Comprehensive identification of terpene synthase genes and organ-dependent accumulation of terpenoid volatiles in a traditional medicinal plant Angelica archangelica L.

Abstract

Angelica archangelica L. is a traditional medicinal plant of Nordic origin that produces an unusual amount and variety of terpenoids. The unique terpenoid composition of A. archangelica likely arises from the involvement of terpene synthases (TPSs) with different specificities, none of which has been identified. As the first step in identifying TPSs responsible for terpenoid chemodiversity in A. archangelica, we produced a transcriptome catalogue using the mRNAs extracted from the leaves, tap roots, and dry seeds of the plant; 11 putative TPS genes were identified (AaTPS1–AaTPS11). Phylogenetic analysis predicted that AaTPS1–AaTPS5, AaTPS6–AaTPS10, and AaTPS11 belong to the monoterpene synthase (monoTPS), sesquiterpene synthase (sesquiTPS), and diterpene synthase clusters, respectively. We then performed in vivo enzyme assays of the AaTPSs using recombinant Escherichia coli systems to examine their enzymatic activities and specificities. Nine recombinant enzymes (AaTPS2–AaTPS10) displayed TPS activities with specificities consistent with their phylogenetics; however, AaTPS5 exhibited a strong sesquiTPS activity along with a weak monoTPS activity. We also analyzed terpenoid volatiles in the flowers, immature and mature seeds, leaves, and tap roots of A. archangelica using gas chromatography-mass spectrometry; 14 monoterpenoids and 13 sesquiterpenoids were identified. The mature seeds accumulated the highest levels of monoterpenoids, with β-phellandrene being the most prominent. α-Pinene and β-myrcene were abundant in all organs examined. The in vivo assay results suggest that the AaTPSs functionally identified in this study are at least partly involved in the chemodiversity of terpenoid volatiles in A. archangelica.

Introduction

Terpenoids are a large and structurally diverse class of plant-specialized metabolites in which more than 50,000 different structures have been identified, including hemiterpenoids (C₅), monoterpenoids (C₁₀), sesquiterpenoids (C₁₅), and diterpenoids (C₂₀). These metabolites are the fragrance and aroma constituents of plants and are involved in responses to biotic and abiotic stresses as well as in the ecological interactions of plants (Tholl 2015). Terpene synthases (TPSs) catalyze the production of terpenoids—either acyclic or cyclic—from prenyl pyrophosphates. Hemiterpenoids, monoterpenoids, sesquiterpenoids, and diterpenoids are derived from dimethylallyl pyrophosphate (DMAPP, C₅), geranyl pyrophosphate (GPP, C₁₀), farnesyl pyrophosphate (FPP, C₁₅), and geranylgeranyl pyrophosphate (GGPP, C₂₀), respectively (Karunanithi and Zerbe 2019). Previous studies have shown that angiosperm TPS genes consist of multigene families in a plant lineage-specific manner and are derived from an ancestral TPS gene via repeated gene duplication and divergence, resulting in TPS paralogs with different substrate and product specificities (Chen et al. 2011; Karunanithi and Zerbe 2019). In the reactions catalyzed by TPSs categorized into class I TPS, an allylic carbocation intermediate is formed by divalent cation-assisted elimination of the diphosphate group from a prenyl diphosphate substrate. The enormous diversity of terpenoid structure, at least in part, results in myriad ways of folding, isomerization, cyclization, and/or rearrangement of the carbocation intermediates generated during TPS catalysis (Christianson 2017; Hare and Tantillo 2016). However, the mechanism and structural determinants governing TPS specificity remain unclear.

Angelica archangelica L. (garden angelica) is a biennial plant belonging to the Apiaceae family with a Nordic origin and was originally a common vegetable and medicinal herb for the Nordic Sami populations (Ojala 1986). Currently, this plant is being commercially cultivated, mainly in Europe, to produce flavor additives, essential oils, and medicinal products. Previous analyses revealed that A. archangelica produces an unusual amount and variety of terpenoids, which are responsible for its unique, complex aroma (Holm et al. 2012; Kerrola and Kallio 1994; Kerrola et al. 1994; Lopes et al. 2004; Nivinskiene et al. 2007). For example, essential oils extracted from the seeds of A. archangelica are composed of more than 60 different terpenoid structures, including monoterpenoids (β-phellandrene as the most abundant) and other terpenoids ubiquitously found in various plants, such as α-pinene and D-limonene (Holm et al. 2012; Lopes et al. 2004; Nivinskiene et al. 2007). The distinctiveness of terpenoid metabolism in A. archangelica is likely due to the involvement of unique TPSs with different specificities and spatial expression patterns. However, to date, none of the TPSs from A. archangelica have been identified and characterized.

In this study, transcriptome analysis of A. archangelica was performed to comprehensively identify TPSs expressed in this plant. Moreover, these TPS homologs were expressed in Escherichia coli cells to determine their TPS activities and specificities. The activity results of the recombinant TPSs were compared with the organ-dependent terpenoid profiles in A. archangelica to understand the unique terpenoid chemodiversity observed in this plant.

Materials and methods

Chemicals

β-Phellandrene, α-phellandrene, (+)-δ-cadinene, trans-β-farnesene, γ-terpinene, camphene, α-terpinene, β-fmyrcene, β-elemene, and isobutylbenzene were obtained from Tokyo Chemical Industry (Tokyo, Japan). (−)-α-Pinene, (−)-β-pinene, D-limonene, β-ocimene, 3-carene, and α-humulene were obtained from Sigma-Aldrich/Merck Japan (Tokyo, Japan). Sabinene was obtained from BLD Pharmatech (Shanghai, China). α-Copaene, germacrene D, and germacrene B were obtained from Cayman Chemical Company (Ann Arbor, MI, USA). t-Butyl methyl ether was obtained from Wako Pure Chemical Industries (Osaka, Japan). All other chemicals used were of analytical grade.

