Skip to main content
NCBI home page
Plant Physiology logoLink to Plant Physiology
. 2018 Aug 10;178(2):535–551. doi: 10.1104/pp.18.00655

An Aromatic Farnesyltransferase Functions in Biosynthesis of the Anti-HIV Meroterpenoid Daurichromenic Acid1

Haruna Saeki a, Ryota Hara a, Hironobu Takahashi b, Miu Iijima a, Ryosuke Munakata c,2, Hiromichi Kenmoku b, Kazuma Fuku d, Ai Sekihara d, Yoko Yasuno d, Tetsuro Shinada d, Daijiro Ueda e, Tomoyuki Nishi e, Tsutomu Sato e, Yoshinori Asakawa b, Fumiya Kurosaki a, Kazufumi Yazaki c, Futoshi Taura a,3,4
PMCID: PMC6181053  PMID: 30097469

A farnesyl diphosphate-preferring aromatic prenyltransferase plays a key role in the biosynthetic pathway of daurichromenic acid, an anti-HIV meroterpenoid, in Rhododendron dauricum.

Abstract

Rhododendron dauricum produces daurichromenic acid, an anti-HIV meroterpenoid, via oxidative cyclization of the farnesyl group of grifolic acid. The prenyltransferase (PT) that synthesizes grifolic acid is a farnesyltransferase in plant specialized metabolism. In this study, we demonstrated that the isoprenoid moiety of grifolic acid is derived from the 2-C-methyl-d-erythritol-4-phosphate pathway that takes place in plastids. We explored candidate sequences of plastid-localized PT homologs and identified a cDNA for this PT, RdPT1, which shares moderate sequence similarity with known aromatic PTs. RdPT1 is expressed exclusively in the glandular scales, where daurichromenic acid accumulates. In addition, the gene product was targeted to plastids in plant cells. The recombinant RdPT1 regiospecifically synthesized grifolic acid from orsellinic acid and farnesyl diphosphate, demonstrating that RdPT1 is the farnesyltransferase involved in daurichromenic acid biosynthesis. This enzyme strictly preferred orsellinic acid as a prenyl acceptor, whereas it had a relaxed specificity for prenyl donor structures, also accepting geranyl and geranylgeranyl diphosphates with modest efficiency to synthesize prenyl chain analogs of grifolic acid. Such a broad specificity is a unique catalytic feature of RdPT1 that is not shared among secondary metabolic aromatic PTs in plants. We discuss the unusual substrate preference of RdPT1 using a molecular modeling approach. The biochemical properties as well as the localization of RdPT1 suggest that this enzyme produces meroterpenoids in glandular scales cooperatively with previously identified daurichromenic acid synthase, probably for chemical defense on the surface of R. dauricum plants.

Plants produce vast arrays of specialized metabolites for various purposes, including self-defense and communication with other organisms in their environment (Bennett and Wallsgrove, 1994; Wittstock and Gershenzon, 2002). Among them, meroterpenoids are a remarkable class of natural products because of their structural diversity and potent activities in both plant physiology and pharmacology (Appendino et al., 2008; Chen et al., 2014). The basic carbon structures of these metabolites are synthesized by aromatic prenyltransferases (PTs) that transfer an isoprenoid chain from a prenyl diphosphate (prenyl donor) to an aromatic substrate (prenyl acceptor; Yazaki et al., 2009). Then, further modifications of the prenyl moiety (e.g. cyclization and hydroxylation) provide a wide variety of meroterpenoid natural products (Barron and Ibrahim, 1996; Bourgaud et al., 2006).

Daurichromenic acid (DCA), isolated from Rhododendron dauricum (Ericaceae), is a meroterpenoid consisting of orsellinic acid (OSA) and sesquiterpene moieties (Fig. 1; Kashiwada et al., 2001). DCA has potent anti-HIV and anti-inflammatory activities (Iwata et al., 2004; Lee, 2010), making this compound an attractive medicinal resource and target for chemical synthesis over the last decade (Bukhari et al., 2015). The structure of DCA is partly similar to those of cannabinoids from hemp (Cannabis sativa), as both contain cyclized isoprenoid attached to alkylresorcylic acid substructures. However, the isoprenoid portion of DCA is sesquiterpene in origin, which is very rare in plant meroterpenoids, whereas those in cannabinoids are derived from monoterpene (Sirikantaramas and Taura, 2017).

Figure 1.

Figure 1.

Open in a new tab

Biosynthetic pathway of DCA and related meroterpenoids in R. dauricum. DCA synthase is an FAD oxidase that catalyzes the oxidative cyclization of grifolic acid (GFA) to produce DCA. Two DCA analogs, cannabichromeorcinic acid and diterpenodaurichromenic acid, also are produced by DCA synthase. The proposed aromatic PT would catalyze the prenylation of OSA at its C-3 position using three different prenyl donor substrates, farnesyl diphosphate (FPP), geranyl diphosphate (GPP), and geranylgeranyl diphosphate (GGPP), to produce GFA and its prenyl analogs as the substrates for DCA synthase. Thick arrows indicate the main metabolic route leading to DCA as the major constituent. aThe absolute configuration of these DCA analogs was not determined (Iijima et al., 2017).

With respect to the biosynthesis of DCA, we previously identified and characterized DCA synthase, a FAD oxidase that catalyzes the stereoselective oxidocyclization of GFA to form DCA (Fig. 1; Iijima et al., 2017). Meroterpenoid cyclase-type FAD oxidases are rare (Sirikantaramas et al., 2004; Taura et al., 2007), and an unusual feature of DCA synthase is that it mediates the cyclization of a farnesyl group in plant products. Apart from GFA, DCA synthase also has detectable enzyme activities for GFA analogs with geranyl or geranylgeranyl side chains to produce DCA analogs (Fig. 1). Furthermore, we confirmed that young leaves of R. dauricum actually contained these DCA analogs and their precursors as minor components (Iijima et al., 2017).

These previous observations suggest that aromatic PT in the DCA pathway would produce GFA as the major reaction product from OSA and FPP and perhaps catalyzes side reactions using GPP and GGPP to activate the three metabolic branches together with DCA synthase, as depicted in Figure 1. We assumed that the identification of a corresponding PT would provide considerable progress in understanding plant secondary metabolism because currently documented aromatic PTs are neither FPP specific nor promiscuous to prenyl donor substrates. In fact, most of the known enzymes are highly specific to dimethylallyl diphosphate (DMAPP), while there are a few exceptions that prefer GPP, such as umbelliferone 8-geranyltransferase, which was isolated from lemon (Citrus limon [ClPT]; Munakata et al., 2014), olivetolate geranyltransferase that produces a cannabinoid precursor in C. sativa (CsPT1; Page and Boubakir, 2014), and p-hydroxybenzoate geranyltransferases, which are implicated in shikonin biosynthesis in purple gromwell (Lithospermum erythrorhizon [LePGT1 and LePGT2]; Yazaki et al., 2002). Thus, we considered that PT in the DCA pathway might have novel structural features underlying the functional differences between these known enzymes.

The PT in question also is intriguing regarding its subcellular localization. Aromatic PTs generally are localized in plastids to utilize DMAPP or GPP derived from the 2-C-methyl-d-erythritol-4-phosphate (MEP) pathway for their specialized functions (Munakata et al., 2014). In contrast, FPP, the precursor of GFA, generally is synthesized in the cytosol through the mevalonic acid (MVA) pathway (Vranová et al., 2013). An irregular example has been reported for LePGTs that are involved in the shikonin pathway; LePGTs are anchored in the membranes of the endoplasmic reticulum to catalyze reactions using GPP, which is synthesized by a cytosolic form of GPP synthase unique in L. erythrorhizon plants (Sommer et al., 1995; Yazaki et al., 2002).

Taking these previous reports into account, we attempted to clarify which isoprenoid pathway contributes to GFA biosynthesis in order to categorize the GFA-producing PT. Using homology-based transcriptome screening and subsequent recombinant protein expression, we provide evidence that one of the candidate cDNAs is a gene that encodes a PT, which we designated RdPT1. In this study, we describe the molecular cloning and biochemical characterization of RdPT1, which is sophisticatedly specialized for meroterpenoid biosynthesis in R. dauricum plants.

RESULTS

The Effects of Isoprenoid Pathway Inhibitors on Meroterpenoid Production

To assess the origin of the isoprenoid portion of GFA, inhibitor experiments were conducted with young shoot cuttings of R. dauricum using clomazone and mevastatin, which are specific inhibitors of 1-deoxy-d-xylulose-5-phosphate synthase in the MEP pathway and 3-hydroxy-3-methylglutaryl-CoA reductase in the MVA pathway, respectively (Han et al., 2013; Zhao et al., 2014). GFA and DCA in the young leaves of treated shoots were then analyzed by HPLC. These meroterpenoids are constitutively accumulated in R. dauricum (Taura et al., 2014; Iijima et al., 2017), but there were apparent changes in their levels observed after 20 d of treatment (Fig. 2). Notably, both GFA and DCA decreased significantly after clomazone treatment, while mevastatin treatment increased the contents of both compounds. This result was unexpected and suggested that the farnesyl moiety of GFA likely originated from the MEP pathway, although FPP generally is regarded as a typical MVA pathway product (Vranová et al., 2013). Thus, we hypothesized that the PT responsible for GFA biosynthesis could be localized in plastids that produce its prenyl donor substrate, as reported for most aromatic PTs involved in plant specialized metabolism (Yazaki et al., 2009).

Figure 2.

Figure 2.

Open in a new tab

Effects of isoprenoid pathway inhibitors on the accumulation of GFA (left) and DCA (right) in young leaves from shoot cuttings treated with clomazone (Clo) or mevastatin (Mev). Treated samples are compared with the water-treated controls (H2O). Bars indicate means ± sd of six biological replicates. Asterisks indicate statistically significant differences from the control, according to Student’s t test (*, P < 0.05 and **, P < 0.01).

