity of EcBBE and TfS9OMTM (Fig. 2c), the six enzymes involved in this artificial biosynthetic pathway were divided and expressed in three different bacterial strains (Supplementary Fig. 7). These strains were organized into three modules: the BIA-MT module, the BBE module, and the S9OMT module. This modular approach will allow for the optimization of reaction conditions for each enzyme, ultimately enhancing the overall production efficiency.

Initially, to alleviate the side reaction of converting aldehyde into the corresponding alcohol by endogenous reductases or dehydrogenases in E. coli (Fig. 2a), an engineered E. coli strain (IAA) was constructed by deleting seven endogenous genes encoding aldo-keto reductases (dkgB, yeaE, and dkgA), alcohol dehydrogenases (yqhD, yahK, and yjgB), and a transcription factor (yqhC)46. Then, the enzymes TpCAR, TfNCS, Ps6OMT and CNMT of the BIA-MT module were introduced into the IAA strain to construct the strain M0a for producing (S)-5a from the substrate 1a and dopamine (Supplementary Fig. 8a). However, the yield of (S)-5a was unsatisfactory (1.2%), with large amounts of substrate 1a and intermediate (S)-3a and (S)-4a accumulated (Supplementary Fig. 8b). Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) analysis of strain M0a indicated that, due to excessive exogenous gene expression, the high metabolic burden reduced the expression level of target proteins (Supplementary Fig. 8c)59. This observation suggests that further division of the BIA-MT module is necessary. Thus, the BIA and MT modules were constructed in separate strains M1a and M2a (Fig. 3a), respectively.

Fig. 3. Biosynthesis of intermediates (S)-3a and (S)-5a through BIA and MT modules.

Fig. 3

Open in a new tab

a Illustration of the designed biosynthetic pathway for the preparation of (S)-3a and (S)-5a through one-pot two-step process (using two strains expressing the BIA module and the MT module, respectively). (S)-3a was produced from 1a and dopamine catalyzed by the strain containing the BIA module, and (S)-5a was produced from the supernatant of step 1 catalyzed by the strain containing the MT module with L-Met and ATP adding exogenously. b Construction of four engineered strains harboring plasmids with different copy numbers expressing TpCAR, BsSfp and TfNCS. The green semicircle represents the ribosome binding site. c Yields of (S)-3a synthesized by engineered strains (M1a-M1d) correspond to a theoretical yield of 1a. d Time course of the production of (S)-3a from substrate 1a. It corresponds to a theoretical yield of 1a in step 1 transforming 13 mM 1a to (S)-3a using the whole-cell catalyst M1d. Black line: 1a; purple line: (S)-3a. e Construction of four engineered strains harboring plasmids with different copy numbers expressing Ps6OMT, MmSAHH, CNMT and EcMAT. f Yields of (S)-5a synthesized by engineered strains (M2a-M2d) correspond to a theoretical yield of 1a. g Time course of yields of (S)-4a and (S)-5a from (S)-3a. They correspond to a theoretical yield of 1a in step 2 converting (S)-3a in the supernatant of step 1 to (S)-5a using the whole-cell catalyst M2c. Purple line: (S)-3a; red line: (S)-4a; yellow line: (S)-5a. All data are presented as mean values of three independent experiments and the error bars indicate ±sd. Source data are provided as a Source Data file.

It showed that the coupling product (S)-3a with a yield of 55.5% was obtained (Fig. 3b, Supplementary Fig. 9), when the engineered strain M1a harboring the BIA module was incubated with the substrate 1a and dopamine. Next, engineered strains with different plasmids (M1a-M1d) were constructed to fine-tune the plasmid copy numbers. In addition, key parameters such as pH, temperature, strain concentration and the ratio of dopamine to 1a were also optimized (Fig. 3b, c, Supplementary Fig. 10). Under optimal conditions, the efficient production of 3.84 g L-1 (12.7 mM) (S)-3a with a yield of 98.0% was achieved when 13 mM 1a was incubated with dopamine at 30 °C and pH 8.0 using strain M1d (OD600 = 15) for 6 h (Fig. 3d). Importantly, (S)-3a was predominantly found in the supernatant (Supplementary Fig. 11), which will simplify the recovery of the intermediate for the subsequent processes by centrifugation.

For the MT module, to solve the issue of insufficient SAM supply in E. coli, the SAM supply system was introduced into the engineered strain M2a, which contained methionine adenosyltransferase from E. coli (EcMAT) and S-adenosylhomocysteine hydrolase from Mus musculus (MmSAHH)60. However, only a yield of 35.5% of (S)-5a was achieved when the supernatant of the BIA module was incubated with strain M2a. To improve this, the engineered strains with plasmids of different copy numbers (M2a-M2d) were constructed. Strain M2c showed the highest yield of 77.1% (Fig. 3e, f, and Supplementary Fig. 12), along with some unconverted intermediate (S)-4a (Supplementary Fig. 13). The reason could be that the change of pH, which was found at around pH 4 due to the addition of ATP, affected the catalytic effect of the MT module. After optimizing pH, temperature and strain concentration (Supplementary Fig. 14), it showed that strain M2c could efficiently produce 12.1 mM (S)-5a within 15 h, achieving a final total yield of 93.3% from 1a (Fig. 3g). This results in an effective production of 3.99 g L-1 of (S)-5a, making a 7-fold improvement over the previously reported titer in E. coli7. Given that (S)-reticuline ((S)-5a) is a crucial intermediate for the biosynthesis of many tetrahydroisoquinolines, this established method will benefit the high production of these important compounds.

Systematic engineering of EcBBE for the efficient biosynthesis of protoberberine alkaloids

Given that the functionally heterologous expression of plant-derived BBE poses an inherent bottleneck in the biosynthesis of protoberberine alkaloids29,34,39,61, as expected, only a yield of 0.2% of (S)-6a was obtained when the supernatant of the MT module (step 2) was incubated with the engineered strain M3a expressing EcBBE (Fig. 4a, Supplementary Fig. 15). To address this, we aimed to systematically engineer EcBBE in E. coli to achieve high production of these alkaloids (step 3) (Fig. 4a).

Fig. 4. Construction and optimization of the BBE module for the biosynthesis of (S)-6a.

Fig. 4

Open in a new tab

a Illustration of the designed biosynthetic pathway for the preparation of (S)-6a from the supernatant of step 2 catalyzed by the strain containing enzyme EcBBE. b Yields of (S)-6a synthesized by engineered strains (M3b-M3e) harboring plasmids with different copy numbers expressing EcBBE correspond to a theoretical yield of 1a. c Structural model of EcBBE with (S)-5a (PDB accession 3D2D)57. (S)-5a is colored in yellow, FAD is colored in cyan, and the designed mutation residues are colored in pink. d Relative activities of EcBBE mutants toward (S)-6a through in vitro catalytic assays. The relative activity is defined as the ratio of the (S)-6a titer of EcBBE mutants to that of EcBBEWT. ND: not detected. e The yield of (S)-6a synthesized by the engineered strain harboring different combinations of the optimal mutant, molecular chaperone, promoter and the FAD supply system. It corresponds to a theoretical yield of 1a. f Time course of the production yield of (S)-6a from (S)-5a. It corresponds to a theoretical yield of 1a in step 3 transforming (S)-5a to (S)-6a using the whole-cell catalyst M3z. Yellow line: (S)-5a; Blue line: (S)-6a. All data are presented as mean values of three independent experiments and the error bars indicate ±sd. Source data are provided as a Source Data file.

