Tandemly duplicated CYP82Ds catalyze 14

May 20, 2023

Nature Communications volume 14, Article number: 875 (2023) Cite this article

3534 Accesses

2 Citations

4 Altmetric

Metrics details

Triptolide is a valuable multipotent antitumor diterpenoid in Tripterygium wilfordii, and its C-14 hydroxyl group is often selected for modification to enhance both the bioavailability and antitumor efficacy. However, the mechanism for 14-hydroxylation formation remains unknown. Here, we discover 133 kb of tandem duplicated CYP82Ds encoding 11 genes on chromosome 12 and characterize CYP82D274 and CYP82D263 as 14-hydroxylases that catalyze the metabolic grid in triptolide biosynthesis. The two CYP82Ds catalyze the aromatization of miltiradiene, which has been repeatedly reported to be a spontaneous process. In vivo assays and evaluations of the kinetic parameters of CYP82Ds indicate the most significant affinity to dehydroabietic acid among multiple intermediates. The precursor 14-hydroxy-dehydroabietic acid is successfully produced by engineered Saccharomyces cerevisiae. Our study provides genetic elements for further elucidation of the downstream biosynthetic pathways and heterologous production of triptolide and of the currently intractable biosynthesis of other 14-hydroxyl labdane-type secondary metabolites.

Tripterygium wilfordii Hook. F., a medicinal plant also known as Lei Gong Teng, has been used in China for >500 years (Ming dynasty, AD 1476) for the treatment of autoimmune diseases, and modern pharmacological studies have shown its extensive antitumor, anti-inflammatory, and immunosuppressive effects1,2. The root is the main medicinal part of T. wilfordii and contains up to 415 chemical components, including terpenoids and alkaloids3. Among this rich treasure trove of compounds, triptolide (1) is undoubtedly an important contributor to pharmaceutical properties4.

Triptolide, an 18(4 → 3) abeo-abietane diterpenoid with a tri-epoxy group and α,β-unsaturated lactone moiety (Fig. 1a), has been designed with multiple structure-based artificial derivatives, particularly through carbon-14 (C-14) modifications, which have been implemented to improve its water solubility, and some of the generated compounds have entered clinical trials (ClinicalTrials.gov), leading to a tremendous demand for triptolide. However, the content of 1 in its native plant is only 139 ng·g−1 dry weight5, and the chemical synthesis approach exhibits low synthesis efficiency and multiple steps due to the stereochemical complexity of its structure (e.g., three epoxy groups, unsaturated lactone, and nine chiral centers)6. Although the cambial meristematic cells of plant tissue culture technology have increased the yield to 138.1 µg·g−1 7, synthetic biology strategies based on elucidating the biosynthetic pathway appear to be a more promising, sustainable, and alternative method8,9 and have been successfully applied to achieve heterologous acquisition of potential intermediates and analog triptonide in microorganisms10,11,12. In contrast, natural products such as tripdiolide (2), triptolidenol (3) and triptriolide (4), which have the same 18(4 → 3) abeo-abietane skeleton and similar biosynthetic pathways as 1 (Fig. 1a), also have multiple pharmaceutical properties13. Elucidating the biosynthetic pathway of 1 will ultimately lead to pharmaceutical, economic and environmental benefits.

a Structures of major bioactive diterpenoids of T. wilfordii. b Proposed triptolide biosynthetic pathway. A solid arrow represents an identified reaction, and a dotted arrow indicates an unknown pathway. GGPP: Geranylgeranyl diphosphate. c Coexpression profiles of CYPs with genes in the triptolide biosynthetic pathway. TwTPS7(v2), TwTPS27(v2) and CYP728B70 are marked in red. The black triangles indicate CYP82D subfamily genes that cluster with identified functional genes. The gene ID of CYP82D274 is TW12G01155.1. Root periderm (RB), root phloem (RP), root xylem (RX), stem vascular bundle (PS), stem periderm (SB), and leaf (L). Source data are provided as a Source Data file.

The core skeleton cyclization and functional decoration in the biosynthesis of 1 are initiated by two types of diterpene synthases, class II TwTPS7(v2) and class I TwTPS27(v2)5,14, which cyclize geranylgeranyl diphosphate (GGPP) to produce olefin miltiradiene (5) and catalyze double-bond rearrangement on the C-ring for spontaneous conversion to stable aromatized abietatriene (6)8,15, and CYP728B70 then catalyzes carboxylation at C-18 to generate dehydroabietic acid (7)16. However, the multiple downstream steps from 7 to 1 remain enigmatic and are mainly divided into C-14 hydroxylation, C-18, 19 lactonization, and triepoxidation processes. According to the proposed biosynthetic pathway (Fig. 1b), cytochrome P450s (CYPs) with hydroxylation, epoxidation, and isomerization functions17 are highly likely to contribute to the biosynthesis of 1, which has been shown to consist of a minimal set of four biosynthetic components of CYPs for the heterologous production of triptonide12.

Here, we reveal that CYP82D274 and CYP82D263 in tandem duplicated CYP82Ds catalyze the C-14 hydroxylation grid in the biosynthesis of 1 through genetic manipulation in native plant cells and functional characterization in a heterologous host. The two CYP82Ds can also transform the aromatization of 5 and promote the production of the rate-limiting 7. In vivo assays and kinetic parameters indicate that CYP82D274 prefers to catalyze 7 in the pathway and thereby advance the biosynthesis of 1. The intermediate 14-hydroxy-dehydroabietic acid (11) is successfully biosynthesized de novo in Saccharomyces cerevisiae. This study provides a systematic introduction to C-14 hydroxylases from evolutionary hypotheses and functional characterization to heterologous biosynthesis and thus paves the way for elucidation of the biosynthetic pathway of 1 and other 14-hydroxyl labdane-/abietane-type secondary metabolites.

Pathway genes undertake the biosynthesis of secondary metabolites together, and these genes often exhibit similar expression patterns18,19,20; therefore, coexpression analysis is one of the most powerful approaches for the initial screening of functional-associated genes from massive transcriptome data. To identify the CYP gene encoding the enzyme responsible for the biosynthesis of 1, we analyzed 416 CYPs in the T. wilfordii transcriptome (NCBI SRA accession SRP199495) of different plant tissues for expression correlation with previously characterized biosynthetic genes, including TwTPS7(v2), TwTPS27(v2) and CYP728B70. The expression profiles in the heatmap were grouped into 67 clusters, and 38 genes from 16 CYP families, including CYP71, CYP72, CYP76, and CYP82, exhibited the same pattern of root periderm-specific high expression and clustered with identified functional genes (Fig. 1c). Notably, the most numerous genes, TW12G01155.1 (CYP82D274), TW12G01159.1, TW12G01162.1, TW12G01165.1, and TW12G01165.2, belonged to the CYP82D subfamily, whose family is reportedly involved in terpenoid biosynthesis21,22. The genes belonging to the CYP82D subfamily were selected as candidate CYPs for further analysis.

A chromosomal localization analysis of all CYP82D genes revealed that they were localized at Chr06, Chr08, Chr11, Chr12, Chr15, and Chr19. Further local BLAST analysis of the whole genome using the nucleic acid and amino acid sequences of CYP82D genes revealed a tandem duplication gene cluster at Chr12 containing 11 complete CYP82D genes, which had a length of 133 kb with 4 additional incomplete CYP82D gene residues (Fig. 2a). This tandem duplication of CYP82Ds occurred in dynamic chromosomal regions enriched in transposable elements (TEs) and was presumably generated by the duplication and rearrangement of long terminal repeat (LTR) elements of retrotransposons within chromosomes. A phylogenetic analysis of orthologous CYP82D genes (Fig. 2b) indicated that all TwCYP82Ds clustered into six clades, and the duplication event and neofunctionalization of Clade I and Clade II appeared to have occurred at 26.52 MYA, which was comparable to the timing of the T. wilfordii whole-genome triplication (WGT) event16. The synonymous substitution rate (Ks) values of orthologous CYP82D genes indicated that the specific expansion of the CYP82D genes occurred via tandem duplication (Supplementary Data 1).

a Chromosomal localization of the 133-kb tandem duplicated CYP82Ds on Chr12 encoding 11 complete genes. The same color indicates orthologous CYP82D genes and the small triangles represent the incomplete gene residues. b Phylogenetic analysis of the TwCYP82D genes. The phylogenetic tree was constructed based on the maximum likelihood method (1000 bootstraps). The numbers represent the predicted divergence time and WGT indicates the whole-genome triplication event of T. wilfordii.

