Functional identification of rice syn-copalyl diphosphate synthase and its role in initiating biosynthesis of diterpenoid phytoalexin/allelopathic natural products


  • Accession number: AY530101 (will be released upon acceptance of this manuscript).

(fax +1 515 294 0453; e-mail


Rice produces a number of phytoalexins, and at least one allelopathic agent, from syn-copalyl diphosphate (CPP), representing the only known metabolic fate for this compound. Thus, the class II terpene synthase that converts the universal diterpenoid precursor geranylgeranyl diphosphate to syn-CPP catalyzes the committed step in biosynthesis of these natural products. Here the extensive sequence information available for rice was coupled to recombinant expression and functional analysis to identify syn-copalyl diphosphate synthase (OsCPSsyn). In addition, OsCPSsyn mRNA was found to be specifically induced in leaves by conditions that stimulate phytoalexin biosynthesis. Therefore, transcription of OsCPSsyn seems to be an important regulatory point for controlling the production of these defensive compounds. Finally, alignments carried out with OsCPSsyn revealed that class II terpene synthases exhibit a sequence conservation pattern substantially different from that of the prototypical class I enzymes. One particularly notable feature is the specific conservation of the functionally cryptic ‘insertional’ sequence element in class II terpene synthases, indicating that this region is important for the corresponding cyclization reaction.


Plants produce large numbers of low-molecular weight organic compounds, the majority of which do not appear to be necessary for growth under optimal conditions and are, thus, considered secondary metabolites (Croteau et al., 2000). Nevertheless, many of these natural products have important ecological roles in plant defense and allelopathy. For example, phytoalexins are produced in response to microbial infection and exhibit potent antimicrobial activity (VanEtten et al., 1994), while allelochemicals are secreted to the rhizosphere to suppress the growth of neighboring plants (Bais et al., 2004). Terpenoids are often found serving in such roles, as these compounds are particularly abundant in plant secondary metabolism and, in fact, comprise the largest class of natural products (Croteau et al., 2000). Notably, a substantial fraction of the known terpenoids can be classified as labdane-related diterpenoids (20 carbon). This is a large group of over 5000 known compounds defined here as minimally containing the bicyclic hydrocarbon structure found in the labdane class of diterpenoids. In addition, this core structure can be further cyclized and/or rearranged, as in the related/derived structural classes (e.g. gibberellins, kauranes, abietanes, and (iso)pimaranes).

Biosynthesis of labdane-related natural products requires two cyclization steps, the first of which gives rise to the characteristic bicyclic backbone in forming labdadienyl/copalyl diphosphate (CPP). This initial cyclization is catalyzed by class II terpene synthases that selectively produce a specific stereoisomer of CPP from the universal diterpenoid precursor (E,E,E)-geranylgeranyl diphosphate (GGPP). CPP is often further elaborated by class I terpene synthases that stereoselectively utilize these bicyclic prenyl diphosphates to produce specific polycyclic compounds (e.g. Figure 1). Thus, class II and class I diterpene synthases act sequentially to form labdane-related skeletal backbones.

Figure 1.

Known cyclization steps in labdane-related diterpenoid biosynthesis in rice.
Reactions catalyzed by class II (tpsII) or class I (tpsI) terpene synthases are indicated, along with the corresponding classes of natural products derived from each of the named polycyclic hydrocarbon structures (dashed arrows indicate multiple enzymatic steps).

The two classes of plant terpene synthases, while clearly related (Bohlmann et al., 1998a) and both catalyzing reactions involving electrophilic cyclization/rearrangement, are biochemically distinct. Each utilizes different active sites and aspartate-rich motifs functionally associated with catalytic activity (Peters et al., 2001). Class I enzymes contain a DDXXD metal binding motif that facilitates the metal ion assisted allylic diphosphate ionization reactions commonly associated with terpene synthases (Davis and Croteau, 2000). In contrast, the class II cyclases use a separately placed DXDD motif to mediate protonation of a carbon–carbon double bond in an acid/base catalyzed reaction (Peters and Croteau, 2002). When the first class II terpene synthase was cloned (Sun and Kamiya, 1994) the corresponding cyclization mechanism and motif was noted to closely resemble those found in squalene-hopene cyclases (Ochs et al., 1992). This led to the common class II designation for the unrelated diterpene (terpene synthase family) and triterpene (lanosterol synthase family) cyclases (Wendt and Schulz, 1998).