To improve the titer of (S)-6a by increasing the soluble and functional expression levels of EcBBE, we first carried out N-terminal signal peptide truncation, a solubilizing tag (MBP) fused with EcBBE34, and the strains optimization with different plasmids (M3b- M3e). It showed that the yield of (S)-6a was improved to 5.3% using strain M3d (Fig. 4b) with increased total and soluble expression level of EcBBE (Supplementary Fig. 16a), but still remained very low. This could be attributed to the lack of an auxiliary protein folding mechanism in E. coli for plant-derived enzymes, as part of MBP-EcBBE still existed in the form of inclusion bodies (Supplementary Fig. 16a). Thus, molecular chaperones and promoter engineering were employed to increase its expression levels of functional fraction. Five engineered strains (M3f, M3g, M3h, M3i and M3j) with different molecular chaperone plasmids (pG-KJE8, pGro7, pKJE7, pTf16, pG-Tf2) were constructed62. It showed that the production of (S)-6a increased to 17.2% when co-expressing the chaperone pGro7 with EcBBE in M3g (Supplementary Fig. 16b and 17). Then, ten constitutive promoters (ssrA, ffs, hupA, cspA, ndk, lysT, mdh, azoR, ycbK, pheM) with different expression strengths were selected from the promoter library to determine their effects for the synthesis of (S)-6a (Supplementary Fig. 18)63. It was found that strain M3k, expressing EcBBE with the promoter ssrA, exhibited the highest catalytic efficiency for (S)-5a, with a yield of 20.3%. However, the yield of (S)-6a remained low.

Next, structure-guided engineering was conducted with the assistance of sequence conservation analysis and structural investigation. Then, the substrate pocket of EcBBE (PDB accession 3D2D) was remodeled to reduce the distance between (S)-5a and the cofactor flavin adenine dinucleotide (FAD)57, aiming to enhance hydride transfer activity for improved catalytic efficiency. Thus, residues within 4.0 Å of (S)-5a were mutated with the exception of Y106 and F356 identified as conserved amino acids crucial for maintaining structural stability64 (Supplementary Fig. 19a), as well as D352, N390, and E417 forming hydrogen bonds with the hydroxyl or methoxy groups of (S)-5a for substrate binding (Supplementary Fig. 19b). Then, S280, L282, L295, R354, F386, A388, and A421 were mutated to tryptophan or phenylalanine to introduce steric hindrance, while W165 was mutated to alanine or glycine to reduce steric hindrance (Fig. 4c). Notably, the EcBBE-R354W mutant (EcBBE#) exhibited a 132-fold increase in catalytic activity for converting (S)-5a to (S)-6a compared to wild-type EcBBE (EcBBEWT) (Fig. 4d, Supplementary Fig. 20 and 21). Based on the results of molecular dynamics (MD) simulations, it suggests that this mutation enhances the protein stability, strengthens the substrate-binding by changing hydrogen bond network (Supplementary Table 1), and improves the activity of deprotonation and C-C coupling processes, in which both the distance between the carboxyl group of E417 and C3′ OH group of (S)-5a (from 2.41 Å to 1.84 Å) and the distance between the C2′ carbon and the carbon atom on the N-methyl group of (S)-5a (from 4.19 Å to 3.64 Å) are shortened to favor the berberine bridge formation (Supplementary Fig. 22-24). The production of (S)-6a significantly increased with the similar trend to that in vitro, achieving a yield of 34.7% (Fig. 4e), when strain M3u was constructed expressing EcBBE#.

Based on the above results, different combinations (M3v-M3y) of the optimal molecular chaperone, promoter, and EcBBE# were used to enhance (S)-6a production. The yield increased to 44.4% with the construction of strain M3y (Fig. 4e). By introducing the FAD supply system for BBE into M3y, including RibH, RibF and RibC65, the resulting strain M3z further increased the yield of (S)-6a to 57.3% (Fig. 4e, Supplementary Fig. 25), similar to the reaction system with the extra addition of FAD. After optimizing pH, temperature and strain concentration, the production of (S)-6a was improved to a remarkable 75.1%, with a titer of 3.19 g L-1 (Supplementary Fig. 26), which is 28-fold higher than that in the previous report29,30 (Fig. 4f). Overall, the above results highlight the effectiveness of the engineered EcBBE for the biosynthesis of protoberberine alkaloids.

Structure-guided engineering of TfS9OMT for the efficient production of Rotundine

As shown in Fig. 2c, the catalytic activity of TfS9OMTM for the 2-O-methylation of (S)-7a (step 4) was very low. To improve its transformation efficiency (Fig. 5a), structure-guided engineering of TfS9OMTM was conducted using the crystal structure model of TfS9OMTWT (PDB accession 6NEJ) (Fig. 5b)28. The goal was to obtain a mutant that not only improved the catalytic activity towards (S)-7a, but also maintained or improved the catalytic activity towards its original substrate (S)-6a. According to the docking results (Supplementary Fig. 27), the methoxy group at 9-position of the D-ring in the substrate (S)-7a is larger than the hydroxy group of the original substrate (S)-6a of TfS9OMT. To reduce the steric hindrance around the D-ring caused by the methoxy group, amino acids within 4.0 Å of TfS9OMTM surrounding the D-ring of (S)-7a were mutated to small (alanine and glycine) or medium-sized (leucine) residues, including L13, G14, S16, L111, T115, N297, T300, P301 and L304 (Fig. 5c, Supplementary Fig. 28). As a result, the mutants TfS9OMTM-S16G, TfS9OMTM-T115A, TfS9OMTM-P301G and TfS9OMTM-P301A showed 11-fold, 5-fold, 12-fold and 13-fold increases in the transformation of (S)-6a to (S)-8a compared with TfS9OMTM, respectively28. These findings demonstrate that these mutants not only kept the activity for the original substrate (S)-6a, but also significantly improved the catalytic activity toward (S)-7a. After combinational mutagenesis, a double mutant TfS9OMTM-S16G/P301A (TfS9OMT#) showed a 19-fold increase in catalytic activity compared to TfS9OMTM (Fig. 5c, Supplementary Fig. 29), without the decrease of catalytic activity for the first 9-O-methylation process. During the two-staged methylation process, another intermediate postulated as (S)-tetrahydropalmatrubine ((S)-7a’) might be formed through initial methylation at 2-hydroxyl position, indicating that both pathways with different methylation sequences are possible28 (Supplementary Fig. 30), but without affecting the production of Rotundine. Molecular docking results suggest that the improved catalytic activity may be due to a larger substrate pocket and lower binding free energy of TfS9OMT# compared to TfS9OMTM, allowing for better substrate binding (Supplementary Fig. 31, Supplementary Table 2 and 3).