It was a bit confusing that mevastatin treatment increased meroterpenoid contents in R. dauricum leaves (Fig. 2). This result might be due to cross talk between the MEP and MVA pathways (Vranová et al., 2013); the MEP pathway could be activated when isoprenoid supply in the cytosol decreased to compensate the overall isoprenoid flux after mevastatin treatment, which could have resulted in an increase of meroterpenoid levels in the treated leaves. Nevertheless, this result does not contradict the possibility that the isoprenoid moiety of GFA is synthesized through the MEP pathway.

Molecular Cloning and Sequence Analysis of RdPT1

Plastid-localized aromatic PTs with specialized functions have an apparent structural similarity with homogentisate phytyltransferases involved in tocopherol biosynthesis (Yazaki et al., 2009). Thus, using a homogentisate PT as a query (AtVTE2-1, NCBI protein ID: AAM10489.1), we performed a tBLASTn similarity search against a young leaf transcriptome of R. dauricum and identified six putative aromatic PTs designated as RdPT1 to RdPT6. The coding sequences of RdPT cDNAs were unequivocally confirmed by sequencing after PCR amplifications as described in “Materials and Methods.” Among these sequences, the deduced primary structures of RdPT1 to RdPT5 had more than 25% overall sequence identity with AtVTE2-1, while RdPT6 had high identities (∼75%) to p-hydroxybenzoate polyprenyltransferases such as AtPPT-1, which is responsible for ubiquinone biosynthesis in mitochondria (Okada et al., 2004). The protein phylogenic analysis of polypeptide sequences with functionally characterized plant PTs provided a more comprehensive overview of the putative functions for each RdPT variant (Fig. 3; Supplemental Table S1). RdPT2 to RdPT6 were included in or positioned close to different PT clusters that participate in the primary metabolic pathway for tocopherol, tocotrienol, plastoquinone, chlorophyll, and ubiquinone biosynthesis, respectively. This result implies that these RdPT enzymes have individual primary functions in planta, as listed in Supplemental Table S2.

Figure 3.

Figure 3.

Open in a new tab

Phylogenetic tree representing the evolution of plant aromatic PTs. The scale indicates 0.1 amino acid substitutions per site. Bootstrap values are presented at each node. RdPT proteins are marked with red diamonds. Species abbreviations are as follows: Ap, Allium porrum; At, Arabidopsis thaliana; Cl, Citrus limon; Cp, Cuphea pulcherrima; Cr, Chlamydomonas reinhardtii; Cs, Cannabis sativa; Gm, Glycine max; Hl, Humulus lupulus; Hv, Hordeum vulgare; La, Lupinus albus; Le, Lithospermum erythrorhizon; Nt, Nicotiana tabacum; Os, Oryza sativa; Pc, Petroselinum crispum; Ps, Pastinaca sativa; Rd, Rhododendron dauricum; Sf, Sophora flavescens; Ta, Triticum aestivum; Zm, Zea mays. Accession numbers of these proteins are listed in Supplemental Table S1.

On the contrary, it is remarkable that RdPT1 was isolated from these primary PT clusters and formed a separate clade together with four recently identified coumarin-specific PT enzymes that mediate species-specific metabolism (Fig. 3): PcPT from parsley (Petroselinum crispum), PsPT1 and PsPT2 from parsnip (Pastinaca sativa), and ClPT from lemon (Karamat et al., 2014; Munakata et al., 2014, 2016). Within this clade, PcPT and PsPT1, both of which are umbelliferone 6-dimethylallyltransferases, are the nearest neighbors and close to PsPT2, which is an umbelliferone 8-dimethylallyltransferase, whereas ClPT, catalyzing the geranylation of umbelliferone at the C-8 position, is placed at a relatively distal branch in relation to RdPT1. GFA is structurally similar to the cannabinoid precursor cannabigerolic acid (Sirikantaramas and Taura, 2017), but none of the RdPTs were closely related to CsPT1, which is functional in the cannabinoid pathway (Page and Boubakir, 2014). Nevertheless, the phylogenetic analysis clearly suggested that RdPT1 is evolutionarily related to coumarin-specific PTs from Apiaceae and Rutaceae and is the most diversified PT member found in R. dauricum. Therefore, we focused on characterizing the structure and function of RdPT1 as a GFA-producing enzyme candidate.

The RdPT1 cDNA contained a 1,155-bp open reading frame encoding a polypeptide of 384 amino acids with a molecular mass of 42.5 kD. The primary structure of RdPT1 shared moderate (40%–48%) sequence identities with aromatic PTs including coumarin-specific enzymes, as shown in the protein sequence alignment (Supplemental Fig. S1). The online software programs ChloroP and iPSORT predicted that the first 54 amino acids constituted a putative N-terminal transit peptide for plastid localization (Fig. 4A). Thus, the mature peptide sequence of RdPT1 is presumed to be of 330 amino acids with a molecular mass of 36.2 kD. In addition, the RdPT1 amino acid sequence contained several transmembrane helices predicted by the TMHMM program (Fig. 4A). Homogentisate PT-related aromatic PTs have conserved first and second Asp-rich motifs (FARM and SARM) with consensus sequences of NQxxDxxID and KDxPDxxGD, respectively, which participate in prenyl diphosphate substrate binding via magnesium ions (Yazaki et al., 2009). RdPT1 also had these conserved regions at the standard positions (Supplemental Fig. S1), but notably, the sequence corresponding to the FARM was NALYDIEID, in which the second amino acid was mutated uniquely from Gln to Ala (Fig. 4B). These structural features are summarized in Figure 4 and suggest that RdPT1 is a membrane-bound aromatic PT that is structurally similar to known PT enzymes. The DCA pathway is constitutively active in R. dauricum (Iijima et al., 2017); accordingly, the responsible PT gene should be highly and constitutively expressed. Indeed, the calculated fragments per kilobase of transcript per million fragments mapped value for the RdPT1 coding sequence in the young leaf transcriptome was 88.8, which was similar to those reported for the previously characterized enzymes involved in the DCA pathway, DCA synthase and orcinol synthase (74.7 and 81.9, respectively; Iijima et al., 2017).

Figure 4.

Figure 4.

Open in a new tab

Structural features of RdPT1 conserved protein motifs. A, Schematic representation of the primary structure of RdPT1. a.a., Amino acids; TM, transmembrane helix; TP, transit peptide. B, Amino acid sequences corresponding to the FARM and SARM. Conserved amino acids are shaded black, dark gray, gray, or light gray depending on the degree of identity. Black shading indicates completely conserved residues. Dark gray, gray, and light gray shading indicate residues conserved in more than five, more than three, and two proteins, respectively. The unique Ala-149 in RdPT1 is highlighted in red. Protein abbreviations and accession numbers are as presented in Supplemental Table S1.

Tissue-Specific Expression of RdPT1 in R. dauricum Plants

R. dauricum is a typical lepidote Rhododendron species bearing numerous glandular scales on both the abaxial and adaxial leaf surfaces (Supplemental Fig. S2; Desch, 1983). The scales are multicellular structures with an epidermal origin that consist of a stalk attached to the leaf epidermis and a circle-shaped expanded cap composed of central and rim cells (Desch, 1983). A previous study reported that DCA synthase has a characteristic distribution pattern; it is expressed specifically in the glandular scales of young leaves rather than the leaf body or other plant organs (Iijima et al., 2017). Thus, we studied the tissue distribution of RdPT1 as a candidate for the penultimate enzyme in the DCA pathway. The expression level of the RdPT1 gene was evaluated by semiquantitative reverse transcription (RT)-PCR using gene-specific primers. This tissue-specific expression analysis demonstrated that the RdPT1 gene was expressed at high levels in young leaves and in trace amounts in other organs (Fig. 5A). In addition, a corresponding band was detected exclusively in samples from leaf scales and was negligible in samples from young leaves that had their scales removed (Fig. 5A), demonstrating that the RdPT1 gene is expressed predominantly in the glandular scales of young leaves, as was reported for DCA synthase. The tissue distribution patterns of RdPT1 transcripts also agreed with the meroterpenoid content accumulation in each tissue, as we reported previously (Taura et al., 2014; Iijima et al., 2017).

Figure 5.

Figure 5.

Open in a new tab

Analyses of the tissue distribution of RdPT1 transcripts and GFA-producing PT activity in R. dauricum. A, Agarose gel electrophoresis of the RdPT1 gene fragment (124 bp) amplified by semiquantitative RT-PCR using gene-specific primers. The 18S rRNA gene fragment (125 bp) was amplified as a housekeeping control gene. Lane numbers indicate young leaves (1), mature leaves (2), twigs (3), flowers (4), roots (5), glandular scales (6), and leaves without scales (7). B, GFA-producing PT activity was assayed using crude protein extracts from each tissue sample and presented as means ± sd of three biological replicates. Numbers indicate the same plant tissues as in A. ND, Not detected.

Furthermore, enzyme assays using crude protein extracts indicated that the GFA-producing PT activity also was localized selectively to glandular scales. Scale extracts showed higher GFA-producing activity than whole leaf extracts, while the activity of extracts from leaves that had their scales removed was under the detection limit (Fig. 5B). These results highlight that the glandular scale is the site for meroterpenoid biosynthesis in R. dauricum plants.

Subcellular Localization of RdPT1

Because in silico sequence analysis suggested a putative transit peptide at the N terminus of RdPT1, a GFP fusion study was conducted to verify the function of this region and to estimate the subcellular localization of RdPT1. We prepared a protein construct that consisted of 80 amino acids of the N-terminal end fused to GFP (TP-GFP), which was expressed transiently in young leaves of Nicotiana benthamiana by agroinfiltration (Karamat et al., 2014). Then, protoplasts were prepared from the leaves 4 d after infiltration and the GFP fluorescence was observed by confocal laser microscopy. As shown in Figure 6, the control cell expressing free GFP showed clear fluorescence in the cytoplasm and nucleus, which is a typical localization pattern of free GFP and did not colocalize with chlorophyll autofluorescence. In contrast, the TP-GFP fusion protein was observed exclusively in chloroplasts, as the GFP fluorescence colocalized within the same organelles where the chlorophyll signal was observed; however, the green and red fluorescence signals had partly different localizations within the chloroplasts (Fig. 6). TP-GFP appeared to localize to the stroma, as this fusion protein did not contain any transmembrane regions, whereas chlorophylls appeared to be anchored in the thylakoid membranes (Köhler et al., 1997). We have demonstrated that the N-terminal domain of RdPT1 functions as a transit peptide and that RdPT1 most likely localizes in plastids in planta. This result also suggests that RdPT1 is a reasonable candidate for the GFA-producing PT that utilizes isoprenoid precursors supplied from the MEP pathway in plastids (Vranová et al., 2013). Based on the tissue and subcellular localization results, RdPT1 would function in the plastids of scale cells.

