Comprehensive characterization of volatile terpenoids and terpene synthases in Lanxangia tsaoko

Abstract

Lanxangia tsaoko is widely utilized in human cuisine as a popular flavoring agent due to its distinctive aroma. It also has a long history of use in traditional Chinese medicine. The edible and medicinal properties of L. tsaoko are primarily attributed to its diverse array of volatile metabolites. Previous research has mainly focused on classifying the constituents and their pharmacological activities in L. tsaoko, leaving gaps in comprehensive identification and elucidation of the biosynthetic mechanisms of these metabolites. In this study, we employed a multi-omics approach and functional characterization to investigate the biosynthesis of volatile terpenoids in L. tsaoko. The results demonstrated that terpenoids constituted the highest proportion of volatile compounds in L. tsaoko. Additionally, 42 terpene synthase (TPS) coding genes were identified through genome-wide analysis. Functional characterization revealed that eight LtTPSs effectively catalyzed geranyl pyrophosphate to produce monoterpenoids, while four LtTPSs converted farnesyl pyrophosphate to generate sesquiterpenoids. Genome-wide and single-gene duplication events contributed to functional diversification among LtTPSs with high identity, promoting the diversity of terpenoids. These findings provide a foundation for understanding the biosynthesis of volatile terpenoids in L. tsaoko, enhance the current knowledge of TPS, and contribute to the broader understanding of the biochemical diversity of terpenoids in plants.

Core

This investigation employed a multidisciplinary approach, integrating transcriptomics, metabolic profiling, and functional characterization of terpene synthase (TPS)-encoding genes, to examine the volatile terpenoids in L. tsaoko. Metabolomic analysis revealed terpenoids as the predominant volatile compounds. Functional identification demonstrated that 10 LtTPSs belonging to the TPS-a/b subfamily could catalyze geranyl pyrophosphate (GPP) or farnesyl pyrophosphate (FPP) to produce mono/sesqui-terpenoids. These findings offer valuable insights into the molecular mechanisms driving the diversity and abundance of volatile terpenoids, which are essential for advancing research in biotechnology and natural product development.

Gene & accession numbers

Information for the genes discussed in this article is available in the GenBank database, maintained by the National Center for Biotechnology Information (NCBI), under accession numbers PQ448317-PQ448325.

Introduction

Lanxangia tsaoko, a perennial herbaceous plant of the Zingiberaceae family, is widely distributed in southeastern China, parts of Vietnam, and northern Laos (Wei et al. 2019). Its fruit, known locally as "Cao Guo" in Chinese, is utilized both as a significant food additive and spice for masking odors and enhancing flavor, and as a traditional Chinese medicine effective in treating various disorders. The distinctive aroma of L. tsaoko is primarily attributed to its rich content of volatile organic compounds (VOCs), including monoterpenoids, sesquiterpenoids, and fatty acids, which have garnered attention due to their diverse biological activities. For instance, monoterpenoids derived from L. tsaoko demonstrate anti-inflammatory (Liang et al. 2024) and anti-Trichomonas vaginalis activities (Dai et al. 2016). Given its high edible and medicinal value, L. tsaoko has been the subject of extensive research in recent years. However, these studies on the active ingredients of L. tsaoko have primarily focused on the identification and classification of constituents (Liu et al. 2018; He et al. 2020; Shi et al. 2021). Further investigation is required into the biosynthetic pathways and regulatory mechanisms governing VOCs in L. tsaoko. In plant tissues, VOCs are synthesized to serve important biological functions, including resistance against pathogenic bacteria, parasitic organisms, and herbivores (Rawat et al. 2024). Therefore, research on VOCs in L. tsaoko could enhance our understanding of its aroma formation mechanisms and provide insights into its growth, development, and ecological adaptation.

VOCs are classified into terpenoids, benzene aromatics, and fatty acid derivatives based on their origin, with terpenoids forming the predominant class (Holopainen and Gershenzon 2010). Terpenoids, also referred to as isoprenoids, represent the largest and most structurally diverse group of natural products in plants (Ahmad et al. 2024). These compounds play crucial roles in chemical ecology, influencing interactions with herbivores, pathogens, and symbiotic organisms (Tholl 2015; Nagegowda and Gupta 2020; Wei et al. 2023; Cao et al. 2024; Jun Yang and Wang, 2024). For example, monoterpenoids and sesquiterpenoids function as indirect defense mechanisms by deterring herbivores or attracting their natural predators (Gershenzon and Dudareva 2007). Furthermore, terpenoids contribute to the production of essential oils and resins, which protect plants against microbial infections and environmental stressors (Pichersky and Raguso 2018; Yadav et al. 2024). Volatile terpenoids, primarily comprising isoprene (C5), monoterpenes (C10), and sesquiterpenes (C15), constitute the largest class of plant volatile compounds (Dudareva et al. 2006). Notably, the more volatile monoterpenoids and sesquiterpenoids play diverse roles in plant growth, development, and ecological fitness (Chen et al. 2011). These functions encompass both direct and indirect forms of defense, the enhancement of flavor and aroma in herbs and fruits, and antioxidant activities (Allen et al. 2019). Due to their significant medicinal value and relatively low abundance in native species, research into the biosynthesis pathways and metabolic engineering of plant-specialized sesquiterpenoids and monoterpenoids has gained increasing attention (Liu et al. 2024).