Prototypical class I terpene synthases are similar in size and are composed of two structurally defined domains, which are largely divided between amino and carboxyl halves of their sequences and have simply been called the N- and C-terminal domains (Starks et al., 1997; Whittington et al., 2002). However, the class II enzymes invariably contain a large amount of additional N-terminal sequence (approximately 240 residues; Peters and Croteau, 2002), which has been termed the ‘insertional’ element (Figure 2). Similar sequences can be found in a few class I enzymes as well, most notably those involved in cyclization of CPP to labdane-related structures (e.g. kaurene synthase; Yamaguchi et al., 1996).

Figure 2.

Domain schematic for prototypical class I (tpsI) and class II terpene synthases (tpsII), as modeled on the structures determined for the typical class I epi-aristolochene and bornyl diphosphate synthases (Starks et al., 1997; Whittington et al., 2002).
Approximate locations of the aspartate-rich motifs and the tpsII-associated ‘insertional’ element are as indicated. For clarity, the transit sequence required in mono- and di-terpene synthases for their plastidal location in planta is not shown.

In addition to the ubiquitous gibberellin growth hormones (Phinney et al., 1957), rice (Oryza sativa) produces a number of other labdane-related diterpenoids that serve as phytoalexins and/or allelochemicals (Figure 1). In particular, four other structurally distinct classes have been defined, momilactones A and B (Cartwright et al., 1981; Kato et al., 1973), oryzalexins A–F (Akatsuka et al., 1985; Kato et al., 1993, 1994; Sekido et al., 1986), oryzalexin S (Kodama et al., 1992), and phytocassanes A–E (Kogaet al., 1995, 1997). All of these natural products exhibit antimicrobial properties and are produced in leaves in response to infection with the blast pathogenic fungus Magneportha grisea and, thus, can be defined as phytoalexins (VanEtten et al., 1994). In addition, momilactones A and B were originally identified as dormancy factors from rice seed husks (Kato et al., 1973), and momilactone B has recently been shown to be constitutively secreted from the roots of rice seedlings (Kato-Noguchi and Ino, 2003), where it acts as an allelopathic agent (Kato-Noguchi et al., 2002). The secretion of antimicrobial agents to the rhizosphere may also provide a competitive advantage for root establishment through local suppression of soil microorganisms (Bais et al., 2004). Notably, rice leaves produce all of these compounds after UV-irradiation as well as blast fungal infection, providing a convenient method for inducing biosynthesis of these natural products and, presumably, transcription of the corresponding enzymatic machinery (Kodama et al., 1988).

UV-irradiation has been found to induce terpene synthase activity producing polycyclic diterpene hydrocarbons corresponding to the putative precursors of momilactones A and B, oryzalexin S, and oryzalexins A–F [syn-pimara-7,15-diene, syn-stemar-13-ene, and ent-sandaracopimaradiene, respectively (Wickham and West, 1992)]. Enzymatic activities specifically converting ent- or syn-CPP to these identified polycyclic hydrocarbons were found in cell-free extracts from chitin-elicited suspension-cultured rice cells (Mohan et al., 1996). A number of other diterpene hydrocarbons were also produced in this cell-free system, presumably one of these corresponds to ent-cassa-12,15-diene, the putative precursor to phytocassanes A–E (Yajima and Mori, 2000). Recently, synthesis of ent-cassa-12,15-diene (Yajima et al., 2004) has enabled identification of this compound and cloning of the relevant class I terpene synthase from rice (Cho et al., 2004). However, the conversion of GGPP to syn-CPP in rice has not been directly demonstrated. Nevertheless, in catalyzing the committed step in biosynthesis of syn-labdane-related diterpenoids, all of which, at a minimum, act as phytoalexins in rice, the corresponding class II terpene synthase would be expected to be a key regulatory target for control of this important metabolic process. Additionally, the corresponding enzyme represents a novel stereochemical outcome relative to the identified class II terpene synthases, as only enzymes producing normal- or ent-CPP are known. Therefore, we utilized a functional genomics approach to identify syn-copalyl diphosphate synthase (OsCPSsyn) from rice (Oryza sativa L. ssp. indica). Expression analysis was then used to demonstrate that transcription of OsCPSsyn is an important point at which the corresponding biosynthetic processes are controlled. We also report here insights from sequence comparisons of OsCPSsyn with other known terpene synthases and speculate about possible functional implication.