Heatmap analysis of the genes in Clade I and Clade II with known functional genes TwTPS7(v2), TwTPS27(v2), and CYP728B70 as well as the content of 1 was achieved by combining the transcriptomic data and metabolite analysis (Supplementary Fig. 1). 1 exhibited a distribution pattern of root periderm and leaf tissue-specific expression. CYP82D274 and TW12G01165.1 in Clade I showed the same high expression pattern in root tissues as the previously characterized genes, whereas CYP82D263 in Clade II showed high expression in the root periderm, stem periderm, and leaves. TW12G01158.1 was highly expressed in the stem periderm and leaves, particularly the stem periderm. In addition, TW12G01156.1 and TW12G01157.1 were not detected and were supposedly selectively silenced by gene redundancy during evolution. In particular, TW08G01036.1 (CYP82D213) exhibited root peridermal-specific expression, and CYP82D213 is reportedly involved in the final step of the analog triptonide biosynthetic pathway. The transcriptional expression of the other five complete CYP82D genes of this tandem duplication gene cluster was also analyzed. In Clade V, TW12G01162.1 was highly expressed in the root and stem periderm, whereas TW12G01159.1 had a base deletion at position 44 in the coding region, causing termination of protein translation at position 26. TW12G01164.1 in Clade VI was highly expressed in the stem periderm, whereas the expression of TW12G01161.1 and TW12G01163.1 was too low to be detected.

Among the CYP82D genes discovered from the transcriptomes, 11 were cloned (Supplementary Table 1). To investigate the biochemical activity of the CYPs, these 11 CYP82Ds were integrated into the plasmid pESC-LEU expressing both CYP and cytochrome P450 oxidoreductase 3 from T. wilfordii (TwPOR3) and then transformed into the yeast strain BY4741. Substrate feeding has been proven to be effective in yeast fermentation16,23; thus, the intermediate 7 was fed into the cultures, and the fermentation products were extracted. Among 11 CYP82Ds, CYP82D274 and CYP82D263 accepted 7 as a substrate with the same three products at m/z 330 (Fig. 3a and Supplementary Fig. 2). The major product 11 was enriched and identified as 14-hydroxy-dehydroabietic acid (11), which was independently confirmed by 1H, 13C, HSQC, 1H-1H COSY, and NOESY NMR analyses (powder purity 98.23%) (Supplementary Figs. 3-7 and Supplementary Note 1). The additional minor product 13 was identified as 15-hydroxy-dehydroabietic acid upon comparison to the authentic standard (Supplementary Fig. 8). And further optimization of the GC‒MS detection by increasing the MS resolution revealed that CYP82D274 and CYP82D263 catalyzed the production of another byproduct, 12-hydroxy-dehydroabietic acid (18)24, which is isomeric with 11 and has identical MS fragments (Supplementary Fig. 9). Altogether, CYP82D274 and CYP82D263 can catalyze 7 to produce 11 by inducing C-14 hydroxylation (Fig. 3b), albeit with promiscuity (C-12, C-15, etc.).

a GC‒MS analysis of methylated products of CYP82D274 and CYP82D263 catalyzing dehydroabietic acid (7) in vivo or in vitro. Empty vector (EV) denotes yeast transformed with an empty vector without CYP. b Catalytic process in a. c Kinetic profiles of CYP82D274 and CYP82D263 catalyzing 7 in vitro. The quantification of 11 was based on the standard curve y = 0.7336x-0.014 (R² = 0.9994) obtained by LC-TQ-MS/MS. The concentration of CYPs was estimated by measuring the reduced CO-difference spectrum. Kinetic parameters were calculated by nonlinear regression analysis using the Michaelis‒Menten model. Data are presented as mean values ± standard deviation SD from three biological independent replicates, and the black circles represent the individual data points. Source data are provided as a Source Data file.

As resulted by in vitro enzyme assays, both CYP82D274 and CYP82D263 could exhibit their major 14-hydroxylase functions and generated 11 in the presence of TwPOR3, NADPH, substrate 7, and cofactors (Fig. 3a). The Km values of CYP82D274 and CYP82D263 were 0.99 ± 0.17 μM and 8.42 ± 1.89 μM, respectively, at the optimal reaction time (Fig. 3c and Supplementary Fig. 10a). These steady-state kinetic parameters indicated that CYP82D274 was significantly superior to CYP82D263 in terms of substrate affinity and catalytic efficacy.

To further investigate the definitive role of CYP82D274 and CYP82D263 in 1 biosynthesis, RNA interference (RNAi) was utilized to knock down the expression of these two genes in T. wilfordii suspension cells. All TwCYP82D nucleic acid sequences in the genome showed a high identity of 69.61%, mainly between CYP82D274 and CYP82D263, with 81.41% concordance. We specifically selected a 459-bp (nucleotides 459-917) fragment of CYP82D274 and a 498-bp (nucleotides 373-870) fragment of CYP82D263 to construct the binary vector pK7GWIWG2D (II) and transformed them into suspension cells by plasmid bombardment25. The electrophoretic bands of the vector-specific fragment indicated successful transformation (Supplementary Fig. 11a), and detection of the expression of CYP82D274 and CYP82D263 showed that they were targeted for disruption as expected (Fig. 4d). Metabolite profiling revealed that the inhibition rate of 1 was 60.37% and a considerable reduction in the accumulation of triptophenolide (10), triptobenzene D (8) and the direct product 11 in the CYP82D274-RNAi cell lines (Fig. 4d and Supplementary Data 2). In particular, the expression of CYP728B70, TwTPS27(v2), and TwTPS7(v2) exhibited opposing changes, which were hypothesized to be intergenic feedback regulation. In the CYP82D263-RNAi cell lines, three metabolites, 1, 10 and 8, were significantly decreased. Furthermore, based on the inhibition rates of individual metabolites, we observed stronger inhibition of metabolites in CYP82D274-RNAi and CYP82D263-RNAi as the pathway progressed downstream, suggesting a cascade accumulation of metabolites.

a MALDI-MSI analysis of the distribution of triptolide intermediates in T. wilfordii root tissue. DHB was used as a matrix for metabolite imaging in the positive ion mode. The colors represent the intensity percentage, and each image is independent. b Relative content of diterpenoids and relative quantification of gene expression in 14-hydroxy-dehydroabietic acid (11)-fed cell lines. Cells that were not fed 11 served as a negative control. P-values of genes with significant differences in expression between groups were 0.0375, 0.0178, 0.0031, 0.0448, and 0.0294 in that order. c Metabolite analysis of 18O-labeled 11-treated cell samples. 18O-11 was fed to T. wilfordii suspension cells that inhibited the biosynthesis of precursors, and wild-type (WT) and unlabeled 11-treated cells were used as negative controls. 18O-11 was converted into 18O-triptinin B (9) in plant cells, and the mass spectra of labeled versus unlabeled metabolites are provided. d Relative content of diterpenoids and relative gene expression in CYP82D274 and CYP82D263 RNAi cell lines. e Relative content of diterpenoids in CYP82D274- and CYP82D263-overexpressing cell lines. The cell lines bombarded with corresponding empty vectors served as controls in d and e, and significant differences (P-values) in metabolites between the groups are shown in Supplementary Data 2 and 3. b, d, e The relative quantification of each metabolite was calculated by dividing each sample by the average content of the control group. The relative expression of genes was determined by the 2−△△Ct method. EFLα was designated as the housekeeping gene, and the corresponding control group was assigned as the reference sample. Data are presented as mean values ± SD (n = 3 biologically independent replicates). ***P < 0.001, **P < 0.01 and *P < 0.05 determined by two-sided Student’s t test. Source data are provided as a Source Data file.

CYP82D274 and CYP82D263 were also specifically overexpressed in plant cells, as evidenced by electrophoretic bands and gene expression (Supplementary Fig. 11). Targeted metabolite analysis indicated that the overexpression of CYP82D274 and CYP82D263 resulted in a significant increase in the accumulation of 1 and 10 as well as the direct product 11 (Fig. 4e and Supplementary Data 3). The results of these in vivo assays suggested that CYP82D274 and CYP82D263 are involved in 1 biosynthesis as 14-hydroxylases.

To provide direct evidence showing whether 11 is implicated in 1 biosynthesis, mass spectrometry imaging and metabolite analysis were employed. Previous studies found that 1 exhibits a build-up property specific to root periderm tissues (Supplementary Fig. 1). The distribution of metabolites in fresh roots was investigated in a targeted manner using MALDI-TOF-MS, and the acquainted intermediates 6 and 10 were found to accumulate within the periderm (Fig. 4a). In addition, the putative intermediates 7 and 826, including the compound 11 discovered in this study, with carboxyl groups also displayed a high distribution in the periderm region, elucidating their spatial correlation. However, 1 was not available for MALDI-MS imaging due to the extremely low content of the compound, the choice of medium, the ionic pattern, and the competition of surrounding ions for laser ionization possibilities.

