A bifunctional geranyl and geranylgeranyl diphosphate synthase is involved in terpene oleoresin formation in Picea abies.

The conifer Picea abies (Norway spruce) defends itself against herbivores and pathogens with a terpenoid-based oleoresin composed chiefly of monoterpenes (C(10)) and diterpenes (C(20)). An important group of enzymes in oleoresin biosynthesis are the short-chain isoprenyl diphosphate synthases that produce geranyl diphosphate (C(10)), farnesyl diphosphate (C(15)), and geranylgeranyl diphosphate (C(20)) as precursors of different terpenoid classes. We isolated a gene from P. abies via a homology-based polymerase chain reaction approach that encodes a short-chain isoprenyl diphosphate synthase making an unusual mixture of two products, geranyl diphosphate (C(10)) and geranylgeranyl diphosphate (C(20)). This bifunctionality was confirmed by expression in both prokaryotic (Escherichia coli) and eukaryotic (P. abies embryogenic tissue) hosts. Thus, this isoprenyl diphosphate synthase, designated PaIDS1, could contribute to the biosynthesis of both major terpene types in P. abies oleoresin. In saplings, PaIDS1 transcript was restricted to wood and bark, and transcript level increased dramatically after methyl jasmonate treatment, which induces the formation of new (traumatic) resin ducts. Polyclonal antibodies localized the PaIDS1 protein to the epithelial cells surrounding the traumatic resin ducts. PaIDS1 has a close phylogenetic relationship to single-product conifer geranyl diphosphate and geranylgeranyl diphosphate synthases. Its catalytic properties and reaction mechanism resemble those of conifer geranylgeranyl diphosphate synthases, except that significant quantities of the intermediate geranyl diphosphate are released. Using site-directed mutagenesis and chimeras of PaIDS1 with single-product geranyl diphosphate and geranylgeranyl diphosphate synthases, specific amino acid residues were identified that alter the relative composition of geranyl to geranylgeranyl diphosphate.

Conifers are frequently subject to attack by herbivorous insects and fungal pathogens (Phillips and Croteau, 1999;Trapp and Croteau, 2001;Franceschi et al., 2005;Keeling and Bohlmann, 2006a). However, the long life span and evolutionary persistence of these trees suggest that they possess effective defense strategies. The best known example of conifer chemical defense is oleoresin, a viscous mixture of terpenoids found in specialized ducts. Oleoresin may be both a constitutive and an inducible defense. For example, in Picea abies (Norway spruce), resin ducts are found constitutively in bark and foliage. However, this species also forms new (traumatic) resin ducts in the wood in response to attack by stem-boring insects and their associated fungi or after trees are sprayed with methyl jasmonate (MJ). Traumatic ducts are believed to help resist attack by augmenting the constitutive resin flow to provide a stronger physical and chemical barrier against herbivores and pathogens (Nagy et al., 2000;Martin et al., 2002;Hudgins et al., 2004;Franceschi et al., 2005;Byun-McKay et al., 2006;Keeling and Bohlmann, 2006a).
Terpenoids are the largest class of plant secondary metabolites, with more than 30,000 structural variants. Oleoresin consists mainly of monoterpenes (C 10 ) and diterpene resin acids (C 20 ) as well as smaller amounts of sesquiterpenes (C 15 ;Langenheim, 2003). The biosynthesis of oleoresin, like all other terpenoids, begins with the synthesis of isopentenyl diphosphate (IPP) via the mevalonic acid pathway or the methylerythritol phosphate pathway (Gershenzon and Kreis, 1999;Fig. 1). IPP and its isomer, dimethylallyl diphosphate (DMAPP), are the five-carbon building blocks of terpenoids that undergo successive condensation reactions to form the larger intermediates geranyl diphosphate (GPP; C 10 ), farnesyl diphosphate (FPP; C 15 ), and geranylgeranyl diphosphate (GGPP; C 20 ). These terpene diphosphate intermediates are in turn the precursors of monoterpenes, sesquiterpenes, and diterpenes, respectively, as well as many larger products (Fig. 1).
In addition to the short-chain IDSs, there are other types. Medium-and long-chain IDSs catalyze the formation of products with more than 20 carbon atoms, such as intermediates in ubiquinone, plastoquinone, dolichol, and rubber biosynthesis (Liang et al., 2002;Kharel and Koyama, 2003;Liang, 2009). IDS proteins can also be classified by the configuration of the double bond formed. The major short-chain prenyl diphosphate intermediates, such as GPP, FPP, and GGPP, all have (E)-double bonds, and most shortchain IDSs reported make products with (E)-double bonds. However, two enzymes making (Z)-products have recently been described (Sallaud et al., 2009;Schilmiller et al., 2009). IDSs producing (Z)-products show little sequence similarity to those making (E)products (Wang and Ohnuma, 2000;Liang et al., 2002;Liang, 2009). The vast majority of (E)-product, shortchain IDSs have a homodimeric architecture, but the existence of heterodimeric GPP synthases has been well documented (Burke et al., 1999;Tholl et al., 2004). A recent study suggests that these enzymes may be widespread in the plant kingdom (Wang and Dixon, 2009).
Most of the short-chain, homodimeric IDSs described make only a single main product, usually GPP, FPP, or GGPP. However, products with one more or one fewer C 5 unit than the main product are occasionally reported at low levels in in vitro assays. For example, FPP synthase from maize (Zea mays) produces up to 10% GGPP (Cervantes-Cervantes et al., 2006), and a GPP synthase from the orchid Phalaenopsis bellini produces nearly 20% FPP (Hsiao et al., 2008). However, it is not clear if these enzymes have such a broad product spectrum in vivo. And no example is Figure 1. Outline of terpenoid biosynthesis leading to the major conifer oleoresin components, monoterpenes and diterpenes, as well as to other classes of terpenes or compounds with terpene components. In the first phase of terpenoid biosynthesis, IPP and DMAPP are formed via the plastidial methylerythritol phosphate pathway and the cytosolic mevalonate pathway. The next phase consists of the reactions catalyzed by short-chain IDSs, GPP synthase, FPP synthase, and GGPP synthase. GPP synthase condenses one molecule of DMAPP and one molecule of IPP. FPP synthase condenses one molecule of DMAPP with two molecules of IPP in succession. GGPP synthase condenses one molecule of DMAPP with three molecules of IPP in succession. During these repeated condensations, the intermediate prenyl diphosphates are normally bound and not released by the enzymes. The PaIDS1 protein is believed to act like a GGPP synthase, but it releases a significant portion of the GPP formed as an intermediate. The remainder of the GPP is converted directly to GGPP without release of FPP. OPP indicates a diphosphate group.
yet known of such a short-chain IDS making two products that differ from each other in size by more than one C 5 unit.
The amino acid sequences of short-chain plant IDSs making products with (E)-double bonds show significant amino acid similarity and contain two highly conserved regions with numerous Asp residues, designated the first and second Asp-rich motifs, respectively (Ashby et al., 1990;Fig. 2). Site-directed mutagenesis studies found that most of the Asp residues in these two highly conserved motifs are critical for substrate binding and catalysis (Koyama et al., 1995(Koyama et al., , 1996. The amino acid sequence features that determine the chain length of the product have been studied by random chemical mutagenesis and experiments on an avian FPP synthase, from which an x-ray structure has been obtained (Tarshis et al., 1994). Here, it was found that the fifth amino acid residue before the first Asp-rich motif is the key residue in determining product chain length. Replacement of the Phe at this position with smaller amino acids resulted in the product specificity shifting to GGPP or longer products. Conversely, replacement of a small amino acid with a larger one shortened the chain length of the product to GPP (Fernandez et al., 2000).
Here, we report the characterization of a new IDS from P. abies that possesses the unique ability to produce both GPP (C 10 ) and GGPP (C 20 ) but not FPP (C 15 ). The product spectrum of this enzyme, its cellular location, and the expression pattern of the corresponding gene all suggest the involvement of this enzyme in oleoresin biosynthesis.