Isolation of a class II terpene synthase cDNA from UV-irradiated rice leaves

Our interest in the functionally distinct class II terpene synthases as potentially significant targets for metabolic engineering and biochemical analysis motivated this work to identify the syn-CPP synthase that has been previously implicated in rice phytoalexin/allelopathic natural product biosynthesis. The extensive genomic (Goff et al., 2002; Yu et al., 2002) and full-length cDNA (Kikuchi et al., 2003) sequence information available for rice was used to identify a number of putative class II terpene synthases, as defined by the presence of a DXDD motif and ‘insertional’ element. One of these was readily cloned from mRNA prepared from UV-irradiated rice and verified by complete sequencing. However, alignment of the corresponding amino acid sequence with other diterpene synthases indicated the absence of the usual substantial N-terminal transit peptide required for the plastidal location of diterpene synthases in plants (see Figure 5a). The presence of additional N-terminal coding sequence was verified by 5′ rapid amplification of cDNA ends (RACE). The additional N-terminal coding sequence (46 aa) clearly forms at least part of the transit peptide, with an alkaline pI of 11.9, relative to the overall pI of 5.5, which is a common feature of plastid targeting sequences (Keegstra et al., 1989). Thus, a full-length cDNA sequence was determined, which has been deposited into the various nucleotide sequence databases under accession number AY530101. The full-length open reading frame was then also cloned and verified as described above for the partial cDNA.

Figure 5.

Sequence comparison of OsCPSsyn with other terpene synthases.
(a) Alignment with a representative class II terpene synthase (AtCPSent) and ‘insertional’ element containing class I terpene synthase, kaurene synthase (AtKS), both from Arabidopsis, as well as bifunctional class II/I cyclase, abietadiene synthase from grand fir (AgAS). The originally predicted start codon is marked with an ‘*’, the boundaries used for domain comparisons are marked with arrowheads, and the requisite class II DXDD motif marked with an overhead line.
(b) Plot of sequence similarity over the entire sequence length for all the known terpene synthases exhibiting class II activity (numbering based on the consensus sequence).

Functional expression and characterization demonstrated the specific production of syn-copalyl diphosphate

Full-length class II terpene synthase preprotein (tpsII) was expressed in Escherichia coli either alone or as a C-terminal fusion to glutathione-S-transferase (GST-tpsII). In addition, the partial cDNA sequence, which presumably resembles the mature native protein, was expressed alone as a potential pseudomature protein. Enzymatic assays with partially purified recombinant preparations were carried out with GGPP as substrate. After the reaction was allowed to proceed for 1 h at room temperature, the solution was extensively extracted with organic solvent to remove any relatively non-polar compounds (i.e. hydrocarbon or alcohol formed from GGPP). Phosphatase treatment was then employed to remove the pyrophosphate from GGPP or any enzymatically formed derivative, to enable straightforward extraction of the resulting alcohol into organic solvent. Enzymatic turnover of GGPP was analyzed by gas chromatography-mass spectrometry (GC-MS) of the resulting extracts. No hydrocarbons or alcohols were produced by any construct. However, conversion of GGPP into an altered prenyl diphosphate structure was detected, although only with the pseudomature construct. The lack of activity with the full-length preprotein provides an extreme example of the previously observed deleterious effect of plant transit peptides on recombinant expression and folding (Williams et al., 1998). Comparison of the enzymatically formed compound to similarly dephosphorylated authentic samples of ent- and syn-CPP demonstrated that the enzyme produces syn-CPP (Figure 3). Therefore, we have functionally identified rice syn-CPP synthase (OsCPSsyn).

