Metabolic Profiling of Terpene Diversity and the Response of Prenylsynthase-Terpene Synthase Genes during Biotic and Abiotic Stresses in Dendrobium catenatum

Dendrobium catenatum is a widely cultivated Chinese orchid herb rich in abundant secondary metabolites, such as terpenes. However, terpene distribution and characterization of terpene biosynthesis-related genes remain unknown in D. catenatum. In this study, metabolic profiling was performed to analyze terpene distribution in the root, stem, leaf, and flower of D. catenatum. A total of 74 terpene compounds were identified and classified. Clustering analysis revealed that terpene compounds exhibited a tissue-specific accumulation, including monoterpenes in the flowers, sesquiterpenes in the stems, and triterpenes in the roots. Transcriptome analysis revealed that the ‘terpenoid backbone biosynthesis’ pathway was only significantly enriched in root vs. flower. The expression of terpene biosynthesis-related genes was spatiotemporal in the flowers. Prenylsynthase-terpene synthases (PS-TPSs) are the largest and core enzymes for generating terpene diversity. By systematic sequence analysis of six species, 318 PS-TPSs were classified into 10 groups and 51 DcaPS-TPSs were found in eight of them. Eighteen DcaPS-TPSs were regulated by circadian rhythm under drought stress. Most of the DcaPS-TPSs were influenced by cold stress and fungi infection. The cis-element of the majority of the DcaPS-TPS promoters was related to abiotic stress and plant development. Methyl jasmonate levels were significantly associated with DcaTPSs expression and terpene biosynthesis. These results provide insight into further functional investigation of DcaPS-TPSs and the regulation of terpene biosynthesis in Dendrobium.


Introduction
Terpenes are the largest and most diverse class of chemical substances that have been found in the plant kingdom, from moss to angiosperm [1]. Plants accumulate terpenes in multiple tissues that serve as defenses against diverse environmental stresses, pathogens, and herbivores [1,2]. Two spatially separated pathways exist for terpene biosynthesis in plants: the mevalonic acid (MVA), in cytosol and peroxisomes, and 2-C-methyl-D-erythritol 4-phosphate (MEP) pathways in plastids [3]. The MVA and MEP pathways are responsible for the formation of the isomeric five-carbon (isoprene) building blocks: isopentenyl pyrophosphate (IPP) and dimethylallyl pyrophosphate (DMAPP) [3,4]. These building blocks are further catalyzed to produce terpene biosynthesis precursors by prenylsynthases (PSs) and terpene synthases (TPSs), which play important roles in the structural diversity of plant terpenes [5].
Plant PSs are a class of enzymes responsible for the synthesis of isoprenoids, which contain three enzymes: geranylgeranyl pyrophosphate synthase (GGPS), farnesyl pyrophosphate synthase (FPS), and geranyl pyrophosphate synthase (GPS) [6]. The PSs are also

Composition and Classification of Terpene in Various Tissues of D. catenatum
The total terpene content was extracted and detected from fresh root, stem, leaf, and blooming flower of D. catenatum (Figure 1a). The terpene level of the root was significantly higher than that of other tissues (Figure 1b). To comprehensively analyze the distributions of terpenes in various tissues, gas chromatography-mass spectrometry (GC-MS) and liquid chromatography-tandem mass spectrometry (LC-MS) were performed to analyze the volatile terpenes and non-volatile terpenes, respectively. Principal component analysis (PCA) was used to analyze the first two characteristic components of the four tissue metabolites under non-supervision. Forty-one volatile terpenes and 33 non-volatile terpenes were separately formed into a cluster (Figure 1c). Of the 74 terpene metabolites identified, 2 were apocarotenoids, 4 were alkaloids, 27 were monoterpenes, 18 were sesquiterpenes, 7 were diterpenes, and 16 were triterpenes (Table S1). Clustering analyses revealed that most terpenes of D. catenatum exhibited a tissue-specific accumulation, with monoterpenes mainly in the flower and triterpenes in the root (Figure 1d). We further selected the differentially accumulated metabolites (DAMs) by a fold change of ≥2 or ≤0.5, and a variable importance in project (VIP) of ≥1 from root vs. flower and stem vs. leaf. Twenty-six DAMs were upregulated and twenty DAMs were downregulated in root vs. flower ( Figure 1e and Table S2). Eleven DAMs were upregulated and twenty DAMs were downregulated in stem vs. flower (Figure 1f and Table S3). Furthermore, the 10 ten DAMs are shown in Figure 1g,h, and most of the terpenes were observed in four tissues, but several terpenes

