Skip to main content
NCBI home page
Frontiers in Plant Science logoLink to Frontiers in Plant Science
. 2026 Sep 11;17:1937259. doi: 10.3389/fpls.2026.1937259

Genome-wide analysis of TPS family reveals kaurene synthase-like genes in Isodon rubescens f. lushanensis

Hao Yang 1,2,3,†, Jinlu Liu 1,†, Conglong Lian 1,2,3,†, Hongwei Yu 4, Le Zhao 1, Ni He 4, Yuanyuan Li 1, Ruohan Chen 1, Xiuyu Liu 1, Shujuan Xue 1,2,3, Xiaoya Sun 1,2,3, Lipeng Zhang 1, Lili Wang 1, Jingfan Yang 1, Yu Fu 1, Rui Ma 1, Bao Zhang 1, Lidan Ye 4,*, Suiqing Chen 1,2,3,*
PMCID: PMC13612420  PMID: 42798568

Abstract

Introduction

Isodon rubescens f. lushanensis, a typical form of I. rubescens, is characterized by its natural deficiency in oridonin and the presence of a unique ent-kaurane diterpenoid, lushanrubescensin.

Methods

To elucidate the molecular basis underlying the distinct chemical profiles between I. rubescens f. lushanensis and I. rubescens (Hemsl.) Hara, we conducted a comprehensive analysis utilizing high-quality genomic and transcriptomic data of I. rubescens f. lushanensis.

Results

Through a genome-wide survey, our study identified 83 terpene synthase (TPS) genes, including 13 TPS-C and 9 TPS-e/f subfamily genes, which are implicated in diterpenoid biosynthesis. By integrating conserved motif analysis, differential expression profiling (FPKM), and RT-qPCR validation, we successfully screened key candidate genes. Subsequent heterologous expression in Saccharomyces cerevisiae enabled the functional characterization of five diterpene synthase genes, specifically CPS and KSL enzymes, involved in the central modules of diterpenoid synthesis.

Discussion

These findings not only expand the repertoire of known biosynthetic genes in I. rubescens but also offer valuable insights into the divergent biosynthetic pathways of oridonin and lushanrubescensin, paving the way for future metabolic engineering and synthetic biology studies.

Keywords: gene family analysis, Isodon rubescens (Hemsl.) Hara, Isodon rubescens f. lushanensis, terpene synthase, yeast heterologous expression

Introduction

Isodon rubescens (Hemsl.) Hara is a medicinal herb belonging to the family Lamiaceae and is primarily distributed in Henan and Shanxi provinces. In Henan Province, I. rubescens comprises three varieties and two forms. In botanical taxonomy, a “form” (forma, abbreviated as f.) is an infraspecific rank used to distinguish individuals with subtle but stable morphological or physiological characteristics The two forms examined in this study are I. rubescens (Hemsl.) Hara (hereafter designated I. rubescens-JY), originally collected from Jiyuan City in the Taihang Mountains, and I. rubescens f. lushanensis (hereafter designated I. rubescens-LS), originally collected from Lushan County in the Funiu Mountains (Gao and Li, 1986; Yang and Liu, 2020). These two forms exhibit distinct diterpenoid profiles: I. rubescens-JY is characterized by oridonin, whereas I. rubescens-LS produces lushanrubescensin (Chen et al., 2011). Both compounds belong to the ent-kaurane diterpenoid class and have been reported to possess anti-tumor activities (Chen et al., 2016; Lian et al., 2022). Their structural differences are primarily attributable to distinct downstream functional-group modifications. Lushanrubescensin contains four acetyl substituents, which may increase its lipophilicity relative to oridonin and highlight its potential pharmaceutical relevance (Wang, 1987; Zhao et al., 2007).

Bioactive diterpenoids, including oridonin and lushanrubescensin, are still predominantly obtained through plant extraction, which is constrained by limited natural resources and low extraction efficiency. The chemical synthesis of complex terpenoids is also challenging because of their structural complexity, stereochemical requirements, and environmental concerns. Synthetic biology provides a promising alternative for producing valuable terpenoids in microbial chassis, as demonstrated by the engineered production of paclitaxel, gastrodin and ferulic acid in Saccharomyces cerevisiae or Escherichia coli (Liu et al., 2026; Wang et al., 2026; Shang et al., 2025). However, the biosynthetic pathway of lushanrubescensin remains largely unsolved, limiting its sustainable production through metabolic engineering.

Previous studies have substantially advanced the understanding of diterpenoid biosynthesis in I. rubescens-JY. The diterpenoid pathway can be broadly divided into upstream precursor-supply reactions, central cyclization steps, and downstream tailoring reactions. Geranylgeranyl diphosphate (GGPP), generated through the cytosolic mevalonate (MVA) pathway or plastidial 2-C-methyl-D-erythritol 4-phosphate (MEP) pathway, is converted by class II copalyl diphosphate synthases (CPSs) into copalyl diphosphate (CPP) intermediates, which are subsequently cyclized by class I kaurene synthase-like (KSL) enzymes to generate diverse diterpene scaffolds. In I. rubescens-JY, several CPS and KSL enzymes have been functionally characterized, including IrKSL5, which catalyzes the formation of ent-kaurene, an early intermediate in the oridonin biosynthetic pathway (Jin et al., 2017). Recent chromosome-level genomic analysis further identified multiple CYP706 family proteins which catalyze early oxidative modifications of the ent-kaurene core during oridonin biosynthesis (Sun et al., 2023). Nevertheless, the TPS repertoire and the early cyclization steps leading to lushanrubescensin formation in I. rubescens-LS remain unexplored. To address this knowledge gap, we performed the first genome-wide analysis of the TPS family in I. rubescens-LS using genomic and transcriptomic resources. We further prioritized candidate diterpene synthase (diTPS) genes by integrating phylogenetic relationships, conserved-motif analysis, expression profiles, and co-expression associations, followed by functional validation in engineered yeast. By identifying and characterizing five CPS/KSL enzymes, this study establishes a molecular framework for the early cyclization steps of diterpenoid biosynthesis in I. rubescens-LS. These findings extend prior knowledge derived from I. rubescens-JY and provide a foundation for comparative analyses of diterpenoid-pathway diversification and for future identification of downstream CYP450s, acyltransferases, and other tailoring enzymes involved in lushanrubescensin biosynthesis.

Materials and methods

Plant materials

The seeds of two distinct chemotypes of I. rubescens were collected from their respective natural producing regions in Henan Province, China: I. rubescens-JY from Jiyuan City and I. rubescens-LS from Lushan City. These plant materials were formally identified by Professor Suiqing Chen from Henan University of Chinese Medicine.