In combination with the abovementioned tissue distribution pattern of the intermediates, we fed 11 to T. wilfordii suspension cell cultures and found that the contents of 1 and 10 were elevated 35.73 and 35.60 times, respectively, compared with the levels in the negative control (Fig. 4b and Supplementary Data 4). Moreover, the catalytic substrate 7 of CYP82D274 and CYP82D263 was markedly reduced. The results further showed that all downstream pathway genes known to be involved in 1 biosynthesis (CYP82D274, CYP82D263, CYP728B70, TwTPS27(v2), TwTPS7(v2), and TwGGPS) showed reductions in expression to 0.42-0.76-fold in different proportions but did not affect genes involved in the upstream MVA and MEP pathways (Fig. 4b). To ensure that the abovementioned changes in metabolites were caused by 11, stable isotope labeling was employed for more in-depth analysis. The use of the MEP inhibitor fosmidomycin and the use of the gene gun to suppress TwTPS7(v2) and TwTPS27(v2) inhibited the biosynthesis of precursors, whereas the elicitor subsequently increased the expression of downstream genes, as revealed when 18O-11 was fed to the cells. Although the native metabolic pathway was not completely inhibited and endogenous 1 and 10 interfered with the determination of isotopically labeled products, we still found the peak of 18O-labeled triptinin B (9) (Fig. 4c), which could be further lactonized by CYP71BE to generate the expected intermediate 1012. The above-described results indicated that 11, which has a carboxyl group, serves as a precursor in the late stage of 1 biosynthesis.

When CYP82D274 and CYP82D263 were overexpressed, a 0.69-fold decrease in 8 and a 1.45-fold increase in ferruginol (15) were also observed, which appeared to indicate other catalytic processes played by these two genes (Fig. 4e and Supplementary Data 3). The fermentation of yeast strains harboring CYP82D274 or CYP82D263 and subsequent incubation with the intermediate 8 revealed that CYP82D274 generated three hydroxylated products with a charge of m/z + 16 (Fig. 5a). The retention time and mass spectral fragmentation for the major product matched the authentic standard for the derivatization of 9 (Supplementary Fig. 12), indicating that CYP82D274 played a C-14 hydroxylation role in catalyzing the formation of 9 from 8. In addition, the incubation of substrate 5 further gave rise to two different ratios of C-14 and C-12 hydroxylated products (Fig. 5b and Supplementary Fig. 13). For CYP82D274, the major product 14 was identified to be 14-hydroxy-abietatriene with small amounts of 15 (12-hydroxy-abietatriene or ferruginol, Supplementary Fig. 14). For CYP82D263, 5 was converted into the hydroxylated major product 15 accompanied by trace amounts of 14.

a GC‒MS analysis of methylated products with triptobenzene D (8) as the substrate. b GC‒MS analysis of the catalytic products of miltiradiene (5) in engineered yeast. The mass spectrum results are shown in Supplementary Figs. 12 and 13. c Relative yield of abietatriene (6) in spontaneous, CYP82D274 and CYP82D263 cultures. Data are presented as mean values ± SD (n = 3 biologically independent replicates). d Catalytic process of the CYP82D274 and CYP82D263 metabolic grid for triptolide (1) biosynthesis. A solid arrow represents an identified reaction, and a dotted arrow indicates an unknown reaction. Source data are provided as a Source Data file.

Double bond rearrangement in the C-ring of olefin 5 to yield stable aromatized 627 has been repeatedly reported to be a spontaneous reaction process14,28,29,30. In our study, we transformed plasmids into engineered yeast BY-PS216, which could self-produce 5 as a substrate. The presence of a double peak for 5 and 6 in CYP82D274 and CYP82D263 compared with the single peak for 5 in the EV (Fig. 5b) indicated that CYP82D274 and CYP82D263 were capable of catalyzing the aromatization of 5 to generate 6. Nevertheless, we demonstrated the existence of spontaneous reactions with small amounts of 6 (Fig. 5c), and if the yield of 6 in the EV was regarded to be 1.00, the relative yield of 6 in CYP82D274 was 180.09 and that in CYP82D263 was 23.89. This finding suggested that CYP82D274 and CYP82D263 can specifically and effectively catalyze the aromatization of 5, whereas other genes in the same CYP82D subfamily cannot (Supplementary Fig. 15).

Based on these results, we conclude that CYP82D274 and CYP82D263 catalyze important hydroxylation and aromatization processes and contribute to the biosynthesis of 1 through multiple pathways (Fig. 5d).

In the previous section, we revealed that the wide substrate range of CYP82D274 enabled the simultaneous catalysis of different substrates, and CYP728B70 catalyzes the reaction of 5 to generate 716. By feeding strains harboring CYP728B70 with 14, the fermentation products were detected to contain both abietatriene-14,18-diol (a CYP728B70 catalyzed intermediate, 19) and 11 (Supplementary Fig. 16)31, suggesting that the biosynthetic pathway is not a linear and specific pathway but rather that the products are generated via multiple pathways together through a metabolic grid, a phenomenon that has also been found in forskolin and tanshinones32,33.

In the face of a metabolic grid with multiple substrates, the substrate affinity and catalytic efficiency of the enzymes are quite different and can be reflected by kinetic parameters34. We found that the Km values of CYP82D274 and CYP82D263 for catalyzing 7 equaled 0.99 ± 0.17 μM and 8.42 ± 1.89 μM (Fig. 3c), indicating their higher sensitivity to 7. We also carried out microsomal experiments with CYP82D274 and CYP82D263 in the presence of NADPH, a redox partner, cofactors, and the substrates 5 and 8. The results showed that only CYP82D274 could catalyze 5 to produce small amounts of 14 with a Km of 47.30 ± 12.98 μM (Supplementary Fig. 10b), but the target product was not detected under the conditions of catalytic 8 or CYP82D263-catalyzed 5. In addition, the Vmax values of CYP82D274 for catalyzing 7 and 5 were 140.30 ± 5.10 umol of product·min−1·(mmol of protein) −1 and 44.08 ± 6.13 umol of product·min−1·(mmol of protein)−1, these kinetic parameters indicated that CYP82D274 has a higher preference for substrate 7.

To further address the unknown biosynthetic pathway of 1, self-produced engineered yeast was first constructed to provide sufficient amounts of the precursor 11 for gene mining. CYP82D274 and CYP82D263 were coexpressed with CYP728B70, suggesting that yeast is capable of expressing multiple CYPs simultaneously to produce 11 (Fig. 6a and Supplementary Fig. 17). To achieve a high yield of 11, three functional modules were designed and investigated for optimization. Multispecies-derived diterpene synthases confirmed that tSmKSL1-CfTPS1 (the fusion of truncated ent-kaurene synthase-like 1 from Salvia miltiorrhiza and terpene synthase from Coleus forskohlii) was the most efficient 5 biosynthase10. The GGPP high-yielding strain BY-HZ1610, tSmKSL1-CfTPS1 (module I) and the more active CYP82D274 (module III) were selected for the production of 11. In this section, we utilized synthetic biology strategies for the optimization of genetic elements and chassis strains.

a GC‒MS analysis of methylated products in CYP82D274 and CYP82D263 coexpressed with CYP728B70. b Screening for optimal combinations of CYP and TwPOR. Optimal genes are marked in red. N.D. indicates not detected. c Yield of 11 in the engineered strains. The images show the biomass of haploid and diploid strains under the same culture conditions. The genotype schematic is shown in the figure, and detailed information is provided in Supplementary Table 3. Data are presented as mean values ± SD (n = 3 biologically independent replicates). Source data are provided as a Source Data file.

The optimization of protein‒protein interactions between CYP and its redox partner POR to enhance the electron transfer efficiency can alleviate the poor coupling of CYP systems and increase terpenoid production35. Because module II (generation of 7) acted as a rate-limiting step hindering the final product yield, this module was first optimized for the genetic elements. Comparison of the yields from multiple species sources revealed that native POR generally exhibits the highest efficiency of electron transfer36,37. Twelve enzyme combinations (Supplementary Table 2) with truncated or nontruncated CYP728B70 and representative TwPORs38 (without TwPOR2, a high identity of 98.73% with TwPOR1) were transformed into the BY-ZY1 strain. LC‒MS/MS quantification of the yield of 11 showed that TwPOR3 and TwPOR4 with intact CYP728B70 had the highest yields among the combinations, presumably due to differences in the electron transfer capacity associated with the phylogenetic clade of POR (Supplementary Fig. 18)39. None of the six combinations with truncated TwPORs detected the final product, suggesting that the intact transmembrane domain is essential for electron transfer. In addition, truncated CYP728B70 showed reduced yields, indicating that the complete transit peptides are important for peptide translocation and folding. TwPOR3, which is more biostable, was selected as the electron shuttle with a yield of 11 equal to 2.23 μg·L−1 (Fig. 6b). Further replacement of CYP728B70 with CYP720B4 from Picea sitchensis resulted in a 17.10-fold increase in the yield to 49.43 μg·L−1 (Fig. 6b).