RESULTS
Isolation of a P. abies IDS cDNA Clone and Sequence Comparison The isolation of IDS genes from P. abies was undertaken by PCR using primers designed from conserved regions of known plant GGPP synthases with P. abies RNA as the template. To find IDSs with a role in Figure 2. Alignment of deduced amino acid sequences of PaIDS1 (accession no. GQ369788) and selected additional conifer IDSs, including PaIDS2 (EU432047), A. grandis GPP synthase1 (AgGPPS1; AAN01133), AgGPPS2 (AAN01134), AgGPPS3 (AAN01135), PaIDS5 (EU432050), PaIDS6 (EU432051), and AgGGPPS (AAL17614). Identical amino acids are boxed in black. The artificial translation initiation site for PaIDS1 is indicated by a triangle. The Asp-rich motifs conserved among IDSs are indicated by the black lines. The amino acid sequence used for the generation of peptide antibodies is marked by the dotted line. Dashes indicate sequence gaps introduced to optimize the alignment. oleoresin formation, the fragments obtained were then employed to screen cDNA libraries constructed with mRNA isolated from saplings that had been treated with MJ to induce oleoresin accumulation. A fulllength clone was obtained (PaIDS1) that was represented by four different sequences differing from each other only in the 3# untranslated region, which consisted of 66 to 132 nucleotides between the stop codon and the poly(A) + tail. The protein encoded by PaIDS1 consisted of 383 amino acids and had a calculated mass of 41.8 kD. The amino acid sequence had the highest similarity to other conifer GPP and GGPP synthase sequences (Table I), reaching 86% identity to the GPP synthase 1 of Abies grandis and 84% and 82% identity to GGPP synthases from P. abies (PaIDS5) and A. grandis, respectively (Burke and Croteau, 2002b;Schmidt and Gershenzon, 2007). Other conifer GPP synthase sequences, like PaIDS2 and PaIDS6 from P. abies and the GPP synthases 2 and 3 from A. grandis, showed 55% to 70% identity to PaIDS1 (Burke and Croteau, 2002b;Schmidt and Gershenzon, 2008). The two Asp-rich motifs, DDxxxxDxDDxRRG and DDxxD, found in other plant IDS proteins forming products with (E)-configurations are both conserved in the gene (Fig. 2). The TargetP 1.1 software (http://www.cbs. dtu.dk/services/TargetP; Emanuelsson et al., 2000) predicted a chloroplast signal peptide for PaIDS1 at the 5# end of the cDNA (Fig. 3). When the PaIDS1 sequence was subjected to a phylogenetic analysis with the known conifer IDS genes showing GPP or GGPP synthase activity, it clustered closely together with the GPP synthase1 gene of A. grandis and associated at a greater distance with the GGPP synthase genes PaIDS5 and AgGGPP synthase (Fig. 3). The other GPP and GGPP synthase gene sequences isolated from P. abies and A. grandis clustered more distantly.

Heterologous Expression in Escherichia coli and Characterization of PaIDS1
To study the enzymatic activity of the isolated clone, the protein encoded by PaIDS1 was expressed in E. coli after truncation of the signal sequence at the 5# end of the coding region for directing chloroplast import. Analysis of crude bacterial extracts of the expressing cultures with SDS-PAGE revealed a recombinant pro-tein band with an apparent molecular mass of approximately 40 kD. These extracts were subsequently purified on a nickel-nitrilotriacetic acid agarose column (Supplemental Fig. S1). When incubated with [1-14 C]IPP and DMAPP as substrates, the purified recombinant PaIDS1 protein exhibited both GPP and GGPP synthase enzyme activity, whereas FPP was not detected (Fig. 4). The ratio of GPP to GGPP produced averaged approximately 9:1, taking into account the fact that 1 mol of GGPP (which is formed from three IPP units and one of DMAPP) should incorporate three times as much radioactivity from [1-14 C]IPP as 1 mol of GPP (formed from one IPP and one DMAPP). Neither the empty vector controls nor heat-treated assays showed any significant enzyme activity.
To explore the mechanism of GGPP (C 20 ) formation, the shorter allylic diphosphates, GPP (C 10 ) and FPP (C 15 ), were offered as substrates. Like DMAPP (C 5 ), both GPP and FPP were incorporated into GGPP in the presence of IPP (Fig. 5), suggesting that reaction proceeds by a sequential addition of IPP units as for other IDS proteins. The kinetic constants of the recombinant PaIDS1 (measured under linear conditions with respect to time and protein concentration) showed an apparent K m value of 390 mM for DMAPP, with the IPP concentration constant at 250 mM, and 170 mM for IPP, with the DMAPP concentration constant at 800 mM (for Lineweaver-Burk plots, see Supplemental Fig. S2). The V max was 4.2 nkat mg 21 and the k cat was 1.4 s 21 . Activity required MgCl 2 and reached its highest value at 10 mM MgCl 2 . At 5 or 15 mM MgCl 2 , activity was  only 65% of wild-type PaIDS1, and no change of the product profile was observed (data not shown).

PaIDS1 Expression in Transgenic P. abies Embryogenic Tissue
To confirm the surprising bifunctional GPP and GGPP synthase activity of PaIDS1 in an appropriate eukaryotic expression system, the full-length PaIDS1 gene was introduced into embryogenic tissue of P. abies under the control of the ubi1 promoter via Agrobacterium tumefaciens transformation. Four transgenic lines were obtained that tested positive for the presence of both the nptII and uidA genes (data not shown). The relative abundance of the PaIDS1 gene in pooled samples from these four lines was about 10-fold higher than in either the vector control or untransformed plants, and PaIDS1 mRNA transcript levels in embryogenic tissue were around 3,000-fold higher than in either the vector control or untransformed controls, indicating the success of the transformation (Fig. 6). Enzyme assays performed with crude protein extracts of PaIDS1-transformed embryogenic tissue exhibited both GPP and GGPP synthase enzyme activity when incubated with IPP and DMAPP as substrates, whereas only traces of FPP were observed, just as for the assay of PaIDS1 protein expressed in E. coli (Fig. 7). Crude protein extracts of empty vector control lines and untransformed embryogenic cultures of P. abies exhibited mainly FPP synthase enzyme activity, with only traces of GPP and GGPP.