For RT-qPCR analysis, 15 individual plants of each form were cultivated synchronously in an artificial climate chamber. The growth conditions were maintained at a constant temperature of 24–26 °C and a relative humidity of 60%, under a 16 h/8 h light/dark photoperiod with a light intensity of 120 µmol m−2s−1. Healthy roots and leaves from both forms were independently harvested, immediately flash-frozen in liquid nitrogen, and stored at -80 °C until RNA extraction. Three independent biological replicates were analyzed for each tissue type and form, with each biological replicate obtained from one individual plant. Three technical replicates were performed for each RT-qPCR reaction.

Data sources

The T2T genome data of I. rubescens-LS were obtained from our previously published study (Yang et al., 2024) and have been deposited in the NCBI database under BioProject accession number PRJNA1089255. The full-length transcriptome data of I. rubescens used for gene annotation were also generated in our previous study and deposited in NCBI under BioProject accession number PRJNA1065080. The RNA-seq transcriptome data of I. rubescens were previously sequenced by our laboratory (Lian et al., 2025) and have been deposited in the Sequence Read Archive (SRA) of NCBI under BioProject accession number PRJNA910296. Protein sequences of Arabidopsis thaliana terpene synthase (TPS) genes were obtained from the A. thaliana Information Resource (TAIR) database (http://www.arabidopsis.org/).

Identification and characteristics of TPSs in I. rubescens-LS

To identify TPS genes in I. rubescens-LS, 33 annotated TPS protein sequences from A. thaliana were retrieved from TAIR and used as queries for BLASTP searches against the predicted I. rubescens-LS proteome. BLASTP searches were conducted using BLAST version 2.5.0+ with an E-value cutoff of 1×10-5). Candidate proteins showing <30% amino acid identity or <50% query coverage relative to the A. thaliana TPS queries were excluded. The remaining candidates were examined for conserved TPS domains using Hidden Markov Model (HMM)-based searches. HMM profiles corresponding to the N-terminal TPS domain (PF01397) and C-terminal TPS domain (PF03936) were obtained from the Pfam database. Candidates were retained when they contained at least one TPS-associated domain; proteins containing both PF01397 and PF03936 were classified as full-length TPS candidates. TPS loci were extracted from the I. rubescens-LS genome using TBtools. Alternative isoforms and duplicated predictions originating from the same genomic locus were manually inspected, and one representative protein sequence was retained per locus. The final TPS set was obtained by intersecting BLASTP and HMM-search results.

The amino-acid length, molecular weight, theoretical isoelectric point, and total average hydropathicity of the predicted IrfTPS (I. rubescens-LS terpene synthases) proteins were calculated using ExPASy (https://web.expasy.org/protparam/). The subcellular localization was predicted using Cell-PLoc (http://www.csbio.sjtu.edu.cn/bioinf/Cell-PLoc/). Transmembrane helices and signal peptides were predicted using TMHMM-2.0 (https://services.healthtech.dtu.dk/service.php?TMHMM-2.0) and SignalP-3.0 (https://services.healthtech.dtu.dk/service.php?SignalP-3.0).

Phylogenetic analysis of TPS genes

The protein sequences of TPS from Salvia miltiorrhiza, S. hispanica, Arabidopsis thaliana, and Perilla frutescens were obtained from NCBI (https://www.ncbi.nlm.nih.gov/). All protein accessions are listed in Supplementary Table 5. Multiple sequence alignments were performed using the MUSCLE method, and the phylogenetic tree was constructed using the neighbor-joining (NJ) method with 1000 bootstrap replications, employing the TBtools software. The resulting tree was visualized using the Chiplot online tool (https://www.chiplot.online/). The NJ tree was used primarily to classify TPS candidates and identify their relationships with previously characterized TPS proteins.

Characteristics and gene expression levels of TPS-c and TPS-e/f family members

Members of the TPS-c and TPS-e/f gene families were subjected to multiple sequence alignment and conserved motif analysis using Jalview software (Jiang et al., 2019; Waterhouse et al., 2009). Proteins lacking the characteristic catalytic motifs required for class II or class I diterpene cyclase activity were excluded from subsequent functional-candidate screening. For phylogenetic placement, the TPS gene of Physcomitrium patens (XP_024380398.1) was used as an outgroup, and functionally characterized TPS genes from I. rubescens were used as reference proteins (Supplementary Table 6). The phylogenetic tree was constructed using TBtools and visualized using Chiplot (https://www.chiplot.online/).

Gene-expression levels were derived from the previously deposited RNA-seq dataset and expressed as fragments per kilobase of transcript per million mapped reads (FPKM) (Hayashi et al., 2006; Mortazavi et al., 2008). FPKM values were used for heatmap visualization and preliminary prioritization of candidate genes because the original RNA-seq dataset was processed using this normalization metric. FPKM values were not considered independent proof of differential expression or gene function. Candidate genes were further evaluated by RT-qPCR and heterologous functional assays.

Validation of the differentially expressed genes by RT-qPCR

Five genes were selected for independent RT-qPCR validation of the RNA-seq expression patterns. In addition, 18 upstream diterpenoid-pathway genes and seven candidate diterpene synthase genes were analyzed to examine tissue-associated expression patterns. Total RNA was extracted from I. rubescens tissues using the FastPure Universal Plant Total RNA Isolation Kit (Vazyme, Nanjing, China), according to the manufacturer’s instructions. RNA quality and purity were verified using NanoDrop spectrophotometry (A260/A280 > 1.8; A260/A230 > 2.0) and 1.5% agarose gel electrophoresis. First-strand cDNA was synthesized from 1 μg of total RNA using HiScript III RT SuperMix for qPCR (+gDNA wiper) (Vazyme, Nanjing, China). Gene-specific primers were designed using the PrimerQuest Tool (Supplementary Table 1). RT-qPCR reactions were performed on a Light Cycler 96 instrument (Roche) with MonAmp SYBR Green qPCR Mix (Monad, Suzhou, China). The actin gene (GenBank: OP321409) was used as an internal reference because of its stable expression across all tested tissues in preliminary stability evaluation (coefficient of variation <5%). Relative expression levels were calculated using the 2-ΔΔCt method based on three biological replicates, each with three technical replicates.

