Biosynthesis of the Major Tetrahydroxystilbenes in Spruce, Astringin and Isorhapontin, Proceeds via Resveratrol and Is Enhanced by Fungal Infection

.

Although much has been learned about stilbene biosynthesis in recent years (Chong et al., 2009;Jeandet et al., 2010), the formation of tetrahydroxy-substituted stilbenes [such as (7), (8), (10), and (11) in Fig. 1] is still unresolved. The stilbene skeleton is synthesized via condensation of three acetate units from malonyl-CoA to a CoA-activated phenolic acid by an enzyme known as stilbene synthase (STS), which results in the formation of a linear tetraketide intermediate. Ring closure to form the stilbene product is achieved via an intramolecular aldol condensation coupled to the loss of CO 2 (Austin et al., 2004). STSs belong to the large polyketide synthase (PKS) enzyme family, whose best known representative in plants is chalcone synthase, catalyst of the first step in flavonoid biosynthesis. For the synthesis of simple dihydroxystilbenes (5) and trihydroxystilbenes (6), the CoA-activated phenolic acid that is incorporated into the tetraketide determines the substitution pattern of the stilbene formed (Austin and Noel, 2003). However, it is not known if complex tetrahydroxystilbenes, such as those produced by spruce, originate from the incorporation of more highly substituted CoA-activated phenolic acids into the initial tetraketide or from initial formation of the basic stilbene ring system and subsequent oxidation.
Within the plant, stilbenes appear to function as protectants against various biotic and abiotic stresses. For example, their roles as constitutive or inducible antifungal defenses have been demonstrated in plants such as grape (Vitis vinifera; Kindl, 1990, 1991;Wiese et al., 1994), sorghum (Sorghum bicolor; Yu et al., 2005), and peanut (Arachis hypogaea; Sobolev, 2008). Stilbenes have also been shown to exhibit antifungal activity in non-stilbene-producing species such as poplar (Populus spp.; Seppänen et al., 2004), wheat (Triticum aestivum; Serazetdinova et al., 2005), and alfalfa (Medicago sativa; Hipskind and Paiva, 2000) when they were introduced by genetic engineering. These compounds inhibit fungal growth by interfering with microtubule assembly (Woods et al., 1995), disrupting plasma membranes, and uncoupling electron transport in fungal spores and germ tubes (Pezet and Pont, 1990). Stilbenes have also been shown to protect plants against oxidative stress (Rosemann et al., 1991;Adrian et al., 1996;He et al., 2008), to deter herbivores (Torres et al., 2003), and to inhibit the growth of competing plants (Fiorentino et al., 2008). However, their functions in conifers are poorly studied despite their abundance in widespread genera, such as spruce and pine. In spruce, it has been suggested that phenolic compounds such as stilbenes may play a pivotal role in defense against herbivores and pathogens due to the appearance of fluorescent inclusion bodies in the phloem parenchyma cells of fungus-treated bark (Franceschi et al., 2005).
The objective of this study was to learn more about the pathway of tetrahydroxystilbene formation in spruce as well as their role in tree defense. We isolated and sequenced two genes encoding STS enzymes (STS1 and STS2) from Picea abies, Picea glauca, and Picea sitchensis. In vitro enzyme assays revealed that these STSs synthesize the trihydroxystilbene resveratrol (6), which is not commonly observed in high concentrations in spruce tissue. However, by overexpressing PaSTS1 in P. abies, we could demonstrate that resveratrol is an intermediate in tetrahydroxystilbene glycoside biosynthesis in spruce. By measuring transcript accumulation and stilbene content after fungal inoculation of bark tissue, as well as testing fungal growth in vitro on stilbene-containing medium, we could also show that STS enzymes in spruce are involved in antifungal defense responses.

Identification of Spruce STS Genes and Their Phylogenetic Relationships
In order to study the stilbene biosynthetic pathway in spruce, STS gene candidates plus putative sequences for the very similar chalcone synthase (CHS) genes were identified using BLAST searches of spruce transcriptome resources with known sequences of the PKS family. Search of more than 180,000 ESTs from P. sitchensis and 250,000 from P. glauca in the Treenomix database (Ralph et al., 2008) revealed two distinct contigs from each spruce species with high similarity to STS from Pinus densiflora (Kodan et al., 2002;Supplemental Fig. S1). Additionally, more than eight distinct contigs were discovered for each spruce species that had high similarity to CHS from Pinus sylvestris (Fliegmann et al., 1992).