The classical cytosolic mevalonate (MVA) pathway synthesizes isopentenyl diphosphate (IPP) from acetyl-CoA (Newman and Chappell 1999), playing a vital role in sesquiterpenoids and sterol production (Nes 2011). Conversely, the plastidial 2-C-methyl-D-erythritol-4-phosphate (MEP) pathway generates IPP and dimethylallyl diphosphate (DMAPP) from pyruvate (Wang et al. 2024), and is fundamental for synthesizing monoterpenoids, diterpenoids, and carotenoids (Lichtenthaler 1999). IPP and DMAPP originate from two distinct biosynthetic pathways localized in different subcellular compartments (Kubeczka 2020; Wu et al. 2024). In some plants, DMAPP formed in plastids serves as a precursor for isoprene synthesis catalyzed by isoprene synthase (ISPS). All terpenoids are synthesized from these universal five-carbon precursors. IPP and DMAPP undergo further condensation by prenyl diphosphate synthases within their respective compartments (Zhao et al. 2023b; Yang et al. 2024). These prenyl diphosphate intermediates then serve as substrates for a diverse array of terpene synthase (TPS) (Wang et al. 2023), encoded by large TPS gene families comprising eight subfamilies. Historically, TPSs were classified into two types based on their structure and catalytic mechanism: type I and type II (Chen et al. 2011; Gao et al. 2012). Recent classifications based on sequence similarities and functional features have expanded this categorization into eight subfamilies: TPS-a, TPS-b, TPS-c, TPS-d (found in gymnosperms only), TPS-e/f, TPS-g, and TPS-h (in Selaginella spp.). Among these, TPS-c belongs to Type II TPS while others belong to Type I (Wang et al. 2019b). Type I TPSs are characterized by an aspartic acid-rich DDXXD motif (α-domain) in their C-terminal domain, which binds metal cofactors (Mg²⁺ or Mn²⁺). These cofactors are essential for interacting with isoprenyl diphosphate substrates and facilitating substrate cation formation (Christianson 2017). In contrast, type II terpene synthases possess a DXDD motif in the β-domain near the N-terminal (Zerbe and Bohlmann 2015). The catalytic activity of TPS enzymes results in the formation of an extensive range of terpenoid compounds, including mono-, sesqui-, and diterpenes, as well as other specialized metabolites crucial for plant adaptation and defense mechanisms (Alicandri et al. 2020).

In recent years, L. tsaoko has attracted significant attention due to its high concentration of bioactive constituents, which show potential for application as food additives and in pharmaceutical development. Among these, terpenoids constitute a major class of compounds in L. tsaoko. Nevertheless, the biochemical and molecular basis of terpenoid biosynthesis remains unclear. Therefore, this study aimed to conduct a comprehensive analysis and investigation of volatile compounds in L. tsaoko, examining their compositional characteristics and biosynthetic pathways. This research endeavored to provide a scientific foundation for the industrial utilization and functional assessment of L. tsaoko. Based on structural domain searches and functional annotations, 42 LtTPSs were identified in this study, with 10 candidate LtTPSs functionally characterized to elucidate the genetic basis of volatile terpenoid enrichment. The identification of terpenoids and the TPS gene family in L. tsaoko provides crucial target genes for in-depth investigation of volatile oil biosynthesis regulation in this species. In conclusion, this study offers novel insights for investigating the biosynthesis mechanisms of diverse volatile terpenoids in Zingiberaceae species, while also providing genetic resources for the improvement of this medicinal plant and the evolutionary analysis of the ginger family.

Results

Volatile metabolite profiles from L. tsaoko using GC–MS

We conducted gas chromatography-mass spectrometry (GC–MS) analysis to evaluate volatile compounds in six distinct tissues: root, stem, flower, leaf, fruit, and podetium (Fig. 1A). Principal Component Analysis (PCA) was employed to assess the metabolic profiles of these samples, revealing consistent patterns within the same group, thus demonstrating the reliability and reproducibility of our detection results. The separation trend between different tissues was evident, indicating significant metabolic differences among these tissues (Fig. 1B). A total of 1009 volatile compounds were detected (Table S1) and categorized into 15 classes. Notably, terpenoids (20.5%) emerged as the predominant class, followed by esters (16.5%), heterocyclic compounds (16.2%), and ketones (8.6%) (Fig. 1C). Furthermore, the Orthogonal Partial Least Squares Discriminant Analysis (OPLS-DA) plot demonstrated significant metabolic alterations among different tissues, with R2X, R2Y, and Q² values of 0.821, 0.992, and 0.996, respectively (Fig. S1, Table S2). The Q2 values among the six tissues ranged from 0.9 to 1, confirming the robustness and efficiency of the results and methods. This approach facilitated the identification of differentially accumulated metabolites (DAMs) using Variables important in the projection (VIP) analysis.

Fig. 1

Six tissues metabolite profiles from L. tsaoko. A Principal Component Analysis (PCA) of metabolites detected in L. tsaoko. The X-axis and Y-axis indicate PC1 and PC2, respectively. B The distribution of the metabolites. C The relative content heatmap (row scale) of the top 30 volatile compounds in six different tissues. R: root, S: stem, L: leaf, Fr: fruit, Fl: flower, P: podetium. D The top 30 KEGG enrichment pathways. The X-axis position and size of the bar indicate the enrichment degree of the DAMs. The color of bars represents the P-value