Yeast strain construction

The S. cerevisiae strain YXWP-120, which features an enhanced supply of geranylgeranyl diphosphate (GGPP), served as the parental strain for constructing all engineered strains in this study (Supplementary Table 2). Key genes involved in the diterpenoid biosynthetic pathway (CPS and KSL) were codon-optimized for S. cerevisiae and synthesized by Qingke Biotechnology Co., Ltd. (Supplementary Table 3). Engineered strains were constructed using a decentralized metabolic pathway integration strategy facilitated by the pUMRI (marker recyclable integrative) plasmid system (Xie et al., 2015; Song et al., 2017). This system is based on the Cre/loxP recombination mechanism, allowing targeted genomic integration and subsequent recycling of the KanMX selection marker (Xie et al., 2014). Genes were assembled into pUMRI vectors and sequentially integrated into the yeast genome at predefined loci. Following each integration step, transformants were selected on YPD medium supplemented with 200 µg/mL geneticin (G418). All constructs were verified by colony PCR and Sanger sequencing before functional characterization. The expression of the integrated genes was controlled using a modified GAL regulatory system to achieve inducible and coordinated expression of the pathway enzymes.

Cultivation in shaking flasks

The engineered yeast strains were pre-cultured in 5 mL YPD (1% yeast extract, 2% peptone, and 2% glucose) at 30 °C, 220 rpm for 24 h. The pre-cultures were then inoculated to an initial OD600 of 0.05 in 200 mL of fresh YPD medium. After 24 h of incubation, dodecane (1%, v/v) was added to the culture medium, and the culture was continued for 4 days.

GC-MS identification of yeast metabolites

The assay was performed using a Trace 1310 series GC equipped with a TSQ8000 MS detector (Thermo Fisher Scientific Co. Ltd.) and a TR-5 MS capillary column (30 m×0.25 mm i.d.; DF = 0.25 μm; Thermo Fisher Scientific). Separation of the injected sample (1 μL) was achieved at a helium flow rate of 1 mL·min-1 with a temperature program of 1 min at 120 °C, followed by a gradient from 120 °C to 250 °C at 10 °C per minute. Commercial standards were not available for ent-atiserene, ent-kaurene, or miltiradiene. Therefore, reference yeast strains expressing previously characterized diterpene synthase combinations—CPS4FB + KSL4FB, CPS4FB + KSL5FB, and CPS1FB + KSL1FB—were constructed or used as reference strains for ent-atiserene, ent-kaurene, and miltiradiene, respectively, according to the established enzyme functions reported by (Jin et al., 2017). Metabolites were assigned by comparison of retention times and mass spectra with authentic standards, where available, or with reference strains producing established diterpene products.

Molecular docking analysis of KSL enzymes with ent-CPP and normal-CPP

To explore the potential structural basis of substrate recognition by KSL homologs, molecular-docking simulations were performed using ent-CPP and normal-CPP as ligands. Before docking, protein structures were selected from the RCSB Protein Data Bank. Water molecules and non-essential ligands were removed, polar hydrogen atoms were added, and partial atomic charges were assigned using AutoDockTools. The ligand structures of ent-CPP and normal-CPP were retrieved from the PubChem database. Docking simulations were carried out using AutoDock Vina 1.2.5. The docking grid for each protein was centered on the predicted catalytic pocket, as defined by conserved active-site residues. The resulting binding poses were analyzed using Discovery Studio 2019 and PyMOL 2.6 to generate two-dimensional interaction diagrams and three-dimensional binding models. The AutoDock Vina docking score was used to evaluate the predicted ligand-receptor interactions. These criteria were applied consistently to all docking complexes.

Results

Identification and characteristics of IrfTPS family in I. rubescens-LS

Based on the results of the HMM search and BLASTP, a total of 83 IrfTPSs (I. rubescens-LS terpene synthases) were identified in the I. rubescens-LS genome (Supplementary Table 4). The physicochemical properties of the predicted IrfTPS proteins were then characterized. The protein lengths varied substantially, ranging from 135 (IrfTPS50) to 1578 (IrfTPS80) amino acids. The predicted molecular weights ranged from 15.49 to 181.08 kDa, while the theoretical isoelectric points (pI) ranged from 4.47 (IrfTPS77) to 7.93 (IrfTPS83), indicating considerable structural diversity within the IrfTPS family. Subcellular localization prediction suggested that most IrfTPS proteins are localized in the cytoplasm or chloroplast, consistent with the compartmentalized biosynthesis of terpenoids in plants. Notably, only IrfTPS61 was predicted to localize to the endoplasmic reticulum. No canonical transmembrane helices or N-terminal signal peptides were predicted for the identified IrfTPS proteins, suggesting that most members are likely soluble enzymes.

Phylogenetic analyses and sequences analysis of IrfTPSs

To classify TPS genes in I. rubescens-LS, a phylogenetic analysis was performed using the TPS classification framework established in A. thaliana (Hu et al., 2019). A total of 281 TPS proteins from S. miltiorrhiza (54), S. hispanica (96), A. thaliana (33), and P. frutescens (98) were included (Figures 1A, B; Supplementary Table 5). Using bifunctional diterpene synthases from Physcomitrium patens as an outgroup (Pelot et al., 2017), a phylogenetic tree was constructed using the neighbor-joining (NJ) method. The IrfTPSs proteins were classified into five subfamilies: TPS-a, TPS-b, TPS-c, TPS-e/f, and TPS-g. Of the 83 genes, 36 belonged to TPS-a, 21 to TPS-b, 13 to TPS-c, 9 to TPS-e/f, and 4 to TPS-g (Figure 1C). The predominance of TPS-a subfamily genes is consistent with the extensive diversification of sesquiterpene synthases reported in other Lamiaceae species, suggesting an expansion of specialized terpenoid metabolism during the evolution of I. rubescens.

Figure 1.

Panel A displays a circular phylogenetic tree with branches colored by TPS gene class, panel B shows a donut chart depicting TPS gene proportions in five plant species, and panel C presents a grouped bar chart showing the number of TPS genes per clade for each species.

Open in a new tab

Phylogenetic classification and subfamily distribution of TPSs in I. rubescens-LS. (A) Neighbor-Joining phylogenetic tree of 364 TPSs from I. rubescens, S. miltiorrhiza, S. hispanica, A. thaliana, and P. frutescens. Branches are color-coded to distinguish five canonical TPS subfamilies (TPS-a, TPS-b, TPS-c, TPS-e/f, TPS-g). The scale bar represents amino acid sequence divergence. (B) Numbers and proportions of TPS genes from each species included in the phylogenetic analysis. (C) Distribution of TPS genes among the five TPS subfamilies in I. rubescens-LS.