By using EST sequences from P. glauca and P. sitchensis as templates for primer design, two full-length STS candidates and eight CHS candidates were amplified from P. abies cDNA by PCR. All of these sequences showed features similar to those of known plant PKS genes, including a malonyl-CoA binding site, conserved amino acid residues in the dimer interface, and a conserved product-binding site (Tropf et al., 1995;Austin and Noel, 2003). Predictive algorithms suggested that N-terminal signal sequences were absent.
Comparisons of these STS and CHS sequences among spruce species demonstrated high conservation within the genus. Deduced amino acid sequences for STS1 and STS2 were 99% identical within species and 98% identical between species (Supplemental Fig.  S1). Outside the genus, spruce STSs showed between 81% and 84% sequence identity to previously characterized pine STSs encoding pinosylvin synthases (Raiber et al., 1995;Kodan et al., 2002).
Amino acid sequences for CHS were 96% to 99% identical within and 95% to 100% identical between spruce species, with the exception of CHS7, which showed only 81% to 84% sequence identity to other CHSs within the genus. Outside the genus, spruce CHSs showed 95% to 97% sequence identity to biochemically characterized CHSs from P. sylvestris (Fliegmann et al., 1992).
Phylogenetic analysis revealed a clear evolutionary divergence between PKS from angiosperms and the Pinaceae (Fig. 2). In addition, the application of neighbor-joining algorithms resulted in separate clusters for CHS and STS in the Pinaceae (P = 0.02). Within the Pinaceae STS cluster, enzymes from spruce grouped with high bootstrap support into a separate subcluster from Pinus STS enzymes (P , 0.001) but gave no resolution of orthologs. Angiosperm PKS separated into clusters according to the taxonomic affinities of their species of origin.

In Vitro Characterization of Recombinant Spruce STSs Expressed in Bacteria
Functional characterization of the two putative STSs from P. abies (PaSTS1 and PaSTS2) was accomplished via heterologous expression in a bacterial system. SDS-PAGE as well as western blotting revealed lowlevel expression of these 45-kD proteins, which form homodimeric enzymes (data not shown). Catalytic activity was determined by incubation with malonyl-CoA and a phenylpropanoid-CoA ester (Fig. 3). Reaction of the two PaSTSs with p-coumaroyl-CoA (2) yielded the stilbenes (E)-and (Z)-resveratrol (6), but neither enzyme converted cinnamoyl-CoA (1), caffeoyl-CoA (3), and feruloyl-CoA (4) to stilbene products (Table I).
In Vivo Characterization of Spruce STSs by Coexpression with 4-Coumaroyl-CoA Ligase in Escherichia coli Enzyme activities of PaSTS1 and PaSTS2 were assayed in vivo in E. coli engineered to produce potential substrates. This was achieved by the addition of phenolic acid starter units to the medium of E. coli coexpressing PaSTS along with 4-coumaroyl-CoA ligase from peanut (Watts et al., 2006), which produced CoA esters in the bacterium. Addition of p-coumaric acid to the growth medium of such cultures resulted in the production of resveratrol (6), which was released back into the medium without the formation of detectable derailment products. However, addition of caffeic acid yielded the derailment products (E)-and (Z)-6-(3,4-dihydroxystyryl)-4-hydroxy-2-pyrone (13) but no stilbenes (Table I). Ferulic acid could not be tested as a substrate in this system due to the substrate specificity of the 4-coumaroyl-CoA ligase.
In Vivo Characterization of Spruce STS Function by Overexpression of PaSTS1 in P. abies To determine the enzymatic activity of spruce STSs in planta, embryogenic tissue from P. abies was transformed with PaSTS1 under the control of the inducible promoter ubi1 using a disarmed Agrobacterium tumefaciens strain. Two transgenic PaSTS1 callus lines (lines 5 and 11) and one empty vector control line were obtained, from which somatic embryos were produced by abscisic acid application under low-light conditions. Shoots and roots were regenerated under moderate light conditions in the absence of plant growth regulators. One-year-old transformed seedlings grown in potting soil were used for transcript and metabolite analysis.