Figure 6.

Figure 6.

Open in a new tab

Subcellular localization of the RdPT1 N-terminal transit peptide fused with GFP (TP-GFP). The fusion protein and free GFP were expressed transiently in N. benthamiana by agroinfiltration, and the fluorescence was observed with confocal microscopy in protoplasts from infiltrated leaves. The top row illustrates the cell expressing free GFP, and the bottom row shows the fluorescence of TP-GFP. Bars = 10 μm.

Heterologous Protein Expression in the Yeast Pichia pastoris

The gene that encodes the mature form of RdPT1 lacking the first 54 amino acids was amplified by PCR and subcloned into the pPICZA expression vector. The expression cassette was then introduced into the P. pastoris strain X-33 by electroporation. The transgenic P. pastoris was cultured in minimal liquid medium, and adding methanol to the cultures induced protein expression. An enzyme assay using the microsomal proteins of the induced cultures produced a sole product peak from OSA and FPP during HPLC analysis with a retention time identical to that of authentic GFA (3-farnesyl OSA; Fig. 7A). This reaction product was confirmed further to be GFA by comparing its mass spectrometry (MS) and UV spectra with those of standard samples using liquid chromatography (LC)-photodiode array (PDA)-electrospray ionization (ESI)-MS analyses (Fig. 7, B and C). Hence, we concluded that the RdPT1 reaction proceeds regioselectively to synthesize GFA as a single product. In addition, the membrane-bound nature of RdPT1 also was confirmed, because the soluble protein fraction of the same P. pastoris culture did not show GFA-producing activity (Fig. 7A). Consequently, the RdPT1 cDNA obtained in this study encoded the active PT, with the formal name of FPP: OSA 3-farnesyltransferase. RdPT1 appears to be a unique enzyme that catalyzes the farnesylation of an aromatic compound involved in plant specialized metabolism.

Figure 7.

Figure 7.

Open in a new tab

Product analysis of the reaction catalyzed by enzyme solutions prepared from transgenic P. pastoris expressing RdPT1. A, HPLC elution profile of the GFA standard (top), the reaction mixture of microsomal proteins (middle), and the reaction mixture of soluble proteins (bottom). B, ESI-MS (negative mode) of the GFA standard (left) and GFA synthesized by the recombinant RdPT1 (right). C, UV spectra of the same samples as in B.

The RdPT1 reaction was dependent on MgCl2 with a saturated concentration of 2.5 mm, whereas EDTA completely abolished the RdPT1 activity (Supplemental Fig. S3). MnCl2 and CaCl2 also activated RdPT1, but their efficiencies were apparently lower than that of MgCl2 (Supplemental Fig. S3). The RdPT1 enzyme was active in a weak basic pH range, with maximum activity at pH 8.6 (Supplemental Fig. S4). A similar pH preference and metal ion dependency were reported in various plant aromatic PTs (Sasaki et al., 2011; Yang et al., 2018). In addition, we also examined the transformed P. pastoris microsomes expressing the full-length RdPT1, but there was less than 30% enzyme activity compared with the mature protein form. Therefore, for our biochemical studies, we used the mature form of RdPT1 suspended in Tris-HCl buffer at pH 8.6 that contained 5 mm MgCl2.

Substrate Specificity of RdPT1 and Its A149Q Mutant

The substrate specificity of RdPT1 was analyzed with enzyme assays that used all combinations of the nine aromatic substrates (nos. 1–9; Fig. 8) and five prenyl diphosphates with different chain lengths from DMAPP (C5) to geranylfarnesyl diphosphate (C25; Fig. 8). RdPT1 was highly selective for OSA (1) as a prenyl acceptor substrate, as it did not react with structurally similar polyketides such as orcinol (2) and olivetolic acid (3; Fig. 8). Phloroacetophenone (4) and umbelliferone (5), which are minor components in R. dauricum (Cao et al., 2004), also were not suitable substrates for RdPT1. Likewise, RdPT1 did not accept phenylpropanoid-related polyphenols (6–8) or homogentisate (9), which is the substrate for identified ancestral enzymes such as AtVTE2-1. On the other hand, we observed that RdPT1 had relatively broad specificity for the prenyl donor substrates; this enzyme also exhibited detectable activities for GPP and GGPP under the standard assay conditions (Fig. 8), which produced GFA analogs with different prenyl side chains, cannabigerorcinic acid (3-geranyl OSA) and 3-geranylgeranyl OSA, respectively (Supplemental Figs. S5 and S6). Such a relaxed substrate preference that can accept three different prenyl donors is a novel feature for RdPT1 as a secondary metabolic PT with plastid localization but seemed to be quite reasonable for a PT involved in secondary metabolism in R. dauricum. As reported previously, R. dauricum plants contain these prenyl analogs as minor constituents along with GFA, which are expected to be side products of aromatic PT implicated in GFA biosynthesis (Iijima et al., 2017).

Figure 8.

Figure 8.

Open in a new tab

Substrate specificity of RdPT1. A, Relative enzyme activity (%) of RdPT1 using combinations of aromatic and prenyl substrates. Enzyme assays were conducted under standard assay conditions as described in “Materials and Methods.” Data are means ± sd of three technical replicates. One hundred percent activity was measured as 70.7 pmol s−1 mg−1 microsomal protein. Dashes indicate no activity detected. B, Structures of the aromatic substrates listed in A.

Using the microsomal protein fraction, we next analyzed the kinetic constants of RdPT1 reactions with OSA and with three prenyl donor substrates to evaluate the affinity and catalytic rate for these substrates (Table 1). RdPT1 had an apparent Km value of 47 μm for OSA, which was severalfold higher than that for FPP but was within the range of Km values for plant PTs and their physiological aromatic substrates (Karamat et al., 2014; Munakata et al., 2016). These data suggest that OSA is likely a suitable prenyl acceptor for RdPT1 reactions, as this enzyme had a relatively high catalytic rate with this substrate (Table 1). With respect to the prenyl donor substrates, RdPT1 unexpectedly exhibited a lower Km value with GGPP (4.16 μm) when compared with FPP (10 μm), whereas its Km with GPP was higher (224 μm; Table 1), suggesting that the binding affinity of RdPT1 for GGPP is higher than those for FPP and GPP. In contrast, RdPT1 had the highest catalytic rate with FPP, followed by GPP, while it was relatively low with GGPP (Table 1). Taken together, the expected catalytic efficiency was highest for RdPT1 with FPP (Table 1), suggesting that FPP is the best-suited prenyl substrate for RdPT1. On the other hand, it also was noted that RdPT1 had the highest affinity for GGPP. Thus, this enzyme might have a hydrophobic pocket large enough to accept the long GGPP hydrocarbon chain, but the binding mode might be somewhat unsuitable for catalysis.

Table 1. Steady-state kinetic parameters of the recombinant RdPT1 and the A149Q mutant enzyme.

Data are presented as means ± sd of three technical replicates.

Enzyme Substrate (Cosubstrate) Apparent Affinity (Km) Catalytic Rate (Vmax) Catalytic Efficiency (Vmax/Km)
μm pmol s−1 mg−1 microsomal protein pmol s−1 mg−1 microsomal protein μm−1
RdPT1 OSA (FPP) 47.0 ± 7.4 140 ± 9 3.00 ± 0.27
GPP (OSA) 224 ± 44 36.9 ± 5.6 0.17 ± 0.01
FPP (OSA) 10.0 ± 2.7 95.0 ± 2.8 9.94 ± 2.61
GGPP (OSA) 4.16 ± 2.11 2.26 ± 0.15 0.65 ± 0.33
RdPT1 A149Q OSA (FPP) 90.5 ± 5.1 45.1 ± 1.1 0.47 ± 0.02
GPP (OSA) 247 ± 49 8.63 ± 1.22 0.035 ± 0.003
FPP (OSA) 9.70 ± 3.82 39.6 ± 3.1 4.71 ± 1.67
GGPP (OSA) 33.7 ± 5.0 1.08 ± 0.04 0.032 ± 0.004
Open in a new tab

RdPT1 exhibited novel prenyl donor preferences, in addition to being an OSA-specific plant PT. To understand these unique features, we next examined the effect of amino acid substitution in the FARM of RdPT1, where a conserved Gln is substituted to Ala-149 (Fig. 4). We constructed a mutant gene by site-directed mutagenesis and prepared microsomes expressing the A149Q mutant enzyme in a manner similar to the wild-type enzyme. The expression level of the mutant enzyme was comparable with that of the wild type based on immunoblot analysis of microsomal proteins using an anti-RdPT1 peptide antibody; the recombinant proteins were of suitable molecular sizes (approximately 35 kD) for the mature form of the RdPT1 protein (Supplemental Fig. S7). Enzyme assays confirmed that the substrate selectivity of the A149Q mutant was not different from that of the wild type; however, kinetic analyses using each substrate provided an interesting insight. As shown in Table 1, the Km values for RdPT1 with OSA, GPP, and FPP were similar to those of wild-type samples. However, the Km with GGPP was 8-fold higher in the A149Q mutant compared with the wild type (Table 1), suggesting that the affinity between the RdPT1 enzyme and GGPP was impaired considerably by the Ala-to-Gln mutation in the FARM region. Therefore, we concluded that Ala-149, which is unique in RdPT1, is partly responsible for the relaxed substrate preference of RdPT1, which allows it to accept GGPP with a long prenyl chain. Moreover, the catalytic rates of the A149Q mutant for substrates were appreciably lower than those of the wild-type enzyme (Table 1). Therefore, the A149Q mutation is not critical for enzyme function but might negatively affect the catalytic action of RdPT1.