Comparison of PaIDS1 Transcript Levels among Organs and after MJ Treatment
Terpene oleoresin is synthesized in ducts found in wood, bark, and needles. To determine whether PaIDS1 is localized in these organs, transcript levels were measured in P. abies saplings by quantitative realtime PCR (qRT-PCR). The highest transcript level was found in wood (Fig. 8). PaIDS1 transcripts in bark were only present at about 20% of the level of the wood, whereas no transcripts were detectable in needles.
To determine whether the PaIDS1 enzyme had a role in induced oleoresin formation in P. abies stems, transcript levels were measured over a 10-d time course after saplings had been sprayed with MJ. In bark (and attached cambium), transcripts of the IDS1 gene increased continuously and reached a maximum at day 8, representing a 350-fold increase from the first time point (Fig. 9). In wood, the PaIDS1 genes also showed a transcript accumulation upon MJ spraying; however, the magnitude was less than in bark. Here, maximum induction, with a relative abundance of almost 14-fold relative to the first time point, occurred 2 d after MJ treatment (Fig. 9). Reaction products were hydrolyzed enzymatically, and the resulting alcohols were analyzed by radio-gas chromatography. The main hydrolysis products detected were geraniol (G) and geranylgeraniol (GG), indicating that GPP and GGPP were the principal enzyme products. No release of FPP (detected as farnesol [F]) was measured (third panel from top). The ratio of GPP to GGPP was calculated as approximately 9:1, taking into account that 1 mol of GGPP incorporates three times as much radioactivity from IPP (three units) as GPP does (one IPP unit). The substrate IPP was hydrolyzed to some extent, but the resulting isopentenol eluted close to the solvent front and is not shown. Purified protein extracts of bacteria expressing the empty vector or boiled protein extracts of bacteria expressing PaIDS1 did not show any measurable activity (two top panels). Compounds were identified by coinjection of standards, as depicted in the thermal conductivity detector trace (bottom panel). Standards included geraniol (23 min), nerol (27 min), linalool (29 min), (E,E)-farnesol (33 min), and (E,E,E)-geranylgeraniol (40.5 min). At least five different replicates of each sample were analyzed, and the variance was below 5%.

Localization of PaIDS1 Protein in Resin Duct Epithelial Cells
Antiserum raised against PaIDS1 gave a single strong band on western blots with extracts of E. coli heterologously expressing the recombinant protein, but it did not yield signals with extracts expressing PaIDS2 or PaIDS5, the most similar isolated IDS enzymes from P. abies Gershenzon, 2007, 2008). The molecular mass of the detected band was approximately 42 kD on SDS-PAGE (data not shown), consistent with the predicted molecular mass of the PaIDS1 protein. In sections of P. abies stems treated with the fungus Ceratocystis polonica to induce oleoresin formation, immunogold labeling of the PaIDS1 antiserum appeared extensively in epithelial cells that line the inner surface of traumatic resin ducts (Fig. 10). These cells are thought to synthesize the oleoresin components and secrete them into the ducts. There was no reproducible immunogold labeling above the background signal in any other cell type or in sections treated with preimmune serum only. For comparison, antiserum raised against a P. abies FPP synthase (PaIDS4) that is not involved in induced oleoresin defense reactions (Schmidt and Gershenzon, 2007) was also tested. This did not show any signals, either in stem cells in C. polonica-treated tissue or in the untreated controls (data not shown).

Alteration of PaIDS1 Product Specificity by Site-Directed Mutagenesis and the Creation of Chimeric IDS Proteins
To explore the contribution of individual amino acid residues to the unusual product spectrum of PaIDS1, site-directed mutagenesis was performed on selected positions in or around the regions previously shown to be important in the product length determination of IDS proteins (Ohnuma et al., 1996a(Ohnuma et al., , 1996b(Ohnuma et al., , 1997Tarshis et al., 1996;Fernandez et al., 2000). In several cases, residues were changed to those found in previously characterized single-product GPP or GGPP synthases from conifers to see if this would make PaIDS1 a single-product enzyme.
The majority of the mutated proteins showed no difference in product composition from that of the wild-type enzyme (Table II). However, changes in residues at some positions within the first Asp-rich motif, known to be involved in substrate binding, did cause significant changes in the products of PaIDS1 catalysis. When a Met residue at position 175 ( Fig. 11) was changed to Ile, as found in the sequence of the P. abies GPP synthase PaIDS2, the product distribution shifted toward more GPP. While the wild-type enzyme produced 90% GPP and 10% GGPP, the M175I mutant formed 95% GPP plus small amounts of both GGPP and FPP (Fig. 12). The immediately adjacent residue 174 is a Pro in the wild-type PaIDS1 and a Cys in PaIDS2. Mutation of Pro to Cys (P174C) caused no change in product composition, but when combined with the previously mentioned mutation (P174C and M175I), the product composition shifted completely to 100% GPP (Fig. 12).
Previous studies on IDS enzymes had shown that the fifth amino acid prior to the first Asp-rich motif (position 165 in PaIDS1) is particularly critical in determining product size (Fig. 11). Larger amino acids block extension of the chain, leading to a decrease in product length. Consistent with this trend, mutation of Figure 5. Catalytic activities of recombinant PaIDS1 protein heterologously expressed in E. coli and assayed with combinations of radioactive and nonradioactive IPP, GPP, and FPP. Substrates and reaction products were hydrolyzed enzymatically, and the resulting alcohols were analyzed by radio-gas chromatography (for details of compound identification, analysis, and abbreviations, see Fig. 4). No formation of GGPP was found after using only [1-3 H]GPP as a substrate (top panel). However, GPP or FPP together (radioactive or nonradioactive) with IPP were accepted as substrates for PaIDS1 and led to the formation of radioactively labeled GGPP (bottom panels), indicating that catalysis proceeds by sequential addition of IPP units. At least three biological replicates of each sample were analyzed, and the variance was below 5%. residue 165 from Met to the bulkier Tyr eliminated all GGPP formation, triggering the formation of 10% FPP with 90% GPP (Fig. 12). When the Met at 165 was mutated to the smaller Cys instead, there was no effect on product distribution. Two other sequence changes based on differences between PaIDS1 and other P. abies or A. grandis IDS, I235T and L273F, completely eliminated the formation of the longer GGPP, leaving only GPP as the product.
Mutations to residues found in single-product GGPP synthases of the conifers P. abies and A. grandis gave mixed results. When Leu at position 180 was changed to Phe, there was a significant increase in GPP and a decrease in GGPP. However, the double mutant, L180F and P174C, gave a significant increase in GGPP from 10% to 35% at the expense of GPP (Fig. 12).
As another approach to investigating the structurefunction relationships of PaIDS1, we created chimeric proteins assembled from PaIDS1 and two other P. abies IDS enzymes, the GPP synthase, PaIDS2, and the GGPP synthase, PaIDS5. Dividing the proteins between the two Asp-rich motifs at residues 213 to 216 (Fig. 2), we produced chimeras in which either the N-terminal or C-terminal portion of PaIDS1 was exchanged for the corresponding portion of PaIDS2 or PaIDS5. The proteins in which the PaIDS1 N-terminal sequence had been exchanged, PaIDS2-1 and PaIDS5-1, had product spectra according to the origin of their N-terminal sequences. PaIDS2-1, which contained the N-terminal sequence of PaIDS2 (a pure GPP synthase) and the C-terminal sequence of PaIDS1, produced 95% GPP and 5% GGPP (Fig. 13). Meanwhile, PaIDS5-1, which contained the N-terminal sequence of PaIDS5 (a pure GGPP synthase) and the C-terminal sequence of PaIDS1, produced 40% GPP and 60% GGPP. However, the PaIDS1-2 and PaIDS1-5 proteins, which contained the N-terminal sequence of PaIDS1 and the C-terminal sequence of either PaIDS2 or PaIDS1-5, did not produce any GGPP, but only 80% GPP and 20% FPP (Fig.  13). All of the chimeric proteins had only 10% to 15% of the activity of wild-type PaIDS1.