Relative transcript accumulation of PaSTS (PaSTS1 and PaSTS2 were measured together by quantitative real-time PCR) was significantly higher in both STS1 overexpression lines than in the vector control line (P , 0.001). PaSTS transcript abundance in bark tissue was on average 30-fold higher in line 5 and 6-fold higher in line 11 than in the vector control line (Fig.  4A). Relative PaSTS transcript abundance was also measured in needles, stems, and roots of line 5 and the vector control line. PaSTS transcript abundance was significantly higher in stems (P , 0.01) than in roots in both lines. Needles contained transcript levels intermediate between stems and roots (Fig. 5A).
Stilbene glycoside concentrations in both PaSTS1 lines were significantly higher than in the vector control line (P , 0.01), providing further evidence that the PaSTS1 gene encodes a functional STS (Fig.  4B). Concentrations of astringin (10) [3,5,3#,4#-tetrahy- Figure 2. Phylogenetic relationships of STS amino acid sequences. A neighbor-joining tree of the PaSTS and representative STS and CHS sequences from angiosperm and gymnosperm species is shown. Branches supported with more than 95% confidence are indicated by asterisks. Some STSs whose products have been previously characterized are resveratrol synthase (RVS) and pinosylvin synthase (PSS). The use of a similar numerical nomenclature for the three different species does not imply that these genes are orthologs, as patterns of orthology could not be resolved in the absence of a genome-wide analysis.
Stilbene Biosynthesis in Spruce droxystilbene 3#-glucopyranoside] and isorhapontin (11) [3,5,4#-trihydroxy-3#-methoxystilbene 3-glucopyranoside] in bark tissue were more than 9-fold higher in line 5 and more than 5-fold higher in line 11 than in the vector control. Stem and root tissue from PaSTS1 line 5 and the vector control contained significantly higher concentrations of stilbene glycosides (Fig. 5B) than the needle tissue (P , 0.0001). Other stilbenes, including resveratrol (6), piceid [resveratrol 3-glucopyranoside (9)], as well as the aglycones of astringin and isorhapontin [piceatannol (7) and isorhapontigenin (8)], could be detected in bark and root tissue of the PaSTS1 lines and the vector control line using liquid chromatography-mass spectrometry (LC-MS) but could not be accurately quantified due to their low concentrations (Table I).
To detect STS activities in transgenic spruce bark, crude extracts were enriched through reactive-red affinity chromatography. As in the previously described in vitro assays with E. coli expressing PaSTS1 and PaSTS2, the partially purified STS fraction from P. abies overexpressing PaSTS1 produced resveratrol (6) with the substrate p-coumaroyl-CoA (2) but only derailment products with the substrates caffeoyl-CoA (3) and feruloyl-CoA (4).

Effect of Transformation with PaSTS1 on Other Pathways of Phenolic Metabolism in Spruce
To investigate whether PaSTS1 overexpression influenced the biosynthesis of other P. abies phenolics, we measured the accumulation of flavan-3-ols, the other major group of phenolics in this species besides stilbenes, and quantified gene transcripts encoding early and late biosynthetic enzymes in flavan-3-ol formation. We measured the transcript abundance of a P. abies gene encoding phenylalanine ammonia lyase (PAL), the first step in the phenylpropanoid pathway (Noel et al., 2005), but transcript levels did not differ between PaSTS1-overexpressing lines and the vector control line (Fig. 6A). However, for leucoanthocyanidin reductase (LAR), which catalyzes the last step in flavan-3-ol biosynthesis (Tanner et al., 2003), relative transcript abundance of one of three putative PaLAR genes was significantly higher in the vector control line and in line 5 (P , 0.0001) than in line 11. No significant differences were observed for transcript levels of the other two PaLAR genes between the two transgenic PaSTS1-overexpressing lines and the vector control. Quantification of the products of LAR enzymes, catechin and proanthocyanidin dimers, revealed that concentrations of these metabolites were significantly higher in PaSTS1 line 5 (P , 0.001) than in line 11 and the vector control (Fig. 6B).