To further analyze the distribution of VOCs across different tissues, we quantified the content of all volatile compounds. The results revealed the highest concentrations in fruits, followed by flowers. This distribution may be closely associated with the culinary applications of L. tsaoko fruits and the floral scents (Fig. S2). A barplot of the 30 most abundant compounds demonstrated the prevalence of terpenoids, indicating their significant contribution to the overall volatile compound composition. Notably, terpenoids were more abundant in flowers and fruits compared to the other four tissues. For instance, α-pinene and β-cis-ocimene were most concentrated in flowers, while citral, o-cymene, pseudolimonene, and trans-anethole were more prevalent in fruits. Eucalyptol was enriched across five tissues, excluding the roots (Fig. S3). A heatmap constructed from this data set further confirmed that the top 30 volatile compounds were primarily detected in flowers and fruits (Fig. 1D). KEGG enrichment analysis of the differential metabolites indicated significant enrichment in monoterpenoid biosynthesis (Fig. 1E). These findings aligned closely with the utilization of L. tsaoko fruits as a spice and in traditional Chinese medicine. The identification of volatile compounds in L. tsaoko revealed terpenoids as the primary active components. Subsequent analysis of volatile terpenoids across six distinct tissues, focusing on the 30 most abundant terpenoids, demonstrated that flowers and fruits possessed the richest repertoire, with most monoterpenoids and sesquiterpenoids highly concentrated in these tissues (Fig. 2A).

Fig. 2

Volatile terpenoids and terpenoid precursor biosynthesis pathways in L. tsaoko. A Heatmap depicting the distribution of volatile terpenoid content across six distinct tissues (R: root, S: stem, L: leaf, Fr: fruit, Fl: flower, P: podetium) of L. tsaoko. B Tissue-specific expression profiles of genes involved in terpenoid precursor biosynthesis (heatmap, row scale). The heatmap illustrates tissue-specific expression patterns of genes participating in the biosynthesis pathways of terpenoid precursors. Different shapes in the lower panel indicate different duplication event origins of the upstream genes

Identification of upstream genes related to terpenoid biosynthesis

The expression patterns of upstream synthases involved in the MEP and MVA pathways play a crucial role in regulating the biosynthesis of flavor terpenoids (Qiao et al. 2021). We initially identified key genes involved in terpenoid backbone biosynthesis, revealing multiple copies of most terpenoid pathway genes. Notably, DXS has eleven copies, DXR has four copies, and HDS has five copies. To investigate the origin of these multiple copies, we identified various gene duplication modes, including whole genome duplication (WGD), tandem duplication (TD), proximal duplication (PD), transposed duplication (TRD), and dispersed duplication (DSD). Among these, TD, PD, TRD, and DSD were classified as single-gene duplication events (Fig. S4). The results indicated that WGD events significantly contributed to the multiplication of upstream genes involved in specific terpenoid biosynthesis (Fig. 2B). Transcriptomic analysis revealed distinctive expression patterns. LtDXS2/3/5/7/8/10 exhibited the highest expression in flowers, while LtDXS1/6/9 showed relatively higher expression in roots compared to the other eight LtDXSs. Three of the four DXR genes demonstrated peak expression in fruits. Among the five HDS genes, three (HDS1/4/5) showed elevated expression levels in flowers. Additionally, although the MCT, CMK, MCS, and HDR genes exhibited only single or double copies in the genome, they were considerably expressed in flowers or fruits. These transcriptomic data aligned with the metabolomic results, suggesting these genes' potential involvement in monoterpenoid biosynthesis. In the MVA pathway, despite the absence of multiple gene copies, most upstream genes were highly expressed in flowers and fruits. For instance, both ACAT and HMGS were present as two copies, primarily derived from WGD events. The three copies of HMGR were generated through TRD. ACAT, HMGS1, and phosphomevalonate kinase (PMK) exhibited high expression levels in both flowers and fruits. The single copy of MVK displayed elevated expression levels in flowers. These highly expressed upstream genes could have facilitated terpenoid accumulation in flowers and fruits. Metabolomic analyses further corroborated this finding by detecting high accumulation of key terpenoids in flowers and fruits (Fig. 2B). Furthermore, the correlation structure at a broader scale for the top 30 volatile terpenoids was presented (Fig. S5). Notable correlations include those between citral and trans-carvone oxide, as well as limonene and β-myrcene, suggesting a potential shared biosynthesis pathway.

Phylogenetic analysis, structural and expression pattern of LtTPSs

TPSs are responsible for the biosynthesis of diverse terpenoid compounds. A total of 42 putative LtTPSs were identified in the genome of L. tsaoko (Table 1). Based on their chromosomal location, the LtTPSs were designated LtTPS1-LtTPS42. The physicochemical properties were analyzed using the ProtParam online tool, including the open reading frame (ORF) length, molecular weight (MW), grand average of hydropathicity (GRAVY), and isoelectric point (pI). The ORF length varied from 294 amino acids (LtTPS11) to 1040 amino acids (LtTPS12), with MW ranging from 33,853.59 Da (LtTPS11) to 118,944.45 Da (LtTPS12). GRAVY values spanned from −0.18 (LtTPS5) to −0.52 (LtTPS2), and pI ranged between 4.9 (LtTPS23) and 6.39 (LtTPS9) (Table 1). Subsequently, a phylogenetic tree of 42 LtTPSs was constructed with 283 representative TPSs from seven other species. The 42 LtTPSs were classified into six subfamilies: TPS-a (15), TPS-b (16), TPS-c (2), TPS-e/f (3), TPS-g (2) and TPS-h (4) (Fig. 3A). Additionally, domain and motif analysis revealed that all LtTPSs of the TPS-a subfamily were characterized by a highly conserved DDXXD motif (α-domain) in their C-terminal domain, along with an additional NSE/DTE motif. The TPS-b subfamily in L. tsaoko primarily comprised monoterpene synthases (monoTPS) characterized by a conserved DDXXD motif and a relatively conserved RR(X)₈W motif, essential for monoterpene cyclization (Fig S6, Tables S3 & S4).