PaIDS1 Is an Unusual Short-Chain IDS That Produces Both GPP and GGPP
The short-chain IDSs combine the basic C 5 units of the terpenoid pathway into C 10 (GPP), C 15 (FPP), and C 20 (GGPP) intermediates and so control the levels of precursors for different classes of terpenes (Fig. 1). We have been investigating the genes encoding the shortchain IDSs of P. abies to learn more about the control of terpene oleoresin formation in conifers, which is a mixture of monoterpenes (C 10 ) and diterpenes (C 20 ). In previous investigations, we have isolated IDSs with GPP synthase, FPP synthase, and GGPP synthase activities Gershenzon, 2007, 2008). Here, we report an unusual IDS (PaIDS1) that produces both GPP and GGPP in substantial amounts, but no FPP. This novel activity was first demonstrated by heterologous expression in E. coli (Fig. 4) and then confirmed by expression in embryonic tissue of P. abies that was transformed via A. tumefaciens with the PaIDS1 gene under the control of a constitutive promoter (Figs. 6 and 7). To our knowledge, this is the first report of A. tumefaciens-mediated gene transfer in conifers involving the overexpression of a characterized endogenous gene. Previously, the overexpression of an endogenous cinnamoyl alcohol dehydrogenase gene was achieved in Pinus radiata by particle bombardment, but this transformation caused cosuppression, resulting in lower activity of the encoded enzyme (Möller et al., 2003). In another study, even though overexpression of an endogenous peroxidase-like gene in P. abies showed a higher level of total peroxidase activity, no correlation to lignin polymer composition was found (Elfstrand et al., 2001). The ratio of GPP to GGPP formed by PaIDS1 was approximately 9:1 (Fig. 4).
PaIDS1 is, to our knowledge, the first short-chain, homodimeric IDS described that makes two major products with such a large difference in size. Reports on other bifunctional IDSs indicate that enzymes mak- Figure 6. Relative abundance of PaIDS1 gene copies and mRNA transcripts in embryogenic tissue transformed with PaIDS1. Gene abundance was measured by quantitative genomic PCR (top), and transcript abundance of the IDS1 gene was measured by qRT-PCR (bottom), both of which employed SYBR Green for detection and the ubiquitin gene for normalization (for details, see text). Samples consisted of embryogenic tissue transformed with PaIDS1, tissue transformed with the empty vector, and wild-type tissue. Abundance of the vector control was set to 1.0. Each value is the average of four biological replicates, each of which is represented by at least four technical replicates.
ing predominantly a single product in in vitro assays can also catalyze the formation of lower amounts of products with one additional or one fewer C 5 unit. For example, there are GPP synthases that also make low amounts of FPP (Burke and Croteau, 2002b;Hsiao et al., 2008), FPP synthases that also make GGPP (Cervantes-Cervantes et al., 2006), and GGPP synthases that produce minor amounts of FPP (Takaya et al., 2003). Interestingly, bifunctionality is not limited to plants. For example, Vandermoten et al. (2008) described a GPP synthase from Myzus persicae producing significant amounts of FPP. Other dual FPP/GGPP synthases have been reported for a protozoan and a hyperthermophilic archaeon (Chen and Poulter, 1994;Ling et al., 2007). However, there are no other reports of a homodimeric, short-chain IDS that makes multiple products differing by 10 or more carbon atoms.
Bifunctional activities observed for short-chain IDSs in general may be examples of "catalytic promiscuity," an expression that was coined to describe the ability of some enzymes to display an adventitious secondary activity at the active site responsible for the primary activity (Copley, 2003). If this adventitious secondary activity becomes useful to the organism at some point, the enzyme may be recruited to provide more of the secondary product (Vandermoten et al., 2009). The existence of a bifunctional protein, such as PaIDS1, that produces precursors for both monoterpene and diterpene formation is not surprising, as monoterpenes and diterpenes occur together as principal constituents of oleoresin.

PaIDS1 Has the Properties of a GGPP Synthase That Releases a Substantial Portion of the Intermediate GPP
The enzymological properties of PaIDS1 are generally similar to those of single-product GPP and GGPP synthases characterized from P. abies and other conifers. All of these enzymes share a requirement for the same divalent metal ion, Mg 2+ . While the K m values of PaIDS1 for the substrates IPP and DMAPP are somewhat greater than those determined for other shortchain conifer IDSs, the ratio between the K m for DMAPP and that for IPP is approximately the same. The substrate specificity of PaIDS1, its ability to use other allylic diphosphates (Fig. 5), also parallels that of GGPP synthases from conifers and other species (Hefner et al., 1998;Ogura and Koyama, 1998;Takaya et al., 2003), suggesting that reaction also involves successive condensation of IPP units (Fig. 1). These successive condensations usually proceed without significant release of the intermediates GPP and FPP. PaIDS1, therefore, may be simply a GGPP synthase in Figure 8. Relative abundance of PaIDS1 mRNA transcripts in different plant organs. Transcript abundance was measured by qRT-PCR using SYBR Green for detection and the ubiquitin gene for normalization (for details, see text). Each value is the average of three biological replicates, each of which is represented by three technical replicates. Values for wood and needles are expressed relative to bark, which was set to 1. n.d., Not detectable. To assess what structural features of PaIDS1 might be responsible for the simultaneous formation of GPP and GGPP, a program of site-directed mutagenesis was undertaken. This effort relied on modeling of PaIDS1 with reference to an avian FPP synthase (Tarshis et al., 1994), the only short-chain IDS for which a crystal structure is available. The sequence of this avian FPP synthase shares 30% amino acid identity to PaIDS1. Its structure is composed of 13 a-helices, of which 10 form a large central cavity (Fig. 11). The active site is located in the central cavity with two conserved Asp-rich motifs facing each other on opposite walls. These motifs have been demonstrated to be involved in substrate binding and catalysis via coordination with Mg 2+ , hence the requirement of this avian FPP synthase and other IDSs for this cofactor (Marrero et al., 1992;Joly and Edwards, 1993;Song and Poulter, 1994;Koyama et al., 1994Koyama et al., , 1995Koyama et al., , 1996. In PaIDS1, the first Asp-rich region spans amino acid residues 170 to 183 and is located in helix D and an adjacent loop, while the second region spans residues 309 to 313 and is located in helix H. These two helices, as well as helix F1, form a significant portion of the active site (depicted in Fig. 11).
In site-directed mutagenesis, selected residues of PaIDS1 were changed to those present in other conifer IDSs that form either GPP or GGPP as a single product. A double mutant at P174C and M175I that changed these residues to those present in single-product GPP synthases (Burke and Croteau, 2002b;Schmidt and Gershenzon, 2008) could catalyze only the formation of GPP and no GGPP. The residues at positions 174 and 175 are thought to have been inserted in the IDS sequence during evolution from bacteria to plants (Wang and Ohnuma, 1999) and are clearly important in determining product specificity. Mutation of the fifth amino acid prior to the first Asp-rich motif (M165Y) also caused the production of shorter products. This residue, which projects into the active site cavity (Fig. 11), has been shown to influence product length in other short-chain IDSs, with large, bulky R groups appearing to block additional condensation with IPP and thus to give shorter chain products (Ohnuma et al., 1996b;Fernandez et al., 2000;Soderberg et al., 2001). Here, mutagenesis from Met to Tyr gave Figure 9. Relative abundance of PaIDS1 mRNA transcripts in MJtreated P. abies saplings. Transcript abundance of the IDS1 gene was measured by qRT-PCR using SYBR Green for detection and the ubiquitin gene for normalization (for details, see text). The time points measured were 0, 0.5, 1, 2, 4, 6, 8, and 10 d after the onset of treatment. Bark samples included the cambium. The time-zero measurement is the untreated control; its abundance was set to 1.0. Each value is the average of four biological replicates, each of which is represented by at least three technical replicates. Figure 10. Immunogold labeling using a PaIDS1 polyclonal antibody applied to stem sections of P. abies that had been treated with the fungus C. polonica. Gold particles were observed only in epithelial cells (ec) lining the inner surface of traumatic resin ducts (TRD), which contain induced oleoresin, and not in the remaining xylem cells, including the ray channel (rc), a site of primary oleoresin secretion (top). The antibody serum did not generate any labeling above the background signal in tissue treated with preimmune serum only (bottom picture). an increased proportion of shorter chain products, while mutagenesis of Met to Cys had no effect on product distribution. Interestingly, there were two other residues where mutation (I235T and L273F) increased the proportion of the short-chain product GPP to 100%. These project into the central cavity from helix F, immediately opposite Met-165, and so may also prevent additional condensations by steric effects.
Mutagenesis of PaIDS1 sequence sites to those present in sequences of similar, single-product GGPP synthases was largely unsuccessful at altering product composition. The one exception was the double mutant, P174C and L180F, which gave a 65% GPP, 35% GGPP distribution compared with 90% GPP, 10% GGPP in the wild-type enzyme. Residues 174 and 180 are located in the first Asp-rich region in the loop attached to helix D and so are likely to be part of the active site. However, it is not obvious how modification at these positions can result in a greater percent-age of additional IPP condensations to give more GGPP and less GPP.
The residues in and around the first Asp-rich region (residues 170-183) of PaIDS1 are clearly important in chain length determination. This conclusion is supported by experiments involving chimeras formed by exchanging the N-or C-terminal portion of PaIDS1 with corresponding portions of similar single-product GPP or GGPP synthases. When the N terminus (including the first Asp-rich region) was replaced by the N terminus of a single-product P. abies GPP synthase or a single-product GGPP synthase, the product distribution shifted accordingly to either GPP or GGPP (Fig.  13). Replacement of the C terminus in the same way did not have such an effect, but the fact that mutagenesis of residues 235 and 273 also affected the length of the product suggests that control of chain length resides at many places in the protein, not all of which are clearly known. Other open questions about PaIDS1 Figure 11. Three-dimensional ribbon model of the complete structure of wild-type PaIDS1 (A) and the substrate-binding region (B) constructed with reference to the avian FPP synthase crystal structure (Tarshis et al., 1994). Depicted are the a-helices D (amino acids 154-181), F1 + F2 + G (amino acids 222-277), and H (amino acids 309-313). Helices that are part of the predicted substrate-binding pocket are colored in blue (helices D + F), and the DDxxD motifs (amino acids 170-176 and 309-313) are indicated in red. Residues used for mutagenesis that are visible in this diagram are shown in turquoise. The model was displayed with the program UCSF Chimera (National Centre for Research Resources). concern why a significant proportion of GPP, the C10 intermediate formed by the enzyme, is released but the next larger intermediate, FPP, remains bound. It is conceivable that as the product elongates from C 10 to C 15 in the active site, it twists in the cavity in a way that prevents release until it reaches C 20 and cannot elongate further. The residues responsible for preventing release might be localized in the C-terminal domain of PaIDS1, because site-directed mutagenesis at positions 235 and 273 and chimeras with replacements of the C-terminal portion of PaIDS1 (residues 213-383) produced FPP as their final product instead of GGPP, albeit with a much lower activity. More work is needed to understand how PaIDS1 creates its unique bifunctional product spectrum.