Effect of Fungal Inoculation on STS Transcription and Stilbene Glycoside Biosynthesis in P. glauca Saplings
To determine if fungal inoculation leads to the activation of stilbene biosynthesis in spruce, 2-yearold P. glauca (white spruce) saplings were wounded and inoculated with an avirulent strain of the bark beetle (Ips typographus)-associated fungus Ceratocystis polonica. Controls included unwounded saplings and those subjected to wounding without fungal inoculation. Given the ability of C. polonica to metabolize  Gershenzon, and C. Paetz, unpublished data), the nonhost P. glauca was used in this experiment to test if fungal inoculation leads to elevated stilbene levels. P. glauca is not a natural host of C. polonica due to nonoverlapping geographic distribution. Bark tissue was harvested for analysis 5, 15, or 25 d after treatment. While no lesions or typical "reaction zones" (Krokene et al., 2001) could be observed in the inner bark of the inoculated saplings 25 d after the onset of the experiment, the level of STS transcript in inoculated P. glauca bark increased significantly between days 5 and 15 (P , 0.001). Compared with the nonwounded control, transcript levels in inoculated saplings were on average 3-fold higher 5 d post inoculation and 8-fold higher 15 d post inoculation. At 25 d post inoculation, transcript levels were only 1.5-fold higher than in the nonwounded control. Saplings that were wounded without inoculation exhibited similar, although less pronounced, changes in STS transcript accumulation. Compared with the nonwounded control, relative transcript levels only reached a maximum 3-fold increase 15 d post wounding (Fig. 7B).
Fungal inoculation resulted in a significant increase in stilbene concentration in bark (P , 0.001) 15 to 25 d post inoculation. Compared with nonwounded controls, total stilbene levels were 3-fold higher in inoculated bark (Fig. 7A). In wounded but not fungus-inoculated saplings, only a small, statistically insignificant increase in total stilbene content could be detected. The increase in stilbenes was principally due to increases in the glycoside astringin, not in isorhapontin, which remained nearly constant over the time course of the experiment (data not shown).

Effect of Stilbene-Containing Spruce Extracts and Pure Tetrahydroxystilbenes on Fungal Growth
In order to determine whether spruce stilbenes had biological activity against the fungus C. polonica, artificial medium was prepared containing extract obtained from either the STS-overexpressing line 5 or the empty vector control line. The fungus was plated on the respective medium and radial growth was measured. The growth rate ( Fig. 8A) of the C. polonica isolate was significantly higher on the medium with extract from the vector control seedlings than on the medium containing extract from the transgenic STS line (P , 0.001).

Stilbene Biosynthesis in Spruce
To determine if the growth effects of the extracts were actually due to differences in their stilbene content, an additional experiment was carried out where C. polonica was grown on artificial nutrient medium amended with stilbene concentrations similar to those found in the two spruce lines. The growth rate (Fig. 8B) of the fungus was significantly higher on the medium with 0.1 mg mL 21 astringin (equivalent to the total stilbene content per gram fresh weight in the bark of the empty vector control line) than on the medium containing 1 mg mL 21 astringin (equivalent to the concentration per gram fresh weight in bark of the transgenic STS line 5; P , 0.001).

Coding Regions of STSs in the Genus Picea Are Highly Conserved
To learn more about the biosynthesis of stilbenes in the genus Picea, we investigated the genes encoding STS, which form the basic stilbene skeleton by condensing three molecules of malonyl-CoA with one phenylpropanoid-CoA molecule. Two genes were identified, STS1 and STS2, that had high sequence similarity within the three species studied, P. abies, P. glauca, and P. sitchensis.
Fossil records and phylogenetic analysis reveal that the genus Picea originated in western North America in the Paleocene era, about 62 million years ago (LePage, 2001), with P. sitchensis as the ancestral species (Ran et al., 2006). Spruce appears to have radiated from North America westward to Asia and Europe, with P. abies originating from a recent speciation event in the Pliocene era, about 5 million years ago (LePage, 2001;Ran et al., 2006). At the amino acid level, STS sequences from P. glauca, P. sitchensis, and P. abies were highly conserved, with 99% similarity within taxa and 98% sequence similarity between taxa (Supplemental Fig. S1; Supplemental Materials and Methods S1). The high similarity of STSs from the ancestral P. sitchensis to that of the more recently evolved P. abies and P. glauca indicates that genes for stilbene biosynthesis originated prior to the diversification of Picea and that these genes most likely fulfill the same function in all three spruce species studied.