Figure 3.

Production of syn-copalyl diphosphate (CPP).
(a) GC-MS analysis (275 m/z extracted ion chromatograph) of the dephosphorylated reaction products from GGPP.
(b) Mass spectrum of GC-MS 275 m/z chromatograph peak (RT = 13.45 min).
(c) Mass spectrum from authentic dephosphorylated syn-CPP (i.e. syn-copalol), which also exhibits RT = 13.45 min and is clearly separable from dephosphorylated ent-CPP (i.e. ent-copalol; RT = 13.61 min) and GGPP (i.e. geranylgeraniol, which is seen as a distinct peak in the total ion chromatograph; RT = 13.37 min).

Expression pattern of OsCPSsynmRNA

The metabolic fate of syn-CPP is thought to be limited to phytoalexin/allelopathic natural products. Therefore, regulation of OsCPSsyn activity, which catalyzes the committed biosynthetic step, provides a logical point for controlling production of these natural products. Previous review of the relevant literature has been used to suggest that plant secondary metabolism is most often regulated at the level of transcription (Peters and Croteau, 2004). Transcriptional control is manifested by upregulation of enzymatic mRNA levels, with subsequent phytochemical accumulation, in response to the appropriate environmental conditions. The convenience of UV-irradiation for induction of phytoalexin biosynthesis in rice leaves has long been appreciated (Cartwright et al., 1977). In fact, this was utilized in the recently reported large-scale cDNA project from which the original sequence for OsCPSsyn is derived (Kikuchi et al., 2003). Further, quantitative analysis of phytochemical accumulation for the detached leaf UV-irradiation induction method used here has been previously reported (Kodama et al., 1988). Therefore, the ability of UV-irradiation to induce expression of OsCPSsyn mRNA in rice leaves was characterized, demonstrating transcriptional upregulation prior to phytoalexin accumulation (Figure 4). In addition, OsCPSsyn transcription is induced by methyl jasmonate, an important plant defense signaling molecule (Farmer and Ryan, 1990), which, as free jasmonic acid, has been previously demonstrated to induce phytoalexin biosynthesis in rice cell culture (Nojiri et al., 1996). Also consistent with a role in defense, five of the six expressed sequence tags (EST) associated with OsCPSsyn in the TIGR Gene Index are from blast pathogen-infected rice EST projects (; riceTC205530). Finally, as expected from its requisite role in constant production of an allelochemical (Kato-Noguchi and Ino, 2003), OsCPSsyn mRNA seems to be constitutively present in roots. These results strongly indicate that biosynthesis of the corresponding syn-labdane-related diterpenoid natural products is controlled, at least in part, by transcriptional regulation of OsCPSsyn.

Figure 4.

syn-Copalyl diphosphate synthase (OsCPSsyn) expression analysis.
(a) Graphical comparison of OsCPSsyn mRNA levels (closed circles) and phytoalexin accumulation [open circles; as described by Kodama et al. (1988)], in UV-irradiated detached leaves. Quantitative RT-PCR analysis of OsCPSsyn mRNA expression levels is shown in (b) and (c). Specific bands corresponding to the 18S rRNA internal control and OsCPSsyn are indicated.
(b) Expression in response to UV-irradiation. Time (h) after exposure is indicated (c = control leaves after approximately 18 h).
(c) Expression in untreated 4-week-old plant roots (R), or in germinated seedlings in response to application of 0.5 mm methyl jasmonate (+MeJA) or water control (−MeJA).