Enrichment Analysis of Terpene Biosynthesis-Related Genes in Various Tissues of D. catenatum
To explore the expression of terpene biosynthesis-related genes in D. catenatum tissues, gene set enrichment analysis (GSEA) was performed on transcriptome data from four tissues. The KEGG pathway 'terpenoid backbone biosynthesis' was only significantly (p < 0.05) enriched in root vs. flower (Figure 2a). Terpene biosynthesis precursors, IPP and DMAPP, were produced by the plastidial MEP pathway and the cytosolic MVA pathway.

Enrichment Analysis of Terpene Biosynthesis-Related Genes in Various Tissues of D. catenatum
To explore the expression of terpene biosynthesis-related genes in D. catenatum tissues, gene set enrichment analysis (GSEA) was performed on transcriptome data from four tissues. The KEGG pathway 'terpenoid backbone biosynthesis' was only significantly (p < 0.05) enriched in root vs. flower (Figure 2a). Terpene biosynthesis precursors, IPP and DMAPP, were produced by the plastidial MEP pathway and the cytosolic MVA pathway. IPP and DMAPP were then modified by PS-TPSs to generate terpene diversity (Figure 2b). A total of 33 enriched genes were associated with terpene biosynthesis (Table S4), and the encoded enzymes are labeled in Figure 2b. Among them, 28 genes showed the highest expression levels in the flowers (Figure 2c). However, one TPS gene (Dca010855) was highly expressed in the stem. Two genes, TPS (Dca026890) and SPS (Dca019501, solanesyl pyrophosphate synthase), were predominantly expressed in the leaf. We further explored the expression of terpene biosynthesis-related genes in early flower buds (F1), medium flower buds (F2), and bloomed flowers (F3). The GSEA showed that 'terpenoid backbone biosynthesis' was only significantly (p < 0.05) enriched in F3 vs. F1 ( Figure S1). A total of 18 enriched genes were found in F3 vs. F1 and root vs. flower comparisons, most of which were highly expressed in bloomed flowers (Figure 2d). The results suggest that the expression of terpene biosynthesis-related genes is spatiotemporal in D. catenatum flowers. IPP and DMAPP were then modified by PS-TPSs to generate terpene diversity (Figure 2b). A total of 33 enriched genes were associated with terpene biosynthesis (Table S4), and the encoded enzymes are labeled in Figure 2b. Among them, 28 genes showed the highest expression levels in the flowers (Figure 2c). However, one TPS gene (Dca010855) was highly expressed in the stem. Two genes, TPS (Dca026890) and SPS (Dca019501, solanesyl pyrophosphate synthase), were predominantly expressed in the leaf. We further explored the expression of terpene biosynthesis-related genes in early flower buds (F1), medium flower buds (F2), and bloomed flowers (F3). The GSEA showed that 'terpenoid backbone biosynthesis' was only significantly (p < 0.05) enriched in F3 vs. F1 ( Figure S1). A total of 18 enriched genes were found in F3 vs. F1 and root vs. flower comparisons, most of which were highly expressed in bloomed flowers (Figure 2d). The results suggest that the expression of terpene biosynthesis-related genes is spatiotemporal in D. catenatum flowers.