The genes most closely related to STS in plants are the CHS genes and members of the PKS gene family. CHS enzymes employ the same substrates as STS but produce the C 15 flavonoid skeleton instead of the C 14 stilbene skeleton. Phylogenetic analysis of the known conifer STSs (from the genera Picea and Pinus) together  with CHS revealed that enzymes from the Pinaceae form separate clusters from angiosperm enzymes with the same function. This separation implies independent origins of STS after divergence of the angiosperm and gymnosperm lineages. In accordance with the phylogeny of the plant PKS family as reconstructed by Tropf et al. (1994), our analysis shows that PKSs in general separate into clusters that are based less on catalytic function and more on the taxonomic affinities of the species from which the genes were isolated. Our results provide further evidence that STS likely evolved from CHS on multiple occasions (Tropf et al., 1995) within stilbene-producing lineages rather than originating from a single common ancestor.
The presence of multiple STS genes in individual species of the pine and spruce genera suggests several levels of control over stilbene formation. For example, in P. sylvestris, four STS genes that are almost identical have been described that are under the control of distinct promoters that activate transcription in response to different environmental cues and at different time intervals after a single stimulus (Preisig-Mü ller et al., 1999). This level of complexity is further enhanced by the existence of multiple copies of the same STS gene in the genome with multiple 3# untranslated regions. For example, P. densiflora contains three conserved STS genes with 12 different 3# untranslated regions (Kodan et al., 2002). Attempts to determine gene copy numbers in P. abies using quantitative realtime PCR suggested the presence of numerous copies of both STS1 and STS2 genes in the spruce genome (Supplemental Fig. S2). It is conceivable, therefore, that, similar to pine, spruce may also possess multiple STS genes that modulate stilbene biosynthesis differently in response to a range of internal and external factors.

Stilbene Biosynthesis in Spruce Is a Defense Mechanism against Fungal Pathogens
One external factor that is known to influence stilbene biosynthesis is fungal challenge. In previous work, STS activity was found to increase substantially in the conifer P. sylvestris when challenged by the fungal pathogens Botrytis cinerea (Gehlert et al., 1990), Leptographium wingfieldii (Chiron et al., 2000), and Lophodermium seditiosum (Lange et al., 1994). Stilbene biosynthesis is also known to be induced in angiosperms by fungal pathogens (for review, see Jeandet et al., 2010). In our study here, inoculation of P. glauca saplings with an avirulent C. polonica isolate led to elevated transcript levels of STS followed by significant increases in tetrahydroxystilbene glycoside concentrations in the bark. Curiously, induction of stilbene biosynthesis in spruce after fungal infection could not be clearly demonstrated in earlier studies (Brignolas et al., 1995;Viiri et al., 2001). This may be due to variation in compatibility among the particular combinations of pathogens and hosts studied, a common occurrence in many plant-pathogen interactions (Vanetten et al., 1989).
It has been shown that the introduction of STSs into non-stilbene-producing species such as poplar (Seppänen et al., 2004), wheat (Serazetdinova et al., 2005), and alfalfa (Hipskind and Paiva, 2000) led to increased fungal resistance. However, it was unclear whether increased stilbene levels in spruce, which already produces a basal level of these compounds, would also lead to increased resistance against the bark beetle-associated fungus C. polonica. Since C. polonica rarely produces spores (Harrington and Wingfield, 1998), the common procedure for testing resistance to such a fungus would be direct inoculation of a wounded spruce stem with a mycelial plug (Christiansen and Solheim, 1990). However, the transgenic spruce seedlings used in this study were too young to create a sufficiently large wound on the stem for fungal inoculation. Therefore, the biological activity of stilbenes against C. polonica was studied on artificial medium.