Subsequently, optimization of the chassis strain was performed. Chromosome integration and diploidization are promising strategies to improve heterologous gene stability and expression levels and thus enhancing the fermentation ability40,41. We introduced the optimal genes (i.e., tSmKSL1-CfTPS1, CYP82D274 or mutants, CYP720B4, and TwPOR3) into the diploid strain BY-ZY2D (Supplementary Table 3) and haploid strain BY-HZ16. Diploid strains have higher cell growth rates, cell yields, and tolerances to various stresses than haploid strains42. In this study, we quantitatively inoculated different haploid and diploid strains and confirmed that the diploid strains exhibited ~2–3 times more biomass than the haploids under identical fermentation conditions (Fig. 6c). The diploid strain BY-ZY16D, in which tSmKSL1-CfTPS1 and CYP82D247L234M were integrated into the yeast chromosome and which carried an inducible plasmid pESC-LEU::(CYP720B4 + TwPOR3), had the highest yield among the compared strains, and the yield of 11 equaled 343.87 μg·L−1, which was 1.34 times (P < 0.05) higher than that of the WT BY-ZY15D strain and 6.78 times (P < 0.01) higher than that of the haploid strain BY-ZY12H of the same genotype (Fig. 6c).

Triptolide (1) is largely insoluble in aqueous solvents; however, chemical structure-bioactivity correlation analyses indicate that the characteristic hydrogen bond at the C-14 position is the key functional group for its antitumor effect by selective alkylation of thiol groups of enzyme-mediated tumor growth13,43,44. Further modification of the C-14 position, which was the earliest and most diverse target, is effective not only for increasing its water solubility45 but also for enhancing its antitumor activity and lower toxicity44; as a result, various compounds, e.g., minnelide (ClinicalTrials.gov: NCT03129139) and 14-succinyl triptolide sodium salt (PG490-88), have entered clinical trials. The biosynthetic pathway of 1 in native plants has advanced to the first step in skeleton modification after several years5,14,16. In this study, we revealed the mechanism for 14-hydroxylation formation and demonstrated the involvement of carboxyl groups in the biosynthetic pathway of 1. CYP82D274 and CYP82D263 act as 14-hydroxylases to catalyze the metabolic grid in 1 biosynthesis and exhibit better affinity and catalytic efficiency for multiple intermediates toward 7 (Fig. 5d). Previous pharmacological studies on the chemical synthesis of 11, the main product of CYP82D274 catalyzing 7, have shown its small level of cytotoxic activity, particularly in human acute T-cell leukemia31. Using synthetic biology strategies, the intermediate 11 was successfully produced in yeast with a shake flask yield of 343.87 μg·L−1 to not only obtain pharmacologically active products in an environmentally friendly manner but also facilitate the mining of enigmatic pathway genes.

In particular, CYP82D274 and CYP82D263 can catalyze the aromatization of 5 to yield 6 (Fig. 5), a process that was previously reported as a spontaneous reaction8,15,28, which results in the widespread presence of tanshinones, carnosic acid, carnosol, and other labdane- or abietane-type diterpenoids. We have demonstrated that functionalized CYP82D274 and CYP82D263 could increase the yield of the rate-limiting enzyme and thus the heterologous production of final products (Fig. 6a). It is reasonable to believe that the introduction of CYP82D274 and CYP82D263 in other 5-derived metabolic pathways could effectively improve access to the target products. Furthermore, the extensive C-14 hydroxylation function of CYP82D274 could effectively provide the necessary alternative genetic element and synthetic precursors for these pharmacologically important natural products and derivatives, for instance, gerardianin A from Isodon lophanthoides46.

CYP82D274 and CYP82D263 identified in this study are located on a tandem duplication gene cluster on Chr12 containing 11 CYP82D subfamily genes (Fig. 2a), presumably arising from LTRs of the retrotransposon (Class I element) generated by intrachromosomal replication rearrangements. According to the evolutionary divergence, the following hypothesis is proposed. In Clade I, the first duplication of neofunctional CYP82D produced CYP82D274, and gene duplication and transfer of TW08G01036.1 (CYP82D213) via transposon from Chr12 to Chr08 then occurred; in addition, the duplication of CYP82D274 produced TW12G01156.1, which then duplicated to generate TW12G01165.1 rearrangement at the other end of the tandem array. CYP82D213 reportedly undertakes the triepoxidation function in the last step of triptonide12, indicating that it underwent neofunctionalization after separation from CYP82D274. In Clade II, the duplication order follows TW12G01157.1, TW12G01158.1 and CYP82D263 (Supplementary Fig. 19).

Genes in the CYP82D subfamily are commonly found in flavonoid biosynthesis. SbCYP82D1.1 in Scutellaria baicalensis functions as a flavone 6-hydroxylase (F6H) in catalyzing chrysin, whereas SbCYP82D2 acts as a flavone 8-hydroxylase (F8H)47. A comparison of eight orthologous gene pairs of CYP82Ds in the genomes of S. barbata and S. baicalensis revealed the presence of the F6H tandem gene clusters SbaiCYP82D1-SbaiCYP82D7-SbaiCYP82D8 and SbarCYP82D1-SbarCYP82D6-SbarCYP82D8 on Chr06 in both species, with chromosomal localization distances of <30 kb. In particular, CYP82D8 showed interspecies gene synteny, indicating conserved and species-specific gene evolution48. In addition to flavonoids, CYP82Ds have been reported to be involved in the biosynthesis of furanocoumarin isopimpinellin49, lignan (-)−4'-desmethylepipodophyllotoxin50, and sesquiterpenoid gossypol21. Both CYP82D213 and the two CYP82D genes reported in the present study are involved in diterpenoids biosynthesis, and the construction of a phylogenetic tree with different functions (Fig. 7) revealed a clear phylogenetic differentiation of CYP82Ds involved in different types of compounds, which can also be used to guide future directions of research on CYP82Ds of unknown function.

A total of 61 CYP82Ds from flavonoids, furanocoumarins, lignan, fatty acid, sesquiterpenoids, diterpenoids, etc., were included and indicated by different colors. Phylogenetic analysis was performed using MEGA 6.0 software with maximum likelihood method (1000 bootstraps). The two functional CYP82Ds in this study are marked with red triangles. The GenBank accession numbers are provided in Supplementary Data 6.

Recently, a minimal set of four genes from the CYP71BE and CYP82D subfamilies required for the formation of triptonide from 5 was reported12, and these undoubtedly constitute an alternative and simple way to bypass the carboxylation step and obtain triptonide in heterologous hosts. However, whether it is an absolute biosynthetic pathway or a branched metabolic pathway in its native plants remains unclear. Here, we confirmed the involvement of carboxyl groups in the biosynthetic pathway of 1 via plant cell experiments. From the current perspective, C-18 carboxyl generation is the rate-limiting step responsible for the extremely low levels of 1 in T. wilfordii, and the Tripterygium genus contains many active natural products with carboxyl or carbonyl groups at C-18 or C-193,26. Improving protein activity is essential for improving the yield of carboxyl-related metabolites. The mechanism of carboxyl shift also needs to be explored for the biosynthesis of 1.

In conclusion, our study revealed the key enzymes for the C-14 hydroxylation metabolic grid of 1, characterized the hydroxylation and aromatization functions of tandemly duplicated CYP82D274 and CYP82D263, and successfully achieved the de novo biosynthesis of precursor 11 in yeast. The elucidation of the formation mechanism of C-14 hydroxylation not only provides important genetic elements for the synthetic biology production of 1 but is also useful for revealing the biosynthetic pathways of other labdane-type compounds with C-14 hydroxylation.

T. wilfordii suspension cells were cultured in Murashige & Skoog (MS) with vitamins (Caisson Lab, USA) medium containing 30 g·L−1 sucrose with 0.5 mg·L−1 indole-3-butyric acid (IBA), 0.5 mg·L−1 2,4-dichlorophenoxyacetic acid (2,4-D) and 0.1 mg·L−1 kinetin (KT), and the final pH was adjusted to 5.8. Suspension cells were grown at 25 °C in darkness with shaking at 120 rpm51. A total of seven four-year-old T. wilfordii plants (No. 164–170) were collected from Taoyuan Town, Datian County, Sanming City, Fujian, China. The geographic coordinates were E117°31′44″, N25°49′29″, and the altitude was 908 m. The chemical standards 7, 10, and 13 were purchased from Shanghai Yuanye Bio-Technology Co., Ltd. (Shanghai, China), and 9 and 15 were procured from BioBioPha Co., Ltd. (Yunnan, China). Compound 8 was kindly provided by Prof. Fei Li from Kunming Institute of Botany, Chinese Academy of Sciences, and 5, 6, 11, and 14 were isolated from enriched fermentation products.

The initial yeast strain used in this study was BY-HZ16 (MATα; trp1Δ0; leu2Δ0; lys2Δ0; ura3Δ0; rox1Δ; yjl064wΔ; ypl062wΔ; erg9::Δ−218-−175; trp1::HIS3-PPGK1-BTS1/ERG20-TADH1-PTDH3-SaGGPS-TTPI1-PTEF1-tHMG1-TCYC1)10. The engineered yeast strains are listed in Supplementary Table 3. Escherichia coli Trans1-T1 (TransGen Biotech, China) was used for cloning and plasmid construction. The shuttle vectors were pESC-LEU, pESC-TRP (Agilent Technologies, USA), and pYES2 (Invitrogen, USA). Synthetic dropout (SD) medium (FunGenome, China) was selected according to the auxotroph and carried plasmids. Yeast extract peptone dextrose (YPD) and YPL media were composed of 1% yeast extract, 2% peptone (OXOID, UK), and 2% D-( + )-glucose (Glc, Sigma‒Aldrich, USA) for YPD or 2% D-( + )-galactose (Gal, Inalco S.p. A, Italy) for YPL.