PaIDS1 Transcript and Protein Are Associated with Oleoresin Biosynthesis
The ability of PaIDS1 to produce both GPP and GGPP suggests that this enzyme could have a biological role in both monoterpene and diterpene production. Since the major components of conifer oleoresin are monoterpenes and diterpenes, we looked for a connection between the occurrence of PaIDS1 in P. abies and oleoresin biosynthesis. The formation of oleoresin in this species can be induced by treatment with MJ, which mimics attack by bark beetles or pathogenic fungi (Franceschi et al., 2002;Martin et al., 2002;Hudgins et al., 2003;Byun-McKay et al., 2006;Erbilgin et al., 2006;Zeneli et al., 2006). MJ Figure 12. Catalytic activities of mutated recombinant PaIDS1 proteins heterologously expressed in E. coli and assayed with [1-14 C]IPP and DMAPP. Reaction products were enzymatically hydrolyzed, and the resulting alcohols were analyzed by radio-gas chromatography. For details of compound identification, analysis, and abbreviations, see Figure 4. At least three replicate assays of each mutant were analyzed, and the variance was below 5%. Figure 13. Catalytic activities of chimeric recombinant PaIDS1 proteins heterologously expressed in E. coli. Either the N-terminal or C-terminal half of PaIDS1 was exchanged for the corresponding portion of the known GPP synthase, PaIDS2, or the known GGPP synthase, PaIDS5, and the resulting proteins were assayed with [1-14 C]IPP and DMAPP. Reaction products were enzymatically hydrolyzed, and the resulting alcohols were analyzed by radio-gas chromatography. For details of compound identification, analysis, and abbreviations, see Figure 4. At least three replicate assays of each mutant were analyzed, and the variance was below 5%. treatment increases the accumulation of resin monoterpenes and diterpenes and stimulates the formation of traumatic resin ducts in stems in the newly formed wood (xylem) just inside the vascular cambium. In this work, MJ treatment also dramatically increased the transcript abundance of PaIDS1 in wood and bark, compared with untreated controls, indicating that the encoded protein is involved in oleoresin formation. In untreated tissue, PaIDS1 transcript is found in wood and bark, but not needles, consistent with a role in resin formation only in traumatic ducts. Although traumatic resin ducts are located in wood, their development is initiated in cells that are still part of the vascular cambium (Martin et al., 2002;Franceschi et al., 2005). Since the cambium layer is usually included with the bark during sampling, it is not surprising to observe expression of a gene involved in traumatic resin duct formation in both the bark and wood.
In previous studies on P. abies short-chain IDSs, we have demonstrated that a single-product GPP synthase (PaIDS2) and a single-product GGPP synthase (PaIDS5) are also associated with oleoresin biosynthesis in stems Gershenzon, 2007, 2008). While these single-product enzymes could suffice for the formation of the major resin terpenes, it is interesting that PaIDS1 transcript is induced much more significantly than those of PaIDS2 or PaIDS5 after MJ treatment and thus may have an especially important function in induced resin formation. Additional evidence for the role of PaIDS1 in induced oleoresin formation comes from our studies that localized this protein to epithelial cells lining the traumatic resin ducts in stems treated with the fungus C. polonica. Epithelial cells are thought to synthesize the resin and secrete it into the lumen of the ducts (Bannan, 1936). Antibodies to PaIDS1 did not react with antigens in any other type of cell in the stem.
At the subcellular level, PaIDS1 appears to be localized in the plastid by virtue of its predicted chloroplast signal peptide. A plastidial location is also consistent with current concepts on the intracellular compartmentalization of GPP-and GGPP-producing or -utilizing enzymes in plant cells (Parmryd et al., 1999;Welsch et al., 2000;Bouvier et al., 2002Bouvier et al., , 2005Bick and Lange, 2003;Martin et al., 2004;Tholl et al., 2004;Keeling and Bohlmann, 2006b;Ro and Bohlmann, 2006).