Gene expression analysis of TPS-c and TPS-e/f in I. rubescens-LS

Diterpene synthase genes are mainly distributed in the TPS-c and TPS-e/f gene families. Class II diterpene cyclases, including CPSs, typically contain a conserved DXDD motif that initiates protonation-dependent cyclization, whereas class I diterpene synthases, including KSLs, possess a DDXXD motif that coordinates divalent metal ions and promotes substrate ionization. In the present analysis, the DDXXD motif was identified in the TPS-e/f candidates, whereas the DXDD motif was present in most TPS-c candidates; IrfTPS74, IrfTPS20, and IrfTPS35 lacked the expected complete class II catalytic motif and were excluded from the functional-candidate set (Supplementary Figure 1).

After excluding proteins lacking conserved catalytic motifs, 10 type II CPS genes and 9 type I KSL genes were retained for further analysis (Figure 2). Based on their phylogenetic relationship with functionally characterized diTPS genes from Isodon (Supplementary Table 6), the type II genes were designated CPS1YC-30073, CPS1YC-30069, CPS5YC-20029, CPS6YC-25791, CPS6YC-25764, CPS6YC-19740, CPS4YC-16872, CPS4YC-16881, CPS4YC-13279, and CPS4YC-16888. The type I genes were designated KSL5YC-15017, KSL4YC-15018, KSL3YC-29831, KSL11YC-9113, KSL11YC-9119, KSL10YC-22810, KSL7YC-18355, KSL9YC-r-9111, and KSL9YC-9118.

Figure 2.

Phylogenetic tree diagram with colored branches categorizing TPS-C and TPS-e/f gene classes, accompanied by a heatmap and an annotated table listing TPS gene names, species, and functions relevant to various plant taxa.

Open in a new tab

Gene expression levels of TPS-c and TPS-e/f candidates from I. rubescens-LS. The left panel shows a phylogenetic tree containing candidate TPS proteins from I. rubescens-LS and functionally characterized diterpene synthase references from Isodon species. Pink and blue branches indicate TPS-c and TPS-e/f clades, respectively. The accompanying table lists gene identifiers, source species, and reported catalytic products of the reference enzymes. Heatmap columns show FPKM-normalized transcript abundance in root and leaf RNA-seq samples from I. rubescens-LS and I. rubescens-JY; darker colors indicate higher transcript abundance. FPKM values were used for visualization and candidate prioritization. The five genes showing striking inter-chemotype expression divergence were selected for yeast-based functional assays, indicated by red fonts.

Candidate genes for functional assays were prioritized using multiple criteria: phylogenetic proximity to characterized Isodon diTPSs, retention of intact catalytic motifs, distinct transcript-abundance patterns between the two forms, tissue-preferential expression confirmed by RT-qPCR, and co-expression associations with other diterpenoid-pathway genes. RNA-seq expression profiles, presented as FPKM heatmaps, showed visually distinct transcript-abundance patterns for CPS1YC-30073, CPS6YC-19740, KSL5YC-15017, KSL4YC-15018, and KSL3YC-29831 between I. rubescens-LS and I. rubescens-JY (Figure 2). These patterns were used for initial candidate prioritization and were subsequently evaluated by RT-qPCR and yeast-based functional assays.

Co-expression analysis of diterpenoid synthesis pathway genes in I. rubescens

Previous studies have shown that diterpenoid biosynthetic genes in I. rubescens exhibit tissue-dependent expression patterns (Sun et al., 2023). Therefore, roots and leaves from I. rubescens-LS and I. rubescens-JY were analyzed by RT-qPCR. Expression profiles were assessed for 18 upstream genes in the diterpene synthesis pathway, along with two previously characterized CPS genes (CPS4FB and CPS1FB), two KSL genes (KSL5FB and KSL1FB) from I. rubescens-JY (Jin et al., 2017), and the five selected I. rubescens-LS diTPS candidate genes. Predicted structural comparisons indicated differences in domain architecture and overall predicted tertiary conformation between KSL9YC-9118/KSL9YC-r-9111 and the JY homolog KSL9FB (Supplementary Figures 3B, S4B). These observations motivated inclusion of the KSL9 homologs in the expression analysis but were not considered direct evidence of altered enzymatic function.

RT-qPCR analysis revealed tissue-associated expression patterns among the candidate genes (Figure 3). CPS6YC-19740, KSL3YC-29831, KSL4YC-15018, and KSL5YC-15017 showed higher transcript abundance in leaves, whereas CPS1YC-30073 showed higher transcript abundance in roots. These expression patterns suggest tissue-associated roles in diterpenoid metabolism but do not independently establish pathway function.

Figure 3.

Metabolic pathway diagram comparing the MVA and MEP pathways, with flow from precursors to terpenoid synthesis enzymes. Heat maps display gene expression (darker blue indicates higher expression) in root and leaf tissues of two I. rubescens cultivars, with labels for each gene, pathway step, and tissue sample.

Open in a new tab

RT-qPCR expression profiles of genes associated with diterpenoid biosynthesis in roots and leaves of I. rubescens-LS and I. rubescens-JY. The heat map shows the relative expression levels of key diterpenoid synthesis pathway genes measured by RT-qPCR. Darker blue rindicates higher relative expression. A and B represent gene expression levels in the roots and leaves of I. rubescens-LS, and C and D represent gene expression levels in the roots and leaves of I. rubescens-JY. CPS6YC-19740, KSL3YC-29831, KSL4YC-15018 and KSL5YC-15017 showed leaf-preferential expression, whereas CPS1YC-30073 showed root-preferential expression in I. rubescens-LS.

FPKM-based co-expression-network analysis showed that CPS1YC-30073 and KSL3YC-29831 displayed multiple co-expression associations with other diterpenoid-pathway-related genes in I. rubescens-LS (Supplementary Figure 2). This network information was used as supportive evidence for candidate prioritization. Co-expression relationships indicate correlated transcript patterns and do not, by themselves, demonstrate direct regulatory or biochemical interactions.

Gene function verification by heterologous expression in yeast

To validate the functions of selected diTPS candidates, leaf-preferential genes (CPS6YC-19740, KSL3YC-29831, KSL4YC-15018, and KSL5YC-15017) and the root-preferential gene CPS1YC-30073 were heterologously expressed in S. cerevisiae. Each gene was individually cloned into the integrative plasmid pUMRI-10 and introduced into the engineered yeast strain YXWP-120, which was previously optimized for enhanced GGPP production (Xie et al., 2014) (Figure 4A).