Identification of PS-TPSs
The precursors of terpenes were modified by PS-TPS, and they played important roles in terpene diversity. Most PS-TPS genes were highly expressed in bloomed flowers (Figure 2d), but the roles of PT-TPS genes remain largely unknown in D. catenatum. A thorough search of the D. catenatum genome sequence led to the identification of 9 PSs and 42 TPSs (Table S4). Based on the phylogenetic analysis of PS-TPSs from D. catenatum and other plants, 71 PSs were divided into three groups (i.e., GGPS, FPS, and GPS), and 247 TPSs were divided into seven groups (i.e., TPS-a, TPS-b, TPS-c, TPS-e/f, TPS-g, TPS-h, and MTPS-like) according to previous studies ( Figure 3) [4,6,13,14].

Identification of PS-TPSs
The precursors of terpenes were modified by PS-TPS, and they played important roles in terpene diversity. Most PS-TPS genes were highly expressed in bloomed flowers ( Figure 2d), but the roles of PT-TPS genes remain largely unknown in D. catenatum. A thorough search of the D. catenatum genome sequence led to the identification of 9 PSs and 42 TPSs (Table S4). Based on the phylogenetic analysis of PS-TPSs from D. catenatum and other plants, 71 PSs were divided into three groups (i.e., GGPS, FPS, and GPS), and 247 TPSs were divided into seven groups (i.e., TPS-a, TPS-b, TPS-c, TPS-e/f, TPS-g, TPS-h, and MTPS-like) according to previous studies ( Figure 3) [4,6,13,14]. The DcaPS family contained five DcaGGPSs, two DcaGPSs, one DcaFPS, and one DcaSPS (Table S4). The aspartate-rich DD(x2-4)D ('x' is any amino acid) motif and CxxxC motif were found in all DcaPSs, except DcaFPS ( Figure 4a). The large subunit of GGPSs (LSU) and GPSs contained the DD(x2-4)D motif, while the GGPS small subunit (SSU) lacked this motif and contained the CxxxC motif [14]. However, DcaGGPS2~4 contained both the DD(x2-4)D and CxxxC motifs ( Figure 4a). Thus, the classification of LUS and SSU The DcaPS family contained five DcaGGPSs, two DcaGPSs, one DcaFPS, and one DcaSPS (Table S4). The aspartate-rich DD(x 2-4 )D ('x' is any amino acid) motif and CxxxC motif were found in all DcaPSs, except DcaFPS ( Figure 4a). The large subunit of GGPSs (LSU) and GPSs contained the DD(x 2-4 )D motif, while the GGPS small subunit (SSU) lacked this motif and contained the CxxxC motif [14]. However, DcaGGPS2~4 contained both the DD(x 2-4 )D and CxxxC motifs ( Figure 4a). Thus, the classification of LUS and SSU according to their motif constitution was not strict in DcaPSs. Another possible reason might be that D. catenatum genome assembly existed in the gaps and resulted in incomplete sequences.
Furthermore, we searched for 30 conserved motifs in DcaTPSs using MEME software. Most subfamily members showed similar conserved motif distributions ( Figure S2). However, several DcaTPSs differed in their homologs, such as DcaTPS09, which might have an incomplete assembly in the genome. Class I activity was found in almost all plant TPSs including the 'DDxxD' and 'NSE/DTE' motifs [4]. All TPS-a and TPS-b members contained 'DDxxD' and 'NSE/DTE', except for DcaTPS17. The N-terminal transit peptides of TPS were presumably cleaved off by a 'RRx8W' motif [3], which was distributed in the TPS-a and TPS-b subfamilies. The class II activity resided in a separate domain and contained 'DxDD' and 'EDxxD' motifs [15]. TPSs containing 'EDxxD' motifs were members of TPS-d, -c, and e/f [4]. However, 'EDxxD' motifs were widely distributed throughout the four DcaTPS subfamilies including TPS-a, TPS-b, TPS-c, and TPS-e/f. Moreover, 'DDxxD' and 'DxDD' motifs were distributed in all TPS-e/f proteins, and 'DxDD' motifs were present in TPS-c with the exception of DoTPS34 (Figures 4b and S2). The genome of Selaginella moellendorffi was found to contain two distinct types of TPS genes, typical plant terpene synthases and microbial TPS-like (MTPS-like) synthases [13]. The phylogenetic analysis showed that DcaTPS41 was classified into the MTPS-like II subgroup ( Figure 3). The above-mentioned motifs were not found in DcaTPS41 ( Figure S2). However, DcaTPS41 contained a conserved 'terpene_synth_C' domain ( Figure S3). MTPS from S. moellendorffi containing only the 'terpene_synth_C' domain represented a type of TPS that has been widely identified in bacteria, fungi, and nonseed plants [13]. Thus, DcaTPS41 might be a new kind of MTPS in D. catenatum.  Furthermore, we searched for 30 conserved motifs in DcaTPSs using MEME software. Most subfamily members showed similar conserved motif distributions ( Figure S2). However, several DcaTPSs differed in their homologs, such as DcaTPS09, which might have an incomplete assembly in the genome. Class I activity was found in almost all plant