Fig. 5. Construction and optimization of the S9OMT module for the efficient biosynthesis of Rotundine and various halogenated protoberberine alkaloids.

Fig. 5

Open in a new tab

a Illustration of the designed biosynthetic pathway for the preparation of Rotundine from the supernatant of step 3 catalyzed by the strain containing S9OMT. b Molecular docking of TfS9OMTM with (S)-7a. (S)-7a is colored in orange, SAM is colored in salmon pink, and the designed mutation residues are colored in green and palecyan. c Relative activities of TfS9OMTM mutants through in vitro catalytic assays. The relative activity is defined as the ratio of the titer of (S)-8a catalyzed by TfS9OMTM mutants to that of TfS9OMTM. d Construction of four engineered strains harboring plasmids with different copy numbers expressing TfS9OMT#, MmSAHH and EcMAT. e Yields of Rotundine synthesized by engineered strains (M4a-M4d) correspond to a theoretical yield of 1a. f Time course of the production yield of Rotundine. It corresponds to a theoretical yield of 1a in step 4 transforming (S)−6a to Rotundine using the whole-cell catalyst M4b. Blue line: (S)−6a; green line: Rotundine. g Biosynthesis of various halogenated protoberberine alkaloids by the optimized six-enzyme cascade. All data are presented as mean values of three independent experiments and the error bars indicate ±sd. Source data are provided as a Source Data file.

To leverage this enhanced activity, engineered strains (M4a to M4d), which expressed TfS9OMT# with different plasmids along with an SAM supply system60, were constructed to investigate their step 3 (Fig. 5d, e). All strains successfully produced the final product Rotundine, with strain M4b ability to biosynthesize Rotundine from the supernatant of exhibiting the highest activity (Fig. 5e, Supplementary Fig. 32). Under optimal conditions of pH 8.0, 40 °C and an OD600 of 21 of strain concentration, strain M4b produced Rotundine at a remarkable gram-scale titer of 2.44 g L-1. The overall production of Rotundine represented 52.8 % of the theoretical yield of 1a (Fig. 5f, Supplementary Fig. 33), over 35 times higher than the previously reported yield29. In all, efficient biosynthesis of Rotundine was realized by an optimized biocatalytic cascade from the readily available substrate 1a (Supplementary Fig. 34).

Exploration of the designed enzyme cascade for the preparation of unnatural halogenated protoberberine alkaloids

Lastly, we tested whether the optimized six-enzyme cascade was able to synthesize various unnatural halogenated protoberberine alkaloids from their corresponding acids (Fig. 5g). We then used the substrate 2b as the model for the biosynthesis of halogenated protoberberine alkaloids. It showed that, although the target product ((S)-8b) could be synthesized, CNMT exhibited low catalytic activity in vitro (Supplementary Fig. 35). Likely, due to the introduction of a larger steric resistance from new substrates33, CNMT-F92A/N332A (CNMT#) was then selected from our previously constructed mutation library46. This variant demonstrated better catalytic activity toward (S)-3b (Supplementary Fig. 35). As a result, the catalytic cascade, with the replacement of CNMT by this variant through a play-and-plug strategy, enabled efficient synthesis of various halogenated protoberberine alkaloids from a number of halogenated 2-(3-hydroxyphenyl) acetic acids, featuring chlorine or fluorine substitutions at the 4-, 5-, and 6- positions (Supplementary Fig. 36-38). These results clearly demonstrate the universality of this biosynthetic process as an efficient platform for protoberberine alkaloid production. Finally, the scaled-up preparation of halogenated protoberberine alkaloids was also studied using the whole-cell catalytic system. Taking (S)-8b as an example, a remarkable gram-scale titer of 1.01 g L-1 was also achieved. Overall, these findings highlight the capability of the designed enzyme cascade to construct various natural and unnatural protoberberine alkaloids, showcasing significant potential for industrial production.

Discussion

In this study, to achieve high production of natural and halogenated protoberberine alkaloids, we streamlined the original biosynthetic pathway by designing a concise enzyme cascade. We then discovered and engineered target enzymes, followed by systematic optimization of BBE to address common manufacturing challenges in protoberberine alkaloid biosynthesis. After modularly assembling these enzymes, we achieved gram-scale production of the drug Rotundine and various unnatural halogenated protoberberine alkaloids. In this concise enzyme cascade, 13 mM 1a, 18.2 mM dopamine, 60 mM ATP, 60 mM L-methionine and 40 mM L-ascorbic acid sodium salt were used to prepare Rotundine, achieving 52.8 % of the theoretical yield of 1a. For the industrial biosynthesis of natural and halogenated protoberberine alkaloids, further engineering of BBE and TfS9OMT# is required. Additionally, the stability of all enzymes should be considered to enable their recycling and reduce costs. Nevertheless, this work not only highlights the advantages of engineered enzyme cascades for scalable, sustainable production of both natural and unnatural products but also opens avenues for exploring their pharmacological potential.

Methods

Chemicals and materials

The genes used in the study, including TpCAR (Genebank accession WP_013126039.1), TfNCS (Genebank accession ACO90248.1), RnCOMT (Genebank accession NP_036663.1), GsOMT1 (Genebank accession QLI49050.1), Ps6OMT (Genebank accession NP_001413547.1), Tf6OMT (Genebank accession Q5C9L7.1), Cj6OMT (Genebank accession KAF9606078.1), CNMT (Genebank accession BAB71802.1), MmSAHH (Genebank accession NP_001291457.1), EcBBE (PDB accession 3D2D), and TfS9OMT (PDB accession 6NEJ), were synthesized by Exsyn-Bio (Wuxi, China) after codon optimization for E. coli, and GDH, EcMAT and BsSfp were amplified from the genome of E. coli and Bacillus subtilis. The promoters including ssrA, ffs, hupA, cspA, ndk, lysT, mdh, azoR, ycbk and pheM were amplified from the genome of E. coli. Nucleotide sequences used in this study are all described in Supplementary Information (Supplementary Data 1). Phanta Flash Super-Fidelity DNA Polymerase and One-Step PAGE Gel Fast Preparation Kit were purchased from Vazyme (Nanjing, China). MultiF Seamless Assembly Mix was purchased from ABclone (Wuhan, China). Gel Extraction Kit and GoldHi Plasmid Mini Kit were purchased from Cwbio (Taizhou, China). Primers used for gene amplification were synthesized in Exsyn-Bio (Wuxi, China), and all constructed plasmids were sequenced by Genewiz (Suzhou, China). 2-(3-hydroxy-4-methoxyphenyl) acetic acid (1a) was purchased from Energy Chemical (Shanghai, China). All other reagents and solvents used in this study were bought from China National Pharmaceutical Group Corp (Shanghai, China), Aladdin (Shanghai, China) and Macklin (Shanghai, China) (Supplementary Table 4).