Table 1 List of identified TPS genes from the L. tsaoko genome with their physiochemical properties

Fig. 3

Analysis of TPS gene family in L. tsaoko. A Phylogenetic tree and subfamily classification of LtTPSs with TPSs from other species. The gene ID in red represents TPS in L. tsaoko. B Heatmap illustrating the expression of LtTPSs across various tissues (R: root, S: stem, L: leaf, Fr: fruit, Fl: flower, P: podetium) based on transcriptome data. The different shapes on the right side of the expression heatmap indicate different replication event origins of the TPS genes. C Schematic map representation of the genomic localization of LtTPSs

Among 42 LtTPSs, the TPS-b (16) subfamily members were significantly expanded in L. tsaoko compared to other representative plants, while the TPS-a (15) subfamily also exhibited multiple gene copies (Table S5). Gene duplication event analysis revealed that DSD and PD, each involving 9 TPS-a/b homologs, primarily influenced TPS-a/b gene expansion, potentially contributing to the abundance and diversity of volatile terpenoids (Fig. 3B&C). As TPS-a and TPS-b subfamilies predominantly function as monoterpene synthases and sesquiterpene synthases (sesTPS), this suggests that these two subfamilies likely contribute to the substantial production of volatile monoterpenoids and sesquiterpenoids in L. tsaoko. Additionally, the expression patterns of LtTPSs were compared across six different tissues (Fig. 3B). Among the TPS-a or TPS-b members, LtTPS8 and LtTPS9, LtTPS13, LtTPS15, LtTPS29, LtTPS30, LtTPS31, and LtTPS42 demonstrated predominant expression in flowers (Fig. 3C, Table S6), indicating that these genes may play a critical role in the biosynthesis of volatile terpenoids in L. tsaoko flowers.

The 42 LtTPSs identified in this study were distributed across nine chromosomes and several unanchored contigs. Notably, 22 of these genes (52.4%) were located within gene clusters. Specifically, seven gene clusters, labeled as TPS-a and TPS-b, ranged in size from two to five genes each. Cluster 6, which contained LtTPS28, LtTPS29, LtTPS30 and LtTPS31, were highly expressed in flowers, underscoring the significance of whole genome and single-gene duplication in floral terpenoid biosynthesis. Similarly, LtTPS35 and LtTPS36, predominantly expressed in fruits, were tandemly duplicated genes clustered together, further emphasizing the crucial role of single-gene duplication in terpenoid production in both fruits and flowers.

Functional characterization of LtTPSs

To further investigate the members of LtTPSs involved in terpenoid biosynthesis, 18 LtTPSs, including 7 TPS-a and 11 TPS-b subfamily members, were successfully cloned. Engineered strains producing farnesyl pyrophosphate (FPP) or geranyl pyrophosphate (GPP) were utilized to verify the function of candidate LtTPSs. The catalytic products from the 18 LtTPS enzymes using FPP or GPP as substrates were analyzed (Table S7). In comparison to the negative control (Fig S7), GC–MS detection revealed that four LtTPS enzymes (LtTPS23, LtTPS31, LtTPS32, and LtTPS42) exhibited sesquiterpene synthase activity, while eight LtTPS enzymes (LtTPS2, LtTPS15, LtTPS20, LtTPS21, LtTPS22, LtTPS23, LtTPS39, and LtTPS42) were found to convert GPP into monoterpenoids. Notably, LtTPS23 and LtTPS42 demonstrated bifunctional activity, underscoring their versatility in terpene synthesis (Figs. 4 and 5).

Fig. 4

Functional analysis of LtTPS using FPP as the terpenoid precursor. A The sesquiterpenoid products of LtTPS23, LtTPS31, LtTPS32, and LtTPS42. Compounds in bold indicate metabolites identified in L. tsaoko, while those marked with an asterisk (*) correspond to the reference standards. Unannotated compounds are based on matches from the NIST standard reference database, and green-colored genes represent bifunctional genes. B The total ion current chromatogram (TIC) of the products from LtTPS23 and LtTPS31 with FPP as the substrate. C The TIC of the products from LtTPS32 and LtTPS42 with FPP as the substrate

Fig. 5

Functional analysis of LtTPS using GPP as the terpenoid precursor. A The sesquiterpenoid products of 8 LtTPSs. Bolded compounds indicate metabolites found in L. tsaoko. Compounds marked with an asterisk (*) represent those that correspond to the reference standards, while the green colored genes represent bifunctional genes. The unlabeled compounds correspond to the NIST standard reference database results. B The TIC of the product from LtTPS2 and LtTPS15 with GPP as the substrate. C The TIC of the product from LtTPS20, LtTPS21 and LtTPS42 with GPP. D The TIC of the product from LtTPS22 and LtTPS23 with GPP. E The TIC of the product from LtTPS39 with GPP

LtTPS23 converted FPP to cis-α-farnesene (1) and trans-α-farnesene (2) (Fig. 4A&B, Fig. S8). LtTPS31 primarily utilized FPP to produce α-copaene (3), germacrene D (4), ð-cadinene (6), and germacrene D-4-ol (7) (Fig. 4A&B, Fig. S9). LtTPS32 and LtTPS42 were found to mainly yield (-)-β-elemene (9), γ-gurjunene (10), δ-selinene (11), unknown sesquiterpenoids (13), and 7 additional byproducts (Fig. 4A&C, Fig. S10, Fig S11). Moreover, extensive functional analyses revealed that LtTPS32 and LtTPS42 exhibited the highest product diversity, highlighting their potential as key enzymes in terpenoid biosynthesis (Fig. 4). This suggests that LtTPS32 and LtTPS42 may play a crucial role in the diversity of volatile sesquiterpenoids in L. tsaoko.