Spatiotemporal Expression Patterns of DcaPS-TPSs in D. catenatum
To investigate the potential roles of DcaPS-TPSs during growth and development, RNA-seq data from different plant tissues and organs were detected [16]. The DcaPS-TPSs showed distinct organ-specific expression patterns, and the half of DcaPS-TPSs were primarily expressed in flowers (Figure 5a). In detail, several genes were highly expressed in specific flower tissues, including DcaTPS25 and -03 in lip (labellum), DcaGGPS4 and DcaTPS23 in flower bud, and DcaTPS17 and -18 in sepal. Furthermore, we explored the expression levels of DcaPS-TPSs during different flowering phases. DcaSPS, DcaGPS2, DcaGGPS1, and most DcaTPSs were highly expressed in F3. The rest of the DcaPS-TPSs were primarily expressed in F1/F2 (Figure 5b). The results were consistent with the tissue expression profiles of DcaPS-TPSs and indicated that the organ-specific expression of DcaPS-TPSs might be important in the terpene biosynthesis of D. catenatum flowers.
The above-mentioned motifs were not found in DcaTPS41 ( Figure S2). However, DcaTPS41 contained a conserved 'terpene_synth_C' domain ( Figure S3). MTPS from S. moellendorffi containing only the 'terpene_synth_C' domain represented a type of TPS that has been widely identified in bacteria, fungi, and nonseed plants [13]. Thus, DcaTPS41 might be a new kind of MTPS in D. catenatum.

Spatiotemporal Expression Patterns of DcaPS-TPSs in D. catenatum
To investigate the potential roles of DcaPS-TPSs during growth and development, RNA-seq data from different plant tissues and organs were detected [16]. The DcaPS-TPSs showed distinct organ-specific expression patterns, and the half of DcaPS-TPSs were primarily expressed in flowers (Figure 5a). In detail, several genes were highly expressed in specific flower tissues, including DcaTPS25 and -03 in lip (labellum), DcaGGPS4 and DcaTPS23 in flower bud, and DcaTPS17 and -18 in sepal. Furthermore, we explored the expression levels of DcaPS-TPSs during different flowering phases. DcaSPS, DcaGPS2, DcaGGPS1, and most DcaTPSs were highly expressed in F3. The rest of the DcaPS-TPSs were primarily expressed in F1/F2 (Figure 5b). The results were consistent with the tissue expression profiles of DcaPS-TPSs and indicated that the organ-specific expression of DcaPS-TPSs might be important in the terpene biosynthesis of D. catenatum flowers.  catenatum is an epiphytic orchid plant that often experiences abiotic and biotic stresses such as drought and low temperatures. We explored the possible roles of DcaTPSs in response to drought stress by analyzing the raw RNA-seq reads from the leaves [17]. For drought stress, the plants were irrigated on the first day, dried from the second to the seventh day, and then recovered on the eighth day. Leaves were sampled at dawn (06:30) and dusk (18:30) on the second (i.e., DR5 and DR8), seventh (i.e., DR6 and DR10), and ninth day (i.e., DR7 and DR15), and on the eighth day, DR11 was sampled at 18:30 (Figure 6a). The results showed that rewatering restored the expression levels of DcaTPS05 and -19. For most of the DcaPS-TPSs, their expression was not always the same under drought stress, such as for DcaGPS1; DcaGGPS2; DcaTPS11, -39, and -40, which were induced on DR5 and repressed on DR8. This suggests that these DcaPS-TPSs might be regulated by the biological clock under drought stress.
ninth day (i.e., DR7 and DR15), and on the eighth day, DR11 was sampled at 18:30 (Fig  6a). The results showed that rewatering restored the expression levels of DcaTPS05 an 19. For most of the DcaPS-TPSs, their expression was not always the same under drou stress, such as for DcaGPS1; DcaGGPS2; DcaTPS11, -39, and -40, which were induced DR5 and repressed on DR8. This suggests that these DcaPS-TPSs might be regulated the biological clock under drought stress.  To understand the role of DcaPS-TPSs in disease resistance, D. catenatum leaves were inoculated with Colletotrichum gloeosporioides. The expression profiles of DcaPS-TPSs showed that 12 DcaPS-TPSs were upregulated and 6 DcaPS-TPSs were downregulated in infected leaves (Figure 6c). A Venn diagram shows that 13 DcaPS-TPSs were identified in three transcriptome data, and most of them were detected in four tissues ( Figure S4). These results suggest that DcaPS-TPSs might be involved in enhancing defense ability under abiotic and biotic stresses.