Plasmids and strains construction

Plasmids, strains and primers used in this study are described in Supplementary Data 2–4. E. coli strain TOP10 was used for cloning and plasmid propagation, and BL21(DE3) was used for constructing engineered strains. Molecular cloning of plasmids were constructed in the following protocol. The expression vectors including pRSFDuet-1, pETDuet-1, pACYCDuet-1, pCDFDuet-1and pET-21b(+) were stored at our laboratory. All fragments were amplified by PCR using the corresponding primers (Supplementary Data 4). The gene fragments and vector fragments were mixed and then assembled by ligase at 50 °C for 25 min. Subsequently, the mixture was transformed into competent cells, which were plated in LB agar with appropriate antibiotics (ampicillin 100 mg L-1, kanamycin 50 mg L-1, streptomycin 50 mg L-1, or chloroamphenicol 34 mg L-1) and then incubated overnight for screening selection. Especially, plasmids contained the BIA module were only transformed into strain IAA to ensure the smooth process of the reaction.

Expression and purification of recombinant proteins

The constructed plasmids containing target genes and a His6 tag at C-terminal were transformed into BL21(DE3) for protein expression. After the optical density (OD600) reaching to 0.6–0.8, protein expression was induced by 0.1 mM β-D-1-thiogalactopyranoside (IPTG) at 18 °C for 12 h. The cells were centrifugated at 8000 g for 5 min, and resuspended with lysis buffer (25 mM Tris-HCl pH 8.0, 300 mM NaCl, 10 mM imidazole and 10% (w/v) glycerol). Next, the sample was lysed by a high-pressure homogenizor (Union-Biotech Co., Ltd., Shanghai, China), and centrifugated at 40000 g for 30 min. The resulted supernatant was loaded onto the NI-NTA agarose column. The desired proteins were eluted and desalted with desalting buffer (25 mM Tris-HCl pH 8.0, 150 mM NaCl and 10% (w/v) glycerol) after extensive washing. The results samples were concentrated to 10 mg mL-1, and stored at −80 °C for in vitro assays. The method of cell growth is the standard procedure for all cell preparation in this paper.

Enzymatic activity assay

To determine the catalytic activity of 6OMT (RnCOMT, GsOMT1, Ps6OMT, Tf6OMT, ang Cj6OMT) toward (S)-1a, a 200 μL reaction system consisting of 6OMT (5 μM), MgCl2 (10 mM), (S)-1a (2 mM), and SAM (10 mM) were mixed with buffer (50 mM Tris pH 8.0 and 100 mM NaCl) in 1.5 mL EP tube, and reacted for 4 h at 30 °C. The reaction mixture was then quenched by adding 800 μL methanol, and filtered through a 0.22 μm filter for HPLC analysis (Waters 2695) with a phenyl reverse-phase column (4.6 × 250 mm, 5 μm) detected by a 2996 photodiode array (PDA) detector. The samples were detected at the wavelength of 280 nm with a flow rate of 1 mL·min-1 at 30 °C by a linear gradient elution method (0–1 min, 10% solvent A/90% solvent B; 1-25 min, linear gradient to 50% solvent A/50% solvent B; 25-27.5 min, 100% solvent A/0% solvent B; 27.5-30 min, 10% solvent A/90% solvent B), in which solvent A was acetonitrile with 0.1% trifluoroacetic acid (TFA), and solvent B was ddH2O with 0.1% trifluoroacetic acid (TFA). The sample preparation method and HPLC method is the standard procedure for the analysis of all reaction in this paper.

To evaluate the catalytic activity of EcBBE and its mutants to (S)-5a, a 200 μL reaction system consisting of EcBBE or its mutants (5 μM) and (S)-5a (2 mM) were mixed with buffer (50 mM Tris pH 9.0 and 100 mM NaCl) in 1.5 mL EP tube, and reacted for 4 h at 30 °C. The reaction solution was quenched by adding 800 μL methanol, and filtered through a 0.22 μm filter for HPLC analysis.

To investigate the catalytic activity of Tf9OMT M and its mutants to (S)-6a, a 200 μL reaction system consisting of Tf9OMT M or its mutants (5 μM), (S)-6a (2 mM) and SAM (10 mM) were mixed with buffer (50 mM Tris pH 8.0 and 100 mM NaCl) in 1.5 mL EP tube, and reacted for 4 h at 37 °C. The reaction solution was quenched by adding 800 μL methanol, and filtered through a 0.22 μm filter for HPLC analysis.

To investigate the feasibility of six-enzyme cascade (in vitro), the one-pot four-step strategy was employed in vitro. For the BIA module, a 200 μL reaction system consisting of TpCAR (5 μM), BsSfp (5 μM), TfNCS (10 μM), GDH (5 μM), MgCl2(10 mM), ATP (4 mM), glucose (10 mM), NADP+ (2 mM), (S)-1a (2 mM), dopamine (2 mM), and L-ascorbic acid sodium salt (10 mM) were mixed with buffer (50 mM Tris pH 7.5 and 100 mM NaCl) in 1.5 mL EP tube, and reacted for 4 h at 30 °C. The above reaction mixture was boiled to quench for HPLC analysis. For the MT module, the supernatant of the BIA module was incubated with Ps6OMT (5 μM) and CNMT (5 μM) with SAM (10 mM) in 1.5 mL EP tube and reacted for 4 h at 30 °C. After boiling, the supernatant of the MT module was adjusted to pH 9.0 and used for determining the catalytic activity of EcBBE# for 4 h at 30 °C. After quenching by boiling, the pH of the resulted supernatant was adjusted to 8.0 and incubated with TfS9OMT# (5 μM) and SAM (10 mM) in 1.5 mL EP tube, reacting for 4 h at 30 °C. For HPLC analysis, the mentioned samples were all quenched by adding 4-fold methanol, and filtered through a 0.22 μm filter. Halogenated substrates were tested at the concentration of 1 mM through HPLC analysis.

Molecule docking

The structure of TfS9OMTM was built by Alphafold2 based on the crystal structure of TfS9OMT (PDB accession6NEJ, scoulerine 9-O-methyltransferase). (S)-7a was docked into TfS9OMTM using CDOCKER protocol in Discovery Studio 2016 (DS) (Dassault Systèmes BIOVIA Discovery Studio Modeling Environment, Release 2017, Dassault Systèmes, San Diego, 2016). All water molecules surrounding the protein were removed. Hydrogen atoms was added into the protein. The docking site sphere was defined according to the coordination of (S)-scourline ((S)-6a) in 6NEJ, and the docking radius was set to 7 Å. Then, the substrate was prepared by generating multi-conformations, which was docked into the TfS9OMT M using the CDOCKER protocol in Discovery Studio 2016. The appropriate complexes were chosen for further analysis after analyzing energy and geometric conformation. The protein engineering of EcBBE was conducted following the same protocol.