Figure 4.

Diagram showing engineered metabolic pathways in a microbial cell for terpenoid biosynthesis, with labeled enzymes and metabolites (panel A), chromatography and mass spectrometry data for geraniol (panel B), linalool (panel C), and several sesquiterpenes (panel D). Each chromatogram is accompanied by mass spectra identifying key molecular structures to illustrate product verification and pathway function.

Open in a new tab

Functional characterization of candidate diterpene synthases from I. rubescens-LS in engineered S. cerevisiae. (A) Schematic overview of yeast strain construction for CPS and KSL gene function verification. Red labels indicate disrupted endogenous genes; yellow and blue labels indicate integrated heterologous genes. Dashed boxes identify target diterpenoid products analyzed by GC-MS. (B) GC-MS analysis of copalol-related products in yeast strains expressing CPS genes (CPS1FB, CPS4FB, CPS6YC-19740, and CPS1YC-30073) from I. rubescens-LS. Peak a, geranylgeraniol; peak b, copalol; b-1: Mirror-plot comparison of the detected compound spectrum (red, upward) and the corresponding NIST library reference spectrum (blue, downward). b-2: NIST library reference mass spectrum and chemical structure of the matched compound, with red peaks indicating relative ion abundance and blue labels denoting m/z values. (C) GC-MS analysis of 13-Epimanool-related products in yeast strains expressing CPS genes from I. rubescens -LS. Peak c, 13-epimanool; c-1: Mirror-plot comparison of the detected compound spectrum (red, upward) and the corresponding NIST library reference spectrum (blue, downward). c-2: NIST library reference mass spectrum and chemical structure of the matched compound, with red peaks indicating relative ion abundance and blue labels denoting m/z values. (D) GC-MS analysis of yeast strains co-expressing CPS and KSL genes from I. rubescens-LS. Peak d, ent-atiserene reference; peak e, ent-kaurene; peak f, 13-epimanool reference; peak g, miltiradiene. Panels e-1 and g-1: Mirror-plot comparison of the detected compound spectrum (red, upward) and the corresponding NIST library reference spectrum (blue, downward).; panels e-2 and g-2: NIST library reference mass spectrum and chemical structure of the matched compound, with red peaks indicating relative ion abundance and blue labels denoting m/z values. All heterologous genes were integrated at the HO locus of S. cerevisiae YXWP-120. The blank control denotes the parental strain carrying an empty vector integrated at the HO locus. Red and blue traces represent the mass spectra of the detected sample peak and the corresponding standard/reference peak, respectively.

Expression of CPS6YC-19740 and CPS1YC-30073 enabled the detection of copalol and 13-epimanool-related products in the engineered yeast strains (Figures 4B, C). Because endogenous yeast phosphatases can hydrolyze CPP intermediates to the corresponding diterpene alcohols (Cao et al., 2010), detection of copalol and 13-epimanool was used as indirect evidence for CPS-mediated formation of CPP intermediates. Geranylgeraniol (GGOH) was detected in all strains (Figure 4B), demonstrating an adequate intracellular supply of GGPP. Peak identities were assigned by comparison with authentic standards and/or reference strains, as described in the Methods. Peak-area profiles suggested different relative abundances of copalol and 13-epimanool among CPS-expressing strains. However, as no absolute calibration or enzyme-kinetic analysis was performed, these data are interpreted as semi-quantitative comparisons of relative product accumulation rather than direct measurements of catalytic efficiency. Copalol and 13-epimanool share the same molecular formula (C20H34O), but they differ in bond orientation and the hydroxyl-group and double-bond positions.

Previous studies have reported that CPS1FB and CPS4FB catalyze the formation of normal-CPP and ent-CPP, respectively (Jin et al., 2017). To characterize the function of KSL genes, leaf-enriched KSL genes were individually co-expressed with either CPS4FB or CPS6YC-19740, whereas root-enriched KSL genes were co-expressed with CPS1FB. Three KSL candidates (KSL5YC-15017, KSL4YC-15018, and KSL3YC-29831) produced diterpene products consistent with ent-kaurene formation (peak e) in the corresponding engineered strains. Notably, the strain expressing KSL3YC-29831 additionally generated a product consistent with miltiradiene (peak g) (Figure 4D).

The identification of ent-kaurene-producing KSL activity in I. rubescens-LS supports the presence of early cyclization reactions shared with the oridonin-producing JY form. Previous work reported that the JY homolog KSL3 produces miltiradiene from normal-CPP (Jin et al., 2017), whereas KSL3YC-29831 in the present yeast assay produced products consistent with both ent-kaurene and miltiradiene. This difference suggests functional divergence between the homologs, although the observed product profile may also be influenced by the available CPP substrate, heterologous-host context, and relative enzyme expression.

Sequence and predicted-structure comparisons indicated that KSL3YC-29831 lacks the predicted γ-domain observed in its JY homolog KSL3FB (Supplementary Figures 3A, S4A). This structural difference may be associated with the distinct product profile, but direct causal evidence will require further structural and mutational analyses.Likewise, the I. rubescens-LS KSL9 homologs, KSL9YC-r-9111 and KSL9YC-9118, did not generate products corresponding to the palustradiene-related activities reported for JY KSL9 (Sun et al., 2023). Predicted differences in domain architecture and tertiary conformation may contribute to these functional differences (Supplementary Figures 3B, S4B). Collectively, these results identify two CPS and three KSL enzymes with detectable activity in engineered yeast and establish a set of candidate diTPSs for the early cyclization steps of diterpenoid biosynthesis in I. rubescens-LS.

Molecular docking to characterize the interaction of KSL enzymes with ent-CPP and normal-CPP

All five KSL3-type proteins showed strong binding affinity for ent-CPP: AFU61898.1-Salvia sclarea (−8.2 kcal/mol, hydrogen-bonded to ASP92), KSL3-FB (−8.8 kcal/mol, hydrogen-bonded to TYR-273, THR-318, HIS-356, GLN-355), KSL3a-FB (−9.4 kcal/mol, hydrogen-bonded to ASN-399, GLU-334), KSL3YC-29831 (−9.0 kcal/mol, hydrogen-bonded to ASN-358, THR-318), and TPS2-FB (−9.0 kcal/mol, hydrogen-bonded to THR-302, ASN-399, HIS-357). Extensive van-der-Waals and alkyl-type hydrophobic interactions further fix ent-CPP within the active-site cavity. These binding characteristics structurally support the ent-CPP recognition and ent-kaurene-forming activity of KSL3 homologs. KSL9 homologs exhibited divergent binding behaviors toward normal-CPP. UVC58051.1-KSL9 had weak binding (−6.2 kcal/mol, hydrogen-bonded to GLN331 、THR336 、LYS378), whereas KSL9YC-9118 (−8.0 kcal/mol, hydrogen-bonded to ASP-493, ARG-634, ASN-637) and KSL9YC-r-9111 (−9.7 kcal/mol, hydrogen-bonded to ARG-566, ASP-525) displayed strong binding. Compared with UVC58051.1-KSL9, KSL9YC-9118 and KSL9YC-r-9118 possessed distinct hydrogen-bond- and hydrophobic-residue profiles, which may alter substrate orientation in the catalytic pocket (Supplementary Figure 5).