To obtain more chromosome information on CYP82Ds, BioEdit software (version 7.0.9.0) was used to search the open reading frame (ORF) of CYP82Ds from the whole genome through the local BLAST function. A sequence with a ‘blastn’ identity greater than or equal to 99% was regarded as one gene, and genes with similarity greater than 50% were recorded. Each CYP82D was ultimately identified by chromosome localization. A phylogenetic analysis was performed using MEGA 6.0 software to align the CYP82D sequences using ClustalW function, and the maximum likelihood method (1000 bootstraps) was used to build a phylogenetic tree. A total of 3n different sites were removed, and the online conversion ALTER (http://www.sing-group.org/ALTER/) and KaKs_Calculator 2.0 software were used to calculate the nonsynonymous substitution rates (Ka), synonymous substitution rates (Ks), and Ka/Ks in the Linux operating environment.

CYPs were obtained from the PFAM database and annotated as cytochrome P450 (PF00067). A heatmap of gene expression was generated using MultiExperiment Viewer (MeV, version 4.9.0) and Java 8 software. The RPKM reads of the root periderm, root phloem, root xylem, stem vascular bundle, stem periderm, and leaf originated from previous tissue transcriptomes (NCBI SRA number SRP199495, Supplementary Fig. 20)16, and the data were normalized and processed by hierarchical cluster analysis with Pearson correlation. CYPs with gene expression profiles similar to those of TwTPS7(v2), TwTPS27(v2), and CYP728B70 were selected as candidate CYPs for further analysis. The candidate CYPs were amplified using 2×Phusion High-Fidelity PCR Master Mix (New England Biolabs, USA) and cDNA of T. wilfordii root as the template, which was consistent with the transcriptome after complete sequencing (Supplementary Data 5).

Candidate CYPs were integrated into the high-copy plasmid pESC-LEU expressing TwPOR3 from the GAL1 promoter and CYP from the GAL10 promoter (Supplementary Data 5). The recombinant plasmids were transformed into the yeast strain BY4741, and pESC-LEU with TwPOR3 was employed as an empty vector control. For the in vivo assay, SD medium without tyrosine and leucine (SD-Trp-Leu) and 2% Glc was used to select positive strains, and 2% Gal was added to 20 mL of SD-Trp-Leu for the fermentation of fresh cells in a shaker at 30 °C and 200 rpm for 12 h. After the feeding of 7 or 8 to a final concentration of 50 μM, the resulting mixture was fermented for another 48-60 h, and ultrasonic extraction of fermentation products with an equal volume of ethyl acetate was then performed twice for 1 h each time. For the in vitro assay, microsomes were extracted according to the following experimental procedure. The base buffer TE (pH 7.5) consisted of 50 mM Tris-HCl and 1 mM EDTA. Cells were collected by centrifugation at 4000 × g for 3 min, resuspended in TEK (0.1 M KCl in TE), and left at room temperature for 10 min. The cells were collected again by centrifugation and resuspended in pre-cooled TESB (0.6 M sorbitol in TE). The cells were completely crushed using a cryogenic homogenizer (ATS, Canada) for 3–5 cycles. After centrifugation at 12,000 × g for 15 min, the microsomes were precipitated by adding NaCl at a final concentration of 0.15 M and polyethylene glycol PEG4000 at a final concentration of 0.1 g·mL−1 to the supernatant. The pellets were resuspended in TEG (20% (v/v) glycerol in TE) for preservation and catalytic reactions32. The enzymatic assay was performed in a 500-μL system containing 100 mM Tris-HCl (pH 7.5), 1 mM nicotinamide adenine dinucleotide phosphate (NADPH), 4 mM glucose-6-phosphate, 1 unit of glucose-6-phosphate dehydrogenase, 5 μM flavin mononucleotide, 5 μM flavin adenine dinucleotide, 1 mM dithiothreitol, 0.5 mg of microsomal protein, and 100 μM substrate, which was incubated for 20 h at 30 °C with shaking at 100 rpm in the dark and then extracted with 500 μL of ethyl acetate three times. The organic phases of both the in vivo and in vitro samples were dried, reconstituted with methanol and then derivatized with trimethylsilyl diazomethane (Tokyo Chemical Industry, Japan)52. The methylated samples were dissolved with ethyl acetate for GC-TQ-MS/MS analysis. For the 5-catalyzed reaction, the plasmid was transformed into engineered yeast BY-PS2 (Supplementary Table 3).

Strains containing the pESC-LEU::(CYP82D274 + TwPOR3) plasmid were inoculated into 500 mL of SD-Trp-Leu + 2% Glc fluid medium and grown at 30 °C and 200 rpm to an OD600 of ~10. The cells were collected by centrifugation, evenly distributed into twenty 2-L flasks containing 500 mL of YPL and 100 μM substrate 7, and then fermented at 30 °C and 200 rpm for 72 h. Then, 15-20 g of HP2MGL resin (Mitsubishi, Japan) was added to each flask in a sterile environment, and the fermentation product was extracted in a shaker at 200 rpm for another 24 h.

Ten liters of fermentation broth was centrifuged at 4000 × g for 5 min, and the yeast cells and resin were collected and dried in an oven at 40 °C. Eight times the volume of anhydrous methanol was added, and the catalytic product was extracted by ultrasonication at low temperature for 30 min. The solids were repeatedly extracted until the anhydrous methanol extract was colorless. The extracts were filtered to remove the resin and cell residues and evaporated under a vacuum at 40 °C. The products were dissolved with 4 mL of chromatography-grade methanol and sampled into a Gilson 281 semipreparative HPLC connected to an H322 pump, GX-281 autosampler, 156 dual wavelength UV detector, and Trilution LC 2.1 workstation. An Xtimate C18 column (30 mm × 75 mm × 3.0 μm, Welch) with 0.1% trifluoroacetic acid-water (mobile phase A) and acetonitrile (mobile phase B) was used. Gradient elution was started with 50% B and then increased to 100% B over 8.0 min, the flow rate was set to 25 mL·min−1, and the injection volume was 500 μL. The product fractions were collected at 7.4–7.8 min and freeze-dried. Three milligrams of product powder was weighed precisely and fully dissolved in 500 μL of deuterated acetone (C3D6O, InnoChem, China). For chemical structure characterization, 1H-NMR (800 M), 13C-NMR (200 M), NOESY, HSQC, and 1H-1H COSY were analyzed using a Bruker Avance III HD (Bruker BioSpin, Germany). MestReNova 14 software was used to analyze the data.

The CYP concentration was estimated by measuring the reduced carbon monoxide difference spectrum using an extinction coefficient (450 versus 490 nm) of 91 mM−1·cm−153. Incubation times from 0.5 to 8 h were used. The Km values were determined using serial concentrations from 0.1 to 100 μM for 7 and 5 to 80 μM for 5 in 300-μL enzyme assays. These also contained ~0.4 mg of microsomal protein, Tris-HCl (pH 7.5), NADPH, and the regenerating system mentioned above. The reactions were initiated by substrate addition and incubated with shaking at 30 °C for 30 min in the dark. Extraction was then terminated three times with 300 μL of ethyl acetate, and the extracts were completely dried with a vacuum concentrator (Eppendorf Concentrator plus, Germany). The product 11 was quantified using 100 μL of methanol dissolved for LC-TQ-MS/MS analysis, and the product 14 was dissolved in 100 μL of ethyl acetate for quantification by GC‒MS. The Km values were calculated by nonlinear regression using GraphPad Prism version 7.

The specific fragments of CYP82D274 and CYP82D263 were chosen via multiple sequence alignment of all CYP82D nucleic acid sequences in T. wilfordii. The specific fragments (Supplementary Data 5) were inserted into the pH7WG2D (for OE) and pK7GWIWG2D (for RNAi) binary vectors. Using the developed gene gun technology25,52, recombinant plasmids were transformed into suspension cells that had been precultured for 7 days and then cultured for another 7 days to collect samples. RNA extraction and reverse transcription were performed using Total RNA Extraction Kit (Promega, China) and FastQuant RT Kit (Tiangen Biotech, China)51, and gene expression was detected using qRT‒PCR (Supplementary Data 5). Twenty milligrams of freeze-dried sample powder was precisely weighed, soaked in 1 mL of 80% (v/v) methanol overnight at 4 °C, and subjected to ultrasonication at 40 kHz for 1 h at 25 °C. After centrifugation at 13,000 × g for 10 min, the supernatant was filtered through a 0.22-μm PTFE microporous membrane to obtain the samples for Q Exactive HF LC‒MS/MS.