Cis-Elements in the Promoter Regions of DcaPS-TPSs
To predict the putative functions of DcaPS-TPSs in response to biotic and abiotic stresses, we analyzed 2 kb upstream of the DcaPS-TPSs (Figure 7a). Potential cis-elements of the DcaPS-TPS promoter were determined using PlantCARE. Twenty-four regulatory sequences were found in the promoters of DcaPS-TPSs including abiotic stress-responsive cis-elements (e.g., MBS in the drought response and LTR in cold stress), plant growth and development-related cis-elements (e.g., circadian, A-box, and O2-site), and phytohormonerelated cis-elements (e.g., the GARE-motif of the gibberellin response, ABRE of the abscisic acid response, and the TCA-element of the salicylic acid response). Most of the DcaPS-TPS promoters contained light-responsive elements, suggesting that DcaPS-TPSs might play important roles in plant photoperiod regulation.  [12]. The endogenous JA and total terpene contents were determined Pr and control plants. Significant increases in the JA and terpene content were observ in Pr (Figure 7b,c). Furthermore, the expression levels of 35 DcaPS-TPSs were detected As plant-specific signaling molecules, jasmonates (JAs) not only steer the response of plant stress but also induce the accumulation of secondary metabolite production [18,19]. Most of the DcaPS-TPS promoters contained CGTCA-motifs of JA response (Figure 7). Our previous study reported a purple (Pr) variety of D. catenatum that enriched more flavonoids in stems [12]. The endogenous JA and total terpene contents were determined in Pr and control plants. Significant increases in the JA and terpene content were observed in Pr (Figure 7b,c). Furthermore, the expression levels of 35 DcaPS-TPSs were detected in Pr and CK by transcriptome analysis. Six DcaPS-TPSs were upregulated and ten DcaPS-TPSs were downregulated (fold change > 2) in Pr vs. CK ( Figure S5). We selected six DcaPS-TPSs to detect the exogenous MeJA effect in D. catenatum stems. DcaTPS05, -21, -28, and -39 were induced under MeJA treatment and DcaTPS30 and -35 were suppressed under MeJA treatment (Figure 7d). These results suggest that JA is involved in the regulation of DcaTPS expression and terpene biosynthesis.