PaIDS1 Has Strong Sequence Identity to Other Conifer GPP and GGPP Synthases and Is Likely to Have Evolved from a Single-Product Enzyme
A phylogenetic analysis of gene sequences for characterized plant short-chain IDSs shows that PaIDS1 clusters closely with most previously reported, singleproduct GPP and GGPP synthase genes from conifers. However, PaIDS1 shares no significant sequence similarity to a gene encoding one conifer GPP synthase that clusters most closely with several angiosperm GPP synthases thought to function in gibberellin biosynthesis (Bouvier et al., 2002;Van Schie et al., 2007;Schmidt and Gershenzon, 2008). It also possesses no significant similarity to the family of heterodimeric GPP synthases known from angiosperms (Burke and Croteau, 2002a;Tholl et al., 2004;Wang and Dixon, 2009).
Given its demonstrated ability to produce both GPP and GGPP, it is not surprising that PaIDS1 has highest sequence similarity to single-product GPP and GGPP synthases from conifers. Its kinetic properties and reaction mechanism also do not distinguish it from other members of this group. PaIDS1 may be considered a GGPP synthase that has the ability to release a substantial portion of the intermediate GPP, giving it a novel product spectrum. Our site-directed mutagenesis experiments showing facile shifts of GPP and GGPP after modification of one or two residues suggest that the evolutionary conversion from singleproduct GPP and GGPP synthases to an enzyme like PaIDS1 may require only a few steps. Having a single short-chain IDS that makes both the GPP and GGPP precursors for conifer resin in a fixed proportion may have been selected to allow better control of the ratio between monoterpenes and diterpenes and to facilitate increased synthesis of both products simultaneously. Other plant species that produce mixtures of terpene natural products may have similar dualfunction short-chain IDSs.

Plant Material
The clonal 4-year-old Picea abies (Norway spruce) saplings used (Samenklenge und Pflanzgarten Laufen) are members of clone 3369-Schongau. Saplings were grown in standard soil in a climate chamber (Vötsch) under a 21°C-day/ 16°C-night temperature cycle, controlled light conditions (16 h per day at 150-250 mmol, obtained from a mixture of cool-white fluorescent and incandescent lamps), and a relative humidity of 70% in a climate chamber (Vötsch). For induction experiments, saplings were sprayed with 100 mM MJ in 0.5% (v/v) Tween 20 as detergent. When stem tissue was harvested, the uppermost internode (representing the previous year's growth) was used. Bark and wood from this zone were separated and frozen in liquid nitrogen in preparation for RNA isolation. The bark included the cambium layer that peeled off from the wood during separation.
The embryogenic culture of P. abies (line 186/3c VIII; kindly provided by Harald Kvaalen, Norwegian Forest and Landscape Institute) was initiated from mature zygotic embryos, dissected from seeds, and cultured on modified Litvay's (MLV) medium (Klimaszewska et al., 2001). Casein amino acids (Becton Dickinson) were added at 1 g L 21 , L-Gln (Duchefa) at 0.5 g L 21 , Suc at 10 g L 21 , and Gelrite (Duchefa) at 4 g L 21 when a semisolid medium was required. The medium was supplemented with 10 mM 2,4-dichlorophenoxyacetic acid and 5.0 mM 6-benzyladenine. The pH of the medium was adjusted with KOH to 5.7 prior to sterilization in the autoclave. The pH of the L-Gln solution was adjusted to 5.7, and the solution was filter sterilized and added to the medium after sterilization. Unless stated otherwise, all cultures were kept in the dark at 24°C.

Amplification and Cloning of the PaIDS1 Sequence
Degenerate primers were designed for conserved regions of GGPP synthase sequences from Taxus canadensis, Helianthus annuus, Croton sublyratus, and Lupinus albus for different conserved regions: 5#-CIATG(A/C)G(G/T)TA (C/T)TCICTICTIGCIGG-3# and 5#-CTTATCIG(C/T)(A/C)AICAAATC(C/T) TT(C/T)CC-3#. RT was carried out with pooled total RNA from MJ-treated P. abies bark harvested at various time points after treatment (6, 12, and 24 h and 2, 4, 6, 10, and 20 d) and oligo-(dT) primers using SuperScript II (Invitrogen), according to the manufacturer's instructions. Subsequently, PCR was performed with 35 cycles of 94°C for 1 min, annealing for 1 min, and then 72°C for 1 min in a thermocycler (Stratagene). To identify the optimal annealing temperature, a temperature gradient was used with 2°C intervals ranging from 42°C to 64°C. A 430-bp fragment was generated at 44°C and subcloned into pCR 4-TOPO (Invitrogen) according to the manufacturer's instructions.

Isolation of the PaIDS1 Clone
Total RNA was isolated according to a method developed by  from P. abies bark harvested at various time points (as above) after being sprayed with 100 mM MJ. From total pooled RNA, poly(A) + RNA was isolated using Dynabeads (Invitrogen). cDNA was synthesized using the SMART cDNA Library Construction Kit (Clontech). In vitro packaging with Gigapack III Gold (Stratagene) yielded a library of 8 3 10 7 plaque-forming units, which was subsequently screened with the [a-32 P]dATP-labeled 430-bp fragment previously amplified by PCR. Single plaques were isolated from positive colonies, and lDNA was converted into pDNA by subcloning into pTriplEx2 vector and transformation in Escherichia coli BM25.8. Sequence analysis was carried out using an ABI 3100 automatic sequencer (Applied Biosystems). One sequence was obtained that was named PaIDS1.

Sequence and Phylogenetic Analyses
The DNAStar Lasergene program version 7.0 (MegAlign) was used to align and to calculate the deduced amino acid sequences of each full-length cDNA or of known sequences from other gymnosperms and angiosperms. The amino acid alignment was conducted by use of ClustalW (gonnet 250 matrix, gap penalty 10.00, gap length penalty 0.20, delay divergent sequences 30%, gap length 0.10, DNA transition weight 0.5). The same software was used to visualize the phylogenetic tree.

Functional Expression of PaIDS1
The entire coding sequence of clone PaIDS1 lacking the predicted signal peptide for chloroplast transport (based on analysis at http://www.cbs.dtu. dk/services) was amplified with primers that included the start and stop codons for translation. The primer combination 5#-ATGTGCTCAAACAC-AAATGCCCAG-3# and 5#-GTTCTGTCTTTGTGCAATGTAATG-3# was used to express PaIDS1. The amplification was carried out with the Expand High Fidelity PCR System (Roche Applied Science). PCR was performed for 2 min at 94°C initial denaturation, 10 cycles of 15 s at 94°C, 30 s at 55°C, and 55 s at 72°C, 20 cycles of 15 s at 94°C, 30 s at 55°C, and 55 s plus 5 s of elongation for each cycle at 72°C, and a final extension for 7 min at 72°C in a Tpersonal thermocycler (Biometra). The resulting cDNA fragments were cloned into the expression vector pCR-T7 CT TOPO (Invitrogen), which adds a tag containing six His residues to the C terminus. Positive clones were first transferred into E. coli strain TOP10F# (Invitrogen) and then into strain BL21 (DE3)pLysS (Invitrogen). Bacterial cultures of 100 mL were grown to an optical density (OD 600nm ) of 0.6, and transformants were induced with 2 mM isopropylthiob-galactoside as described by the manufacturer's instructions, except that they were grown exclusively at 18°C. Bacterial pellets were resuspended in 3 mL of buffer containing 20 mM 3-(N-morpholino)-2-hydroxypropanesulfonic acid (MOPSO), pH 7.0, 10% (v/v) glycerol, and 10 mM MgCl 2 and sonicated using a Sonopuls HD 2070 (Bandelin) for 4 min, cycle 2, power 60%. The PaIDS1 protein was purified over nickel-nitrilotriacetic acid agarose columns (Qiagen) according to the manufacturer's instructions. Protein was eluted with 20 mL of 250 mM imidazole in the assay buffer, and 1.5-mL fractions were collected. After adding 2 mM dithiothreitol, each fraction was checked for purity by SDS-PAGE. Fraction 2, which contained the highest amount of recombinant protein, was used for the assay. The added 250 mM imidazole was not removed before the assay, as it did not affect the activity.