Molecular dynamics simulation

The pmemd module of Amber 20 software was used to perform all molecular dynamic (MD) simulations66. Initially, the structural model of EcBBE# (EcBBE-R354W)-FAD was manually constructed using PyMOL, based on the crystal structure of EcBBEWT-FAD-(S)-6a complex (PDB accession 3D2D). Then, a classical minimization was performed to eliminate the potential steric hindrance between the newly built residue and surrounding residues. Next, the substrate (S)-6a was docked into the structural model of the EcBBE#-FAD to obtain EcBBE#-FAD-(S)-6a complex using Schrödinger 2021 software (Schrödinger Release 2021-4). Before MD simulations, the parameters for the protein residues and ligands (FAD and (S)-6a) were obtained using the Amber ff14SB force field67 and the General Amber Force Field (GAFF)68, respectively. The EcBBEWT-FAD-(S)-6a and EcBBE#-FAD-(S)-6a complexes were solvated within a cubic TIP3P water box69, which has at least a minimum distance of 10 Å from the protein surface to the water box boundary. Na+ ions were added to neutralize the system using the Leap module of Amber20. Then, MD simulations was carried out as follows: (i) A minimization with 2500 steps using the steepest descent algorithm and 2500 steps using the conjugate gradient algorithm with a restraint of 50 kcal mol-1 Å-2 on EcBBEWT-FAD-(S)-6a and EcBBE#-FAD-(S)-6a. (ii) A same minimization but without any restraints. (iii) A 300 ps heating procedure from 0 to 300 K with a restraint of 50 kcal mol-1 Å-2. (iv) A 500 ps equilibration procedure with a restraint of 5 kcal mol-1 Å-2 for backbone of protein residues, FAD and (S)-6a. (v) An additional 1 ns equilibration procedure with a restraint of 2 kcal mol-1 Å-2 for FAD and (S)-6a. (vi) A 100 ns unrestrained productive sampling procedure. All MD trajectories were analyzed using Cpptraj module of Amber20. The Root Mean Square Deviation (RMSD) and the Root Mean Square Fluctuation (RMSF) were calculated according to the backbone of protein residues. Other parameters, such as Langevin thermostat, integration step and nonbonded cut-off, were consistent with our previous studies70,71.

Whole-cell biocatalysis

The cells were collected by centrifugation and washed with buffer for whole-cell reaction. The reaction of each module of the artificial biosynthetic pathway of Rotundine were conducted with a final volume of 5 mL in a 50 mL falcon tube under optimal conditions. The reaction of the time course and the characterization of products were conducted with a final volume of 50 mL in a 250 mL conical flask.

For the BIA module, fresh cells prepared by the standard procedure of cell growth were resuspended with buffer (50 mM Tris pH 8.0, 100 mM NaCl) to achieve OD600 of 15. Then the reaction was initiated by adding MgCl2(10 mM), glucose (20 mM), (S)-1a (13 mM), dopamine (18.2 mM) and L-ascorbic acid sodium salt (40 mM) at 30 °C, and was shaken at 250 rpm for 6 h. 200 μL sample was taken after 6 h, quenched by adding 800 μL methanol, and filtered through a 0.22 μm filter for HPLC analysis using the standard procedure of reaction analysis.

For the MT module, fresh cells were resuspended with the resulted supernatant of the BIA module to OD600 of 21, and was adjusted to pH 8.5. The reaction was initiated by adding ATP (30 mM) and L-methionine (30 mM) at 30 °C, and shaken at 250 rpm for 15 h. Similarly, 200 μL sample was taken after 6 h, quenched by adding 800 μL methanol, and filtered through a 0.22 μm filter for HPLC analysis using the standard procedure of reaction analysis.

For the BBE module, fresh cells expressed EcBBE# were resuspended with the resulted supernatant of the MT module to OD600 of 21, and was adjusted to pH 9.0. The reaction was initiated at 35 °C and shaken at 250 rpm for 9 h. Sample preparation for HPLC analysis using the standard procedure of reaction analysis.

Last, for the S9OMT module, fresh cells expressed TfS9OMT# were resuspended with the supernatant of the BBE module to OD600 of 21, and was adjusted to pH 8.0. The reaction was initiated by adding ATP (30 mM) and L-methionine (30 mM) at 40 °C and shaken at 250 rpm for 11 h. Sample preparation for HPLC analysis using the standard procedure of reaction analysis.

Preparation and characterization of intermediates and Rotundine

Similar to the whole-cell reaction section, a 50 mL reaction was carried out to prepare of the intermediates and Rotundine. Generally, the reaction mixture was centrifugated to collect the supernatant at 8000 g for 5 min. Then, the sample was freeze-dried and dissolved in methanol, and purified by semi-preparative HPLC (Welch, Ultimate XB-phenyl 10 × 250 mm, 5 µm) after filtering through a 0.22μm filter. The semipreparative HPLC conditions were shown in Supplementary Information. NMR data were determined by Bruker Avance spectrometers (400 or 600 MHz for 1H NMR, and 101 or 151 MHz for 13C NMR), which were shown in Supplementary Note 1. NMR spectra are presented in the Supplementary Fig. 39-50. The sample for LC-MS was prepared by filtration of the whole-cell reaction solution after adding 4-times volume of methanol. HRMS data were determined once by a Waters Maldi Synapt UPLC-MS system with a BEH C18 column (2.1×150 mm, 1.7 µm) detected by a WATERS ACQUITY photodiode array (PDA) detector at the range of 200-800 nm with a flow rate of 0.3 mL·min-1 at 45 °C by a linear gradient elution method (0-5 min, linear gradient to 50% solvent A/50% solvent B; 5-7 min, linear gradient to 100% solvent A; 7–7.5 min, 100% solvent A; 7.5-10 min, 0% solvent A/100% solvent B), in which solvent A was acetonitrile with 0.1% formic acid (FA), and solvent B was ddH2O with 0.1% formic acid (FA). The WATERS MALDI SYNAPT Q-TOF mass spectrometer was used to collect MS data in positive ion mode (parameters: mass range, 50–2000 m/z; capillary, 3500 V; cone, 40 V; source block, 100 °C; desolvation gas, 400 °C, 700 L/h; cone gas, 50 L/h; collision energy, 6-20 eV; detector, 2200 V). The HRMS data were analyzed by MassLynx (MassLynx Version 4.1) and presented in the Supplementary Note 1.