Discussion
As it is a valuable traditional herb in China, the extraction of D. catenatum displays strong immune modulatory activity, and is used for relieving stomach upsets and its antitumor and antipyresis properties [10,20]. To date, research on the secondary metabolites in D. catenatum primarily focuses on phenylpropanoid biosynthesis such as flavonol and anthocyanin [12,21]. MeJA treatment and mycorrhizal fungi infection could induce dendrobine biosynthesis in Dendrobium [11,22]. Dendrobine, a sesquiterpene alkaloid, is regarded as the standard of quality for Dendrobium stem [10]. However, no reports have been published on the distributions of terpenes in D. catenatum. In this study, 41 volatile terpenes and 33 non-volatile terpenes were detected (Figure 1d). Most of them were first reported in D. catenatum. For instance, isolongifolene, which is more highly accumulated in the root (Figure 1g), has been previously demonstrated to have a neuroprotective effect against Parkinson's disease [23]. Bisabolene, epizonarene, and muurolene primarily accumulate in the stem (Figure 1h), but their function is still unknown. Interestingly, we found that most of the terpene biosynthesis-related genes were positively correlated with tissue-specific accumulation of terpenes, such as monoterpenes, primarily in the flowers ( Figure S7). Dendrobium flower has been used as a health care tea for its light scent and medicinal value, but its aroma composition is still unclear. Myrcene, ocimene, α-pinene, and α-terpineol are considered to be the major floral fragrance compositions [24]. These monoterpenes were also detected in D. catenatum (Table S1). Moreover, we found two terpene derivatives, 6-hydroxydendroxine and α-amyrenone, which were only detected in D. catenatum flowers (Figure 1h). 6-Hydroxydendroxine and α-amyrenone have an important medicinal value. 6-Hydroxydendroxine is a derivative of dendroxine, and it is especially common in orchidaceae [25]. Recently, α-amyrenone has been found to have anti-hyperglycemic, lipid-lowering, and anti-obesity effects [26]. Thus, the medicinal value of D. catenatum is yet to be discovered.
Previous research has focused on the putative upstream elements of the alkaloid biosynthetic pathway, which overlap with the terpene biosynthesis pathway [11]. Terpenes originated via de novo synthesis from IPP and DMAPP, and they were further catalyzed to produce short-chain products by PSs early in the isoprenoid pathway [27]. PSs belong to the chain elongation family (PF00348) within the terpene synthase superfamily, whose members also include squalene/phytoene synthase and TPS [28]. All proteins in the terpene synthase superfamily contain a characteristic all α-helix motif [28]. However, none of the previous studies grouped PS and TPS together in Orchidacea [29]. In this study, DcaPT-TPSs contained five conserved regions within a six-helix bundle that contained aspartate-rich DD(x 2-4 )D pyrophosphate-binding motifs (Figure 4). However, the conserved motifs of the DcaFPS protein were not found with other DcaPTs by MEME motif analysis. We further aligned the sequence of DcaFPS with AtFPS1 and -2, which all contained aspartate-rich DDxxD motifs ( Figure S6). Phylogenetic analysis showed that DcaFPS is grouped with FPSs of Arabidopsis and rice (Figure 3). This suggests that FPSs are evolutionarily conserved in dicots and monocots. Moreover, previous studies reported 34 TPSs in D. officinale and 35 TPSs in D. catenatum [29,30]. In this study, 42 TPS genes were found in the D. catenatum genome v.2.2 (Table S4). Genome v.2.2 (NCBI_Assembly:GCA_001605985.2) sequences are not complete and many gaps exist [31]. This is probably why several TPS motifs were not detected in the