The substrate feeding was performed by soaking 0.2 g of suspension cells with 1 mL of MS medium containing 0.8 mM 11 for 7 days before sampling, and the negative control was soaked using an equal volume of MS liquid. For stable isotope labeling assay, 18O-11 was obtained by expanding the above microsomal assay system to 600 mL and completely replacing 16O2 with 18O2 (purity > 97%) before the addition of substrate 7, and after 1 h of reaction, an equal volume of ethyl acetate was extracted three times and then subjected to product enrichment and isolation. The purified 18O-11 was determined by LC-qTOF-MS. After gene gun bombardment of TwTPS7&TwTPS27-RNAi (Supplementary Data 5) for 5 days, ~0.2 g of cells were soaked in 1 mL of MS medium with 0.4 mM 18O-11 and 100 μM fosmidomycin for 5 days and then sampled after 2 days of induction with 50 μM methyl jasmonate. Equal concentrations of unlabeled 11 with identical treatment and WT cells were used as controls. Three biological replicates of each group were included in the experiment. The metabolite extraction was performed as described above, with Q Exactive HF LC‒MS/MS for quantitative assays and LC-qTOF-MS for qualitative assays.

For CYP coexpression, the recombinant plasmids pESC-LEU::(CYP82Ds + TwPOR3) and pESC-TRP:: (CYP728B70 + TwPOR3) were chemically transformed into the yeast strain BY-PS2 together or separately. A corresponding auxotrophic medium was employed for selection and fermentation, and products were extracted with ethyl acetate and detected via GC‒MS. For the screening of CYP and POR combinations, the recombinant plasmid pYES2:: (PTDH3-tSmKSL1-CfTPS1-TTPI1-PADH1-CYP82D274-TPGI) was constructed (Supplementary Data 5) using BY-HZ16 strain homologous recombination to generate the BY-ZY1 strain. Truncated transit peptides (50 amino acids at the N-terminus predicted by TargetP Server v2.0) or complete CYP728B70 was constructed at multiple cloning site (MCS) 1, and a truncated transmembrane domain (67 amino acids for TwPOR1 and 46 amino acids for TwPOR3 and TwPOR4, predicted by TMHMM Server v2.0) or complete TwPOR was integrated at MCS2 of pESC-LEU (Supplementary Data 5). Twelve enzyme combinations (Supplementary Table 2) were transformed into the BY-ZY1 strain. For optimization of the chassis strain, the haploid strains BY4741H (Supplementary Table 3) and BY-HZ16 were inoculated in the respective media and incubated at 30 °C and 200 rpm for 24 h. The two strains were mixed in SD-His-Trp supplemented with 2% Glc liquid medium at an equal OD and cocultured for another 24 h. After centrifugation, the cells were collected and coated on SD-His-Trp with 2% Glc solid plates and incubated at 30 °C for 72 h, and the monoclonal strains were selected. The newly constructed diploid strain was identified and named BY-ZY2D. The tSmKSL1-CfTPS1 and CYP82D274 (or mutants) modules were cloned and integrated into the haploid strain BY-HZ16 and diploid strain BY-ZY2D chromosome YPRCΔ15 locus by the modularized two-step chromosome integration technique54 using Ura3 as a selection marker (Supplementary Fig. 21 and Supplementary Data 5). Subsequently, pESC-LEU::(CYP720B4 + TwPOR3) was chemically transformed to generate BY-ZY11H to ZY14H (haploid strains) and BY-ZY15D to ZY18D (diploid strains) (Supplementary Table 3). For quantitative inoculation, the seed strain was transferred into 20 mL of fresh medium with a quantitative OD of 0.1, and three biological replicates of each group were included in the experiment. The product was extracted with an equal volume of n-hexane and dissolved in 100 μL of methanol for LC-TQ-MS/MS.

Matrix-assisted laser desorption/ionization-mass spectrometry imaging (MALDI-MSI) is a label-free technique used for tissue sample imaging that allows the localization of known or unknown biological molecules in a spatial region within subcellular compartments. The sample is rasterized while a laser is used as a stationary ionization source, and the mass spectrum is utilized for recording and converting the ion signal intensity and then drawing the XY coordinate, enabling the spatial distribution of the ion of interest for visualization55,56. High-coverage MALDI-MSI was employed to map the target molecules in the root of T. wilfordii. The fresh root was flash-frozen in liquid nitrogen and then maintained at −80 °C for over 8 h. The root tissue was cryosectioned into 20-μm sections on a cryostat microtome (CM3050S, Leica Biosystems, Germany) in a −20 °C environment and attached to the conductive side of an indium tin oxide (ITO)-coated glass slide (Bruker, Germany). The glass slide was kept in a vacuum desiccator and dried for 30 min. Then, 2,5-dihydroxybenzoic acid (DHB) was selected as the matrix under the positive ion mode after the pre-experiment and was sprayed evenly on the surface of the slide by an HTXTM-Sprayer (HTX Technologies, USA). MALDI-MSI was conducted with a timsTOF fleXTM mass spectrometer (Bruker), and the parameter settings were as follows. The laser frequency and accumulated shots were set to 10,000 Hz and 500 shots, respectively, the spatial resolution was 50 μm, and the target profile was 18.70 μm. The raw mass spectra data were acquired over the m/z range of 200-1000. A Bruker SCiLS Lab 2020a instrument and MetaboScape software were used to analyze the MALDI imaging data and visualize the m/z values.

GC-TQ-MS/MS was performed on an Agilent 7890B GC system equipped with a 7000 C GC/MS Triple Quad. For the DB-5MS (15 m × 0.25 mm × 0.1 μm) capillary column, splitless injection was performed at 250 °C. The GC oven temperature was adjusted according to the following program: an initial temperature of 50 °C was maintained for 1 min, followed by an increase to 200 °C at a rate of 40 °C·min−1, an increase to 240 °C at a rate of 20 °C·min−1, an increase to 246 °C at a rate of 1.5 °C·min−1, and an increase to 300 °C at a rate of 40 °C·min−1, with the final temperature being maintained for 3 min. For the DB-5MS (30 m × 0.25 mm × 0.1 μm) column, the oven temperature program was as follows: 50 °C isothermal for 1 min, increased to 240 °C at 50 °C·min−1, increased to 255 °C at 1.5 °C·min−1, increased to 300 °C at 50 °C·min−1 and maintained at the final temperature for 1 min. Helium was used as the carrier gas at a flow rate of 1 mL·min−1. The inlet temperature was set to 300 °C, the ion trap temperature was 250 °C, the electron energy was 70 eV, and the spectra were recorded over a range of 10-400 m/z. Qualitative Analysis software (B.07.00) was used for data analysis.

LC-TQ-MS/MS was performed with an Agilent 1290 Infinity LC tandem QTRAP 6500 MS (AB SCIEX). We used an ACQUITY UPLC HSS T3 column (2.1 mm × 100 mm × 1.8 μm, Waters) with 0.1% formic acid-water as mobile phase A and acetonitrile as mobile phase B. A flow rate of 0.3 mL·min−1 was used, and the gradient program was as follows: 0–1 min, 40% B; 2–3 min, 90% B; and 5–6 min, 40% B. In each run, 3 μL of the sample was analyzed, and the column temperature was 40 °C. The quantitative ions for multiple reaction monitoring (MRM) 11 were 315 and 269 m/z.

Q Exactive HF LC‒MS/MS (Thermo Scientific) was performed on an ACQUITY UPLC HSS T3 column (2.1 mm × 100 mm × 1.8 μm, Waters) with 0.1% formic acid-water as mobile phase A and acetonitrile as mobile phase B. A flow rate of 0.4 mL·min−1 was used, and the gradient program was as follows: 0–4 min, 30–34% B; 4–8 min, 34–52% B; 8–12 min, 52% B; 12–23 min, 52–77% B; and 23–26 min, 90% B. In each run, 2 μL of the sample was analyzed, and the column temperature was 40 °C. When the parent ion quantification method was applied, either positive or negative ion mode was used for different metabolites.

LC-qTOF-MS (Waters) was performed on an ACQUITY UPLC HSS T3 column (2.1 mm × 100 mm × 1.8 μm, Waters) with 0.1% formic acid-water as mobile phase A and acetonitrile as mobile phase B. The gradient program was as follows: 0–0.5 min, 20% B; 0.5–21 min, 20–85% B; 21–23 min, 85–100% B; 23–26 min, 100% B; 26–27 min, linear decrease from 100 to 20% B; and 27–30 min, isocratic 20% B. The flow rate was set at 0.4 mL·min−1. Mass spectra were acquired in positive/negative ion mode over a scan range of m/z 50-1200 with a scan time of 0.2 s. The MS settings were as follows: resolution analyzer mode; normal dynamic range; ramp collision energy 10-50 V; cone voltage 40 V. MassLynx v4.2 was used for data analysis.