Sequence comparison of OsCPSsyn with other class II terpene synthases

As selected for by the identification strategy, OsCPSsyn contains not only the requisite DXDD motif, but also the class II synthase associated ‘insertional’ element (Figure 2). In addition, OsCPSsyn clearly does not contain the C-terminal domain DDXXD metal ion binding motif required for class I catalysis (Davis and Croteau, 2000), explaining its inability to catalyze the corresponding diphosphate ionization reaction. Amino acid alignment of OsCPSsyn with the other known class II terpene synthases from plants demonstrates the expected homology (Figure 5). Specifically, OsCPSsyn is 38–44% identical with ent-copalyl diphosphate synthases (CPSent) (required for gibberellin plant growth hormone biosynthesis) from various angiosperms (Ait-Ali et al., 1997; Bensen et al., 1995; Rebers et al., 1999; Richman et al., 1999; Smith et al., 1998; Sun and Kamiya, 1994) and approximately 28% identical with the known gymnosperm bifunctional class II/I terpene synthases (Schepmann et al., 2001; Stofer Vogel et al., 1996), which produce normal CPP. Significantly, this homology is not evenly distributed throughout the entire sequence, as can be seen in Figure 5(b), where similarity is plotted over the length of these proteins. On the basis of previous modeling and truncation studies of abietadiene synthase (AgAS) from Abies grandis (grand fir), approximate boundaries for the ‘insertional’ element (AgAS aa107–344), central region (AgAS aa345–582), and C-terminal domain (AgAS aa583–868) have been assigned (Peters et al., 2003). Given these definitions it can be readily appreciated that the class II enzymes are most similar in the area of the ‘insertional’ element and central region, and least similar in the C-terminal domain. In particular, the ‘insertional’ element and central region are ≥40% identical, while the C-terminal domain is much less conserved (<20% identity in some cases), in terpene synthases with class II activity. This contrasts to the situation with class I terpene synthases, where those few that do contain an ‘insertional’ element show lower levels of conservation across this portion of their sequences, exhibiting identities as low as 25%.


Previous cell-free biosynthetic tracer studies have implied that rice produces syn-CPP as the first committed intermediate in the production of syn-labdane-related diterpenoids (Mohan et al., 1996). These compounds all act as phytoalexins or allelopathic agents. Therefore, the corresponding class II terpene synthase is expected to be a key regulatory point in biosynthesis of these natural products. We are also interested in the class II terpene synthases as model enzymatic systems for the analysis of electrophilic cyclization/rearrangement reactions, and none of the previously identified class II enzymes results in this specific stereochemical outcome. This is of particular biochemical interest because, whereas the bicyclization to normal- and ent-CPP catalyzed by the known class II terpene synthase proceeds from a chair–chair conformation, GGPP must be folded in the energetically less stable chair–boat conformation to form syn-CPP (Figure 6). In addition, the extensive sequence information available for rice was expected to assist functional identification of the gene for this putative enzymatic activity.

Figure 6.

Illustration of the chair–chair and chair–boat conformations required of GGPP to undergo cyclization to ent- and syn-CPP, respectively.

A predicted class II terpene synthase was found in the rice large-scale full-length cDNA project (Kikuchi et al., 2003). Corresponding sequence was readily amplified from rice leaves that had been UV-irradiated, which has been shown to induce phytoalexin biosynthesis (Kodama et al., 1988). This putatively full-length cDNA was cloned, and then extended by 5′ RACE. Expression and biochemical analysis of a recombinant pseudomature construct demonstrated the production of syn-CPP from GGPP (Figure 3) and we have designated the gene syn-CPP synthase (OsCPSsyn). In addition, mRNA expression analysis (Figure 4) demonstrated that OsCPSsyn is constitutively transcribed in roots, but is only transcribed in leaves under conditions that stimulate phytoalexin production (i.e. UV-irradiation or exposure to methyl jasmonate). This matches the pattern of biosynthesis of the known syn-labdane-related diterpenoid natural products. Thus, transcription of OsCPSsyn seems to be required for phytochemical production, strongly indicating that the transcriptional control of OsCPSsyn represents an important regulatory point in these important metabolic processes.