Homology Modeling of RdPT1

We hypothesized that, besides an aromatic binding site, a hydrophobic pocket could reside in RdPT1 to accommodate large prenyl substrates such as FPP and GGPP, based on its biochemical properties. To confirm this hypothesis, a homology model for RdPT1 was prepared using the crystal structure of a bacterial protein from Aeropyrum pernix, ApUbiA (Cheng and Li, 2014). ApUbiA is a p-hydroxybenzoate polyprenyltransferase that shares ∼19% overall amino acid identity with RdPT1. Although the homology is not very high, we considered ApUbiA as a suitable modeling template because, similar to RdPT1, this enzyme is a membrane protein composed of nine transmembrane helices. In addition, the reported ApUbiA structure in complex with aromatic and prenyl ligands (PDB ID: 4OD5) facilitated the deduction of the substrate-binding cavity in RdPT1, as described below.

The RdPT1 structure was modeled in a range from the first to the eighth putative transmembrane helices, of which the overall fold resembled that of ApUbiA (Supplemental Fig. S8). The structural model suggested that the conserved Asp residues in the FARM and SARM are clustered in the putative active site cavity, as in the case of ApUbiA, suggesting their suitable positioning. In addition, it appeared that the unique Ala-149 also constituted the active site cavity (Supplemental Fig. S8). Hence, we next attempted docking simulations between protein and ligands, and the complex structures are presented as closeup views of the putative active site cavity (Fig. 9).

Figure 9.

Figure 9.

Open in a new tab

Substrate-docking simulation of RdPT1. A to C, Putative substrate-binding cavities docked with FPP (A), GPP (B), and GGPP (C) in the presence of OSA and magnesium ions, which are shown in the closeup views. In each part, OSA is shown in yellow and prenyl diphosphate in light orange as a stick model. Magnesium ions are presented as dark gray spheres. Only selected amino acid residues that are predicted to participate in substrate binding are shown as white stick models, with nitrogen atoms colored blue and oxygen atoms colored red. Hydrogen bonds are shown as green dotted lines. The distance between OSA C-3 and prenyl diphosphate C-1 is indicated with a red dotted line, while that between each terminal carbon of prenyl diphosphate and Ala-149 is shown with a blue dotted line. D, Protein surface representation of the active site cavity docked with GGPP and OSA as in C.

In general, the RdPT1 model could accept OSA and FPP reasonably within the putative aromatic and hydrophobic binding sites (Fig. 9A), respectively, as reported in ApUbiA (Cheng and Li, 2014). First, in the energy-minimized complex, OSA was surrounded by the amino acids Asn-141, Thr-144, and Arg-236 (Fig. 9A). Among these residues, Arg-236 interacted with the carboxyl group of OSA via two hydrogen bonds that direct the orientation of OSA; the reactive C-3 position faces the center of the active site (Fig. 9A). As expected, the diphosphate moiety of FPP protruded to the active site floor, implying an interaction with the conserved Asp residues via magnesium ions (Fig. 9A). In addition, the two basic residues, Arg-163 and Lys-280, seemed to stabilize the diphosphate moiety via multiple hydrogen bond formations (Fig. 9A). Notably, the interaction between the diphosphate moiety and the basic residues also was reported for the ApUbiA ligand complex (Cheng and Li, 2014). The farnesyl group of FPP interacted with two distinct wall-like hydrophobic regions in the active site cavity. The first hydrophobic wall consisted of Ile-100, Gly-145, and Ala-149, whereas the second one was composed of Ile-232, Phe-276, and Ala-277 (Fig. 9A). Thus, Ala-149 probably contributes to the hydrophobic environment in the active site, although the distance between the methyl carbon of Ala-149 and the terminal carbon of FPP was not very close (7.7 Å; Fig. 9A). With respect to the reaction possibility, the FPP C-1 atom was set close to the OSA C-3 atom (4.8 Å), which is reasonable for C-C bond formation between these atoms during the biosynthesis of GFA.

We next prepared RdPT1 models in complex with prenyl substrates other than FPP. First, the docked GPP protrudes straight to the first hydrophobic wall to make a close interaction with Ile-100 and Gly-145, whereas this substrate is oriented distally to the second hydrophobic wall (Fig. 9B). The distance between GPP C-1 and OSA C-3 was 5.1 Å (Fig. 9B), which was similar to that between FPP and OSA (Fig. 9A), indicating a similar reaction potential. On the other hand, the docking energy of GPP to the active site was calculated to be −673 kcal mol−1, which was much higher than that of FPP (−1,200 kcal mol−1). These in silico results may be reasonable predictions for substrate interactions because RdPT1 has a relatively low affinity for GPP, judging from the Km value (Table 1).

The docking simulation also suggested that RdPT1 could accept the bulky substrate GGPP (Fig. 9C). The hydrocarbon moiety of GGPP was suitably folded to be set close to amino acid residues in the first and second putative hydrophobic walls. The protein surface model clearly illustrated that GGPP occupied the entire substrate-binding cavity (Fig. 9D). It also was noted that the docking energy for GGPP was calculated to be the lowest (−1,628 kcal mol−1) among the prenyl substrates we tested. These results suggest a tight interaction between GGPP and the hydrophobic binding walls of RdPT1 and were consistent with the small Km value RdPT1 had for GGPP (Table 1). Furthermore, it also was of interest that the terminal methyl group of GGPP was relatively proximal to Ala-149 (5.9 Å; Fig. 9C). We suggest that the increase in the Km value observed for the A149Q mutant enzyme probably was due to steric and electrostatic hindrance during GGPP binding by the hydrophilic Gln residue introduced in the first hydrophobic wall. Moreover, in the docking model, GGPP C-1 was distal (7.5 Å) from the reactive C-3 atom of OSA (Fig. 9C), suggesting a lower reactivity with GGPP, which also is consistent with the relatively low catalytic rate RdPT1 exhibited for this prenyl substrate (Table 1).

DISCUSSION

Molecular and Biochemical Characterization of RdPT1

Plant meroterpenoids also are referred to as prenylated polyphenols, which are hybrid natural products composed of isoprenoid and aromatic moieties. Versatile aromatics, such as flavonoids, isoflavonoids, stilbenoids, and coumarins, are employed as phenolic cores in a species-specific manner and can be decorated with linear or cyclized isoprenoid moieties with different chain length and position selectivity, leading to synergetic increases in the chemical diversity of meroterpenoid natural products (Bourgaud et al., 2006; Veitch and Grayer, 2011; Rivière et al., 2012; Veitch, 2013). During their biosynthesis, aromatic PTs play key roles in the construction of carbon skeletons by coupling phenolics from the polyketide or shikimate pathways with isoprenoids from the MEP or MVA pathways (Yazaki et al., 2009). More than a dozen aromatic PTs with specialized functions have been identified, most of which are evolutionarily derived from homogentisate PTs and localized in plastids, where they can exploit isoprenoids from the MEP pathway (Li et al., 2015; Munakata et al., 2016; Yang et al., 2018). It also is notable that all specialized PTs prefer either DMAPP or GPP as prenyl donor (Munakata et al., 2014), consistent with the fact that farnesylated meroterpenoids are fairly rare as plant natural products. For example, only a few farnesylated flavonoids have been reported to date (Veitch and Grayer, 2011), while at least 700 prenylated flavonoids have been isolated from plants (Barron and Ibrahim, 1996). Therefore, the identification of PT enzymes with novel prenyl donor selectivity, as was expected for the PT involved in the DCA pathway, would facilitate structure-function studies of this class of enzymes and could be an invaluable step in the metabolic engineering of plants with unnatural meroterpenoid composition.

In this study, we cloned the cDNA coding for RdPT1, which is evolved for GFA biosynthesis, from young leaves of R. dauricum. RdPT1 catalyzes the farnesylation of aromatic compounds in plant secondary metabolism processes. Despite its catalytic activity, RdPT1 shared various structural features with known plant PTs. The primary structure contains an N-terminal transit peptide for plastid localization, several transmembrane helices, and the two Asp-rich motifs, FARM and SARM. However, we noticed a small difference in the RdPT1 protein sequence: the second residue of the FARM was Ala instead of the highly conserved Gln that is at this position in other aromatic PTs reported thus far. One exception in the previous literature was reported in Sophora flavescens isoliquiritigenin dimethylallyltransferase, in which the corresponding position is Glu (Sasaki et al., 2011). However, the change from a large hydrophilic (Gln) to a small hydrophobic (Ala) amino acid at this position was unique to RdPT1 and encouraged us to study the active site architecture of RdPT1 in this study.

As we reported previously, R. dauricum plants produce DCA as the major constituent, whereas it also contained DCA analogs with different isoprenoid chain lengths (i.e. C10 and C20) as minor constituents (Iijima et al., 2017). In addition, their precursors also were identified in young leaves along with GFA. In this study, we showed that RdPT1 is responsible for the production of meroterpenoids in R. dauricum. RdPT1 has promiscuous specificity to prenyl donor substrates and produced three OSA derivatives with different prenyl side chains (Fig. 1). In addition, our kinetic study demonstrated that this enzyme has a high catalytic efficiency for FPP and produces GFA as its major reaction product, which is the predominant biosynthetic flow leading to the high accumulation of DCA observed in R. dauricum plants (Taura et al., 2014; Iijima et al., 2017). It is of interest that RdPT1 and DCA synthase belong to different biosynthetic enzyme categories that underwent convergent evolution to produce DCA as a major component, along with two prenyl analogs that might have physiological functions as minor components in R. dauricum.