Among the 11 TPS-b family genes cloned, LtTPS2, LtTPS15, LtTPS20, LtTPS21, LtTPS22, LtTPS23, and LtTPS39 demonstrated the ability to catalyze GPP, while LtTPS42 from TPS-a also exhibited monoterpene synthase activity. Specifically, LtTPS2 catalyzed the conversion of GPP into eucalyptol (18). LtTPS15 and LtTPS42 catalyzed the conversion of GPP into β-myrcene (19), which predominantly accumulated in flowers and fruits (Table S1), along with another unidentified monoterpenoid (20) (Fig. 5A&B, Fig. S12). LtTPS20 and LtTPS21, sharing 98.976% identity, exhibited identical catalytic products, producing linalool (22) (Fig. 5A&C, Fig. S13). Both LtTPS22 and LtTPS23 catalyzed GPP to yield neo-alloocimene (23) and β-cis-ocimene (24), with β-cis-ocimene predominantly accumulating in flowers and fruits (Fig. 5A&D, Fig. S14, Table S1). LtTPS22 was predominantly expressed in flowers, aligning with the observation that β-cis-ocimene accumulated most significantly in flowers (Table S2). Additionally, the main product of LtTPS39, γ-terpinene (26), accumulated at higher levels in flowers (Fig. 5A&E, Fig. S14, Table S1). These catalytic results indicate diverse catalytic activities among the TPS-b family members, contributing to the complexity of terpenoid biosynthesis in L. tsaoko. The TPS-a subfamily member LtTPS42 and the TPS-b subfamily member LtTPS23 exhibited the ability to catalyze both GPP and FPP. When catalyzing GPP, LtTPS42 produced linalool (22) as its main product, similar to the outputs of LtTPS20 and LtTPS21 (Fig. 5A&C, Fig. S13).

LtTPS32 and LtTPS42 exhibited a remarkably high sequence similarity of 99.091%, with only five amino acid differences (Fig S15). Despite this close genetic relationship, LtTPS32 lacked the monoterpene synthase activity observed in LtTPS42, underscoring the latter's unique bifunctionality. A similar pattern was observed with LtTPS22 and LtTPS23. These tandemly duplicated genes displayed a high degree of similarity, with LtTPS22 containing an additional 85 amino acids in the N-terminal region and 28 amino acid substitutions compared to LtTPS23 (Fig S16). Notably, while LtTPS23 catalyzed FPP to produce multiple sesquiterpenoids, LtTPS22 was limited to catalyzing GPP. These enzymatic results demonstrate that even minor alterations in amino acid sequences can result in significant variations in catalytic activity among TPS family members, reflecting the complex evolution of terpenoid biosynthesis pathways.

Discussion

L. tsaoko is a medicinally and economically significant spice widely distributed in south-west China. The distinctive aroma of L. tsaoko is primarily attributed to its abundance of volatile compounds, particularly monoterpenoids and sesquiterpenoids, which enhance its flavor and culinary value. While previous research has focused mainly on the extraction of volatile oils and identification of certain bioactive compounds from dried fruits (Yang et al. 2008; Feng et al. 2010), this study provides a deeper molecular understanding of flavor terpenoids. In the biosynthesis of flavor terpenoid compounds, the expression patterns of the upstream synthases involved in the MEP and MVA pathways and TPS genes play critical roles. Upstream genes, particularly those involved in the synthesis of precursor compounds such as IPP and DMAPP, directly influence the efficiency of terpenoids biosynthesis (Wang et al. 2019a). Previous studies have shown that DXS overexpression in Litsea cubeba (Zhao et al. 2023a) and DXR regulation in Salvia miltiorrhiza can significantly impact monoterpenoids and sesquiterpenoids production (Zhou et al. 2016). This suggests that adjusting DXR and DXS can greatly affect the amount and type of terpenoid metabolites. In our comprehensive annotation of all genes of L. tsaoko involved in the MVA and MEP pathways, we identified specific genes with high expression levels in flowers and fruits, which may suggest tissue-specific precursor accumulation. For example, in the MEP pathway, such as DXS2, DXS3, DXS5, DXS7, DXS10, HDS1, HDS4, HDS5 and DXR1, which showed predominant expression in flowers and fruits, potentially indicating terpenoid precursors accumulation in these tissues. Although we lack direct evidence of precursor accumulation, the high expression of upstream genes, including HDR in flowers, indicates a potential for efficient terpenoid synthesis in these tissues. In the MVA pathway, genes such as ACAT1, HMGS, HMGR, MVK, PKM, and FPPS also exhibited specific expression patterns, implying a similar pattern of precursor accumulation. Furthermore, our metabolite profiling identified monoterpenoids and sesquiterpenoids among the top 30 compounds with the highest concentrations in the analyzed samples, indicating a strong correlation between the elevated expression of MVA and MEP pathway genes and the high levels of terpenoids. The integration of gene expression data with metabolite profiles highlights the crucial role of the MEP pathway in monoterpenoid biosynthesis and emphasizes the differential accumulation of terpenoid precursors across various plant tissues. Additionally, KEGG enrichment analysis of the differential metabolites revealed that these compounds were primarily associated with the monoterpene synthesis pathway.