Putative class II terpene synthases were initially identified by the presence of the catalytically requisite DXDD motif, as well as the functionally cryptic N-terminal ‘insertional’ element that has been associated with class II enzymes. Thus, OsCPSsyn contains both, with the DXDD motif then falling within a central region that is homologous to the N-terminal domain of typical class I terpene synthases (Figure 2). Alignment of OsCPSsyn with the other known plant class II terpene synthases reveals striking conservation of both the central region and ‘insertional’ element (Figure 5b). In contrast, the C-terminal domain, which contains the class I active site in the corresponding synthases (Starks et al., 1997), is relatively divergent. As the class II central region contains the DXDD motif and other residues important for class II activity (Peters and Croteau, 2002), its conservation is not surprising. However, the role of the ‘insertional’ element is not well defined. Modeling and mutational analysis of AgAS was used to suggest a role for this region in catalysis (Peters and Croteau, 2002). Unfortunately, truncation analysis was only able to demonstrate that the ‘insertional’ element is required for correct folding (Peters et al., 2003). In particular, it was not possible to separate the class II and I activities into separate, or even overlapping, polypeptides, as was demonstrated with a, at best, very distantly related bifunctional class II/I fungal ent-kaurene diterpene synthase (Kawaide et al., 2000). Nevertheless, recognition of the highly conserved nature of the ‘insertional’ region, specifically in class II terpene synthases, strengthens the suggestion that this sequence element serves a functional role in class II catalysis.

In conclusion, we have used a functional genomics approach to identify syn-CPP synthase (OsCPSsyn), a previously putative enzymatic activity, from rice. Notably, production of syn-CPP represents a novel stereochemical outcome relative to the previously known class II terpene synthases. In addition, OsCPSsyn initiates production of syn-labdane-related diterpenoids and its transcription seems to be an important control point for production of these natural products. Specifically, OsCPSsyn mRNA expression is induced in leaves by conditions that stimulate phytoalexin biosynthesis, but is constitutive in roots where it is required for the constant production of at least one allelochemical. Finally, alignment of OsCPSsyn with other known terpene synthases revealed that the class II enzymes exhibit a sequence conservation pattern substantially different from that of the class I synthases. In particular, the selective conservation of the ‘insertional’ element in class II terpene synthases seems to indicate a functional role in catalysis for this region.

Experimental procedures


The preparation of GGPP, and ent- and syn-CPP, from the corresponding copalol stereoisomers (Yee and Coates, 1992), has been previously described (Mohan et al., 1996). Unless otherwise noted, all other chemicals were purchased from Fisher Scientific (Fairlawn, NJ, USA).

Plant material

Rice plants (Oryza sativa L. ssp. indica cv. IR24) were cultivated from seed under standard greenhouse conditions. After 4 weeks the plants were harvested for analysis. Detached leaves were UV-irradiated for 25 min using a Spectroline UV lamp (set to emit at 254 nm wavelength) suspended 15 cm above the leaf surfaces, then incubated 18 h at 30°C with high humidity in the dark. These tissues were frozen by liquid N2 and total RNA extracted using Concert Plant Reagent, following the manufacturers instructions (Invitrogen, Carlsbad, CA, USA), and stored at −80°C. Where indicated, purification of mRNA was performed with Dynabeads Oligo(dT)25, again following the recommended procedure (Dynal Biotech, Oslo, Norway) and also stored at −80°C.

Isolation of cDNA

Three putative class II terpene synthases were identified by BLAST queries of the rice genomic and cDNA databases with the nucleotide sequence of ent-CPP synthase from Arabidopsis (Sun and Kamiya, 1994), with manual screening for the presence of the DXDD motif and ‘insertional’ element. RT-PCR reactions were performed to generate 5′ fragments of the corresponding open reading frames, which exhibit limited homology to each other in this region (<55% nucleotide identity). However, for only one of these genes was it possible to amplify this region (efforts are underway to clone the remaining two sequences using differentiated internal sequences). The obtained 5′ terminal fragment (599 bp) was cloned into pCR-Blunt II-TOPO (Invitrogen) and sequenced to verify its identity. To amplify the corresponding (partial) cDNA it was necessary to perform a 3′ RACE reaction, which was done following the recommended protocol for the utilized GeneRacer kit (Invitrogen) and using the forward primer from the original 5′ fragment isolation. This was also cloned into pCR-Blunt II-TOPO and verified by sequencing. The corresponding (partial) open reading frame was then subcloned, by PCR amplification for directional topoisomerization, into pENTR/SD/D-TOPO (Invitrogen). This was again verified by complete sequencing, demonstrating approximately 99% identity to the expected coding sequence (nucleotide sequence database accession number AK100631). Presumably the few observed differences reflect the intersubspecies variation between the ssp. japonica used by Kikuchi et al. (2003) and the ssp. indica used here. The true full-length cDNA was then identified through a 5′ RACE reaction, using the reverse primer from the original 5′ fragment isolation. Identical sequence was found in the 599 bp overlap between the originally cloned partial cDNA and the 5′ RACE product. The full-length open reading frame was simply defined as running from the first occurrence of ATG to an in-frame stop codon (a total of 2304 bp). 3′ RACE was again used to clone the open reading frame and 3′ untranslated region into pCR-Blunt II-TOPO. The full-length open reading frame was then also subcloned into pENTR/SD/D-TOPO and the resulting construct verified by complete sequencing. This was then transferred, by directional recombination into the pDEST14 and pDEST15 expression vectors, while the partial cDNA was transferred into pDEST14 only (Gateway Technology, Invitrogen).