In this study, the homology modeling and docking simulations provided valuable information toward understanding the unique substrate specificities of RdPT1 at the protein structure level. For example, the regioselective nature of RdPT1 reactions could be explained by the docking mode of OSA in the putative aromatic binding site with suitable orientation for prenylation at its C-3 position. It also was interesting that the prenyl donor preferences of RdPT1 were reasonably interpreted by the docking energy and docking mode of each prenyl substrate. Furthermore, the modeling study also suggested that the unique Ala-149 in RdPT1 could interact closely with GGPP, consistent with the increased Km value selectively of the A149Q mutant for GGPP. Consequently, the consecutive agreements between the biochemical and structural features of RdPT1 support our protein model structures. Further mutational studies of the putative aromatic binding site and hydrophobic walls will provide more detailed information about the unique substrate selectivity of RdPT1.

Plant aromatic PTs are membrane-bound proteins and, thus, difficult targets for structural elucidation based on protein crystallographic analyses. The molecular modeling study we conducted may be a practical approach for obtaining insights into the structure-function relationships of aromatic PTs, many of which produce pharmacologically valuable prenylated polyphenols (Yazaki et al., 2009). For example, the molecular modeling studies of a group of coumarin-specific PTs also might provide interesting information, because currently characterized coumarin PTs have variable regioselectivity and prenyl donor preferences (Karamat et al., 2014; Munakata et al., 2014, 2016). Furthermore, O-farnesylated as well as C-farnesylated coumarins have been reported as plant natural products (Miski and Jakupovic, 1990; Gliszczyńska and Brodelius, 2012), suggesting that there are additional unidentified PTs with novel catalytic functions in their biosynthesis. In this way, coumarin-specific PTs may be the most diversified group of aromatic PTs involved in plant secondary metabolism, along with RdPT1, which has a close evolutionary relationship with coumarin PTs.

It is noteworthy that GFA also is a fungal secondary metabolite (Geris and Simpson, 2009), while farnesylated natural aromatic substances are rather rare in plants, and the gene for GFA-producing PT, StbC, was identified recently in Stachybotrys bisbyi (Li et al., 2016). StbC catalyzes the same reaction as RdPT1; however, it is very difficult for us to compare the structural features of these enzymes because StbC shares very low sequence similarity with RdPT1 (Li et al., 2016).

Glandular Scales Are the Meroterpenoid-Producing Factories in R. dauricum

With respect to the tissue distribution of RdPT1, we demonstrated that this enzyme is expressed mainly in the glandular scales of young leaves. This result is consistent with the previous observation that DCA synthase, which works together with RdPT1, also is localized to glandular scales (Iijima et al., 2017). Therefore, we suggest that the glandular scales are the specialized factories for the biosynthesis of meroterpenoids, including DCA, which is an anti-HIV principal component that is produced effectively in R. dauricum. We also reported previously that DCA accumulates in the apoplastic space of glandular scales, suggesting an extracellular biosynthesis of DCA based on the fact that DCA synthase is a secreted FAD oxidase (Iijima et al., 2017).

Many plant natural products are biosynthesized in specialized tissues, and the final products, especially for lipophilic metabolites, often are sequestered in apoplastic spaces like the secretory cavities in glandular trichomes (Lange, 2015; Yazaki et al., 2017). Such an accumulation of secondary metabolites in the apoplastic compartments at the plant surface would contribute to self-defense mechanisms, because many specialized metabolites exhibit defense-related activities (Bennett and Wallsgrove, 1994; Wittstock and Gershenzon, 2002). For example, furanocoumarins, a group of meroterpenoids derived from prenylated coumarins, are antiherbivore metabolites (Zangerl and Berenbaum, 1990) that accumulate in oil ducts of herbs in the Apiaceae family and oil glands in fruit skin of the Citrus spp. belonging to Rutaceae (Voo et al., 2012; Karamat et al., 2014). In addition, furanocoumarins also are toxic to plants (Araniti et al., 2015) and need to be sequestered from cells to avoid self-poisoning. These observations hold true for meroterpenoid biosynthesis in R. dauricum. GFA and DCA exhibit nonselective antibacterial and antifungal activities (Hashimoto et al., 2005; Okada et al., 2017), suggesting that these meroterpenoids can participate in antimicrobial defense at the plant surface. In addition, we also have reported that they are phytotoxic metabolites that induce cell death in R. dauricum cell cultures with potency similar to that of camptothecin, which can trigger programmed cell death (Taura et al., 2018). It is reasonable, therefore, that DCA is biosynthesized and stored in the apoplast of specialized tissues and that GFA has to be effectively secreted once it is synthesized by RdPT1 in the plastids of R. dauricum scale cells.

Many hydrophobic specialized metabolites, such as terpenoids and meroterpenoids, are secreted to specific extracellular compartments or the rhizosphere, although only a few studies have reported the transport mechanism of these natural products (Shitan, 2016). For example, a recent study demonstrated that shikonin, a naphthoquinone meroterpenoid, is secreted from L. erythrorhizon hairy roots through pathways common to the ADP-ribosylation factor/guanine nucleotide-exchange factor system and actin filament polymerization (Tatsumi et al., 2016). In addition, a fungitoxic diterpenoid sclareol is secreted from Nicotiana spp. by the ATP-binding cassette transporters NpPDR1/NpABC1 and NtPDR1 (Hwang et al., 2016). However, it is still unclear whether GFA is secreted from scale cells by a similar molecular mechanism, as GFA is not structurally similar to sclareol or shikonin. Thus, additional research will be required to clarify the machinery mediating GFA transportation from plastids to the apoplast.

A possible intracellular metabolic flow of intermediates included in the DCA pathway is illustrated in Figure 10. We suggest that the aromatic precursor OSA is synthesized in the cytosol and then enters the plastids, because plant polyketides generally are produced in the cytosol (Winkel-Shirley, 2001; Gagne et al., 2012). Indeed, orcinol synthase, which is the polyketide synthase in this pathway, does not contain sorting signals (Taura et al., 2016). The biosynthetic mechanism of FPP is a new topic explored in this study; we suggest that the isoprenoid moiety of GFA originated from the MEP pathway, whereas most FPP synthases characterized thus far are located in the cytosol and the provided FPP is the product of the MVA pathway, similar to yeast and mammals (Vranová et al., 2013). At present, we cannot exclude the possibility that FPP is synthesized by FPP synthase in the cytosol, despite using isoprenoid precursors supplied from the MEP pathway in plastids. Such a dynamic cross talk of isoprenoid intermediates has been proposed for the artemisinin biosynthetic pathway: FPP synthase in sweet wormwood (Artemisia annua) synthesizes FPP for the artemisinin pathway using GPP supplied by GPP synthase in plastids (Schramek et al., 2010). However, it is a rather simple presumption that a plastid-localized FPP synthase has undergone convergent evolution to support meroterpenoid biosynthesis in R. dauricum. Studies on isoprenoid-producing enzymes in R. dauricum are now in progress to address these additional questions.

Figure 10.

Figure 10.

Open in a new tab

Model of the possible subcellular metabolic flow of the DCA pathway. GFA is produced by RdPT1 in the plastid of glandular scale cells and then transported to the apoplast for the subsequent DCA synthase reaction. The isoprenoid moiety of GFA originates from the MEP pathway, although the detailed mechanism of FPP biosynthesis in this pathway has yet to be clarified.

In conclusion, we conducted molecular and biochemical characterizations of RdPT1, which plays a key role in the DCA biosynthetic pathway and may contribute to the defense of R. dauricum at the plant surface by supplying GFA in the glandular scales. As a farnesyltransferase involved in a specialized plant metabolic pathway, the sequence and structural information of RdPT1 will be valuable for the structure-function studies of aromatic PT enzymes. In addition, it is still unknown why RdPT1, together with DCA synthase, drives three parallel biosynthetic routes to produce meroterpenoids with C10, C15, and C20 isoprenoid moieties. We argue that it is essential to conduct comprehensive studies on various biological activities of meroterpenoid constituents to establish their physiological importance in plants. This is especially true for understanding the insect-deterrent potential of DCA-related meroterpenoids, which are of interest because lepidote species of Rhododendron such as R. dauricum are more resistant to insect pests than elepidote species (Doss, 1984).

MATERIALS AND METHODS

Plant Materials and Reagents

Rhododendron dauricum plants were cultivated at the Experimental Station for Medicinal Plant Research at the University of Toyama. Nicotiana benthamiana plants were grown in a growth chamber at 22°C with a photoperiod of 16 h of light/8 h of dark. Meroterpenoid standards, DCA, GFA, cannabigerorcinic acid, and 3-geranylgeranyl OSA were from our laboratory collection (Iijima et al., 2017). Olivetolic acid was purchased from Santa Cruz Biotechnology. Olivetol, DMAPP, GPP, FPP, and GGPP were purchased from Sigma-Aldrich. Geranylfarnesyl diphosphate was synthesized chemically as described previously (Sato et al., 2013). Other chemical reagents were purchased from Wako Pure Chemicals unless stated otherwise.

Inhibitor Treatments and Meroterpenoid Analysis

Treatments with mevastatin and clomazone (each 1% (w/v) in distilled water) were performed on young shoot cuttings of R. dauricum plants at room temperature. Inhibitor solutions were replaced every 24 h. Control samples were treated with distilled water in a similar manner. After 20 d of incubation, the youngest leaf tissue was harvested from six independent shoot cuttings for each treatment. Then, methanol extract was prepared from respective samples, and the GFA and DCA levels were analyzed by HPLC as described previously (Taura et al., 2014). The data were calculated as means ± sd of six biological replicates. Statistical differences were analyzed by Student’s t test for two-group comparisons.