Assay of Recombinant PaIDS1 and Product Identification
Small-scale PaIDS1 assays were carried out in a final volume of 50 mL containing 20 mM MOPSO (pH 7.0), 10 mM MgCl 2 (for PaIDS5), 10% (v/v) glycerol, 2 mM dithiothreitol, and 800 mM IPP or 250 mM DMAPP at saturated concentrations. For radioactive detection of products, 10 mM [1-14 C]IPP (55 Ci mol 21 ) was added to each assay. The reaction was initiated by the addition of recombinant protein, and the assay mixture was overlaid with 1 mL of hexane and incubated for 20 min at 30°C. Assays were stopped by the addition of 10 mL of 3 N HCl and incubated for an additional 20 min at 30°C to solvolyze the acid-labile allylic diphosphates formed during the assay to their corresponding alcohols. Solvolysis products were extracted into the hexane phase by vigorous mixing for 15 s. After centrifugation, 300 mL of the hexane phase was used for total radioactivity determination by liquid scintillation counting. Protein concentration was measured according to Bradford (1976) using the Bio-Rad reagent with bovine serum albumin (BSA) as standard. The reaction was linear for up to 30 min at protein concentrations of 40 mg mL 21 . Kinetic parameters were calculated with Michaelis-Menten kinetics and Lineweaver-Burk plotting using the enzyme kinetics program EkI 3 (Wiley-VCH). All assays were carried out in triplicate.
For the identification of reaction products of PaIDS1, larger scale assays were carried out in a final volume of 500 mL. The assay mixture was as described above, but concentrations of 40 mM [1-14 C]IPP, 40 mM DMAPP, and a 1-mL layer of pentane were used. Assays were incubated overnight at 30°C. To stop the assay and hydrolyze all diphosphate esters, a 1-mL solution of 2 units of calf intestine alkaline phosphatase (Sigma) and 2 units of potato apyrase (Sigma) in 0.2 M Tris-HCl, pH 9.5, was added to each assay and followed by overnight incubation at 30°C. After enzymatic hydrolysis, the resulting isoprenyl alcohols were extracted into 2 mL of diethyl ether. Then, following the addition of a standard terpene mixture, the organic extracts were evaporated under N 2 and used for radio-gas chromatography (GC) measurements. Radio-GC analysis was performed on a Hewlett-Packard HP 6890 gas chromatograph (injector at 220°C, thermal conductivity detector at 250°C) in combination with a Raga radioactivity detector (Raytest) using a DB 5-MS capillary column (J&W Scientific; 30 m 3 0.25 mm with a 0.25-mm phase coating). Separation of the injected concentrated organic phase (1 mL) was achieved under an H 2 flow rate of 2 mL min 21 with a temperature program of 3 min at 70°C, followed by a gradient from 70°C to 240°C at 6°C min 21 with a 3-min hold at 240°C. Products measured by radio-GC were identified by comparison of retention times with those of coinjected, nonradioactive, authentic terpene standards monitored via a thermal conductivity detector.
To study the reaction mechanism, substrates were used as follows in the assays depicted in Figure

Agrobacterium tumefaciens Strain and Culture Preparation
The disarmed A. tumefaciens strain C58/pMP90 (Koncz and Schell, 1986) containing either the binary vector pCAMGW::IDS1 or pCAMBIA2301 (as a control) was used in the transformation experiment. For construction of the transformation vector, attR recombination sites (Gateway Technology, Invitrogen) were cloned into the vector pMJM (McKenzie et al., 1994), creating pHJMGW. A 4-kb-long cassette from pMJMGW containing the maize (Zea mays) ubiquinone promoter (ubi1) and the attR recombination sites was cloned into pCAMBIA2301. The resulting plasmid pCAMGW carries the neomycin phosphotransferase gene (nptII), which confers kanamycin resistance under the control of the double 35S cauliflower mosaic virus promoter and terminator and the GUS reporter gene (uidA), coding for GUS, under the control of the 35S cauliflower mosaic virus promoter and the nopaline synthase terminator. The complete open reading frame of IDS1 was cloned into pCAMGW upstream of the ubi1 promoter using Gateway Technology.
The bacterial culture was grown in yeast extract peptone liquid medium containing 50 mg mL 21 rifampicin, 50 mg mL 21 kanamycin sulfate, and 20 mg mL 21 gentamicin sulfate (all Duchefa) on a shaker at 250 rpm at 28°C for 16 h. Subsequently, the bacterial cells were pelleted by centrifugation and resuspended in liquid MLV medium to an OD 600nm of 0.6.

Transformation Procedure for Embryogenic Tissue
Prior to the transformation experiment, embryogenic tissue was prepared and cultured on filter papers as described by Klimaszewska and Smith (1997). For the transformation experiments, the embryogenic tissue was harvested from the filter papers 7 d after subculture and resuspended in liquid MLV medium. Subsequently, an equal volume of A. tumefaciens culture (OD 600nm = 0.6) in MLV was added to the cell suspension, resulting in 100 mg (fresh weight) of embryogenic tissue suspended in 1 mL of bacterial culture at OD 600nm of 0.3 in MLV medium in a 100-mL Erlenmeyer flask. Control embryogenic tissue was treated the same way except that A. tumefaciens was omitted.
From the culture, 1 mL (100 mg of fresh material) was poured over a 5.5-cm sterile filter paper (Whatman no. 2) in a Bü chner funnel. A short vacuum pulse (5 s, 45.7 cm of mercury) was applied to drain the liquid, and the filter was placed on a semisolid MLV medium in a 100-mm 3 15-mm petri dish. A total of 10 petri dishes were used for each transformation experiment and the controls.
After 2 d of coculture on the semisolid medium supplemented with 50 mM acetosyringone, filter papers with cells from five petri dishes were placed in an Erlenmeyer flask (250 mL) with 50 mL of liquid MLV medium. The cells were then dislodged by manual shaking and collected on the new filter papers in a Bü chner funnel. The filters papers with cells were then placed on fresh MLV medium with 250 mg L 21 cefotaxime (Duchefa). At the first signs of embryogenic tissue growth (approximately 2 d), the filters with cells were transferred to selective medium. Selective medium contained cefotaxime and 35 mg L 21 kanamycin. For maintenance, the transgenic embryogenic cultures of P. abies were subcultured every 14 d and cultivated as described above.

Quantitative Genomic PCR
For quantitative genomic PCR of embryogenic tissue, genomic DNA was isolated using the Plant DNA Isolation Kit (Qiagen) according to the manufacturer's instructions. Genomic PCR was performed with Brillant SYBR Green QPCR Master Mix (Stratagene) as described in the section on qRT-PCR below, but with the following changes. The denaturation step at the beginning was elongated to 13 min at 95°C. Gene abundance was normalized to the abundance of the ubiquitin gene, amplified with the forward primer (5#-CCCTCGAGGTAGAGTCATCG-3#) and the reverse primer (5#-CCAGA-GTTCTCCCATCCTCC-3#). This quantity was the average of three independent biological replicates, each represented by four measurements on three technical replicates. Relative gene abundance was obtained by comparison with the gene abundance of the pCAMGW vector control.