Recombinant expression and preparation

Expression was carried out with the BL21-derived C41 strain (Miroux and Walker, 1996), using freshly transformed cells. Initial 1 ml NZY (1% NZ amine, 1% NaCl, 0.5% yeast extract, 0.2% MgSO4) cultures, with 50 μg ml−1 ampicillin, were inoculated from three to five colonies to average over previously observed variations in expression level from single colony inoculation. The initial cultures were grown overnight (200 rpm at 37°C) to saturation and 0.5 ml used to inoculate 50 ml NZY cultures, with 50 μg ml−1 ampicillin. These expression cultures were grown to A600 approximately 0.6, transferred to 16°C (200 rpm) for 1–2 h prior to induction with 1 mm IPTG, then incubated a further 16–20 h. The cells were harvested by centrifugation (4000 g, 20 min, 4°C), resuspended in 1 ml of lysis buffer (50 mm Bis–Tris, pH 6.8, 150 mm KCl, 10 mm MgCl2, 1 mm DTT) and lysed by mild sonication on ice (Branson sonifier 450: 10 sec, continuous, output, setting 5). The resulting lysates were cleared by centrifugation (15 000 g, 30 min, 4°C) and filtration (0.45 μ) to yield soluble extracts. The recombinant protein was then partially purified as previously described (Peters et al., 2003). Briefly, the soluble extract was adsorbed to 0.2 ml ceramic hydroxyapatite type II beads (Bio-Rad, Hercules, CA, USA), which was washed four times with 1 ml buffer (50 mm Bis–Tris, pH 6.8) to remove unbound material. Recombinant terpene synthase was then eluted with 0.5 ml of 200 mm sodium phosphate (pH 6.8), glycerol added to 10% (v/v), DTT to 5 mm, and the resulting solution spin filtered (0.2 μ).

Enzymatic assay and analysis

Enzymatic analysis was carried out in assay buffer (50 mm HEPES, pH 7.2, 100 mm KCl, 7.5 mm MgCl2, 5% glycerol, and 5 mm DTT), with 25 μl of recombinant soluble extract, and initiated by the addition of GGPP to 200 μM for a final volume of 0.2 ml. Reactions were incubated for 1 h in the dark at room temperature and then extracted three times with 0.5 ml of hexanes. Residual organic solvent was removed by evaporation under N2 to enable enzymatic dephosphorylation by calf intestinal phosphatase (10 U; New England Biolabs, Beverly, MA, USA), which was allowed to proceed approximately 4 h at 37°C, largely as described (Wise et al., 2001). Similar dephosphorylation was carried out with authentic ent- and syn-CPP. The dephosphorylated compounds were then extracted two times with 0.5 ml of hexanes. All extracts were completely dried under N2 and then resuspended in 100 μl of hexanes for GC-MS analysis. GC-MS was performed using an HP-1 column on an Agilent (Palo Alto, CA, USA) 6890N GC instrument with 5973N mass selective detector. Samples (5 μl) were injected at 40°C in the splitless mode. After a 3 min isothermal hold at 40°C, the temperature was increased at 20°C min−1 to 300°C, where it was held for an additional 4 min. MS data from 40 to 500 m/z were collected during the temperature ramp and high isothermal hold.