Plant materials

Seeds of A. archangelica were purchased from Mikasa Engei (Tokyo, Japan). The seeds were sown in potting soil (Dio Chemicals, Tokyo, Japan) and vernalized at 4°C for 10 days. For transcriptome analysis, A. archangelica plants were grown for 8 months—before flowering—at 22°C under long-day conditions (16-h light and 8-h dark) of white light (approximately 200 µmol m⁻² sec⁻¹) in a plant growth cabinet (model KCLP-1400ICT, Nippon Medical & Chemical Instruments, Osaka, Japan). The flowers and roots were harvested, immediately frozen in liquid nitrogen (LN₂), and stored at −80°C until total RNA was extracted from the samples. For terpenoid extraction, A. archangelica plants were grown for 26 months until they reached stem heights of 30–40 cm. Flowers, immature and mature seeds, leaves, and tap roots (Figure 1) were harvested, immediately frozen in LN₂ after their fresh weights were measured, and stored at −80°C until further use.

Figure 1. A. archangelica plant used in this study.

Total RNA extraction

Total RNA was extracted from the mature leaves of A. archangelica using the phenol/SDS method (Palmiter 1974). Total RNA in the dry seeds and tap roots of A. archangelica was extracted through the method developed by Suzuki et al. (Suzuki et al. 2004). The extracted RNA was treated with DNase I (recombinant, RNase-free; Roche, Basel, Switzerland). The total RNA integrity was assessed using the RNA 6000 Nano Kit with the Agilent 2100 Bioanalyzer System (Agilent Technologies, Santa Clara, CA, USA).

Transcriptome analysis and identification of TPS genes

Preparation of cDNA sequencing libraries using the SureSelect Strand-Specific RNA Library Prep System (Agilent Technologies), pair-end sequencing using the HiSeq 2000 sequencing system (Illumina, San Diego, CA, USA) and de novo assembly were outsourced to Medical & Biological Laboratories (Tokyo, Japan). After pair-end sequencing, the raw reads of each RNA sample were assessed using FastQC v0.11.2 and subjected to quality control by removing low-quality bases using PRINSEQ 0.20.4 (Schmieder and Edwards 2011); clipping an adapter sequence (5′-AGATCGGAAGAGC-3′); and removing overly short reads (<100 bases) using FASTX-Toolkit 0.0.14 (http://hannonlab.cshl.edu/fastx_toolkit/). The remaining high-quality trimmed read pairs from three samples (leaves, tap roots, and dry seeds) were collectively assembled using Trinity r2013-02-25 (Grabherr et al. 2011), which yielded 376,899 contigs. After removing contaminated sequences, a non-redundant transcript dataset was generated from the original contigs by mapping each read using RSEM v1.2.30 (Li and Dewey 2011) and removing splicing variants. The resultant unigene dataset (273,968 contigs) was functionally annotated by a BlastX search against a non-redundant National Center for Biotechnology Information protein database, the protein database of Arabidopsis thaliana (TAIR10, https://www.arabidopsis.org), and the Universal Protein Resource database (https://www.uniprot.org). For transcriptome analysis, the trimmed read pairs from each sample were mapped to the 273,968 unigenes using RSEM v1.2.30. The expression level of each transcript was estimated based on the fragments per kilobase of exon per million mapped reads (FPKM) values.

Unigenes, including a coding sequence highly similar to TPS, were searched by TBLASTN using the TPSs of A. thaliana as queries and collectively termed A. archangelica TPS (AaTPS). Among the 11 unigenes of AaTPS, all, except comp128661 (AaTPS5), comp123204 (AaTPS10), and comp119361 (AaTPS11), were considered to include their full-length coding sequences (CDSs) (Table 1). The full-length CDSs of AaTPS5 and AaTPS11 were clarified by means of 3′-rapid amplification of cDNA ends (3′-RACE) and 5′-RACE strategies, respectively, using the total RNA prepared from the tap root (see above) and the GeneRacer Kit for full-length, RNA ligase-mediated rapid amplification of 5′ and 3′ DNA ends (RLM-RACE) (Thermo Fisher Scientific, Waltham, MA, USA). The full-length coding sequence of AaTPS10 was clarified by means of 5′-RACE using the total RNA prepared from the mature leaves as a template. The total nucleotide sequences of AaTPS cDNAs and their deduced amino acid sequences are provided in Supplementary Figures S1 and S2, respectively, and are summarized in Table 1.

Table 1. TPS homologs identified in this study.

Name	Contig ID	CDS (bp)	Protein (aa)	Predicted localization3)	BlastX top hit against uniprot database4)	Predicted TPS cluster based on phylogenetics	Major products identified in in vivo enzyme assays
AaTPS1	comp121029	1,854	617	Other	Isoprene synthase from Populus alba (Q50L36, 0.0)	hemiTPS	No product
AaTPS2	comp128785	1,785	594	Chloroplast (42)	(−)-α-Terpineol synthase from Vitis vinifera (Q6PWU2, 0.0)	monoTPS	α-Phellandrene and β-phellandrene
AaTPS3	comp129373	1,770	589	Other	(+)-Limonene synthase from Citrus sinensis (A0A1C9J6A7, 0.0)	monoTPS	α-Pinene
AaTPS4	comp115931	1,758	585	Chloroplast (29)	(+)-α-Terpineol synthase from Vitis vinifera (F6M8I0, 0.0)	monoTPS	α-Pinene
AaTPS5	comp128661 1)	1,6502)	5492)	Other	(R)-Limonene synthase 1 from Vitis vinifera (A0A1C9J6A7, 0.0)	monoTPS	β-Bisabolene
AaTPS6	comp136494	1,713	570	Other	α-Copaene synthase from Eleutherococcus trifoliatus (Η9L9Ε5, 0.0)	sesquiTPS	β-Cedrene, β-barbatene, and β-farnesene
AaTPS7	comp120360	1,692	563	Other	α-Farnesene synthase from Ricinus communis (Β9RXW0, 3e-145)	sesquiTPS	α-Bergamotene
AaTPS8	comp129101	1,701	566	Other	Kunzeaol synthase from Thapsia garganica (K4LMW2, 0.0)	sesquiTPS	α-Humulene and germacrene D
AaTPS9	comp127914	1,641	546	Other	Valerianol synthase from Camellia sinensis (A0A167V661, 5e-167)	sesquiTPS	δ-Cadinene
AaTPS10	comp1232041)	1,551	516	Other	(3S,6E)-Nerolidol synthase 1 from Fragaria×ananassa (P0CV94, 0.0)	sesquiTPS	β-Farnesene
AaTPS11	comp1193611)	2,3252)	7742)	Other	ent-Kaurene synthase from Vitex agnus-castus (A0A2K9RFY0, 0.0)	diTPS	No product