MEME analysis ( Figure S2). These TPSs were divided into four subfamilies ( Figure S2), which is in accordance with previous studies [29], except for DcaTPS41. TPSs have additional β and/or γ domains in plants, not only an α motif [4,13,28]. However, these motifs were not found in DcaTPS41 ( Figure S2). Phylogenetic analysis showed that DcaTPS41 and its P. equestris homolog were more closely related to SmMTPS (Figure 3). Plant MTPSs were first found in S. moellendorfi and are regarded as a special group in nonseed plants [13], but the sequence alignment showed that Phytolacca bogotensis and Opuntia sp. contained one and four MTPSs, respectively [32]. Orchidaceae seed is well known for its symbiotic germination with mycorrhizal fungi [33]. Dendrobium seeds will not germinate without the fungus to feed them [34]. However, no studies have focused on the symbiotic relationship between Dendrobium and fungus in vivo. We believe that high-quality genome assembly and omics analysis will facilitate MTPS study in Dendrobium.
D. catenatum is an epiphytic plant that grows in a warm and humid environment through its special aerial roots. Aerial root development is associated with a high degree of IAA accumulation and is promoted by the light signaling pathway [35]. Some AtTPSs exhibited root zone-specific expression patterns in Arabidopsis, such as the duplicated genes AtTPS23 and AtTPS27, which produced the monoterpene 1,8-cineole and were expressed in the stele and epidermis of the root [4]. AtTPS08 encoding a putative diterpene synthase was expressed in the stele, and AtTPS22 encoding a putative sesquiterpene synthase was expressed in the lateral root cap [36]. Interestingly, we found that three DcaPS-TPSs were expressed specifically in green root tips, and two were especially expressed in white root (Figure 5a). This could be associated with root growth and development. Moreover, monoterpenes are major volatile constituents of many plants, and they are catalyzed by TPS [24]. The GSEA indicated that terpene biosynthesis-related genes were mainly expressed in floral organs (Figure 2a). Recently, one TPS gene from D. officinale showed the highest transcript level in floral organs and could only covert GPP to linalool in vitro [30]. Linalool is one of the predominant components of floral scents [24], but GC-MS did not detect it in the flower (Table S1). This suggests that terpene biosynthesis regulation is more complex in vivo.
Terpenes play significant ecological roles in the interactions between plants and environmental stresses. As the core genes in terpene biosynthesis, the expression levels of PS-TPS genes in different plants under various stress conditions have been widely reported. Five strawberry TPSs were significantly upregulated in response to anthracnose at 3-6 h post-infection [37]. TPS21 encoding sesquiterpene synthase was negatively correlated with dendrobine accumulation after infection with mycorrhizal fungus in D. nobile [22]. Both the AtTPS03 and AtTPS04 genes were induced locally at wound sites to promote biosynthesis of volatile terpenes [9,38]. In addition, terpene metabolism is related to phytohormone biosynthesis, such that gibberellins (GAs) are tetracyclic diterpenoids that are biosynthesized from GGPP. Except for GA 20 , GA 7 , and GA 8 , which increased in Pr, the others were all reduced in Pr ( Figure S8). Moreover, the endogenous JA and total terpene levels increased in Pr (Figure 7b,c). Not all DcaTPS genes were induced by JA treatment (Figure 7d). These results suggest that terpene biosynthesis was induced by JA crosstalk with GA. The regulation of DcaTPSs expression was influenced by multiple factors, and the functionality of DcaTPSs needs to be studied further.