Transcriptome Screening of Candidate Genes

The transcriptome assembly data, which were composed of 71,944 cDNA contigs with an average length of 535 bp, were established previously based on RNA sequencing of R. dauricum young leaves (Iijima et al., 2017) and used here as a local cDNA database for our tBLASTn homology search. Six nucleotide sequences that encoded aromatic PTs were designated RdPT1 to RdPT6 and mined from the database using the amino acid sequence of AtVTE2-1 (NCBI protein ID: AAM10489.1) as a query. These cDNA contigs contained putative full-length coding sequences for the respective proteins and were used in the design of gene-specific primers for PCR (Supplemental Table S3).

cDNA Cloning of RdPT1

RdPT1 was defined as a PT in this study; therefore, its cDNA sequence was defined to include the untranslated regions based on 3ʹ- and 5ʹ-RACE reactions and subsequent PCR of the full-length coding sequence. The gene-specific oligonucleotide primers and reaction conditions are listed in Supplemental Table S3, which were used to amplify cDNA fragments. First, total RNA was extracted from young R. dauricum leaves using an RNAqueous kit following the manufacturer’s instruction (Thermo Fisher Scientific). Then, first-strand cDNAs were synthesized with SMARTScribe Reverse Transcriptase using the SMARTer RACE 5ʹ/3ʹ Kit and protocol (Clontech). The 5ʹ-RACE-ready cDNA was prepared with an oligo(dT) primer (5′-CDS ) in the presence of the SMARTer IIA Oligonucleotide, whereas an adapter-linked oligo(dT) primer (3′-CDS) was used to synthesize 3ʹ-RACE-ready cDNA. The 3ʹ-RACE reaction was conducted using the Advantage 2 polymerase mix with a gene-specific primer (RdPT1-3R) and the universal primer mix as indicated by the manufacturer (Clontech). The 5ʹ-end region was amplified with the RdPT1-5R primer and the universal primer mix. The RACE products were subcloned into the pMD19 T-vector (Takara Bio) and sequenced on a 3500 DNA sequencer (Thermo Fisher Scientific). Next, the coding sequences for full-length RdPT1 and the mature form without the coding sequence for the first 54 amino acids were amplified using KOD-plus Neo (Toyobo) with the respective gene-specific primer pairs, RdPT1-Fw and RdPT1-Rv (or RdPT1-mature-Fw and RdPT1-Rv). PCR products were subcloned into the pPICZA expression vector (Thermo Fisher Scientific), predigested with BspT104I and SalI (Takara Bio), using the In-Fusion HD cloning reagent (Clontech). In addition, the gene variant for the RdPT1 A149Q mutant was prepared using site-directed mutagenesis of the mature form of the coding sequence cloned into pPICZA with the mutational primers, A149Q-Fw and A149Q-Rv, and the KOD-plus mutagenesis kit (Toyobo). The coding sequences in the respective expression constructs were confirmed by nucleotide sequencing.

cDNA Cloning of RdPT2 to RdPT6

The respective gene-specific primer pairs, RdPT(2–6)-Fw and RdPT(2–6)-Rv, which anneal to the 5ʹ- and 3ʹ-untranslated regions adjacent to the start and stop codons of each RdPT gene (Supplemental Table S3), were used to amplify the coding sequences of RdPT2 to RdPT6 genes, respectively. The PCR products were subcloned into the pMD19 T-vector, and the nucleotide sequences were confirmed as described above. The structural features of RdPT2 to RdPT6 are listed in Supplemental Table S2.

Computational Sequence Analyses

The subcellular location and protein-sorting signal were predicted with the online programs ChloroP (http://www.cbs.dtu.dk/services/ChloroP/) and iPSORT (http://ipsort.hgc.jp). Protein transmembrane domains were scanned using TMHMM version 2.0 (http://www.cbs.dtu.dk/services/TMHMM/). Multiple alignments of protein sequences were carried out using the Clustal Omega program (http://www.ebi.ac.uk/Tools/msa/clustalo/). A neighbor-joining phylogenetic tree was drawn with 1,000 bootstrap tests using MEGA6.06 software (http://www.megasoftware.net). The p-distance method was used to compute the evolutionary distances with the option of complete deletion of gaps. The fragments per kilobase of transcript per million fragments mapped value for the RdPT1 coding sequence in the young leaf transcriptome was calculated as described previously (Iijima et al., 2017).

Expression Analysis of the RdPT1 Gene by Semiquantitative RT-PCR

Total RNA was isolated from R. dauricum young leaves, mature leaves, twigs, flowers, roots, glandular scales, and young leaves with the scales removed using an RNAqueous kit. Prior to RNA extraction, the glandular scales were collected from both sides of young leaves using adhesive tape as described previously (Iijima et al., 2017) and suspended in RNA extraction buffer with gentle sonication. First-strand cDNA was then synthesized from 0.5 μg of RNA from each sample as described above, except that a random hexamer primer (Toyobo) was used. The RdPT1 gene fragment (124 bp) was amplified with the primers RdPT1-RT-Fw and RdPT1-RT-Rv, whereas the 18S rRNA gene fragment (GenBank accession no. AB973224; 125 bp) was amplified with the 18S-RT-Fw and 18S-RT-Rv primers and used as a housekeeping gene control for the PCR protocol. The PCR products were resolved on a 2% (w/v) agarose gel and stained with ethidium bromide.

GFA-Producing Activity in R. dauricum Tissues

Crude protein extracts were prepared from fresh tissues by homogenization in protein extraction buffer (100 mm Tris-HCl [pH 8] containing 1 mm mercaptoethanol and 0.1% (v/v) Triton X-100) with a pestle and mortar. The glandular scales were suspended in protein extraction buffer and disrupted using a microtube BioMasher homogenizer (Nippi). The protein amount in each extract was determined with the Bradford method (Bradford, 1976) using bovine serum albumin as the standard. GFA-producing activity was measured as described below.

GFP Protein Fusion Construct Cloning and Agroinfiltration

The binary vector pBI121 with a GFP cassette (pBI121-GFP) was prepared previously (Iijima et al., 2017) and used for the protein fusion construction. The gene fragment that encoded the N-terminal 80 amino acids of RdPT1 was amplified with RdPT1-TP-Fw and RdPT1-TP-Rv primers and then subcloned into pBI121-GFP by the use of XbaI and BamHI restriction sites, to produce the pBI121-TP-GFP expression construct. The pBI121-GFP (control) and pBI121-TP-GFP constructs were introduced individually into Agrobacterium tumefaciens LBA4404 by electroporation. Bacterial cultures were grown, collected by centrifugation, and resuspended in 10 mm MES (pH 5.6) containing 10 mm MgCl2 and 200 µm acetosyringone to a cell density of OD600 = 0.6. These suspensions were then incubated for 2 h. The A. tumefaciens C5851 strain containing the pBIN61-p19 vector (Karamat et al., 2014) also was cultured and treated similarly before it was mixed in equal parts with either of the LBA4404 strains. Then, this bacterial mixture was infiltrated with a syringe into the lower surface of N. benthamiana leaves. After 96 h of incubation, protoplasts were prepared from the infiltrated leaves using an enzyme solution of 20 mm MES (pH 5.6), 5 mm MgCl2, 0.6 m sorbitol, 1% (w/v) Cellulase Onozuka RS (Yakult), and 0.5% (w/v) Macerozyme R-10 (Yakult). Confocal images were taken with an LSM 700 laser-scanning fluorescence microscope (Carl Zeiss). The GFP fluorescence was detected using excitation at 488 nm and 505- to 530-nm emission laser lines; the chlorophyll signal was excited at 488 nm and collected with a long-pass filter > 615 nm.

Heterologous Expression of RdPT1 in P. pastoris

The pPICZA expression constructs containing the RdPT1 genes were introduced individually to P. pastoris strain X-33 (Thermo Fisher Scientific) by electroporation. The transformants were selected for on YPD agar plates containing 0.5 mg mL−1 zeocin (Invivogen). The three minimal media used herein contained 200 mm potassium phosphate buffer (pH 6), 1.34% (w/v) yeast nitrogen base (Difco Laboratories), and 4 × 10−5% (w/v) d-biotin and differed with respect to the carbon source, which was 10 g L−1 Glc, 1% (v/v) methanol, or 5% (v/v) methanol for BMD, BMM2, or BMM10, respectively. A single colony was used to inoculate Erlenmeyer flasks containing 100 mL of BMD and cultivated at 25°C for 60 h. Then, 100 mL of BMM2 was added to the culture to initiate protein expression. We supplied 20 mL of BMM10 every 24 h after the induction. After 96 h of induction, the culture was centrifuged to obtain the cell pellet. The cells were washed twice with water and suspended in 20 mm potassium phosphate buffer (pH 7.5) that contained 1 m sorbitol, 1 mm EDTA, and 1 mm mercaptoethanol. Zymolyase-20T (Nacalai Tesque) was added at a final concentration of 1 mg mL−1 and incubated at 30°C for 1 h. Then, cells were disrupted by vigorous vortexing with glass beads. The homogenate was centrifuged at 10,000g for 5 min and at 100,000g for 1 h to pellet the microsomes. The microsomal pellet was washed once with 100 mm Tris-HCl buffer (pH 8) that contained 1 mm mercaptoethanol and then resuspended in the same buffer. Protein concentrations of the microsomal and soluble (i.e. 100,000g supernatant) fractions were determined with the Bradford method. Enzyme activity in each fraction was assayed as described below.

Standard Assay Conditions for RdPT1

The standard reaction mixture consisted of 100 mm Tris-HCl buffer (pH 8.6), 5 mm MgCl2, 100 μm OSA (or other aromatic substrate), 100 μm FPP (or other prenyl substrate), and 30 μg of protein from the microsomal fraction in a total volume of 100 μL. The reaction was incubated at 30°C for 10 min and arrested by adding 100 μL of methanol. The incubation time and protein amount were chosen to ensure the linearity of the reaction. The 10-μL aliquots were analyzed by HPLC and LC-PDA-ESI-MS to quantify and characterize the reaction products.