¹⁾Contigs including a partial CDS. ²⁾The sequences for comp119361 and comp128661 were deduced from the full-length cDNA sequences amplified by RT-PCR after determination of the terminal sequences by means of 5′- and 3′-RACE, respectively. ³⁾Predicted by ChloroP, Figures in parenthesis show length (in amino acids) of predicted transit peptide. ⁴⁾Top hit results with accession number and e-value in parentheses.

Chemical synthesis of codon-optimized coding sequences of AaTPS

The codon-optimized CDSs of AaTPS (except for AaTPS10) for heterologous expression in E. coli, having NdeI and XhoI sites on the 5′ and 3′-end, respectively, were designed using the GeneArt algorithm (Thermo Fisher Scientific), chemically synthesized by GenScript Biotech Corporation (Tokyo, Japan), and collectively termed synthetic AaTPS (sAaTPS). It must be mentioned that N-terminal 42- and 29-residue sequences of AaTPS2 and AaTPS4, respectively, were predicted to serve as transit peptides for translocation to chloroplast, as analyzed by ChloroP 1.1 Prediction Server (Table 1). Thus, for the expression of these two enzymes, codon-optimized CDSs encoding N-terminally truncated forms of AaTPS2 and AaTPS4 without the predicted transit peptide were synthesized and termed sAaTPS2-deltaN (sAaTPS2dn) and sAaTPS4dn, respectively. The nucleotide sequences of these codon-optimized CDSs are shown in Supplementary Figure S3.

In vivo enzyme assays of AaTPSs using E. coli expression systems

System A

Each of the codon-optimized sAaTPS CDSs (except for AaTPS10) was co-expressed with Mycobacterium tuberculosis GPP synthase (Rv0989c) (Mann et al. 2011) in E. coli JM109(DE3) cells (Promega, Madison, WI, USA) as follows. The synthesis of a codon-optimized Rv0989c gene (sRv0989c) with 5′-BamHI and 3′-EcoRI sites was outsourced to Medical & Biological Laboratories. The sRv0989c fragment digested with BamHI and EcoRI was cloned into multi-cloning site 1 (BamHI/EcoRI) of pCOLADuet™-1 (Merck, Darmstadt, Germany) to obtain pCOLADuet-sRv. One of the sAaTPS CDSs (except for AaTPS10) was digested with NdeI and XhoI and cloned into multi-cloning site 2 (NdeI/XhoI) of pCOLADuet-sRv to obtain a series of plasmids: pCOLADuet-sRv-sAaTPS1 through pCOLADuet-sRv-sAaTPS11 (see Supplementary Figure S4 for general structure). The resulting plasmid was used to transform E. coli DH5α cells (Takara Bio, Shiga, Japan) and purified from the cultured cells using the FastGene® Plasmid Mini Kit (Nippon Genetics, Tokyo, Japan).

Cells of E. coli JM109(DE3) were transformed with pCOLADuet-sRv-AaTPS. The transformant cells were grown in LB medium (4 ml) containing 50 mg l⁻¹ ampicillin at 37°C overnight. One hundred microliters of the culture was then inoculated into 40 ml of fresh LB medium and further grown until its optical turbidity at 600 nm reached approximately 0.5. Five milliliter aliquots of the culture were then transferred to sterile 13 ml gas chromatography (GC) glass vials (Agilent Technologies, Santa Clara, CA, USA), and isopropyl β-D-1-thiogalactopyranoside (Nacalai Tesque, Kyoto, Japan) was added to the culture at a final concentration of 1 mM. The vials were sealed with a crimp cap with a polytetrafluoroethylene/silicone septum (Agilent Technologies) and the transformant cells were further grown at 180 rpm for 24 h at 18°C. Terpenoid volatiles in the headspace of the vials were analyzed by gas chromatography-mass spectrometry (GC-MS) using analytical GC-MS condition 1 (see below).

System B

The AaTPS10 CDS was amplified by polymerase chain reaction (PCR) using KOD-Plus-Neo DNA polymerase (Toyobo, Osaka, Japan) with a set of primers (123204NdeIFw and 123204XhoIRv; see Supplementary Figure S5). The amplified cDNA was gel-purified using the UltraClean® 15 DNA Purification Kit (Mo Bio Laboratories, Carlsbad, CA, USA), dA-attached using the 10x A-attachment mix (Toyobo), and cloned into the pGEM®-T Easy Vector (Promega). After validating the AaTPS10 sequence, the resulting plasmid was digested with NdeI and XhoI to obtain a fragment containing the AaTPS10 CDS, which was cloned into multi-cloning site 2 (NdeI/XhoI) of pCOLADuet™-1 and pCOLADuet-sRv to obtain pCOLADuet-AaTPS10 and pCOLADuet-sRv-AaTPS10, respectively. The resulting plasmids were used to transform E. coli Rosetta2(DE3) (Merck) bearing pBR-MScodon, which was introduced to enhance the synthesis of isopentenyl pyrophosphate (IPP) in E. coli (see below). Cultivation of transformant cells and GC-MS analysis were carried out as in system A, except that GC-MS analysis was performed with analytical GC-MS condition 1 (see below).