Discussion

The biosynthesis of bioactive ingredients in traditional Chinese medicine represents a critical frontier in pharmacological research, yet progress is frequently constrained by incomplete knowledge of the enzymes and pathways responsible for the formation of structurally complex natural products (Hu et al., 2019). Lushanrubescensin is a distinctive ent-kaurane diterpenoid that contains four acetyl substituents and displays greater lipophilicity than oridonin—the characteristic diterpenoid of I. rubescens-JY (Wang, 1987). These structural characteristics highlight its potential pharmaceutical relevance and motivate elucidation of its biosynthetic pathway. In the present study, we identified 22 diterpene synthase (diTPS) genes in I. rubescens-LS, comprising 13 type II copalyl diphosphate synthase (CPS) genes and 9 type I kaurene synthase-like (KSL) genes. By comparison, 26 diTPS genes, including 13 CPS and 13 KSL genes, have been reported in I. rubescens-JY. Thus, the two forms exhibit broadly comparable diTPS repertoires but differ in the composition of their KSL subfamilies. CPS enzymes from both forms generated normal-CPP and ent-CPP, whereas the characterized KSL enzymes from I. rubescens-LS produced two diterpene products, miltiradiene and ent-kaurene (Supplementary Figure 6). In contrast, KSL enzymes from I. rubescens-JY reportedly generated a broader range of products, including nezukol, palustradiene, abieta-7,13-diene, miltiradiene, isopimaradiene, ent-sandaracopimaradiene, ent-kaurene, and ent-atisirene (Jin et al., 2017; Sun et al., 2023). These observations indicate lower characterized KSL product diversity in I. rubescens-LS than in I. rubescens-JY. Furthermore, heterologous expression in S. cerevisiae functionally validated five diTPSs (CPS6YC-19740, CPS1YC-30073, KSL5YC-15017, KSL4YC-15018, KSL3YC-29831). This research Together, these results expand the currently available enzymatic toolkit for Isodon diterpenoid biosynthesis and provide a foundation for reconstructing the early cyclization steps of lushanrubescensin biosynthesis.

Comparative functional analysis indicated that the CPS complement and core cyclization activities are relatively conserved between I. rubescens-LS and I. rubescens-JY. Both forms possess CPS enzymes capable of producing stereoisomeric precursors (ent-CPP and normal-CPP), as supported by the formation of copalol and 13-epimanool in yeast strains expressing the newly identified I. rubescens-LS CPS genes. In contrast, the KSL family exhibited a reduced gene complement and a narrower experimentally characterized product spectrum in I. rubescens-LS. I. rubescens-LS contains 9 functional KSL genes, compared to the 13 KSL genes documented in I. rubescens-JY (Han et al., 2003) (Supplementary Figure 7). This difference may contribute to the distinct diterpenoid profiles of the two forms, although causal links between KSL gene-copy variation and chemotypic divergence require further investigation.

Sequence and domain-architecture comparisons further suggest possible functional diversification among KSL homologs. For instance, KSL3YC-29831 in I. rubescens-LS lacks the predicted γ-domain present in its JY homolog, KSL3FB (Sun et al., 2023). Whereas KSL3FB was reported to produce miltiradiene from normal-CPP, KSL3YC-29831 produced products consistent with both miltiradiene and ent-kaurene in the yeast assay. This association between domain architecture and product profile raises the possibility that structural variation contributes to functional diversification among KSL homologs. Independent loss of the γ-domain has been reported during plant diterpene synthase evolution, supporting the evolutionary plausibility of domain-architecture variation among TPS enzymes (Hillwig et al., 2011; Wang and Peters, 2023). Similarly, the I. rubescens-LS homologs of I. rubescens-JY KSL9 (KSL9YC-r-9111 and KSL9YC-9118) displayed predicted differences in domain architecture and tertiary conformation, particularly within the β-domain, and did not produce the palustradiene-related products reported for I. rubescens-JY KSL9 (Sun et al., 2023). Collectively, these observations suggest that variation in KSL gene complement, domain architecture, and protein sequence may contribute to diterpene scaffold diversity between the two forms.

To further investigate the potential structural basis of functional divergence, molecular-docking simulations were performed to compare the predicted binding modes of KSL homologs with ent-CPP and normal-CPP. Within the same docking workflow, the calculated docking scores were used as relative indicators of the favorability of the predicted enzyme-substrate binding poses (Madeddu et al., 2022). The simulations suggested that KSL3 homologs could accommodate ent-CPP and that KSL9 homologs could accommodate normal-CPP within their respective predicted active-site pockets. The top-ranked poses were stabilized by predicted hydrogen bonds, van der Waals contacts, and hydrophobic interactions involving alkyl-containing residues. Comparative docking analysis further indicated that amino-acid substitutions surrounding the predicted substrate-binding pocket may alter CPP orientation and local noncovalent interactions, thereby potentially contributing to the divergent product profiles observed among KSL paralogs. However, the docking results should be interpreted as model-dependent, hypothesis-generating evidence rather than direct measurements of substrate affinity, complex stability, or catalytic efficiency. Future structural studies, site-directed mutagenesis, and biochemical kinetic analyses will be necessary to verify the functional roles of the identified candidate residues. Nevertheless, these docking results provide a rational basis for prioritizing active-site residues for subsequent mutational analysis of CPP recognition and product specificity in KSL diterpene synthases.