Synthesis of halogenated substrates

Synthesis of halogenated substrates was followed previous reported procedure72,73. First, boronic acid (1 equiv), P(1-nap)3 (15 mol%), Pd(OAc)2 (5 mol%), α-bromo ethyl acetate (1.2 equiv), and K3PO4 (4 equiv) were dissolved in anhydrous THF/water (v/v = 2/1), and charged in a round-bottom flask. The reaction mixture was then stirred under nitrogen at room temperature overnight. After the consumption of substrate, the solvent was removed in vacuo and extracted 3 times with EtOAc. The combined EtOAc layers were dried over Na2SO4, concentrated in vacuo and dissolved in anhydrous THF, to which LiOH (2 equiv) in water was added. The reaction mixture was stirred at room temperature. After the reaction completed, the mixture was washed with EtOAc, and the aqueous layer was acidified with HCl to pH 1 and extracted 3 times with CH2Cl2. The combined CH2Cl2 layers were dried over Na2SO4 and concentrated in vacuo. Next, 40% hydrobromic acid and AcOH were added to the residue, which was then refluxed overnight. The reaction mixture was concentrated under freeze drying to obtain the crude product, which was further purified by column chromatography (Petroleum ether: EtOAc = 10:1) to yield 1b, 1c, 1 d, 1e, 1 f and 1 g as white solids.

Statistics and reproducibility

Sample size was not predetermined using any statistical method. All data were included in the analyses. At least three biological replicates were performed for each quantification. The average values were noted in the figures, and error bars were noted to show standard deviations of the experimentally measured values.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary information

Supplementary Information (3.1MB, pdf)
Peer Review file (2.8MB, pdf)
41467_2025_57280_MOESM3_ESM.pdf (73.4KB, pdf)

Description of Additional Supplementary Files

Supplementary Data 1 (15.6KB, xlsx)
Supplementary Data 2 (11.7KB, xlsx)
Supplementary Data 3 (11.6KB, xlsx)
Supplementary Data 4 (17KB, xlsx)
Reporting Summary (88.1KB, pdf)

Source data

Source data (4.1MB, xlsx)

Acknowledgements

This work is supported by the Natural Science Foundation of Jiangsu Province (BK20202002), National Natural Science Foundation of China (32270082, 22207044, 22477047 and 22108122), the Basic Research Program of Jiangsu and the Jiangsu Basic Research Center for Synthetic Biology (BK20233003), the Fundamental Research Funds for the Central Universities (JUSRP124020), and Postgraduate Research & Practice Innovation Program of Jiangsu Province (KYCX24_2579).

Author contributions

Y.R. supervised and designed the project; F.L., Z.Y., Y.G., Z.D., Y.Z. and Z.L. performed research and data analysis; F.L., Z.Y., and Y.R. wrote the paper.

Peer review

Peer review information

Nature Communications thanks Jiazhang Lian and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

Previously reported PDB accession codes for EcBBE (3D2D [https://www.ncbi.nlm.nih.gov/Structure/pdb/3D2D])57 and TfS9OMT (6NEJ [https://www.ncbi.nlm.nih.gov/Structure/pdb/6NEJ])28 were used for data analysis. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

These authors contributed equally: Fei Li, Zhenbo Yuan, Yue Gao.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-57280-0.