The pBR-MScodon plasmid was constructed as follows. The plasmid pBR322 (Takara Bio) was digested with SapI and BsaAI, treated with T4 DNA polymerase (Takara Bio) for blunting, and self-ligated to eliminate the NdeI site in the plasmid. The resulting plasmid was designated as pBR322N. A DNA fragment containing IPP isomerase of Saccharomyces cerevisiae (ScIDI) fused with the rrnB terminator (TrrnB) was amplified from the plasmid pAC-Mev/Scidi, carrying the genes for the mevalonate pathway enzymes of Streptomyces sp. strain CL190 and ScIDI (Harada et al. 2009), by PCR using primers (ScidiFw4 and tacterRv4) and KOD-Plus-Neo DNA polymerase to attach 5′-EcoRV and 3′-ClaI sites (see Supplementary Figure S5 for nucleotide sequences of primers and thermal cycling conditions). The amplified fragment was purified, digested with restriction enzymes EcoRV and ClaI, and ligated with EcoRV/ClaI-digested pBR322N using Ligation high Ver. 2 (Toyobo). The resulting plasmid was used to transform dam⁻/dcm⁻ E. coli cells (New England Biolabs Japan, Tokyo, Japan) and purified from the cultured cells to obtain a pBR322N derivative with ScIDI and TrrnB—pBR-Scidi-TrrnB. The tac promoter region (P_tac) of pAC-Mev/Scidi was also amplified by PCR using primers (tacproFw4 and tacproRv3) and KOD-Plus-Neo DNA polymerase to attach 5′-BspEI and 3′-NdeI-EcoRV sites (see Supplementary Figure S5). The amplified fragment was purified, digested with BspEI and EcoRV, and ligated with BspEI/EcoRV-digested pBR-Scidi-TrrnB to obtain pBR-Ptac-Scidi-TrrnB. pAC-Mev/Scidi was digested with NdeI and HindIII to obtain a DNA fragment coding for the mevalonate pathway enzymes of Streptomyces sp. strain CL190 and ScIDI. That DNA fragment was subsequently ligated with NdeI/HindIII-digested pBR-Ptac-Scidi-TrrnB to produce pBR-MS. To replace ScIDI in pBR-MS with a codon-optimized ScIDI, SD-ScIDIopt, which is a DNA fragment with 5′-EcoRV and 3′-SpeI-HindIII sites that encodes a codon-optimized ScIDI gene attached to a Shine–Dalgarno sequence (AGGAGG), was designed using the GeneOptimizer algorithm (Thermo Fisher Scientific) and chemically synthesized by Integrated DNA Technologies (Tokyo, Japan). Finally, the EcoRV/HindIII-digested SD-ScIDIopt and EcoRV/HindIII-digested pBR-MS fragments were ligated to obtain the pBR-MScodon.

Analysis of terpenoids in A. archangelica

Each harvested organ and tissue (i.e., flowers, immature and mature seeds, leaves, and tap roots) of A. archangelica (Figure 1) was homogenized in LN₂ using a motor and pestle. Each homogenized plant material (0.3 g of fresh weight) was suspended in 1.5 ml of t-butyl methyl ether supplemented with isobutylbenzene (1.2 mM) as an internal standard and shaken overnight at room temperature using a microtube mixer (Model MT-360, TOMY Seiko, Tokyo, Japan) to extract terpenoids. After the mixture was centrifuged, an aliquot of its organic phase (500 µl) was collected and vigorously mixed with an equal volume of 0.1 M ammonium carbonate. This step was followed by another round of centrifugation. The resulting organic phase (400 µl) was collected, diluted with 800 µl of t-butyl methyl ether (with 1.2 mM isobutylbenzene), and subjected to GC-MS analysis (condition 2, see below).

GC-MS analysis

GC-MS analysis was performed using a 6890N GC apparatus (Agilent Technologies) and a quadrupole mass selective detector (MSD; Model 5979C, Agilent Technologies) under one of the following conditions.

Condition 1

The vials containing the E. coli culture were kept at 80°C and shaken at a low speed using a network headspace sampler G1888 (Agilent Technologies) for 15 min. The headspace sample was injected for 0.15 min in splitless mode, where the temperatures of the loop, transfer line, and injector were maintained at 110, 120, and 250°C, respectively. Separation was achieved using a DB-1701 column (30 m long and 0.25 mm in diameter; Agilent Technologies). The flow rate of the carrier gas (helium) was set to 1.0 ml min⁻¹. The column oven temperature was programmed as follows: the initial oven temperature at 50°C was maintained for 1 min; it was increased to 150°C (at a rate of 5°C min⁻¹) and then to 250°C (at a rate of 100°C min⁻¹), which was sustained for 1 min. The MSD was operated in scan mode, where data acquisition was set between m/z 10 and 500 and the solvent delay was set at 3 min. The instrumental parameters were set as follows: the transfer line, source, and quadrupole temperatures were maintained at 250, 230, and 150°C, respectively, and the electron energy was set at 70 eV.

Condition 2

One microliter of the t-butyl methyl ether extracts from A. archangelica tissues, including isobutylbenzene as an internal standard, was injected into HP-INNOWAX polyethylene glycol (30 m long and 0.25 mm in diameter; Agilent Technologies). The injector was set to a temperature of 250°C for injection at a split ratio of 20 : 1. The column oven temperature was programmed as follows: the initial temperature at 75°C was maintained for 4 min; it was increased to 250°C (at a rate of 4°C min⁻¹), which was sustained for 10 min. The MSD was operated in scan mode, where data acquisition was set between m/z 20 and 300, and the solvent delay was set at 1.9 min. The instrumental parameters were set as follows: the transfer line, source, and quadrupole temperatures were maintained at 250, 230, and 150°C, respectively, and the electron energy was set at 70 eV.

Monoterpenoids and sesquiterpenoids were identified by comparing their retention times and mass spectrums to those for authentic standards or by referencing their mass spectral data to the National Institute of Standards and Technology (NIST) database (https://webbook.nist.gov/chemistry/name-ser/) with a threshold coincidence rate of 90% and comparing their retention times to GC retention indexes in the NIST database. The standard curves of isobutylbenzene, β-phellandrene, and β-elemene were obtained by GC-MS analysis of the authentic samples after appropriate dilutions with t-butyl methyl ether (condition 2). After normalizing each extract with isobutylbenzene using the standard curve, β-phellandrene and β-elemene levels in each tissue were determined from the peak areas obtained by GC-MS analysis and the corresponding standard curves; the concentrations of other monoterpenoids and sesquiterpenoids were estimated using the β-phellandrene and β-elemene standard curves, respectively.