HPLC and LC-PDA-ESI-MS Analyses of the Reaction Products

The enzyme reaction products were analyzed and quantified in an HPLC system (Tosho) equipped with a Cosmosil 5C18-MS-II column (4.6 mm × 150 mm; Nacalai Tesque) as described previously (Taura et al., 2014). Elution was performed isocratically with aqueous acetonitrile that contained 0.1% (v/v) formic acid at a flow rate of 1 mL min−1. The acetonitrile concentration was set for each product: 75% (v/v) for GFA, 60% (v/v) for cannabigerorcinic acid, and 90% (v/v) for 3-geranylgeranyl OSA. The products were detected by absorption at 254 nm and quantified using calibration curves of the standard compounds. The samples also were analyzed with an LC-PDA-ESI-MS system (Thermo Fisher Scientific) that included an Accela 600 HPLC pump, an Accela PDA detector, and an LTQ-Orbitrap-XL ETD Hybrid Ion Trap-Orbitrap Mass Spectrometer. The column and mobile phase were the same as those used for the HPLC analysis. The ESI-MS (negative) and UV spectra were collected as described previously (Taura et al., 2016).

Enzyme Kinetics

The enzyme reactions were conducted in a similar manner to the standard assay conditions. To evaluate the reaction kinetics for OSA, enzyme activities were determined using six concentrations of OSA (25, 50, 100, 150, 200, and 300 μm) in the presence of 100 μm FPP. To obtain the kinetic parameters for each prenyl substrate, the reactions were conducted using varying concentrations of GPP, FPP, or GGPP (25, 50, 100, 150, 200, and 300 μm) in the presence of 100 μm OSA. The reaction products were quantified by HPLC, and the kinetic constants were calculated by fitting the velocity data at each substrate concentration to Hanes-Woolf plots. Experiments were conducted in three technical replicates.

Immunoblot Analysis of Pichia pastoris Microsomes Expressing RdPT1

A rabbit polyclonal antibody was raised against the synthetic peptide SEDWHLSDPKKENG that corresponds to the Ser-63 to Gly-76 sequence of RdPT1. This antibody was affinity purified by Eurofins Genomics. For immunoblot analysis, P. pastoris microsomal proteins were separated by SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Millipore). The membrane was incubated at room temperature for 1 h with the anti-RdPT1 primary antibody and then treated with a horseradish peroxidase-conjugated secondary antibody (MBL). The protein bands were visualized using a Peroxidase Stain DAB (3,3′-diaminobenzidine) Kit (Nacalai Tesque).

Homology Modeling of RdPT1

The homology model of RdPT1 was generated with the SWISS-MODEL software (https://swissmodel.expasy.org) using the crystal structure of ApUbiA in complex with p-hydroxybenzoate and a prenyl substrate analog, geranyl thiolodiphosphate (PDB ID: 4OD5) as a template. The model quality was validated with a Ramachandran plot using RAMPAGE (http://mordred.bioc.cam.ac.uk/∼rapper/rampage.php) to confirm that most amino acid residues were grouped in the energetically allowed regions. Molecular docking simulations were performed using Molecular Operating Environment software (MOE; Chemical Computing Group). The RdPT1 homology model conserved the putative active site cavity that is sterically capable of accommodating aromatic and prenyl substrates. Thus, p-hydroxybenzoate, geranyl thiolodiphosphate, and magnesium ions in the ApUbiA crystal structure were fitted to the corresponding positions of the active site cavity in RdPT1. Then, OSA replaced p-hydroxybenzoate and the docking was optimized with the ASEDock program in the MOE software, which assisted in the selection of the model with the most stable docking mode. Likewise, potential prenyl donor substrates, GPP, FPP, or GGPP, were introduced at the active site instead of geranyl thiolodiphosphate. The docking simulation was then performed for each prenyl substrate with the ASEDock program. The orientation with the lowest docking energy was selected for each substrate as the docking structure. All protein figures were rendered with PyMOL (http://www.pymol.org).

Accession Numbers

The nucleotide sequences encoding RdPT1 to RdPT6 were deposited into the GenBank/EMBL/DDBJ databases under accession numbers LC381857 to LC381862, respectively.

Supplemental Data

The following supplemental materials are available.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

Acknowledgments

We are grateful to Dr. T. Koeduka (Yamaguchi University) and I. Umebara (University of Toyama) for helpful discussions on plant isoprenoid biosynthesis. We sincerely thank H. Fujino, Y. Tatsuo, Y. Takao, and Y. Murakami at the Experimental Station for Medicinal Plant Research of the University of Toyama for R. dauricum plant breeding. We also thank Dr. D. Baulcombe (University of Cambridge) for providing the pBIN61-p19 vector and Dr. M. Takeshita (Miyazaki University) for N. benthamiana seeds.

Footnotes

1

This study was supported in part by Japan Society for the Promotion of Science (JSPS) KAKENHI Grant 15K07994 (to F.T.), by Ministry of Education, Culture, Sports, Science, and Technology KAKENHI Grants 17H05436 (to F.T.) and 17H05448 (to Te.S), by Grant-in-Aid for JSPS Fellows Number 17J10178 (to M.I.), and by a Grant-in-Aid for the Cooperative Research Project from the Institute of Natural Medicine, University of Toyama, in 2017 (to F.T.).