Primers (Supplementary Data 5), cDNA, and reagents were added according to the TransStart Top Green qPCR SuperMix (TransGen Biotech) instructions. QuantStudio5 (Applied Biosystems) was employed to assay the expression of genes, and each reaction was repeated three times. The mixed reagents were first activated for 30 s at 94 °C, and then denaturation occurred for 5 s at 94 °C, followed by annealing for 15 s at 56 °C and extension for 10 s at 72 °C. This procedure was repeated for 45 cycles, followed by the dissociation stage. EFLα (Supplementary Data 5) was used as the housekeeping reference gene. The relative expression of genes was analyzed using the 2-ΔΔCt method57.

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

The genome and transcriptome sequence data are available at NCBI BioProject PRJNA542587. Full sequences of CYP82D274 (accession XP_038717985) and CYP82D263 (accession XP_038716720) are deposited in GenBank. The databases of TargetP v2.0 (https://services.healthtech.dtu.dk/service.php?TargetP-2.0) and TMHMM v2.0 (https://services.healthtech.dtu.dk/service.php?TMHMM-2.0) are used for data analyses. Source data are provided with this paper.

Chen, S. R. et al. A mechanistic overview of triptolide and celastrol, natural products from Tripterygium wilfordii Hook F. Front. Pharmacol. 9, 104 (2018).

Article PubMed PubMed Central Google Scholar

Zhang, Y. Q. et al. Tripterygium wilfordii: An inspiring resource for rheumatoid arthritis treatment. Med. Res. Rev. 41, 1337–1374 (2020).

Article PubMed Google Scholar

Lange, B. M. et al. Integrative approaches for the identification and localization of specialized metabolites in Tripterygium roots. Plant Physiol. 173, 456–469 (2017).

Article CAS PubMed Google Scholar

Gao, J. et al. Triptolide: pharmacological spectrum, biosynthesis, chemical synthesis and derivatives. Theranostics 11, 7199–7221 (2021).

Article CAS PubMed PubMed Central Google Scholar

Hansen, N. L. et al. The terpene synthase gene family in Tripterygium wilfordii harbors a labdane-type diterpene synthase among the monoterpene synthase TPS-b subfamily. Plant J. 89, 429–441 (2017).

Article CAS PubMed Google Scholar

Zhang, X., Xiao, Z. & Xu, H. A review of the total syntheses of triptolide. Beilstein J. Org. Chem. 15, 1984–1995 (2019).

Article CAS PubMed PubMed Central Google Scholar

Song, Y. et al. A novel strategy to enhance terpenoids production using cambial meristematic cells of Tripterygium wilfordii Hook. f. Plant Methods 15, 129 (2019).

Article ADS PubMed PubMed Central Google Scholar

Scheler, U. et al. Elucidation of the biosynthesis of carnosic acid and its reconstitution in yeast. Nat. Commun. 7, 12942 (2016).

Article ADS CAS PubMed PubMed Central Google Scholar

Paddon, C. J. et al. High-level semi-synthetic production of the potent antimalarial artemisinin. Nature 496, 528–532 (2013).

Article ADS CAS PubMed Google Scholar

Hu, T. et al. Engineering chimeric diterpene synthases and isoprenoid biosynthetic pathways enables high-level production of miltiradiene in yeast. Metab. Eng. 60, 87–96 (2020).

Article CAS PubMed Google Scholar

Forman, V., Callari, R., Folly, C., Heider, H. & Hamberger, B. Production of putative diterpene carboxylic acid intermediates of triptolide in yeast. Molecules 22, 981 (2017).

Article PubMed PubMed Central Google Scholar

Hansen, N. L. et al. Tripterygium wilfordii cytochrome P450s catalyze the methyl shift and epoxidations in the biosynthesis of triptonide. Nat. Commun. 13, 5011 (2022).

Article ADS CAS PubMed PubMed Central Google Scholar

Zhou, Z. L., Yang, Y. X., Ding, J., Li, Y. C. & Miao, Z. H. Triptolide: structural modifications, structure-activity relationships, bioactivities, clinical development and mechanisms. Nat. Prod. Rep. 29, 457–475 (2012).

Article CAS PubMed Google Scholar

Su, P. et al. Identification and functional characterization of diterpene synthases for triptolide biosynthesis from Tripterygium wilfordii. Plant J. 93, 50–65 (2018).

Article CAS PubMed Google Scholar

Ignea, C. et al. Carnosic acid biosynthesis elucidated by a synthetic biology platform. Proc. Natl Acad. Sci. USA 113, 3681–3686 (2016).

Article ADS CAS PubMed PubMed Central Google Scholar

Tu, L. et al. Genome of Tripterygium wilfordii and identification of cytochrome P450 involved in triptolide biosynthesis. Nat. Commun. 11, 971 (2020).

Article ADS CAS PubMed PubMed Central Google Scholar

Bathe, U. & Tissier, A. Cytochrome P450 enzymes: A driving force of plant diterpene diversity. Phytochemistry 161, 149–162 (2019).

Article CAS PubMed Google Scholar

Wong, J., d’Espaux, L., Dev, I., van der Horst, C. & Keasling, J. De novo synthesis of the sedative valerenic acid in Saccharomyces cerevisiae. Metab. Eng. 47, 94–101 (2018).

Article CAS PubMed Google Scholar

Rao, X. & Dixon, R. A. Co-expression networks for plant biology: why and how. Acta Biochim. Biophys. Sin. 51, 981–988 (2019).

Article PubMed Google Scholar

Tai, Y. et al. Gene co-expression network analysis reveals coordinated regulation of three characteristic secondary biosynthetic pathways in tea plant (Camellia sinensis). BMC Genomics 19, 616 (2018).

Article PubMed PubMed Central Google Scholar

Tian, X. et al. Characterization of gossypol biosynthetic pathway. Proc. Natl Acad. Sci. USA 115, e5410–e5418 (2018).

Article CAS PubMed PubMed Central Google Scholar

Lee, S. et al. Herbivore-induced and floral homoterpene volatiles are biosynthesized by a single P450 enzyme (CYP82G1) in. Arabidopsis. Proc. Natl Acad. Sci. USA 107, 21205–21210 (2010).

Article ADS CAS PubMed Google Scholar

Yang, Z. & Zhang, Z. Engineering strategies for enhanced production of protein and bio-products in Pichia pastoris: A review. Biotechnol. Adv. 36, 182–195 (2018).

Article CAS PubMed Google Scholar

Kinouchi, Y. et al. Potential antitumor-promoting diterpenoids from the stem bark of Picea glehni. J. Nat. Prod. 63, 817–820 (2000).

Article CAS PubMed Google Scholar

Zhao, Y. et al. Genetic transformation system for woody plant Tripterygium wilfordii and its application to product natural celastrol. Front. Plant Sci. 8, 2221 (2017).

Article PubMed Google Scholar

Kutney, J. P. & Han, K. Studies with plant-cell cultures of the Chinese herbal plant, Tripterygium wilfordii. Isolation and characterization of diterpenes. Red. Trav. Chim. Pays-Bas 115, 77–93 (1996).

Article CAS Google Scholar

Gao, W. et al. A functional genomics approach to tanshinone biosynthesis provides stereochemical insights. Org. lett. 11, 5170–5173 (2009).

Article CAS PubMed PubMed Central Google Scholar

Zi, J. C. & Peters, R. J. Characterization of CYP76AH4 clarifies phenolic diterpenoid biosynthesis in the Lamiaceae. Org. Biomol. Chem. 11, 7650–7652 (2013).

Article CAS PubMed PubMed Central Google Scholar

Qiu, Y., Diao, H., Zheng, Y. & Wu, R. Multiscale simulations on the catalytic plasticity of CYP76AH1. Front. Chem. 9, 689731 (2021).

Article CAS PubMed PubMed Central Google Scholar

Heskes, A. M. et al. Biosynthesis of bioactive diterpenoids in the medicinal plant Vitex agnus-castus. Plant J. 93, 943–958 (2018).

Article CAS PubMed PubMed Central Google Scholar

Zapata, B. et al. Cytotoxic, immunomodulatory, antimycotic, and antiviral activities of semisynthetic 14-hydroxyabietane derivatives and triptoquinone C-4 epimers. MedChemComm 4, 1239 (2013).

Article CAS Google Scholar

Ma, Y. et al. Expansion within the CYP71D subfamily drives the heterocyclization of tanshinones synthesis in Salvia miltiorrhiza. Nat. Commun. 12, 685 (2021).

Article ADS CAS PubMed PubMed Central Google Scholar

Pateraki, I. et al. Total biosynthesis of the cyclic AMP booster forskolin from Coleus forskohlii. Elife 6, e23001 (2017).

Article PubMed PubMed Central Google Scholar

Su, D., Kosciuk, T., Yang, M., Price, I. R. & Lin, H. Binding affinity determines substrate specificity and enables discovery of substrates for N-myristoyltransferases. ACS Catal. 11, 14877–14883 (2021).

Article CAS PubMed PubMed Central Google Scholar

Urlacher, V. B. & Girhard, M. Cytochrome P450 monooxygenases in biotechnology and synthetic biology. Trends Biotechnol. 37, 882–897 (2019).