Results and discussion

Identification of TPS homologs via transcriptome analysis

As the first step in identifying the TPSs responsible for terpenoid chemodiversity in A. archangelica, we produced a transcriptome catalogue using the mRNAs collected from its leaves, tap roots, and dry seeds. For the transcriptome analysis, A. archangelica dry seeds were commercially obtained for RNA extraction because A. archangelica is a biennial plant; therefore, seeds were not immediately harvested. Leaves and tap roots were harvested from A. archangelica plants grown in our laboratory. The high-quality read pairs obtained from RNA sequencing of these three samples were collectively assembled to provide 376,899 contigs and 273,968 unigene sequences, and the resultant unigene dataset was functionally annotated as described in the section of Materials and methods. As a result, we identified 11 expressed TPS paralogs: AaTPS1–AaTPS11 (Figure 2, Table 1). Among the 11 TPS homologs, the full-length CDSs for AaTPS5, AaTPS10, and AaTPS11 were clarified by cloning the corresponding cDNAs by means of 5′- and 3′-RACE; other TPS paralogs were predicted to encode full-length CDSs (see Supplementary Figures S1 and S2, respectively, for the nucleotide and predicted amino acid sequences). The following characteristics of the 11 AaTPSs are summarized in Table 1: the lengths of the CDS (in base pairs) and polypeptide chains (in amino acids), predicted subcellular localizations (based on ChloroP 1.1), BlastP top-hit results, and predicted TPS clusters based on phylogenetics (see below). The FPKM values of these AaTPS genes in dry seeds, tap roots, and leaves are shown in Supplementary Figure S6.

Figure 2. Phylogenetic analysis of TPSs from A. archangelica. The tree was constructed by the neighbor-joining method. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the Poisson correction method and are in the units of the number of amino acid substitutions per site. The phylogenetic TPS clades (TPS-a, TPS-b, TPS-c, TPS-d, and TPS-e/f) are indicated by blue, red, gray, green arcs, and yellow, respectively. Bar=0.1 amino acid substitution per site. Abbreviated names of MonoTPSs, sesquiTPS, and diTPS in the entries are colored in red, blue, and purple, respectively. Abbreviations and accession numbers of TPS sequences used in this analysis are shown in Supplementary Table S2.

The primary structures of these 11 AaTPSs shared the structural characteristics conserved among class I plant TPSs, which include the DDXXD and NSE/DTE motifs for divalent cation-mediated holding and elimination of a diphosphate group from a substrate (see Supplementary Figure S7) (Chen et al. 2011); however, AaTPS1, having DXXDD instead of DDXXD, is an exception. Plant TPS have been classified into seven phylogenetic clades: TPS-a through TPS-d, TPS-e/f, TPS-g, and TPS-h (Chen et al. 2011; Falara et al. 2011). From the perspective of TPS specificity, most TPSs belonging to the clades TPS-b and TPS-g have been shown to be monoterpene synthases (monoTPSs) and can be grouped into the monoTPS cluster. Clade TPS-a mostly consists of sesquiterpene synthases (sesquiTPSs), forming the sesquiTPS cluster, while TPS-c and TPS-e/f consist mostly of diterpene synthases (diTPSs) and can be put together into the diTPS cluster (Falara et al. 2011). Phylogenetic analysis revealed that each of the 11 AaTPSs could be assigned to clades TPS-a (AaTPS6–AaTPS10), TPS-b (AaTPS1–AaTPS5), and TPS-e/f (AaTPS11), allowing us to predict the specificity of each AaTPS (Table 1 and Figure 2). Some AaTPSs (AaTPS2–AaTPS6) also share the RRX₈W motif, which has been proposed to be important in monoterpenoid ring closure—a process catalyzed by monoTPSs and some sesquiTPSs (see Supplementary Figure S7) (Chen et al. 2011).

Organ/tissue-dependent compositions of terpenoid volatiles

A. archangelica plants were grown as described in the section of Materials and methods; mono- and sesquiterpenoids were extracted from the flowers, immature and mature seeds, leaves, and tap roots (see Figure 1) and analyzed by GC-MS analysis. The results are presented in Figures 3–5, and Supplementary Figures S8 and S9, and summarized in Supplementary Table S1.

Figure 3. GC chromatograms of terpenoid volatiles extracted from different tissues and organs in A. archangelica. (A) From top to bottom panels, leaf, flower, immature seeds, mature seeds, and tap root. Left panels, monoterpenoids and right panels, sesquiterpenoids. The monoterpenoids and sesquiterpenoids were identified by referencing their mass spectral data to the NIST database with a threshold coincidence rate of 90% (see Supplementary Figure S8 for details). Peak at 5.87 min in left panels corresponds to isobutylbenzene added as a spike control. (B) The identified metabolites, numbered (1 through 13 for monoterpenoids and derivatives, and 14 through 27 for sesquiterpenoids), and their structures.