References

  • 1.Bui, V. H., Rodríguez-López, C. E. & Dang, T. T. T. Integration of discovery and engineering in plant alkaloid research: recent developments in elucidation, reconstruction, and repurposing biosynthetic pathways. Curr. Opin. Plant Biol.74, 102379 (2023). [DOI] [PubMed] [Google Scholar]
  • 2.Xiang, Z.-d et al. Protoberberine alkaloids: a review of the gastroprotective effects, pharmacokinetics, and toxicity. Phytomedicine126, 155444 (2024). [DOI] [PubMed] [Google Scholar]
  • 3.Niu, Z. X., Wang, Y. T. & Wang, J. F. Recent advances in total synthesis of protoberberine and chiral tetrahydroberberine alkaloids. Nat. Prod. Rep.41, 1457 (2024). [DOI] [PubMed] [Google Scholar]
  • 4.Debnath, B. et al. Role of plant alkaloids on human health: a review of biological activities. Mater. Today Chem.9, 56–72 (2018). [Google Scholar]
  • 5.Chou, T. Q. The alkaloids of the Chinese Corydalis ambigua Cham, et Sch.(yen-hu-so). VI. Identification of corydalis D and corydalis M. Chin. J. Physiol10, 507–511 (1936). [Google Scholar]
  • 6.Du, Q. Y., Meng, X. L. & Wang, S. H. A comprehensive review on the chemical properties, plant sources, pharmacological activities, pharmacokinetic and toxicological characteristics of tetrahydropalmatine. Front. Pharmacol.13, 890078 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Wang, J. B. & Mantsch, J. R. L-tetrahydropalamatine: a potential new medication for the treatment of cocaine addiction. Future Med. Chem.4, 177–186 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Chu, H. Y., Jin, G. Z., Friedman, E. & Zhen, X. C. Recent development in studies of tetrahydroprotoberberines: mechanism in antinociception and drug addiction. Cell. Mol. Neurobiol.28, 491–499 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Liu, L., Liu, M., Zhao, W., Zhao, Y. L. & Wang, Y. Levo-tetrahydropalmatine: a new potential medication for methamphetamine addiction and neurotoxicity. Exp. Neurol.344, 113809 (2021). [DOI] [PubMed] [Google Scholar]
  • 10.Zhang, H. L. & Tang, X. Y. Combining microbial and chemical syntheses for the production of complex natural products. Chin. J. Nat. Med.20, 729–736 (2022). [DOI] [PubMed] [Google Scholar]
  • 11.Liu, S. Q., Yu, B. L., Dai, J. G. & Chen, R. D. Targeting the biological activity and biosynthesis of hyperforin: a mini-review. Chin. J. Nat. Med.20, 721–728 (2022). [DOI] [PubMed] [Google Scholar]
  • 12.Sun, S. J., Wei, Y. F., Wang, H., Tang, L. F. & Deng, B. Y. Simultaneous electrochemiluminescence determination of sinomenine, cepharanthine and tetrahydropalmatine in Stephania epigaea by capillary electrophoresis coupled with ultrasonic-assisted aqueous two-phase extraction. Int. J. Electrochem. Sci.15, 5002–5017 (2020). [Google Scholar]
  • 13.Tarabasz, D. & Kukula-Koch, W. Palmatine: a review of pharmacological properties and pharmacokinetics. Phytother. Res.34, 33–50 (2020). [DOI] [PubMed] [Google Scholar]
  • 14.Neag, M. A. et al. Berberine: botanical occurrence, traditional uses, extraction methods, and relevance in cardiovascular, metabolic, hepatic, and renal disorders. Front. Pharmacol.9, 557 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Zhong, F. R. et al. Jatrorrhizine: a review of sources, pharmacology, pharmacokinetics and toxicity. Front. Pharmacol.12, 783127 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Li, W. J. et al. Total synthesis of (-)-Canadine, (-)-Rotundine, (-)-Sinactine, and (-)-Xylopinine using a last-step enantioselective Ir-catalyzed hydrogenation. J. Org. Chem.86, 8143–8153 (2021). [DOI] [PubMed] [Google Scholar]
  • 17.Gatland, A. E., Pilgrim, B. S., Procopiou, P. A. & Donohoe, T. J. Short and efficient syntheses of protoberberine alkaloids using palladium-catalyzed enolate arylation. Angew. Chem. Int. Ed.53, 14555–14558 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Galanie, S., Thodey, K., Trenchard, I. J., Filsinger Interrante, M. & Smolke, C. D. Complete biosynthesis of opioids in yeast. Science349, 1095–1100 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Mehta, N., Meng, Y., Zare, R., Kamenetsky-Goldstein, R. & Sattely, E. A developmental gradient reveals biosynthetic pathways to eukaryotic toxins in monocot geophytes. Cell187, 5620–5637 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Nett, R. S., Lau, W. & Sattely, E. S. Discovery and engineering of colchicine alkaloid biosynthesis. Nature584, 148–153 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Wang, Y. J. et al. Discovery and engineering of the cocaine biosynthetic pathway. J. Am. Chem. Soc.144, 22000–22007 (2022). [DOI] [PubMed] [Google Scholar]
  • 22.Wang, Y. J. et al. Genomic and structural basis for evolution of tropane alkaloid biosynthesis. Proc. Natl Acad. Sci. USA120, e2302448120 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.He, S. M. et al. Identification and characterization of genes involved in benzylisoquinoline alkaloid biosynthesis in Coptis Species. Front. Plant Sci.9, 731 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Caputi, L. et al. Missing enzymes in the biosynthesis of the anticancer drug vinblastine in Madagascar periwinkle. Science360, 1235–1238 (2018). [DOI] [PubMed] [Google Scholar]
  • 25.Kunjapur, A. M., Tarasova, Y. & Prather, K. L. J. Synthesis and accumulation of aromatic aldehydes in an engineered strain of Escherichia coli. J. Am. Chem. Soc.136, 11644–11654 (2014). [DOI] [PubMed] [Google Scholar]
  • 26.Liao, D. Q. et al. Identification and developmental expression profiling of putative alkaloid biosynthetic genes in Corydalis yanhusuo bulbs. Sci. Rep.6, 19460 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Galanie, S., Thodey, K., Trenchard, I. J., Interrante, M. F. & Smolke, C. D. Complete biosynthesis of opioids in yeast. Science349, 1095–1100 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Valentic, T. R., Payne, J. T. & Smolke, C. D. Structure-guided engineering of a scoulerine 9-O-methyltransferase enables the biosynthesis of tetrahydropalmatrubine and tetrahydropalmatine in yeast. ACS Catal.10, 4497–4509 (2020). [Google Scholar]
  • 29.Jiao, X. et al. De novo production of protoberberine and benzophenanthridine alkaloids through metabolic engineering of yeast. Nat. Commun.15, 8759 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Minami, H. et al. Microbial production of plant benzylisoquinoline alkaloids. Proc. Natl Acad. Sci. USA105, 7393–7398 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Gao, J. C. et al. Biosynthesis of catharanthine in engineered Pichia pastoris. Nat. Synth.2, 231–242 (2023). [Google Scholar]
  • 32.Zhang, J. et al. A microbial supply chain for production of the anti-cancer drug vinblastine. Nature609, 341–347 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Han, J. N. & Li, S. J. De novo biosynthesis of berberine and halogenated benzylisoquinoline alkaloids in Saccharomyces cerevisiae. Commun. Chem.6, 27 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Zhao, W. L. et al. Biosynthesis of plant-specific alkaloids tetrahydroprotoberberines in engineered Escherichia coli. Green Chem.23, 5944–5955 (2021). [Google Scholar]
  • 35.Galanie, S. & Smolke, C. D. Optimization of yeast-based production of medicinal protoberberine alkaloids. Microb. Cell Fact.14, 1–13 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Chang, M. C. Y., Eachus, R. A., Trieu, W., Ro, D. K. & Keasling, J. D. Engineering Escherichia coli for production of functionalized terpenoids using plant P450s. Nat. Chem. Biol.3, 274–277 (2007). [DOI] [PubMed] [Google Scholar]
  • 37.Li, S. J., Li, Y. R. & Smolke, C. D. Strategies for microbial synthesis of high-value phytochemicals. Nat. Chem.10, 395–404 (2018). [DOI] [PubMed] [Google Scholar]
  • 38.Hawkins, K. M. & Smolke, C. D. Production of benzylisoquinoline alkaloids in Saccharomyces cerevisiae. Nat. Chem. Biol.4, 564–573 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Zhang, Q., Wu, Y. H., Huang, X. S., Liu, H. L. & Wang, Y. Design and optimization for efficient production of (S)‑Canadine in Escherichia coli. ACS Sustainable Chem. Eng.12, 6941–6951 (2024). [Google Scholar]