Assay of PaIDS1 in Transgenic Embryogenic Tissue
Embryogenic tissue culture was dried on a filter paper, resuspended in assay buffer (see above), and treated with an Ultraturrax with maximal speed for 20 s at 4°C. After centrifugation at 20,000g for 10 min at 4°C, supernatant containing crude protein extract of embryogenic tissue was assayed. Protein concentration was measured according to Bradford (1976) using the Bio-Rad reagent with BSA as standard. Identical amounts of protein extracts were used in the larger scale assays as described for the assay of recombinant PaIDS1 and product identification.

qRT-PCR
Total RNA from plants at different time points after MJ treatment and from different organs was isolated according to the method of , except that an additional DNA digestion step was included (RNase Free DNase Set; Qiagen). Using identical amounts of total RNA, template cDNA for subsequent PCR was generated using SuperScript III (Invitrogen) according to the manufacturer's instructions. qRT-PCR was performed with Brillant SYBR Green QPCR Master Mix (Stratagene). Specific qRT-PCR primers, forward (5#-GACATCTGGTATCATCACTC-3#) and reverse (5#-GTGACCT-TCCCCTTTACTTG-3#), were designed using the following criteria: predicted melting temperature of at least 58°C, primer length of 22 to 24 nucleotides, guanosine-cytosine content of at least 48%, and an amplicon length of 120 to 150 bp. Primer specificity was confirmed by melting curve analysis, by an efficiency of product amplification of 1.0 6 0.1, and by sequence verification of at least eight cloned PCR amplicons for each PCR procedure. Reactions with water instead of cDNA template were run with each primer pair as controls. A standard thermal profile of 95°C for 10 min, then 45 cycles of 95°C for 30 s, 53°C for 30 s, and 72°C for 30 s, was used. The fluorescence signal was captured at the end of each cycle, and a melting curve analysis was performed from 53°C to 95°C with data capture every 0.2°C during a 1-s hold. qRT-PCR was performed as described in the operator's manual using a Stratagene MX3000P. Transcript abundance was measured as the average of three biological replicates; each is represented by three determinations of three technical replicates. All amplification plots were analyzed with the MX3000P software to obtain threshold cycle values. Transcript abundance was normalized with the transcript abundance of the ubiquitin gene (GenBank accession no. EF681766), amplified with the forward primer (5#-GTTGATTTTTGCTGG-CAAGC-3#) and reverse primer (5#-CACCTCTCAGACGAAGTAC-3#). Relative transcript abundance was obtained by calibration against the transcript abundance at the onset of treatment.
For qRT-PCR of embryogenic tissue, RNA was isolated using the Plant RNA Isolation Kit (Invitek) according to the manufacturer's instructions, including an additional DNA digestion step (RNase Free DNase Set; Qiagen). Further steps were carried out as above. Relative transcript abundance was calibrated against the transcript abundance of pCAMGW vector control plants.

Microscopy and Immunogold Labeling of PaIDS1
Stems of P. abies trees (5 years old) were inoculated with Ceratocystis polonica as described by Christiansen et al. (1999). Bark samples (2 3 3 mm, including phloem, vascular cambium, and xylem with traumatic resin ducts) were collected 36 d after inoculation. All samples were isolated from areas 5 cm above the inoculation site, well above the edge of necrotic lesions (Nagy et al., 2000). Fixation, dehydration, and embedding were performed as described previously (Nagy et al., 2000). Semithin sections (1 mm) were cut using a LKB 2128 Ultratome (Leica Microsystems) and dried onto Superfrost Plus slides (Menzel-Gläzer) for light microscopy.
Immunogold labeling was performed as described previously (Nagy et al., 2000). Polyclonal peptide antibodies for PaIDS1 were synthesized against amino acids 285 to 300 with the sequence 5#-GASDDEIERVRKYARC-3# (Fig.  2) and affinity purified (Eurogentec). Polyclonal peptide antibodies for PaIDS4 were synthesized against the amino acids 172 to 187 with the sequence 5#-HEGATDLSKYKMPT-3# (Schmidt and Gershenzon, 2007). The primary antibodies were diluted 1:4,000 in TBST/BSA buffer (10 mM Tris, pH 7.2, 0.15 M NaCl, 0.1% [w/v] Tween 20, and 2% [w/v] BSA), and the secondary antibody goat anti-rabbit IgG conjugated to 10-nm gold (British Biocell International) was diluted 1:4,000 in TBST/BSA buffer. The sections were counterstained with 0.5% Safranin O, air dried, mounted with Biomount (British Biocell International) for evaluation by a Leitz Aristoplan light microscope, and documented with a Leica DC300 digital camera. To ensure the specificity of the antibody labeling, controls included (1) incubation with the rabbit preimmune serum instead of the primary antibody and (2) incubation with the secondary antibody omitting the primary antibody step.

Three-Dimensional Structural Modeling of PaIDS1
The ribbon model of PaIDS1 was created with the program Swiss-Model (version 36.0003). The crystallographic structure of avian FPP synthase was used as a template to build the PaIDS1 model structure. The PaIDS1 sequence was entered under the Web address http://swissmodel.expasy.org and created as an image with DeepView SwissPdbViewer 3.7 (Sp5). Pairwise alignments were made between the PaIDS1 model structure and avian crystal structure to establish positions and nearby contacts of analogous catalytic residues located near the substrate-binding site. The image depicted (Fig. 11) was created with the program UCSF Chimera (National Centre for Research Resources).

Construction of Mutated PaIDS1
Site-directed mutagenesis was performed with the QuikChange XL Site-Directed Mutagenesis Kit (Stratagene) following the manufacturer's instructions. The cloned PaIDS1 in the expression vector pCR-T7 CT TOPO served as a template for PCR amplification. Sequences of the primers used are listed in Supplemental Table S2. PCR was performed for 60 s at 95°C initial denaturation, 18 cycles of 60 s at 95°C, 60 s at 55°C, and 4 min at 68°C, and a final annealing for 7 min at 68°C in a Tpersonal thermocycler (Biometra). After amplification, the parental (methylated) DNA was digested with DpnI (10 units mL 21 ) for 1 h at 37°C. The DpnI-treated DNA was transformed in XL10-Gold Ultracompetent Cells (Stratagene). All mutations were confirmed by DNA sequencing, and the plasmid was transformed into E. coli BL21 (DE3) pLysS (Invitrogen).

Generation of Chimeric IDS
Chimeric versions of PaIDS1 assembled with the N-terminal or C-terminal portion of the coding sequence exchanged for the corresponding portion of the coding sequence of PaIDS2 or PaIDS5 were generated by separate PCR amplification followed by ligation. Amplification of the partial PaIDS2 and PaIDS5 sequences was performed using primers that included the BssSI restriction site at nucleotide position 652 for PaIDS2, the BanII site at position 717 for PaIDS5, and primers with the added appropriate restriction site for amplification of the PaIDS1 partial cDNA (Supplemental Table S2). The PCR conditions are as described for the functional expression of PaIDS1. Amplicons were restricted with the appropriate restriction enzyme, gel purified using of Wizard SV Gel and PCR Clean-Up System (Promega) following the manufacturer's instructions, and ligated using of T4 DNA Ligase (Invitrogen). The chimeric constructs were cloned into the vector pCR-T7 CT TOPO (Invitrogen) and transformed into One Shot Top 10 (Invitrogen) chemically competent cells. After sequencing, the constructs were transformed into E. coli BL21 (DE3)pLysS (Invitrogen).
The PaIDS1 sequence was deposited in GenBank with accession number GQ369788.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. SDS-PAGE analysis with Coomassie Brilliant Blue staining of PaIDS1 recombinant protein expressed in E. coli and partially purified.
Supplemental Figure S2. Lineweaver-Burk plots of partially purified recombinant PaIDS1 protein expressed in E. coli for determination of K m and V max. for DMAPP and IPP.
Supplemental Table S1. Sequences of primers used for site-directed mutagenesis, including the melting temperatures and the positions of the changed amino acids.
Supplemental Table S2. Sequences of primers used for the construction of chimeric IDS, including their melting temperatures.