Analysis of Phylogenetic Relationship, Motif Architecture, and Cis-Elments of Promoters
The sequences of PS-TPS proteins from these species were aligned with ClustalW software and then an un-rooted tree was constructed using FastTree with options by default [39,40]. The motif analysis was performed in MEME (https://meme-suite.org/ meme/, accessed on 2 April 2021). The cis-elements of DcaPS-TPSs promoters were analysed in PlantCARE (http://bioinformatics.psb.ugent.be/webtools/plantcare/html/, accessed on 2 April 2021).

In Silico Expression Profiling of DcaPS-TPSs
For expression profiling of DcaPS-TPSs, the raw RNA-seq data of tissues (PRJNA348403 and PRJNA680456) were downloaded from the NCBI [16,41]. For drought stress, 8-monthold seedlings were transferred into an incubator (12/12 h light/dark, light intensity 100 µmol m −2 s −1 ; 28/22 • C day/night; and relative humidity 60/70% day/night) and adapted to the controlled conditions for 10 days. Plants were irrigated on the 1st day, dried from the 2nd to the 7th day, and then resumed on the 8th day. The raw RNA-seq data of drought stress were obtained from Sci Data [17]. The RNA-seq and metabolome data of Pr were described in our previous studies [12]. For fungus treatments, two-year-old cultivated plants were infected with C. gloeosporioides for 25 d in greenhouse (natural condition in April 2021, Zhejiang University, Hangzhou, China). The raw RNA-seq data of C. gloeosporioides infected plants were submitted to the BIG Data Center of the Chinese Academy of Sciences (https://bigd.big.ac.cn (accessed on 6 May 2022)) with accession number CRA002691. For cold stress, two-year-old cultivated plants were placed at 0 • C and then subjected to a gradual drop from 0 to −6 • C within 3 h. The temperature was held at −6 • C for 3 h and then held at 8 • C for 24 h without lighting. The raw RNA-seq data of cold stress have been submitted to the BIG Data Center with accession number CRA003229 and CRA005177 [42]. The quality control of raw data and mapping of reference genomes were performed according to previous work [12]. In brief, the raw data were first qualified using trimmomatic (v.0.39, accessed on 6 July 2021) with the parameter 'ILLUMINACLIP: TruSeq3-PE-2 to obtain clean data. Clean data were aligned to the reference genome using hisat2 (v.2.2.1, accessed on 7 July 2021) and then were counted using featureCount (v.2.0.3, accessed on 7 July 2021) to obtain the expression matrices with the default parameters. Subsequently, the expression levels of unigene were calculated as FPKM (fragments per kilo bases of exons for per million mapped reads). DcaPS-TPSs expression matrices were treated by hierarchical clustering using the R package ggplot2 (v.3.3.5, accessed on 7 July 2021).

MeJA Treatment and Real-Time Quantitative PCR
Two-year-old cultivated D. catenatum were grown at 20 ± 2 • C with a light/dark cycle of 12/12 h and a 65-75% relative humidity as a control treatment. For MeJA treatment, the leaves were sprayed with 1 mM MeJA for 24 h. The leaves were harvested and immediately frozen in liquid nitrogen.
Total RNA was isolated from leaf samples using TransZol reagent (TransGen Biotech, Beijing, China). First-strand cDNA was reverse transcribed using the TIANscript RT Kit according to the manufacturer's instructions (TransGen Biotech, Beijing, China). Quantitative real-time PCR analyses were performed using a SYBRGreen qPCR kit (TransGenBiotech) with a MyiQ system (Bio-Rad, Hercules, CA, USA) as described previously [43]. The primers for amplification are listed in Table S6.

Metabolomics and Analysis
Two-year-old cultivated D. catenatum were grown in greenhouse (natural condition in May 2021, Zhejiang University, Hangzhou, China). The D. catenatum flowers were in full bloom in May. The tissue samples were collected from six independent plants (three biological replicates) and immediately ground into powder in liquid nitrogen.
For volatile organic compounds analysis, 1 g powder was transferred to an SPME vial containing 10 µL heptanone as an internal standard. A preconditioned SPME fiber (120 µm DVB/CAR/PDMS, Agilent, Santa Clara, CA, USA) was then exposed to the headspace of the capped vial at 100 • C for 15 min. The fiber coating was carried out in the injection chamber of the GC (Agilent 8890) in splitless mode at 250 • C for 5 min. Quantification of volatiles was carried out using GC (Agilent 8890) and mass spectrometer (Agilent 5977B), and equipped with a capillary DB-5MS column (30 m × 0.25 mm inner diameter with 0.25 µm film thickness). Helium was used as the carrier gas at a linear velocity of 1.0 mL/min. The injector temperature was kept at 250 • C and the detector at 280 • C. The GC oven temperature was programmed at 40 • C for 2min, increased from 40 to 280 • C at 7 • C/min, and then held for 5 min. The mass spectra were taken at electron ionization at 70 eV, and the mass range was 50-500 m/z. The quadrupole mass detector, ion source and transfer line temperatures were set at 150, 230, 280 • C, respectively. Compounds were identified by matching the data from the MWGC library.
For non-volatile organic compounds analysis, 100 mg powder was dissolved in 1.2 mL 70% methanol solution at 4 • C overnight. UPLC-ESI-MS/MS conditions were performed according to our previous work [12]. The metabolite counts were treated by hierarchical clustering using the R package ggplot2 (v.3.3.5).  Data Availability Statement: All data generated or analyzed during this study are included in this published article and its Additional files. The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.