Expression analysis

OsCPSsyn mRNA transcription in response to UV-irradiation was measured using ssp. japonica rice, also grown from seed for 4 weeks under standard greenhouse conditions. Roots were directly harvested, while detached leaves either underwent the UV-irradiation treatment described above (Plant material), or the same procedure without exposure to UV-irradiation (control leaves). At the indicated times leaves were frozen in liquid N2 and stored at −80°C prior to RNA extraction. Transcription in response to methyl jasmonate was measured in ssp. indica seedlings. Seeds were sterilized for 2–3 h in 10% bleach, rinsed five times in sterile water, then germinated in Petri dishes on three layers of filter paper moistened with 3 ml sterile water (this was maintained by the addition of 1 ml of sterile water every other day) at 30°C in the dark for a week. Methyl jasmonate treatment was based on previous reports (Farmer and Ryan, 1990; Martin et al., 2002). Briefly, one plate (seven seedlings) was then sprayed with 2 ml 0.5 mm (±)-methyl jasmonate (Bedoukian Research, Danbury, CT, USA) in carrier solution (0.1% Tween 20 in sterile deionized water), while another (control plate) was only treated with 2 ml of the carrier. These plates were then incubated in separate air-tight containers for 48 h, also at 30°C in the dark, prior to RNA extraction from the whole seedlings. Quantitative RT-PCR was performed on 0.5 μg total RNA (isolated as described above) from each of the indicated tissues, using OsCPSsyn specific primers (from the 5′ fragment isolation) or QuantumRNA 18S standard primers (Ambion, Austin, TX, USA). Appropriate preliminary studies were performed to ensure that the final amplification conditions were in the linear response range. RT-PCR reaction products were separated on a 2.5% agarose gel, stained with ethidium bromide, visualized over a UV-light source, and quantified (where indicated and with normalization against the 18S standard) with NIH Image software. Notably, our preliminary results with the other two putative rice class II terpene synthases indicate that these each exhibit a different expression pattern than OsCPSsyn (i.e. one is non-inducible while the other is inducible but not expressed in roots).

Sequence analysis/alignments

Sequence alignments and identity/similarity comparisons were performed with the AlignX program in the Vector NTI software package (Invitrogen), using standard parameters. OsCPSsyn was the reference sequence in all cases. CPSent sequences were those from Arabidopsis thaliana (Sun and Kamiya, 1994), Zea maize (Bensen et al., 1995), Pisum sativum (Ait-Ali et al., 1997), Cucurbita maxima (Smith et al., 1998), Lycopersicon escclentum (Rebers et al., 1999), and Stevia rebaudiana (Richman et al., 1999). Bifunctional class II/I sequences for abietane synthases were those from Abies grandis (Stofer Vogel et al., 1996) and Ginkgo biloba (Schepmann et al., 2001). Kaurene synthase sequences were those from Cucurbita maxima (Yamaguchi et al., 1996), Arabidopsis thaliana (Yamaguchi et al., 1998), and Stevia rebaudiana (Richman et al., 1999). Other strictly class I sequences included in sequence comparisons were specifically those containing ‘insertional’ elements; linalool synthase from Clarkia breweri (Dudareva et al., 1996), taxadiene synthase from Taxus brevifolia (Wildung and Croteau, 1996), and bisabolene synthase from Abies grandis (Bohlmann et al., 1998b). Domain level comparisons were carried out by manually trimming the relevant sequences based on whole sequence alignment with abietadiene synthase from grand fir (e.g. Figure 5), for which approximate domain boundaries have been defined (Peters et al., 2003).


The authors thank Dr Adam Bogdanove and his laboratory for supplying rice plants and seeds, Dr Robert Thornburg for methyl jasmonate, Dr Nathan K. N. Yee for synthesis and Yinghua Jin for assistance with the syn- and ent-CPP samples used in this study, and Dana J. Morrone and P. Ross Wilderman for critical reading of the manuscript. This work was generously supported by funds from Iowa State University and grant no. 03-190 from the Roy J. Carver Charitable Trust.