References

  1. Appendino G, Gibbons S, Giana A, Pagani A, Grassi G, Stavri M, Smith E, Rahman MM (2008) Antibacterial cannabinoids from Cannabis sativa: a structure-activity study. J Nat Prod 71: 1427–1430 [DOI] [PubMed] [Google Scholar]
  2. Araniti F, Mancuso R, Lupini A, Giofrè SV, Sunseri F, Gabriele B, Abenavoli MR (2015) Phytotoxic potential and biological activity of three synthetic coumarin derivatives as new natural-like herbicides. Molecules 20: 17883–17902 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Barron D, Ibrahim RK (1996) Isoprenylated flavonoids: a survey. Phytochemistry 43: 921–982 [Google Scholar]
  4. Bennett RN, Wallsgrove RM (1994) Secondary metabolites in plant defence mechanisms. New Phytol 127: 617–633 [DOI] [PubMed] [Google Scholar]
  5. Bourgaud F, Larbat HR, Doerper S, Gontier E, Kellner S, Matern U (2006) Biosynthesis of coumarins in plants: a major pathway still to be unravelled for cytochrome P450 enzymes. Phytochem Rev 5: 293–308 [Google Scholar]
  6. Bradford MM. (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 72: 248–254 [DOI] [PubMed] [Google Scholar]
  7. Bukhari SM, Ali I, Zaidi A, Iqbal N, Noor T, Mehmood R, Chishti MS, Niaz B, Rashid U, Atif M (2015) Pharmacology and synthesis of daurichromenic acid as a potent anti-HIV agent. Acta Pol Pharm 72: 1059–1071 [Google Scholar]
  8. Cao Y, Chu Q, Ye J (2004) Chromatographic and electrophoretic methods for pharmaceutically active compounds in Rhododendron dauricum. J Chromatogr B Analyt Technol Biomed Life Sci 812: 231–240 [DOI] [PubMed] [Google Scholar]
  9. Chen X, Mukwaya E, Wong MS, Zhang Y (2014) A systematic review on biological activities of prenylated flavonoids. Pharm Biol 52: 655–660 [DOI] [PubMed] [Google Scholar]
  10. Cheng W, Li W (2014) Structural insights into ubiquinone biosynthesis in membranes. Science 343: 878–881 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Desch C., Jr (1983) The Rhododendron leaf scale. J Am Rhododendr Soc 37: 78–80 [Google Scholar]
  12. Doss RP. (1984) Role of glandular scales of lepidote rhododendrons in insect resistance. J Chem Ecol 10: 1787–1798 [DOI] [PubMed] [Google Scholar]
  13. Gagne SJ, Stout JM, Liu E, Boubakir Z, Clark SM, Page JE (2012) Identification of olivetolic acid cyclase from Cannabis sativa reveals a unique catalytic route to plant polyketides. Proc Natl Acad Sci USA 109: 12811–12816 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Geris R, Simpson TJ (2009) Meroterpenoids produced by fungi. Nat Prod Rep 26: 1063–1094 [DOI] [PubMed] [Google Scholar]
  15. Gliszczyńska A, Brodelius PE (2012) Sesquiterpene coumarins. Phytochem Rev 11: 77–96 [Google Scholar]
  16. Han M, Heppel SC, Su T, Bogs J, Zu Y, An Z, Rausch T (2013) Enzyme inhibitor studies reveal complex control of methyl-D-erythritol 4-phosphate (MEP) pathway enzyme expression in Catharanthus roseus. PLoS ONE 8: e62467. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Hashimoto T, Quang DN, Nukada M, Asakawa Y (2005) Isolation, synthesis and biological activity of grifolic acid derivatives from the inedible mushroom Albatrellus dispansus. Heterocycles 65: 2431–2439 [Google Scholar]
  18. Hwang JU, Song WY, Hong D, Ko D, Yamaoka Y, Jang S, Yim S, Lee E, Khare D, Kim K, et al. (2016) Plant ABC transporters enable many unique aspects of a terrestrial plant’s lifestyle. Mol Plant 9: 338–355 [DOI] [PubMed] [Google Scholar]
  19. Iijima M, Munakata R, Takahashi H, Kenmoku H, Nakagawa R, Kodama T, Asakawa Y, Abe I, Yazaki K, Kurosaki F, et al. (2017) Identification and characterization of daurichromenic acid synthase active in anti-HIV biosynthesis. Plant Physiol 174: 2213–2230 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Iwata N, Wang N, Yao X, Kitanaka S (2004) Structures and histamine release inhibitory effects of prenylated orcinol derivatives from Rhododendron dauricum. J Nat Prod 67: 1106–1109 [DOI] [PubMed] [Google Scholar]
  21. Karamat F, Olry A, Munakata R, Koeduka T, Sugiyama A, Paris C, Hehn A, Bourgaud F, Yazaki K (2014) A coumarin-specific prenyltransferase catalyzes the crucial biosynthetic reaction for furanocoumarin formation in parsley. Plant J 77: 627–638 [DOI] [PubMed] [Google Scholar]
  22. Kashiwada Y, Yamazaki K, Ikeshiro Y, Yamagishi T, Fujioka T, Mihashi K, Mizuki K, Cosentino LM, Fowke K, Morris-Natschke SL, et al. (2001) Isolation of rhododaurichromanic acid B and the anti-HIV principles rhododaurichromanic acid A and rhododaurichromenic acid from Rhododendron dauricum. Tetrahedron 57: 1559–1563 [Google Scholar]
  23. Köhler RH, Cao J, Zipfel WR, Webb WW, Hanson MR (1997) Exchange of protein molecules through connections between higher plant plastids. Science 276: 2039–2042 [DOI] [PubMed] [Google Scholar]
  24. Lange BM. (2015) The evolution of plant secretory structures and emergence of terpenoid chemical diversity. Annu Rev Plant Biol 66: 139–159 [DOI] [PubMed] [Google Scholar]
  25. Lee KH. (2010) Discovery and development of natural product-derived chemotherapeutic agents based on a medicinal chemistry approach. J Nat Prod 73: 500–516 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Li C, Matsuda Y, Gao H, Hu D, Yao XS, Abe I (2016) Biosynthesis of LL-Z1272β: discovery of a new member of NRPS-like enzymes for aryl-aldehyde formation. ChemBioChem 17: 904–907 [DOI] [PubMed] [Google Scholar]
  27. Li H, Ban Z, Qin H, Ma L, King AJ, Wang G (2015) A heteromeric membrane-bound prenyltransferase complex from hop catalyzes three sequential aromatic prenylations in the bitter acid pathway. Plant Physiol 167: 650–659 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Miski M, Jakupovic J (1990) Cyclic farnesyl-coumarin and farnesyl-chromone derivatives from Ferula communis subsp. communis. Phytochemistry 29: 1995–1998 [Google Scholar]
  29. Munakata R, Inoue T, Koeduka T, Karamat F, Olry A, Sugiyama A, Takanashi K, Dugrand A, Froelicher Y, Tanaka R, et al. (2014) Molecular cloning and characterization of a geranyl diphosphate-specific aromatic prenyltransferase from lemon. Plant Physiol 166: 80–90 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Munakata R, Olry A, Karamat F, Courdavault V, Sugiyama A, Date Y, Krieger C, Silie P, Foureau E, Papon N, et al. (2016) Molecular evolution of parsnip (Pastinaca sativa) membrane-bound prenyltransferases for linear and/or angular furanocoumarin biosynthesis. New Phytol 211: 332–344 [DOI] [PubMed] [Google Scholar]
  31. Okada K, Ohara K, Yazaki K, Nozaki K, Uchida N, Kawamukai M, Nojiri H, Yamane H (2004) The AtPPT1 gene encoding 4-hydroxybenzoate polyprenyl diphosphate transferase in ubiquinone biosynthesis is required for embryo development in Arabidopsis thaliana. Plant Mol Biol 55: 567–577 [DOI] [PubMed] [Google Scholar]
  32. Okada M, Saito K, Wong CP, Li C, Wang D, Iijima M, Taura F, Kurosaki F, Awakawa T, Abe I (2017) Combinatorial biosynthesis of (+)-daurichromenic acid and its halogenated analogue. Org Lett 19: 3183–3186 [DOI] [PubMed] [Google Scholar]
  33. Page JE, Boubakir Z (2014) Aromatic prenyltransferase from Cannabis. November 11, 2014. Patent No. US 8884100 B2
  34. Rivière C, Pawlus AD, Mérillon JM (2012) Natural stilbenoids: distribution in the plant kingdom and chemotaxonomic interest in Vitaceae. Nat Prod Rep 29: 1317–1333 [DOI] [PubMed] [Google Scholar]
  35. Sasaki K, Tsurumaru Y, Yamamoto H, Yazaki K (2011) Molecular characterization of a membrane-bound prenyltransferase specific for isoflavone from Sophora flavescens. J Biol Chem 286: 24125–24134 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Sato T, Yamaga H, Kashima S, Murata Y, Shinada T, Nakano C, Hoshino T (2013) Identification of novel sesterterpene/triterpene synthase from Bacillus clausii. ChemBioChem 14: 822–825 [DOI] [PubMed] [Google Scholar]
  37. Schramek N, Wang H, Römisch-Margl W, Keil B, Radykewicz T, Winzenhörlein B, Beerhues L, Bacher A, Rohdich F, Gershenzon J, et al. (2010) Artemisinin biosynthesis in growing plants of Artemisia annua: a 13CO2 study. Phytochemistry 71: 179–187 [DOI] [PubMed] [Google Scholar]
  38. Shitan N. (2016) Secondary metabolites in plants: transport and self-tolerance mechanisms. Biosci Biotechnol Biochem 80: 1283–1293 [DOI] [PubMed] [Google Scholar]
  39. Sirikantaramas S, Taura F (2017) Cannabinoids: biosynthesis and biotechnological applications. In Chandra S, Lata H, ElSohly M, eds, Cannabis sativa L.: Botany and Biotechnology. Springer, Cham, Switzerland, pp 183–206 [Google Scholar]
  40. Sirikantaramas S, Morimoto S, Shoyama Y, Ishikawa Y, Wada Y, Shoyama Y, Taura F (2004) The gene controlling marijuana psychoactivity: molecular cloning and heterologous expression of ∆1-tetrahydrocannabinolic acid synthase from Cannabis sativa L. J Biol Chem 279: 39767–39774 [DOI] [PubMed] [Google Scholar]
  41. Sommer S, Severin K, Camara B, Heide L (1995) Intracellular localization of geranyl pyrophosphate synthase from cell cultures of Lithospermum erythrorhizon. Phytochemistry 38: 623–627 [Google Scholar]
  42. Tatsumi K, Yano M, Kaminade K, Sugiyama A, Sato M, Toyooka K, Aoyama T, Sato F, Yazaki K (2016) Characterization of shikonin derivative secretion in Lithospermum erythrorhizon hairy roots as a model of lipid-soluble metabolite secretion from plants. Front Plant Sci 7: 1066. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Taura F, Sirikantaramas S, Shoyama Y, Yoshikai K, Shoyama Y, Morimoto S (2007) Cannabidiolic-acid synthase, the chemotype-determining enzyme in the fiber-type Cannabis sativa. FEBS Lett 581: 2929–2934 [DOI] [PubMed] [Google Scholar]
  44. Taura F, Iijima M, Lee JB, Hashimoto T, Asakawa Y, Kurosaki F (2014) Daurichromenic acid-producing oxidocyclase in the young leaves of Rhododendron dauricum. Nat Prod Commun 9: 1329–1332 [PubMed] [Google Scholar]
  45. Taura F, Iijima M, Yamanaka E, Takahashi H, Kenmoku H, Saeki H, Morimoto S, Asakawa Y, Kurosaki F, Morita H (2016) A novel class of plant type III polyketide synthase involved in orsellinic acid biosynthesis from Rhododendron dauricum. Front Plant Sci 7: 1452. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Taura F, Iijima M, Kurosaki F (2018) Daurichromenic acid and grifolic acid: phytotoxic meroterpenoids that induce cell death in cell culture of their producer Rhododendron dauricum. Plant Signal Behav 13: e1422463. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Veitch NC. (2013) Isoflavonoids of the Leguminosae. Nat Prod Rep 30: 988–1027 [DOI] [PubMed] [Google Scholar]
  48. Veitch NC, Grayer RJ (2011) Flavonoids and their glycosides, including anthocyanins. Nat Prod Rep 28: 1626–1695 [DOI] [PubMed] [Google Scholar]
  49. Voo SS, Grimes HD, Lange BM (2012) Assessing the biosynthetic capabilities of secretory glands in Citrus peel. Plant Physiol 159: 81–94 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Vranová E, Coman D, Gruissem W (2013) Network analysis of the MVA and MEP pathways for isoprenoid synthesis. Annu Rev Plant Biol 64: 665–700 [DOI] [PubMed] [Google Scholar]
  51. Winkel-Shirley B. (2001) Flavonoid biosynthesis: a colorful model for genetics, biochemistry, cell biology, and biotechnology. Plant Physiol 126: 485–493 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Wittstock U, Gershenzon J (2002) Constitutive plant toxins and their role in defense against herbivores and pathogens. Curr Opin Plant Biol 5: 300–307 [DOI] [PubMed] [Google Scholar]
  53. Yang T, Fang L, Sanders S, Jayanthi S, Rajan G, Podicheti R, Thallapuranam SK, Mockaitis K, Medina-Bolivar F (2018) Stilbenoid prenyltransferases define key steps in the diversification of peanut phytoalexins. J Biol Chem 293: 28–46 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Yazaki K, Kunihisa M, Fujisaki T, Sato F (2002) Geranyl diphosphate:4-hydroxybenzoate geranyltransferase from Lithospermum erythrorhizon: cloning and characterization of a ket enzyme in shikonin biosynthesis. J Biol Chem 277: 6240–6246 [DOI] [PubMed] [Google Scholar]
  55. Yazaki K, Sasaki K, Tsurumaru Y (2009) Prenylation of aromatic compounds, a key diversification of plant secondary metabolites. Phytochemistry 70: 1739–1745 [DOI] [PubMed] [Google Scholar]
  56. Yazaki K, Arimura GI, Ohnishi T (2017) ‘Hidden’ terpenoids in plants: their biosynthesis, localization and ecological roles. Plant Cell Physiol 58: 1615–1621 [DOI] [PubMed] [Google Scholar]
  57. Zangerl AR, Berenbaum MR (1990) Furanocoumarin induction in wild parsnip: genetics and population variation. Ecology 71: 1933–1940 [Google Scholar]
  58. Zhao S, Wang L, Liu L, Liang Y, Sun Y, Wu J (2014) Both the mevalonate and the non-mevalonate pathways are involved in ginsenoside biosynthesis. Plant Cell Rep 33: 393–400 [DOI] [PubMed] [Google Scholar]

Articles from Plant Physiology are provided here courtesy of Oxford University Press

RESOURCES