TPS serves as the crucial enzyme determining the structural diversity and skeleton of terpenoids (Chen et al. 2011). Investigations into the TPS gene families of W. villosa (Yang et al. 2022), Eucalyptus grandis (Myburg et al. 2014), and Cinnamomum kanehira (Chaw et al. 2019) have revealed significant expansions within the TPS-a and TPS-b subfamilies, likely playing a pivotal role in the biosynthesis and diversity of volatile sesquiterpenoids and monoterpenoids. This study similarly observed a substantial expansion in the TPS gene family, particularly within the TPS-b subfamilies. Prior research has demonstrated that tandem duplication acts as a major evolutionary force driving the expansion of the TPS gene family across various plant species (Karunanithi and Zerbe 2019; Chen et al. 2020; Wang et al. 2021). The current analyses of enzymes involved in terpenoid biosynthesis pathways further suggest that this expansion, predominantly facilitated by WGD and single-gene duplication, may have contributed to the observed terpenoid diversity in L. tsaoko. Transcriptome analysis revealed that more than half of the LtTPS genes were predominantly expressed in flowers and fruits, aligning with the high concentration of volatile terpenoids in these tissues and the enzymatic function of the genes in question. Consequently, a detailed investigation into the expression regulation of these upstream genes and TPS genes not only aids in elucidating the biosynthetic pathways of terpenoid compounds but also provides a theoretical foundation for genetic engineering approaches aimed at enhancing the yield of specific terpenoid compounds.

Previous studies have indicated that the catalytic products of TPSs are closely related to their sequence similarity (Karunanithi and Zerbe 2019). For instance, in rice (Oryza sativa) and maize (Zea mays), homologous TPS genes contribute to the biosynthesis of sesquiterpenoids such as trans-β-farnesene, which functions as a signal molecule in plant defense against insect herbivores (Schnee et al. 2006). In tomato, LeTPS28 and LeTPS48, sharing 85% amino acid sequence identity, both utilize E, E-FPP substrate to produce elemol and also catalyze the formation of an unidentified major sesquiterpenoid from Z, Z-FPP. Similarly, LeTPS33 and LeTPS35, with 91.7% sequence identity, catalyze the production of guaia-1(10),11-diene from E, E-FPP (Zhou and Pichersky 2020). In L. tsaoko, we identified a TPS gene cluster, which contained two pairs of duplication genes: LtTPS20 and LtTPS21(PD), as well as LtTPS22 and LtTPS23(TD), exhibiting a high degree of sequence similarity, respectively. Both LtTPS20 and LtTPS21 displayed identical catalytic activity for GPP. Interestingly, LtTPS23 possessed dual functionality and could utilize both GPP and FPP as a substrate, while LtTPS22 was restricted to catalyzing GPP (Figs. 4A, E, and 5). Consequently, predicting the products of a specific TPS based on homology with known TPS sequences is generally unreliable (Alquézar et al. 2017). It has been suggested that the product specificity of mono/sesTPS is influenced more by specific critical regions rather than overall sequence similarity. For example, β-pinene synthases in lemon and Satsuma mandarin (Lücker et al. 2002; Shimada et al. 2004) exhibit 95% and 97% sequence identity, respectively, with the sabinene synthase isolated from rough lemon. However, due to a single amino acid substitution, the latter cannot predominantly produce β-pinene (Kohzaki et al. 2009). Additionally, dispersed duplication resulted in copies of LtTPS32 and LtTPS42, while LtTPS32 and LtTPS42 exhibited the same catalytic activity with FPP. Furthermore, LtTPS42 could also catalyze GPP to produce a variety of monoterpenoids, whereas LtTPS32 was limited to FPP (Figs. 4B, D, and 5). Similarly, CsTPS5 and CsTPS31 in Cannabis sativa are terpene synthases that arose through a gene duplication event. Although both enzymes can utilize FPP and GPP as substrates to produce various sesquiterpenoids and monoterpenoids, they exhibit distinct expression patterns and produce different specific terpenoids. This functional divergence likely enhances the chemical diversity of volatile compounds (Booth et al. 2020). These findings indicate that both WGD and single-gene duplication events play important roles in the functional diversification of TPS genes, thereby contributing to the abundance and diversity of terpenoids in L. tsaoko and other plants. The significant accumulation of monoterpenoids in L. tsaoko fruits likely results from the diverse catalytic activities of monoTPS, where multiple TPSs are capable of producing the same monoterpenoids. For instance, LtTPS15 and LtTPS42 both synthesized β-myrcene. LtTPS20, LtTPS21, and LtTPS42 produced linalool. Compared to sesquiterpene synthases, there was a greater abundance of monoterpene synthases in L. tsaoko, with most capable of producing various types of monoterpenoids. This observation aligned with the higher detection rates of monoterpenoids in L. tsaoko. These results suggest that monoterpene synthases may play a dominant role in the biosynthesis of volatile compounds in L. tsaoko. Moreover, transcriptome analysis revealed predominant expression of LtTPS genes in flowers and fruits of L. tsaoko, correlating with the observed enrichment of volatile terpenoids in these tissues and corresponding enzyme activities. These findings underscore the pivotal role of these genes in shaping the profile and diversity of volatile terpenoids in L. tsaoko flowers and fruits.

The findings of this study provide a foundation for future research utilizing advanced technologies to further explore this topic. Recent technological developments in relevant fields present significant opportunities to elucidate the biosynthesis and regulation of active compounds in medicinal plants. Spatial metabolomics, for instance, can provide detailed mapping of bioactive compound localization within tissues, offering insights into their ecological and functional roles. Integrating metabolomics with transcriptomics, proteomics, and epigenomics facilitates a comprehensive, systems-level understanding of the pathways governing the synthesis and accumulation of key compounds. Emerging single-cell omics technologies further refine these approaches by identifying cell-specific metabolic activities (Zhang et al. 2023), while CRISPR-based functional genomics offers precise tools for gene function validation and pathway engineering. Collectively, these technologies possess substantial potential to illuminate the intricate biosynthetic networks in medicinal plants and to drive the development of innovative biotechnological strategies for enhancing valuable compound production.