GC-MS analysis identified a total of 13 monoterpenoids and 14 sesquiterpenoids in the A. archangelica plant, with varying compositions in the organs and tissues examined (Figures 3 and 4). The total content of monoterpenoids was higher than that of sesquiterpenoids in all organs and /tissues examined (Figure 5A), concordant with terpenoid compositions of root and seed oils (Holm et al. 2012; Lopes et al. 2004; Nivinskiene et al. 2007) (Figure 5A). Among the organs and tissues examined, the immature and mature seeds accumulated the highest levels of monoterpenoids; β-phellandrene 10 was the most prominent terpenoid [73% and 71% (mol/mol) of the total terpenoid volatiles, respectively] (Figures 3, 4A, 5B), as previously reported (Lopes et al. 2004; Nivinskiene et al. 2007). α-Pinene 1 and β-myrcene 6 were relatively abundant in all organs examined in this study (Figures 3, 4A, 5B). More specifically, α-pinene 1 was the most prominent terpenoid in the tap roots and flowers [55% and 35% (mol/mol), respectively] and the second major constituent in the mature and immature seeds [10% and 11% (mol/mol), respectively] (Figure 5B). β-Myrcene 6 accounted for 25%, 25%, and 8.6% (mol/mol) of the total terpenoid volatiles in the leaves, flowers, and tap roots, respectively, but only 1.5%–1.7% (mol/mol) in the mature and immature seeds (Figure 5B). Sesquiterpenoids were the most abundant in the mature seeds (Figure 5A); among which, germacrene D 22 and α-humulene 20 were the most prominent (Figure 4B). These two sesquiterpenoids also accumulated in all other organs and tissues examined; interestingly, the organ/tissue-dependent accumulation patterns of these sesquiterpenoids were very similar to each other (Figure 4B). It must be mentioned that β-elemene and γ-elemene (Figure 4B, 18 and 19, respectively) are known to be formed in the Cope rearrangement from germacrene A and germacrene B, respectively, under GC conditions (Adio 2009; de Kraker et al. 1998). Thus, the detection of β-elemene and γ-elemene in the GC-MS analysis potentially indicates the occurrence of germacrene A and germacrene B in the harvested organs. Moreover, some terpenoid volatiles identified were not considered as primary enzymatic reaction products. For example, pseudolimonene (Figure 4A, 8) can be formed non-enzymatically from β-pinene (Meehan-Atrash et al. 2021; Sun et al. 2014). m-Cymene (Figure 4A, 13) was considered to be formed by the spontaneous dehydration of a monoterpene alcohol formed by a TPS and a subsequent cytochrome P450 enzyme, similar to p-cymene formation (Krause et al. 2021).

Figure 5. Organ/tissue-dependent distributions of terpenoids in A. archangelica. (A) Total mono- and sesquiterpenoid contents. (B) Organ/tissue-dependent terpenoid compositions. Values are calculated using the data presented in Figure 4.

Figure 4. Terpenoid contents in different tissues and organs in A. archangelica. (A) Monoterpenoids and derivatives. (B) Sesquiterpenoids. In panels (A) and (B), the contents in tap roots, mature seeds, immature seeds, flower, and leaf (from top to bottom; see also Figure 1) are shown with yellow, red, pink, blue, and green bars, in which data are presented as the average of three independent determinations (±standard deviation). Amounts of monoterpenoids and sesquiterpenoids were expressed as equivalents for β-phellandrene and β-elemene, respectively. Note that the horizontal axis is logarithmic. The identified metabolites are numbered as in Figure 3.

In vivo assay of AaTPSs

We heterologously expressed AaTPSs in E. coli cells to examine their enzymatic activities and specificities. To this end, the codon-optimized genes encoding N-terminally truncated forms of AaTPS2 and AaTPS4 (sAaTPS2dn and sAaTPS4dn, respectively) and the full-length amino acid sequences of other AaTPSs (sAaTPS), except for AaTPS10, were chemically synthesized (Supplementary Figure S3 for the nucleotide sequences). Each codon-optimized AaTPS or AaTPS10 cDNA was co-expressed with a codon-optimized M. tuberculosis GPP synthase gene (sRv0989c) (Mann et al. 2011) in E. coli cells to furnish GPP as the substrate for monoTPS using assay systems A or B (see the section of Materials and methods; see also Supplementary Figure S4 for the metabolic pathways). In these systems, FPP, as the substrate of sesquiTPS, was provided by the activity of the endogenous FPP synthase. To facilitate the expression of codon-optimized AaTPS genes, E. coli strain JM109(DE3) was utilized as the host for system A; E. coli strain Rosetta2(DE3) was used as the host for AaTPS10 expression in system B to enhance the translation efficiency of AaTPS10, which contained codons rarely used in E. coli. Moreover, in system B, the host E. coli cells also expressed actinomycetous mevalonate pathway enzymes and yeast ScIDI (Harada et al. 2009) to amplify the supply of IPP, a common precursor of terpenoids (see the section of Materials and methods; see also Supplementary Figure S4 for the metabolic pathways designed for system B), resulting in enhanced production of terpenoids by AaTPS10 whose expression level in E. coli was lower than those of codon-optimized AaTPSs. In both systems, the E. coli transformant cells expressing each AaTPS were grown in a sealed GC vial (see the section of Materials and methods for cultivation conditions), and the terpenoid volatiles in the headspace of the vials were analyzed by GC-MS (see the section of Materials and methods for GC-MS conditions 1). The results are presented in Figure 6, and Supplementary Figures S10–S12. When compared with the volatile metabolites from a control strain that did not bear AaTPS CDS, all transgenic strains expressing each AaTPS, except for AaTPS1 and AaTPS11, accumulated terpenoids derived from the exogenous enzyme.

Figure 6. Terpenoids detected in cultures of the transgenic E. coli strains expressing AaTPSs. Left panels show GC chromatograms of in vivo enzyme assays of AaTPS1 through AaTPS11 (from top to bottom), with a table right to each GC chromatogram showing peak identification by referencing their mass spectral data to the NIST database with a threshold coincidence rate of 90%. Total ion chromatograms in red and blue corresponding data for the E. coli strains expressing each AaTPS and that no exogenous TPS.

The results showed that the AaTPSs of clade TPS-b (i.e., AaTPS2, AaTPS3, AaTPS4, and AaTPS5) and those of clade TPS-a (i.e., AaTPS6, AaTPS7, AaTPS8, AaTPS9, and AaTPS10) consistently produced monoterpenoids and sesquiterpenoids (Figure 6 and Table 1), respectively. Interestingly, AaTPS5, a member of clade TPS-b, displayed a strong sesquiTPS activity to produce β-bisabolene 24. Its sesquiTPS activity was much higher than its monoTPS activity (Figure 6), strongly suggesting that AaTPS5 was a sesquiTPS that had evolved from a monoTPS-related ancestor. Detailed analyses of the phylogenetics, specificity, biochemical properties, and subcellular localization of AaTPS5 that would provide important information concerning the evolutionary aspects of TPS specificity and functions will be reported elsewhere.