Article CAS PubMed Google Scholar

Zhao, Y., Fan, J., Wang, C., Feng, X. & Li, C. Enhancing oleanolic acid production in engineered Saccharomyces cerevisiae. Bioresour. Technol. 257, 339–343 (2018).

Article CAS PubMed Google Scholar

Dai, Z. et al. Identification of a novel cytochrome P450 enzyme that catalyzes the C-2α hydroxylation of pentacyclic triterpenoids and its application in yeast cell factories. Metab. Eng. 51, 70–78 (2019).

Article CAS PubMed Google Scholar

Su, P. et al. Probing the single key amino acid responsible for the novel catalytic function of ent-kaurene oxidase supported by NADPH-cytochrome P450 reductases in Tripterygium wilfordii. Front. Plant Sci. 8, 1756 (2017).

Article PubMed PubMed Central Google Scholar

Jensen, K. & Møller, B. L. Plant NADPH-cytochrome P450 oxidoreductases. Phytochemistry 71, 132–141 (2010).

Article CAS PubMed Google Scholar

Yamada, R., Tanaka, T., Ogino, C., Fukuda, H. & Kondo, A. Novel strategy for yeast construction using delta-integration and cell fusion to efficiently produce ethanol from raw starch. Appl. Microbiol. Biotechnol. 85, 1491–1498 (2010).

Article CAS PubMed Google Scholar

Yamada, R., Tanaka, T., Ogino, C. & Kondo, A. Gene copy number and polyploidy on products formation in yeast. Appl. Microbiol. Biotechnol. 88, 849–857 (2010).

Article CAS PubMed Google Scholar

Yamada, R. et al. Direct ethanol production from cellulosic materials using a diploid strain of Saccharomyces cerevisiae with optimized cellulase expression. Biotechnol. Biofuels 4, 8 (2011).

Article CAS PubMed PubMed Central Google Scholar

Kupchan, S. M. & Schubert, R. M. Selective alkylation: a biomimetic reaction of the antileukemic triptolides. Science 185, 791–793 (1974).

Article ADS CAS PubMed Google Scholar

Noel, P. et al. Triptolide and its derivatives as cancer therapies. Trends Pharmacol. Sci. 40, 327–341 (2019).

Article CAS PubMed Google Scholar

Li, Z. et al. Design and synthesis of novel C14-hydroxyl substituted triptolide derivatives as potential selective antitumor agents. J. Med. Chem. 52, 5115–5123 (2009).

Article CAS PubMed Google Scholar

Lin, C. Z. et al. Gerardianin A, a new abietane diterpenoid from Isodon lophanthoides var. gerardianus. J. Asian Nat. Prod. Res. 10, 841–844 (2008).

Article CAS PubMed Google Scholar

Zhao, Q. et al. Two CYP82D enzymes function as flavone hydroxylases in the biosynthesis of root-specific 4’-deoxyflavones in Scutellaria baicalensis. Mol. Plant 11, 135–148 (2018).

Article CAS PubMed Google Scholar

Xu, Z. et al. Comparative genome analysis of Scutellaria baicalensis and Scutellaria barbata reveals the evolution of active flavonoid biosynthesis. Genom. Proteom. Bioinf. 18, 230–240 (2020).

Article Google Scholar

Limones-Mendez, M. et al. Convergent evolution leading to the appearance of furanocoumarins in citrus plants. Plant Sci. 292, 110392 (2020).

Article CAS PubMed Google Scholar

Lau, W. & Sattely, E. S. Six enzymes from mayapple that complete the biosynthetic pathway to the etoposide aglycone. Science 349, 1224–1228 (2015).

Article ADS CAS PubMed PubMed Central Google Scholar

Zhang, Y. et al. The expression of TwDXS in the MEP pathway specifically affects the accumulation of triptolide. Physiol. Plant. 169, 40–48 (2020).

Article CAS PubMed Google Scholar

Zhang, Y. et al. The gibberellin 13-oxidase that specifically converts gibberellin A9 to A20 in Tripterygium wilfordii is a 2-oxoglutarate-dependent dioxygenase. Planta 250, 1613–1620 (2019).

Article ADS CAS PubMed Google Scholar

Omura, T. & Sato, R. The carbon monoxide-binding pigment of liver microsomes. II. solubilization, purification, and properties. J. Biol. Chem. 239, 2379–2385 (1964).

Article CAS PubMed Google Scholar

Li, S., Ding, W., Zhang, X., Jiang, H. & Bi, C. Development of a modularized two-step (M2S) chromosome integration technique for integration of multiple transcription units in Saccharomyces cerevisiae. Biotechnol. Biofuels 9, 232 (2016).

Article PubMed PubMed Central Google Scholar

Feenstra, A. D. et al. Spatial mapping and profiling of metabolite distributions during germination. Plant Physiol. 174, 2532–2548 (2017).

Article CAS PubMed PubMed Central Google Scholar

Sun, C. et al. Development of a high-coverage matrix-assisted laser desorption/ionization mass spectrometry imaging method for visualizing the spatial dynamics of functional metabolites in Salvia miltiorrhiza Bge. J. Chromatogr. A 1614, 460704 (2020).

Article CAS PubMed Google Scholar

Livak, K. J. & Schmittgen, T. D. Analysis of relative gene expression data using real-time quantitative PCR and the 2-△△Ct Method. Methods 25, 402–408 (2001).

Article CAS PubMed Google Scholar

Download references

The authors thank Prof. Juan Guo from the National Resource Center for Chinese Materia Medica for writing and experimental guidance. We thank Dr. Pengfei Li and Kairong Wu from Bruker (Beijing) Scientific Technology Co., Ltd for MALDI-MSI. We appreciate the guidance of Dr. Yaqiu Zhao from National Resource Center for Chinese Materia Medica in the isotopically labeled compound. We thank Dr. David Nelson from the University of Tennessee for assigning CYP nomenclature. We thank Prof. Fei Li from the Kunming Institute of Botany for providing chemical standard. This work was supported by the National Key R&D Program of China (2020YFA0908000) to L.H. and W.G., the Innovation Team and Talents Cultivation Program of National Administration of Traditional Chinese Medicine (ZYYCXTD-D-202005) to W.G., the National Natural Science Foundation of China (82204547), the Key Project at central government level: The ability establishment of sustainable use for valuable Chinese medicine resources (2060302), and the Fundamental Research Funds for the Central public welfare research institutes (ZZXT202206) to W.G. and Yi.Z.

Deceased: Wei Gao.

State Key Laboratory of Dao-di Herbs, National Resource Center for Chinese Materia Medica, Chinese Academy of Chinese Medical Sciences, Beijing, China

Yifeng Zhang, Jie Gao, Ping Su, Yujun Zhao & Luqi Huang

School of Traditional Chinese Medicine, Capital Medical University, Beijing, China

Yifeng Zhang, Jie Gao, Lin Ma, Xiaoyi Wu, Yuan Liu, Yan Yin, Huan Zhao, Yun Lu, Jiadian Wang & Wei Gao

Beijing Shijitan Hospital, Capital Medical University, Beijing, China

Yifeng Zhang & Wei Gao

Key Laboratory of Novel Targets and Drug Study for Neural Repair of Zhejiang Province, School of Medicine, Zhejiang University City College, Hangzhou, China

Lichan Tu

School of Pharmacy, Hangzhou Normal University, Hangzhou, China

Tianyuan Hu

School of Pharmaceutical Sciences, Capital Medical University, Beijing, China

Dan Li & Yuru Tong

College of Biotechnology and Bioengineering, Zhejiang University of Technology, Hangzhou, China

Jiawei Zhou

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

Yi.Z., W.G., and L.H. conceived and initiated the study. Yi.Z., J.G., L.M., Y.Liu, J.Z., Y.Lu, and J.W. performed the experiments. L.T. performed the phylogenetic analysis. Yu.Z., Y.T., and H.Z. provided the resources and materials. Yi.Z., X.W., and P.S. analyzed the quantitative data. D.L. conducted the NMR analysis. T.H., and Y.Y. established yeast strains used. Yi.Z. and W.G. wrote the manuscript. W.G. and L.H. revised the manuscript.

Correspondence to Wei Gao or Luqi Huang.

Yi.Z., J.G., L.M., and W.G. are inventors of Chinese invention patents (application No. CN202210059108.2 and CN202210059174.X) related to the cytochrome P450 enzymes described in the paper. Other authors declare no competing interests.

Nature Communications thanks Young Jin Lee and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

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

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 license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license 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 license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

Zhang, Y., Gao, J., Ma, L. et al. Tandemly duplicated CYP82Ds catalyze 14-hydroxylation in triptolide biosynthesis and precursor production in Saccharomyces cerevisiae. Nat Commun 14, 875 (2023). https://doi.org/10.1038/s41467-023-36353-y

Download citation

Received: 02 August 2021

Accepted: 27 January 2023

Published: 16 February 2023

DOI: https://doi.org/10.1038/s41467-023-36353-y

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.