  • 40.Gou, Y. et al. Intein-mediated temperature control for complete biosynthesis of sanguinarine and its halogenated derivatives in yeast. Nat. Commun.15, 5238 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Pinto, N. D. C. et al. Antimicrobial Annona muricata L. (soursop) extract targets the cell membranes of Gram-positive and Gram-negative bacteria. Ind. Crops Prod.107, 332–340 (2017). [Google Scholar]
  • 42.Huffman, M. A. et al. Design of an in vitro biocatalytic cascade for the manufacture of islatravir. Science366, 1255–1259 (2019). [DOI] [PubMed] [Google Scholar]
  • 43.Moody, E. R., Obexer, R., Nickl, F., Spiess, R. & Lovelock, S. L. An enzyme cascade enables production of therapeutic oligonucleotides in a single operation. Science380, 1150–1154 (2023). [DOI] [PubMed] [Google Scholar]
  • 44.Hou, X. et al. Structural analysis of an anthrol reductase inspires enantioselective synthesis of enantiopure hydroxycycloketones and β-halohydrins. Nat. Commun.14, 353 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Mukherjee, G. et al. Inspiration from Nature: Influence of Engineered Ligand Scaffolds and Auxiliary Factors on the Reactivity of Biomimetic Oxidants. ACS Catal.11, 9761–9797 (2021). [Google Scholar]
  • 46.Gao, Y. et al. Modular assembly of an artificially concise biocatalytic cascade for the manufacture of phenethylisoquinoline alkaloids. Nat. Commun.15, 30 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Medema, M. H., Breitling, R., Bovenberg, R. & Takano, E. Exploiting plug-and-play synthetic biology for drug discovery and production in microorganisms. Nat. Rev. Microbiol.9, 131–137 (2011). [DOI] [PubMed] [Google Scholar]
  • 48.Smith, D. R. M., Grüschow, S. & Goss, R. J. M. Scope and potential of halogenases in biosynthetic applications. Curr. Opin. Chem. Biol.17, 276–283 (2013). [DOI] [PubMed] [Google Scholar]
  • 49.Müller, K., Faeh, C. & Diederich, F. Fluorine in pharmaceuticals: looking beyond intuition. Science317, 1881–1886 (2007). [DOI] [PubMed] [Google Scholar]
  • 50.Trenchard, I. J., Siddiqui, M. S., Thodey, K. & Smolke, C. D. Production of the key branch point benzylisoquinoline alkaloid reticuline in yeast. Metab. Eng.31, 74–83 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Pyne, M. E. et al. A yeast platform for high-level synthesis of tetrahydroisoquinoline alkaloids. Nat. Commun.11, 3337 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Sheng, X. & Himo, F. Enzymatic Pictet-Spengler reaction: computational study of the mechanism and enantioselectivity of norcoclaurine synthase. J. Am. Chem. Soc.141, 11230–11238 (2019). [DOI] [PubMed] [Google Scholar]
  • 53.Finnigan, W. et al. Characterization of carboxylic acid reductases as enzymes in the toolbox for synthetic chemistry. ChemCatChem9, 1005–1017 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Samanani, N., Alcantara, J., Bourgault, R., Zulak, K. G. & Facchini, P. J. The role of phloem sieve elements and laticifers in the biosynthesis and accumulation of alkaloids in opium poppy. Plant J.47, 547–563 (2006). [DOI] [PubMed] [Google Scholar]
  • 55.Robin, A. Y., Giustini, C., Graindorge, M., Matringe, M. & Dumas, R. Crystal structure of norcoclaurine-6-O-methyltransferase, a key rate-limiting step in the synthesis of benzylisoquinoline alkaloids. Plant J.87, 641–653 (2016). [DOI] [PubMed] [Google Scholar]
  • 56.Subrizi, F. et al. Multienzyme one-pot cascades incorporating methyltransferases for the strategic diversification of tetrahydroisoquinoline alkaloids. Angew. Chem. Int. Ed.60, 18673–18679 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Winkler, A. et al. A concerted mechanism for berberine bridge enzyme. Nat. Chem. Biol.4, 739–741 (2008). [DOI] [PubMed] [Google Scholar]
  • 58.Steffens, P., Nagakura, N. & Zenk, M. H. Purification and characterization of the berberine bridge enzyme from Berberis beaniana cell cultures. Phytochemistry24, 2577–2583 (1985). [Google Scholar]
  • 59.Wang, F. et al. One-pot biocatalytic route from cycloalkanes to α,ω‐dicarboxylic acids by designed Escherichia coli consortia. Nat. Commun.11, 5035 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Popadic, D. et al. A bicyclic S-adenosylmethionine regeneration system applicable with different nucleosides or nucleotides as cofactor building blocks. RSC Chem. Biol.2, 883–891 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Winkler, A., Hartner, F., Kutchan, T. M., Glieder, A. & Macheroux, P. Biochemical evidence that berberine bridge enzyme belongs to a novel family of flavoproteins containing a bi-covalently attached FAD cofactor. J. Biol. Chem.281, 21276–21285 (2006). [DOI] [PubMed] [Google Scholar]
  • 62.Kim, S. A., Shin, K. C. & Oh, D. K. Complete biotransformation of protopanaxadiol-type ginsenosides into 20-O-β-Glucopyranosyl-20(S)-protopanaxadiol by permeabilized recombinant Escherichia coli cells coexpressing β-glucosidase and chaperone genes. J. Agric. Food Chem.67, 8393–8401 (2019). [DOI] [PubMed] [Google Scholar]
  • 63.Zhou, S. H. et al. Obtaining a panel of cascade promoter-5′-UTR complexes in Escherichia coli. ACS Synth. Biol.6, 1065–1075 (2017). [DOI] [PubMed] [Google Scholar]
  • 64.Capra, J. A. & Singh, M. Predicting functionally important residues from sequence conservation. Bioinformatics23, 1875–1882 (2007). [DOI] [PubMed] [Google Scholar]
  • 65.Lin, Z. Q. et al. Metabolic engineering of Escherichia coli for the production of riboflavin. Microb. Cell Fact.13, 1–12 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 66.Case, D. A. et al. AMBER2020, (University of California, San Francisco, 2020).
  • 67.Maier, J. A. et al. ff14SB: improving the accuracy of protein side chain and backbone parameters from ff99SB. J. Chem. Theory Comput.11, 3696–3713 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 68.Wang, J. M., Wolf, R. M., Caldwell, J. W., Kollman, P. A. & Case, D. A. Development and testing of a general amber force field. J. Comput. Chem.25, 1157–1174 (2004). [DOI] [PubMed] [Google Scholar]
  • 69.Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W. & Klein, M. L. Comparison of simple potential functions for simulating liquid water. J. Chem. Phys.79, 926–935 (1983). [Google Scholar]
  • 70.Deng, Z. W. et al. Mechanistic investigation on C-C bond cleavage of anthraquinone catalyzed by an atypical nonheme Iron-dependent dioxygenase BTG13. ACS Catal.14, 797–811 (2024). [Google Scholar]
  • 71.Guo, B. D. et al. Enhancement of rebaudioside M production by structure-guided engineering of glycosyltransferase UGT76G1. J. Agric. Food Chem.70, 5088–5094 (2022). [DOI] [PubMed] [Google Scholar]
  • 72.Kinney, W. A. et al. Discovery of KLS−13019, a cannabidiol-derived neuroprotective agent, with improved potency, safety, and permeability. ACS Med. Chem. Lett.7, 424–428 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 73.Schubert, T. J. et al. Structure-activity relationship studies of the aryl acetamide triazolopyridazines against reveals remarkable role of fluorine. J. Med. Chem.66, 7834–7848 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Information (3.1MB, pdf)
Peer Review file (2.8MB, pdf)
41467_2025_57280_MOESM3_ESM.pdf (73.4KB, pdf)

Description of Additional Supplementary Files

Supplementary Data 1 (15.6KB, xlsx)
Supplementary Data 2 (11.7KB, xlsx)
Supplementary Data 3 (11.6KB, xlsx)
Supplementary Data 4 (17KB, xlsx)
Reporting Summary (88.1KB, pdf)
Source data (4.1MB, xlsx)

Data Availability Statement

Previously reported PDB accession codes for EcBBE (3D2D [https://www.ncbi.nlm.nih.gov/Structure/pdb/3D2D])57 and TfS9OMT (6NEJ [https://www.ncbi.nlm.nih.gov/Structure/pdb/6NEJ])28 were used for data analysis. Source data are provided with this paper.

Articles from Nature Communications are provided here courtesy of Nature Publishing Group

RESOURCES