In this study, distinct TPS activities for AaTPS1 and AaTPS11 were not detected. Based on amino acid sequence similarity, AaTPS1 was expected to encode an isoprene synthase (IspS) (Table 1). We attempted to detect the formation of isoprene from DMAPP in an E. coli strain expressing sAaTPS1 under various GC-MS conditions, but failed (data not shown). The fact that the IspS responsible for isoprene emission is not conserved in all plant species (Sharkey et al. 2008), and that isoprene could not be detected in any of the terpenoid tissues of A. archangelica analyzed in this study, even though the transcript levels of AaTPS1 in leaves were comparable to those of other AaTPS paralogs (Supplementary Figure S6), suggest that AaTPS1 is not likely to encode IspS. Although the putative transit peptide was not predicted in the AaTPS1 sequence using the TargetP algorithm, an N-terminal extension of the amino acid sequence of AaTPS1 was found in comparison with those of sesquiTPSs (i.e., AaTPS6, AaTPS7, AaTPS8, AaTPS9, and AaTPS10; Supplementary Figure S7). This suggests that AaTPS1 may have an N-terminal transit peptide. N-terminal extensions, including transit peptides, seldom negatively affect the functional expression of plastidial TPSs in the E. coli expression system, causing inclusion body formation (Bohlmann et al. 1998) and interference with the catalytic activity (Lima et al. 2013; Williams et al. 1998) of recombinant TPSs. Therefore, removing the N-terminal extension is a promising approach for the functional expression of recombinant AaTPS1.

AaTPS11 belongs to clade TPS-e/f and is closely related to TPS4 from Vitex agnus-castus (VacTPS4; 58% amino acid sequence identities) and ent-kaurene synthase (Heskes et al. 2018). In the gibberellin biosynthesis of vascular plants, GGPP is converted to ent-kaurene in a sequential reaction catalyzed by two enzymes; GGPP is converted to ent-copalyl diphosphate (CPP) by a class II diTPS: ent-CPP synthase (TPS-c clade). This step is followed by the conversion of ent-CPP to ent-kaurene by a class I diTPS: ent-kaurene synthase (TPS-e/f clade) (Chen et al. 2011; Karunanithi and Zerbe 2019). The TBLASTN search against the A. archangelica contigs using the ent-CPP synthase from A. thaliana as a query revealed the existence of three contigs encoding partial protein sequences that showed high similarity to the different regions of ent-CPP synthase (data not shown); however, these partial sequences could not be assembled as one unigene covering a full-length CDS. Therefore, AaTPS11 might encode an ent-kaurene synthase that functions in gibberellin biosynthesis along with an unidentified ent-CPP synthase. Another possibility that AaTPS11 might function as a TPS that could use oligoprenyl diphosphate as a substrate was also raised by a report on the identification of β-phellandrene synthase from Solanum lycopersicum PHS1(Schilmiller et al. 2009). PHS1 belongs to the TPS-e/f clade and primarily consists of class I diTPSs (Figure 2); however, its catalysis of a reaction forming some monoterpenoids, with β-phellandrene as the major product, indicates its unusual substrate preference. PHS1 scarcely accepts GPP but prefers neryl diphosphate (NPP), the cis-isomer of GPP. To examine the monoTPS activity of AaTPS11, we heterologously expressed the full-length AaTPS11 as well as a few N-terminal truncated forms of AaTPS11—lacking residues expected to include transit peptides—in E. coli and conducted in vitro TPS assay using the crude protein extracts prepared from E. coli cells and various substrates: DMAPP, GPP, FPP, GGPP, and NPP. However, no distinct TPS activity was detected under any reaction conditions (data not shown).

Possible involvement of AaTPSs in terpenoid chemodiversity in A. archangelica

The enzymatic formation of terpenoid volatiles identified in A. archangelica plants could be explained in terms of the generally accepted reaction pathways of TPS catalysis. Specifically, all identified monoterpenoids were considered to be produced via the proposed monoTPS reaction pathways involving geranyl and α-terpinyl cations as key intermediates (Figure 7). The formation of sesquiterpenoids in A. archangelica can also be explained by the proposed sesquiTPS reaction pathways, which involve farnesyl cations and their cyclized isomers (e.g., cadinanyl, germacradienyl, humulyl, and bisabolyl cations) as key intermediates (Figure 8). The present in vivo assay results strongly suggest that the AaTPSs identified in this study are at least partly involved in the chemodiversity of terpenoid volatiles in A. archangelica plants. Especially for the AaTPSs in TPS-b clade, all of monoterpenoids derived from heterologously expressed TPSs in E. coli, except for γ-terpinene, were identified in A. archangelica tissues (Figure 7). On the other hand, some sesquiterpenoids detected in the E. coli culture systems expressing sAaTPS6, sAaTPS7, or sAaTPS9, i.e., β-cedrene, β-himachalene, β-chamigrene, β-barbatene,α-bergamoteme, and β-caryophyllene, could not be detected in any A. archangelica tissues, suggesting lower catalytic activity than other sesquiTPSs or lower expression levels in the tissues analyzed. In this study, we were unable to identify TPSs that were capable of producing the following terpenoid volatiles found in A. archangelica plants: monoterpenoids (see Figure 4A), trans- and cis-ocimenes (11 and 12, respectively), camphene 2, and 3-carene 5; and sesquiterpenoids (Figure 4B), α-farnesene 25, α-muurolene 23, γ-elemene 20 (germacrene B 27, see above), bicyclosesquiphellandrene 17, and β-cubebene 16. Other unknown TPSs should be responsible for the accumulation of these terpenoid volatiles. This should be addressed in future studies.

Figure 7. Proposed reaction pathways of TPS-catalyzed formation of monoterpenoids from GPP. Structures of A. archangelica terpenoid volatiles identified in this study are enclosed by dashed squares. TPS names are shown in parenthesis below the metabolite structures if the enzymes produce the metabolite.

Figure 8. Proposed reaction pathways of TPS-catalyzed formation of sesquiterpenoids from FPP. Structures of A. archangelica terpenoid volatiles identified in this study are enclosed by dashed squares. TPS names are shown in parenthesis below the metabolite structures if the enzymes produce the metabolite.