Materials and methods

Sample collection and preparation

The root (R), stem (S), leaf (L), flower (Fl), fruit (Fr) and podetium (P) of L. tsaoko were collected from Honghe, Yunnan province. The collected materials were immediately weighed, frozen in liquid nitrogen, and stored at −80 °C until further processing. Subsequently, the samples were ground to a fine powder under cryogenic conditions using liquid nitrogen. Approximately 500 mg of the powder was then transferred to a 20 mL headspace vial (Agilent, Palo Alto, CA, USA) containing a saturated NaCl solution to inhibit enzymatic activity. The vials were sealed using crimp-top caps with TFE-silicone headspace septa (Agilent, Agilent Technologies, Santa Clara, CA, USA). For solid-phase microextraction (SPME) analysis, each sealed vial was incubated in a 60 °C water bath for 5 min. Following equilibration, a 120 µm DVB/CWR/PDMS fiber (Agilent, Agilent Technologies, Santa Clara, CA, USA) was inserted into the headspace of the sample for 15 min at 60 °C.

GC–MS conditions

After sampling, desorption of the VOCs from the fibre coating was carried out in the injection port of the GC apparatus (Model 8890; Agilent, Agilent Technologies, Santa Clara, CA, USA) at 250 °C for 5 min in the splitless mode. The identification and quantification of VOCs were carried out using an Agilent Model 8890 GC and a 7000D mass spectrometer (Agilent, Agilent Technologies, Santa Clara, CA, USA), equipped with a 30 m × 0.25 mm × 0.25 μm DB-5MS (5% phenyl-polymethylsiloxane) capillary column. Helium was used as the carrier gas at a linear velocity of 1.2 mL⋅min⁻¹. The injector temperature was kept at 250 °C and the detector at 280 °C. The oven temperature was programmed from 40 °C (3.5 min), increasing at 10 ℃⋅min⁻¹ to 100 °C, at 7 ℃⋅min⁻¹ to 180 °C, at 25 ℃⋅min⁻¹ to 280 °C, and held for 5 min. Mass spectra were recorded in electron impact (EI) ionization mode at 70 eV. The quadrupole mass detector, ion source and transfer line temperatures were set, respectively, at 150, 230 and 280 °C. Selected ion monitoring (SIM) mode was used for the identification and quantification of analytes in GC–MS analysis. Each experiment was performed in three independent replicates.

Qualitative, quantitative, quality control and statistical analysis of metabolites

The method described by Yuan et al. was used for the acquisition and processing of MS data (Yuan et al. 2022). Metabolite identification utilized the proprietary database of Wekemo company. Qualitative and quantitative analyses of the raw mass spectrometry data were conducted using MassHunter software (Zhu et al. 2013).

PCA, OPLS-DA, and hierarchical clustering analysis (HCA) were conducted using R software (https://www.r-project.org/) with the MetaboAnalyst package (Chong et al. 2018; Chong and Xia 2018). The correlation heatmap of metabolites was generated using R with the ggplot2 packages (Wickham 2011). VIP values (VIP ≥ 1) and fold change (FC) of metabolites (|Log2FC|≥ 1.0) were utilized to identify DAMs. The DAMs were subsequently mapped to the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database (https://www.kegg.jp/kegg/pathway.html) for functional and pathway association analysis.

Transcriptome library construction, sequencing and gene expression

Total RNA was extracted from six tissues (R, S, L, Fr, Fl, and P). The quality and purity of the RNA samples were evaluated using an RNA 6000 Nano LabChip Kit (Agilent Technologies, Santa Clara, CA, USA), with an RNA integrity number (RIN) of > 7.0. Poly-(A)-containing mRNA was isolated using a NEBNext® Poly(A) mRNA Magnetic Isolation Module (New England Biolabs, Ipswich, MA, USA). The mRNA was fragmented and used as templates to synthesize the first-strand cDNA using reverse transcriptase. Second-strand cDNA synthesis was performed using buffer and Second Strand Synthesis Enzyme Mix. The resulting double-stranded cDNA fragments underwent end-repair and adapter ligation. A USER enzyme excised uracil bases in the adapter. Adapter-modified fragments were selected through gel purification and amplified via PCR to generate the final cDNA library. The cDNA library preparation utilized the NEBNext® Ultra™ RNA Library Prep Kit for Illumina® (New England Biolabs, Ipswich, MA, USA). Following this methodology, six libraries representing different tissues were constructed, with each experiment conducted in triplicate. These libraries were sequenced on the NovaSeq 6000 platform (Illumina, San Diego, CA, USA). Raw reads underwent quality control using Trimmomatic (Bolger et al. 2014), and clean reads were aligned to the reference genome (https://doiorg.publicaciones.saludcastillayleon.es/https://doiorg.publicaciones.saludcastillayleon.es/10.6084/m9.figshare.27600624.v1) using HISAT2 (Kim et al. 2015). The L. tsaoko genome assembly and annotation were previously completed by our research group. Gene expression levels were quantified as fragments per kilobase of transcript per million fragments mapped (FPKM) using StringTie 2 (Kovaka et al. 2019). Data normalization was performed using TBtools (Chen et al. 2018), employing the z-score method.