The identification of functional ent-kaurene synthase activity (KSL5YC-15017, KSL4YC-15018, KSL3YC-29831) in I. rubescens-LS establishes ent-kaurene as a likely central intermediate in the biosynthetic routes to oridonin in I. rubescens-JY and lushanrubescensin in I. rubescens-LS. Because the two forms produce distinct oxidized and acetylated ent-kaurane diterpenoids, their metabolic divergence is likely to arise mainly from downstream tailoring reactions after ent-kaurene formation. Cytochrome P450 (CYP450) enzymes are expected to mediate sequential hydroxylation and oxidation reactions, while acyltransferases are likely to participate in the acetylation steps required for lushanrubescensin formation. In I. rubescens-JY, the tandemly duplicated enzymes IrCYP706V2 and IrCYP706V7 have been shown to catalyze early oxidative modifications of the ent-kaurene core during oridonin biosynthesis (Sun et al., 2023). Accordingly, comparative genomic and transcriptomic analyses of CYP450s—including CYP706V candidates—as well as candidate acyltransferases between I. rubescens-LS and I. rubescens-JY will be essential for identifying the downstream enzymes responsible for lushanrubescensin formation (Chen et al., 2016). The co-expression network generated in this study (Supplementary Figure 2) offers a preliminary basis for prioritizing such candidates, particularly genes co-expressed with the functionally validated CPS and KSL genes.

Although this study establishes the TPS repertoire of I. rubescens-LS and functionally validates five CPS/KSL enzymes in yeast, the downstream enzymatic steps responsible for conversion of ent-kaurene to lushanrubescensin remain to be elucidated. In particular, the CYP450s, acyltransferases, and other potential tailoring enzymes involved in the oxidation and acetylation of the ent-kaurane scaffold have not yet been functionally characterized. Moreover, the association between KSL structural variation and catalytic divergence is inferred from sequence/domain comparisons and product profiles and will require further validation through structural and mutational analyses. Nonetheless, the engineered yeast strains capable of producing ent-kaurene provide a useful chassis for the functional screening of downstream CYP450s and acyltransferases. Future work integrating comparative multi-omics, co-expression-guided candidate selection, enzyme assays, and pathway reconstruction will be required to establish the complete lushanrubescensin biosynthetic pathway. An integrated pathway diagram summarizing the validated CPS/KSL enzymes, their products, and the proposed downstream CYP450- and acyltransferase-mediated reactions has been added as Figure 5. Collectively, this work establishes a genetic and enzymatic foundation for dissecting diterpenoid diversification in I. rubescens and for future microbial production of lushanrubescensin.

Figure 5.

Chemical pathway diagram showing the biosynthesis of lushanrubescensin from GGPP through multiple intermediates including LPP, normal-CPP, ent-CPP, miltiradiene, and ent-kaurene, with specific enzymes labeled at each step and red X marks denoting blocked normal-CPP conversion steps.

Open in a new tab

Proposed biosynthetic pathway of lushanrubescensin in I. rubescens-LS.Red crosses indicate that no detectable diterpene products were formed from normal-CPP by the two KSL9 enzymes. The experimentally validated CPS and KSL reactions are indicated by solid arrows. Putative downstream oxidation and acetylation steps catalyzed by CYP450s and acyltransferases are indicated by dashed arrows. CPS, copalyl diphosphate synthase; KSL, kaurene synthase-like enzyme; CYP450, cytochrome P450 monooxygenase; AT, acyltransferase.

Funding Statement

The author(s) declared that financial support was received for this work and/or its publication. This study was funded by the National Natural Science Foundation of China (82574543, 81173486), Natural Science Foundation of Henan (252300423157, 232301420078), Chinese Herbal Medicine Industry Technology System of Henan Province (Grant No.14 [2018]).

Footnotes

Edited by: Yunpeng Cao, Chinese Academy of Sciences (CAS), China

Reviewed by: Muzammil Shah, Emerson University Multan, Pakistan

Shu Wang, Southwest Forestry University, China

Data availability statement

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

Author contributions

HY: Project administration, Data curation, Writing – original draft, Methodology. JL: Data curation, Writing – original draft, Investigation, Methodology. CL: Methodology, Project administration, Data curation, Writing – original draft. HWY: Writing – original draft, Investigation, Methodology. LZ: Writing – review & editing. NH: Writing – review & editing. YL: Writing – review & editing. RC: Writing – review & editing. XL: Writing – review & editing. SX: Writing – review & editing. XS: Writing – review & editing. LPZ: Writing – review & editing. LW: Writing – review & editing. JY: Writing – review & editing. YF: Writing – review & editing. RM: Writing – review & editing. BZ: Writing – review & editing. LY: Writing – original draft. SC: Methodology, Writing – original draft, Project administration.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2026.1937259/full#supplementary-material

DataSheet1.docx (14.8MB, docx)
Table1.xlsx (155.8KB, xlsx)