Although the precise roles of the respective AaTPSs in organ/tissue-dependent accumulation of terpenoid volatiles in A. archangelica remain to be clarified in future studies, it would be useful to discuss the possible involvement of AaTPSs in the accumulation of the major terpenoid volatiles in this plant.

β-Phellandrene

β-Phellandrene 10 is the most prominent terpenoid in the seeds, with up to 71% of total terpenoids (see above). Among the 11 AaTPSs identified in this study, AaTPS2 was the only TPS capable of producing β-phellandrene 10, as determined by in vivo enzyme assays (Figure 6). However, AaTPS2 was unlikely to be responsible for the accumulation of β-phellandrene 10 in the seeds because it produced α-phellandrene 7 and β-phellandrene 10 at an approximate molar ratio of 2 : 1 (Figure 6); this product ratio did not account for the exclusive accumulation of β-phellandrene 10 over α-phellandrene 7 in the seeds. Moreover, the FPKM values of AaTPS2 obtained from transcriptome analysis (Supplementary Figure S6) did not support the organ/tissue-dependent accumulation pattern of β-phellandrene 10. The FPKM values and TPS assay results also suggest that neither AaTPS1 nor AaTPS11 may be the predominant β-phellandrene synthase in seeds. Thus, another unknown TPS that is responsible for the accumulation of a large amount of β-phellandrene 10 remains to be identified in future studies.

α-Pinene and β-myrcene

The in vivo enzyme assay results revealed that AaTPS3 and AaTPS4 were capable of producing α-pinene 1 as their major product; thus, they were considered as the TPSs responsible for the accumulation of α-pinene 1 in this plant (Figure 6). The FPKM values of these two TPSs support their role in α-pinene accumulation in A. archangelica leaves and tap roots (Supplementary Figure S6). The in vivo enzyme assay results also showed that AaTPS4 was capable of producing β-myrcene 6 as its major product, suggesting that the enzyme is potentially involved in the accumulation of β-myrcene 6 in this plant. The FPKM values of AaTPS4 support its role in β-myrcene accumulation in the plant’s leaves and tap roots (Supplementary Figure S6). AaTPS5, which exhibited sesquiTPS activity to produce β-bisabolene 24 as its major product, also produced monoterpenoids, including β-myrcene 6, as its minor product in the E. coli assay system. Nevertheless, AaTPS5 is unlikely to contribute to monoterpenoid formation in A. archangelica because AaTPS5, which does not have an N-terminal extension (Supplementary Figure S7), may not be localized in plastids, where GPP/NPP synthases are generally found. The GPP acceptability of AaTPS5 strengthens the idea that AaTPS5 might be derived from a monoTPS ancestor, as speculated from the phylogenetic analyses (Figure 2).

Germacrene D and α-humulene

AaTPS8 was capable of producing germacrene D and α-humulene, which are sesquiterpenoids, as its major products. This suggests that it might be at least partly responsible for the observed coexistence of germacrene D 22 and α-humulene 20 in this plant. The FPKM values of AaTPS8 support its role in the accumulation of these two sesquiterpenoids in the plant’s leaves and tap roots (Supplementary Figure S6). TPSs that catalyze formation of germacrene D as a major product have been identified other Apiaceae, such as AkTPS1 and AkTPS2 from Angelica keiskei (Iimura et al. 2020) and DcTPS07, DcTPS11, and DcTPS42 form Daucus carota (Muchlinski et al. 2020). AkTPS1 and AkTPS2 show the highest amino acid sequence identity to AaTPS8 (75% and 71%, respectively) among TPSs from A. archangelica, suggesting that AaTPS8 might be an orthologue of AkTPS1 and AkTPS2.

In this study, we successfully elucidate the distributions of mono- and sesquiterpenoid volatiles in various tissues of A. archangelica. To the best of our knowledge, this is the first report of terpenoid composition analysis of non-stressed physiological plant samples, as previous studies have analyzed terpenoid compositions of A. archangelica using processed plant samples, such as essential oils (Lopes et al. 2004; Nivinskiene et al. 2007) and air-dried roots (Kerrola and Kallio 1994; Kerrola et al. 1994). It has been reported that many TPS genes are transcriptionally induced in response to various biotic and abiotic stresses, accompanying with dynamic changes in terpenoid compositions (Pichersky and Raguso 2018). Therefore, the terpenoid composition of the A. archangelica tissues, as well as enzymatic properties of AaTPSs, clarified in this study would be useful fundamental information for investigation of physiological roles of terpenoids in response to environmental stresses and identification of other stress-inducible TPS genes. Furthermore, comparable functional analyses of AaTPSs would provide important information for elucidating key residues that regulate the substrate and product specificities of TPS.

Abbreviations

AaTPS

Angelica archangelica terpene synthase

DMAPP

dimethylallyl pyrophosphate

FPKM

fragments per kilobase of exon per million mapped reads

FPP

farnesyl pyrophosphate

GC

gas chromatography

GC-MS

gas chromatography-mass spectrometry

GGPP

geranylgeranyl pyrophosphate

GPP

geranyl pyrophosphate

IPP

isopentenyl diphosphate

monoTPS

monoterpene synthase

sesquiTPS

sesquiterpene synthase

TPS

terpene synthase

Conflict of interest

The authors are not aware of any affiliations, memberships, funding, or financial holdings that might affect the objectivity of this review.

Author contribution

TS, TW, TN, MF, and ST planned and designed this study; MS-H, DM, YK, JT, NK, YA, TS, and ST annotated the transcriptome data and cloned cDNAs. MS-H, DM, and HA extracted and analyzed the terpenoids in A. archangelica plants. MS-H, DM, and HA constructed E. coli assay systems and analyzed the terpenoids. TN and ST wrote a draft; All authors reviewed the manuscript.

Accession numbers

The transcriptome data of this article are available from the DNA Data Bank of Japan at https://www.ddbj.nig.ac.jp/index-e.html and can be accessed with the accession numbers DRR378897-DRR378899. The other data underlying this article are available in the article and online supplementary files.