Identification of genes related to terpenoid biosynthesis

To investigate the genes involved in terpenoid skeleton synthesis pathways, we initially retrieved protein sequences from the Arabidopsis thaliana genome, including 1-deoxy-D-xylulo se-5-phosphate synthase (DXS), 1-deoxy-D-xylulose-5-phosphate reductoisomerase (DXR), 2-Cmethyl-D-erythritol 4-phosphate cytidylyltransferase (MCT), 4-diphosphocytidyl-2-C-methyl-D-erythritol kinase (CMK), 2-C-methyl-D-erythritol-2,4-cyclodiphosphate synthase (MCS), (E)−4-hydroxy-3-methylbut-2-enyl-diphosphate synthase (HDS), 4-hydroxy-3-methylbut-2-enyl-diphosphate reductase (HDR), acyl-coenzyme A-cholesterol acyltransferase (ACAT), hydroxymethylglutaryl-CoA synthase (HMGS), hydroxymethylglutaryl-CoA reductase (HMGR), mevalonate kinase (MVK), phosphomevalonate kinase (PMK), mevalonate diphosphate decarboxylase (MVD), isopentenyl-diphosphate isomerase (IDI), geranyl diphosphate synthase (GPPS), and farnesyl diphosphate synthase (FPPS) from the NCBI database. We created a BLAST database for the selected sequences using the makeblastdb tools of BLAST + . These sequences served as queries and were blasted against the protein sequences of L. tsaoko with an E-value cutoff of 1e⁻⁵. Candidates with over 40% identity and alignment protein length exceeding 200 amino acids were selected for further analysis.

To identify candidate TPS genes, we employed Hidden Markov Model (HMM) profiles of Terpene_synth (PF01397) and Terpene_synth_C (PF03936) obtained from the Pfam database (http://pfam.xfam.org). These profiles were subsequently used to search against L. tsaoko protein sequences using HMMER (Johnson et al., 2010), with an E-value threshold of 1e⁻⁵. To categorize the TPS genes into distinct subfamilies, we acquired TPS protein sequences from various plants, including Vitis vinifera, Oryza sativa, A. thaliana, Populus trichocarpa, Sorghum bicolor, Selaginella tamariscina and Physcomitrium patens (Chen et al. 2011). Subsequently, the TPS sequences from different species and L.tsaoko were aligned using MAFFT, and a maximum likelihood tree was constructed using IQ-TREE (Nguyen et al. 2015) with 1000 bootstrap replicates. The visual enhancement of the phylogenetic tree was performed using ITOL (https://itol.embl.de/). Different modes of gene duplication in L. tsaoko were classified into five categories using DupGen_Finder (Qiao et al. 2019) (v1.12) with default parameters.

Gene structure analysis, motif identification and chromosomal locations of LtTPS

The length and chromosomal location of LtTPSs were derived from the L. tsaoko genome annotation file. Chromosomal mapping of genes was conducted using TBtools (Chong and Xia 2018). The Gene Structure Display Server 2.0 (http://gsds.gao-lab.org/) was utilized to determine the intron and exon structures of LtTPSs using genomic and cDNA sequences (Hu et al. 2015). Protein motif analysis of LtTPSs was performed using the online MEME Suite (http://meme-suite.org/) with default parameters and a set motif number of twenty. CDD (https://www.ncbi.nlm.nih.gov/Structure/bwrpsb/bwrpsb.cgi) was employed to identify conserved domains within the predicted TPS protein sequences.

Gene clone and functional identification of LtTPSs

Several LtTPS genes identified in this study were manually refined due to the presence of long introns and multiple exons using Softberry (https://www.softberry.com). mRNA was extracted from L. tsaoko composite samples across six tissues RNAprep Pure Plant Plus Kit (TIANGEN Biotech (Beijing, China)), and cDNA was isolated using the Reverse Transcription System (Promega, Madison, WI, USA) according to the manufacturer's protocol. Specific primers (Table S8) were designed by TaKaRa (https://www.takarabio.com/learning-centers/cloning/primer-design-and-other-tools). LtTPSs were subsequently cloned into pET28a/ pRSFDuet-MispA vectors (Lei et al. 2021). Recombinant pET28a-LtsaTPSs plasmids and pMevT-MBIS (Glasscock et al. 2021) were transformed into Escherichia coli strain BL21 (DE3) for sesquiterpene protein expression. pRSFDuet-MispA-LtTPSs were transformed into E. coli strain EIP81 (DE3) (Lei et al. 2021) for monoterpene protein expression. The transformants were inoculated into 5 mL of TB medium and cultivated overnight at 37 °C and 200 rpm. A 2 mL aliquot of each preculture was transferred to 50 mL of fresh TB medium in a flask and incubated with shaking (200 rpm) at 37 °C. Protein expression was induced by 1 mM of Isopropyl β-D-thiogalactopyranoside (IPTG) and 1 mM MgCl₂ when the optical density at 600 nm (OD₆₀₀) reached 0.6 − 0.8, followed by incubation at 16 °C for an additional 76 h. A 2 mL sample of induced bacterial culture was collected for GC analysis.

GC analysis was conducted using an Agilent System 8890 GC/Q-tof equipped with a flame ionization detector (FID) and an HP-5MS UI column (15 m × 0.25 mm, 0.25 μm film thickness). The headspace extraction vial was equilibrated at 50 °C for 10 min (rotational speed 300 rpm). The Carbon WR/PDMS SPME Arrow underwent conditioning at 250 °C for 10 min, adsorption for 40 min, and desorption for 10 min. The inlet temperature was set to 250 °C. The column temperature program was initiated at 40 °C for 5 min, then increased at 2 ℃⋅min⁻¹ to 120 °C, and finally raised at 10 ℃⋅min⁻¹ to 290 °C, where it was maintained for 29 min. Mass spectrometry scanning was performed within the mass range of 35 m/z to 500 m/z, with a scan rate of 781 u/s.