References

  1. Cao R., Zhang Y., Mann F. M., Huang C., Mukkamala D., Hudock M. P., et al. (2010). Diterpene cyclases and the nature of the isoprene fold. Proteins 78, 2417–2432. doi:  10.1002/prot.22751 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Chen S. Q., Cui C., Pei L. X., Dong C. M., Feng W. S. (2011). Qualitative evaluation of Rabdosiae rubescentis herba from different place and origin. Chin. J. Exp. Tradit. Med. Formulae 17, 122–126. doi:  10.13422/j.cnki.syfjx.2011.17.052 [DOI] [Google Scholar]
  3. Chen S. Q., Yin L., Song J., Cui C., Meng C. (2016). Molecular analysis of different origin of Rabdosia rubescens germplasm resources. Asia Pac. Tradit. Med. 12, 5. [Google Scholar]
  4. Gao Z. Y., Li Y. R. (1986). New taxa of Rabdosia rubescens. J. Syst. Evol. 24, 15–16. [Google Scholar]
  5. Han Q. B., Xiang W., Li R. T., Li M. L., Sun H. D. (2003). Ent-kaurane diterpenoids from Isodon rubescens var. lushanensis. Chem. Pharm. Bull. 51, 269–272. doi:  10.3987/com-02-9688 [DOI] [PubMed] [Google Scholar]
  6. Hayashi K., Kawaide H., Notomi M., Sakigi Y., Matsuo A., Nozaki H. (2006). Identification and functional analysis of bifunctional ent-kaurene synthase from the moss Physcomitrella patens. FEBS Lett. 580, 6175–6181. doi:  10.1016/j.febslet.2006.10.018 [DOI] [PubMed] [Google Scholar]
  7. Hillwig M. L., Xu M., Toyomasu T., Tiernan M. S., Wei G., Cui G., et al. (2011). Domain loss has independently occurred multiple times in plant terpene synthase evolution. Plant J. 68, 1051–1060. doi:  10.1111/j.1365-313X.2011.04756.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Hu H. Y., Shen X. F., Liao B. S., Luo L., Xu J., Chen S. L. (2019). Herbgenomics: a stepping stone for research into herbal medicine. Sci. China Life Sci. 62, 913–920. doi:  10.1007/s11427-018-9472-y [DOI] [PubMed] [Google Scholar]
  9. Jiang S. Y., Jin J. J., Sarojam R., Ramachandran S. (2019). A comprehensive survey on the terpene synthase gene family provides new insight into its evolutionary patterns. Genome Biol. Evol. 11, 2078–2098. doi:  10.1093/gbe/evz142 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Jin B. L., Cui G., Guo J., Tang J., Duan L., Lin H., et al. (2017). Functional diversification of kaurene synthase-like genes in Isodon rubescens. Plant Physiol. 174, 943–955. doi:  10.1104/pp.17.00202 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Lian C. L., Yang H., Lan J. X., Zhang X. Y., Zhang F., Yang J. F., et al. (2022). Comparative analysis of chloroplast genomes reveals phylogenetic relationships and intraspecific variation in the medicinal plant Isodon rubescens. PloS One 17, e0266546. doi:  10.1371/journal.pone.0266546 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Lian C. L., Zhang B., Zhang F., Lan J. X., Yang J. F., Yang H., et al. (2025). Multi-combination comparative transcriptome analysis response to MeJA and identifies genes putatively involved in oridonin biosynthesis in Isodon rubescens. BMC Plant Biol. 25, 16. doi:  10.1186/s12870-025-07460-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Liu X., Zhang Y., Li J., Jiang H. (2026). Emerging technologies empowering the biosynthesis of paclitaxel. Synth. Biol. Eng. 4, 10006. doi:  10.70322/sbe.2026.1000639040700 [DOI] [Google Scholar]
  14. Madeddu F., Di Martino J., Pieroni M., Del Buono D., Bottoni P., Botta L., et al. (2022). Molecular docking and dynamics simulation revealed the potential inhibitory activity of new drugs against human topoisomerase I receptor. Int. J. Mol. Sci. 23, 14652. doi:  10.3390/ijms232314652 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Mortazavi A., Williams B. A., McCue K., Schaeffer L., Wold B. (2008). Mapping and quantifying mammalian transcriptomes by RNA-Seq. Nat. Methods 5, 621–628. doi:  10.1038/nmeth.1226 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Pelot K. A., Hagelthorn D. M., Addison J. B., Zerbe P. (2017). Biosynthesis of the oxygenated diterpene nezukol in the medicinal plant Isodon rubescens is catalyzed by a pair of diterpene synthases. PloS One 12, 1–17. doi:  10.1371/journal.pone.0176507 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Shang H. Q., Yang Q. B., Qiang S., Zheng R., Zhang C. Q., Hu C. Y., et al. (2025). Engineering caffeic acid O-methyltransferase for efficient de novo ferulic acid synthesis. Eng. Life Sci. 25, e70018. doi:  10.1002/elsc.70018 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Song T. Q., Ding M. Z., Zhai F., Liu D., Liu H., Xiao W. H., et al. (2017). Engineering Saccharomyces cerevisiae for geranylgeraniol overproduction by combinatorial design. Sci. Rep. 7, 14991. doi:  10.1038/s41598-017-15005-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Sun Y. W., Shao J., Liu H. L., Wang H., Wang G. Y., Li J. H., et al. (2023). A chromosome-level genome assembly reveals that tandem-duplicated CYP706V oxidase genes control oridonin biosynthesis in the shoot apex of Isodon rubescens. Mol. Plant 16, 517–532. doi:  10.1016/j.molp.2022.12.007 [DOI] [PubMed] [Google Scholar]
  20. Wang Q. D. (1987). Study of the antitumor action of lushanrubescensin. Henan Inst. Med. Sci. 22, 372. doi:  10.13705/j.issn.1671-6825.1987.04.012 [DOI] [Google Scholar]
  21. Wang X., Jing Z., Wang F., Wang Z., Li X. (2026). Advances in gastrodin production: from native to engineered biosynthesis. Synth. Biol. Eng. 4, 10009. doi:  10.70322/sbe.2026.1000939040700 [DOI] [Google Scholar]
  22. Wang Z., Peters R. J. (2023). Dynamic evolution of terpenoid biosynthesis in the Lamiaceae. Mol. Plant 16, 963–965. doi:  10.1016/j.molp.2023.04.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Waterhouse A. M., Procter J. B., Martin D. M. A., Clamp M., Barton G. J. (2009). Jalview Version 2--a multiple sequence alignment editor and analysis workbench. Bioinformatics 25, 1189–1191. doi:  10.1093/bioinformatics/btp033 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Xie W., Liu M., Lv X., Lu W., Gu J., Yu H. (2014). Construction of a controllable beta-carotene biosynthetic pathway by decentralized assembly strategy in Saccharomyces cerevisiae. Biotechnol. Bioeng. 111, 125–133. doi:  10.1002/bit.25002 [DOI] [PubMed] [Google Scholar]
  25. Xie W., Ye L., Lv X., Xu H., Yu H. (2015). Sequential control of biosynthetic pathways for balanced utilization of metabolic intermediates in Saccharomyces cerevisiae. Metab. Eng. 28, 8–18. doi:  10.1016/j.ymben.2014.11.007 [DOI] [PubMed] [Google Scholar]
  26. Yang H., Lian C. L., Liu J. L., Yu H. W., Zhao L., He N., et al. (2024). High-quality assembly of the T2T genome for Isodon rubescens f. lushanensis reveals genomic structure variations between 2 typical forms of Isodon rubescens. GigaScience 13, giae075. doi:  10.1093/gigascience/giae075 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Yang H., Liu D. (2020). Current situation and prospective on resource evaluation and sustainable utilization of Rabdosiae rubescentis Herba. Tradit. Chin. Med. 9, 506–514. doi:  10.12677/tcm.2020.96076 [DOI] [Google Scholar]
  28. Zhao Y. P., Qian Y. X., Liu X. H., Sun J., He Z. G. (2007). Freeze-drying of oridonin-loaded solid lipid nanoparticles. Chin. J. Pharm. 5, 241. doi:  10.14146/j.cnki.cjp.2007.05.015 [DOI] [Google Scholar]

Associated Data

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

Supplementary Materials

DataSheet1.docx (14.8MB, docx)
Table1.xlsx (155.8KB, xlsx)

Data Availability Statement

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

Articles from Frontiers in Plant Science are provided here courtesy of Frontiers Media SA

RESOURCES