Medium amended with extracts from a transgenic STS-overexpressing line inhibited fungal growth when compared with medium amended with extracts from empty vector control seedlings. That this result was due to differences in stilbene content was confirmed by comparing medium amended with stilbene concentrations similar to those observed in bark of the STS-overexpressing line with medium amended with stilbene concentrations similar to those in the bark of the empty vector control seedlings. Thus, the tetrahydroxystilbenes of spruce are not only induced on fungal infection, but higher concentrations can help defend against fungal pathogens, just as they have been previously shown to defend various angiosperms and pine against pathogenic fungi (Jeandet et al., 2010). The most abundant stilbenes in spruce are the 3,5,3#,4#-tetrahydroxystilbene glycosides astringin (10) and isorhapontin (11). STSs condense phenylpropanoid-CoA esters with three molecules of malonyl-CoA to form the stilbene ring system (Austin and Noel, 2003). Several possible biosynthetic pathways can be envisioned that differ according to the substrate specificity of the STS reaction and the timing of oxidation (Fig. 9). The first possibility (I) involves STS-catalyzed formation of a 3,5-dihydroxystilbene, pinosylvin, from cinnamoyl-CoA. This 3,5-dihydroxystilbene product could then be further modified by 3#-and 4#-hydroxylation and glucosylation to give astringin (10), while additional 3#-O-methylation is required to yield isorhapontion (11). Another possibility (II) involves STS formation of a 3,5,4#-trihydroxystilbene, resveratrol (6), from p-coumaroyl-CoA, which could be further modified by 3#-hydroxylation and glucosylation (and 3#-Omethylation) to give astringin (and isorhapontin). A third possibility (III) is STS formation of a 3,5,3#,4#tetrahydroxystilbene, piceatannol (7), from caffeoyl-CoA (Chong et al., 2009), which could be further transformed by glucosylation (and 3#-O-methylation). Feruloyl-CoA could also serve as a STS substrate to give a product that does not require 3#-O-methylation.
Of the phenylpropanoid-CoA substrates offered, the recombinant spruce STSs, PaSTS1 and PaSTS2, accepted only p-coumaroyl-CoA, producing the 3,5,4#-trihydroxy product resveratrol. Cinnamoyl-CoA, caffeoyl-CoA, and feruloyl-CoA were not converted to stilbene products. However, both caffeoyl-CoA and feruloyl-CoA, but not cinnamoyl-CoA, were converted to styrylpyrones. These compounds are considered to arise from the condensation of the phenylpropanoid-CoA ester with two malonyl-CoA units, followed by premature release from the active site and cyclization before the third condensation can occur. The formation of such derailment products indicates that the STS cannot convert the respective phenylpropanoid-CoA substrate to a stilbene, probably due to limitations in the size of the active site cavity (Jez et al., 2002). Styrylpyrones can also arise from STSs due to suboptimal in vitro conditions that lead to distortions in the active site cavity (Jez et al., 2002). In our work, styrylpyrone products were also formed in assays with p-coumaroyl-CoA along with stilbenes. However, the substrates caffeoyl-CoA and feruloyl-CoA gave only styrylpyrone products and no stilbenes; thus, they are unlikely to be native substrates of spruce STSs. Styrylpyrones have never been observed in planta. Moreover, the PaSTS enzymes in genetically engineered E. coli expressing a 4-coumaroyl-CoA ligase assayed in vivo and the PaSTS enzymes in the bark of P. abies overexpressing PaSTS1 assayed in vitro both converted p-coumaroyl-CoA only to the stilbene resveratrol without any styrylpyrone formation, while caffeoyl-CoA and feruloyl-CoA were only converted to styrylpyrones.
In a previous characterization of spruce STS, it was demonstrated that resveratrol as well as small amounts of piceatannol (7) and isorhapontigenin (8) were formed by partially purified enzyme from cell culture extracts of Picea excelsa (Rolfs and Kindl, 1984). Our results confirm that resveratrol is the major product formed in vitro by STS from spruce. However, we could not detect piceatannol or isorhapontigenin.
Although P. abies STS1 and STS2 produced the trihydroxystilbene aglycone resveratrol in vitro and in vivo, the major stilbenes accumulated in most spruce accessions are the tetrahydroxystilbene glycosides astringin and isorhapontin (Toscano-Underwood and Pearce, 1991;Lieutier et al., 2003). To determine if resveratrol is the precursor of the major stilbenes, P. abies was genetically engineered to overexpress PaSTS1. Overexpression lines had higher PaSTS transcript levels and produced significantly more astringin and isorhapontin than a control line transformed with an empty vector. Resveratrol levels, on the other hand, remained consistently low in both overexpression and vector control lines. Thus, PaSTS1 and PaSTS2 are involved in the biosynthesis of the major P. abies stilbenes, astringin and isorhapontin. Our in vitro characterization of PaSTS indicates that the first step of the pathway is the formation of resveratrol. This intermediate is then modified by 3#-hydroxylation, 3#-O-methylation, and 3-O-glycosylation to yield the major stilbene glycosides accumulated in spruce.
In rhubarb (Rheum rhaponticum), a similar pathway for the production of the major stilbene, 3,5,3#-trihydroxy-3#-methoxystilbene-3-O-glucoside, from resveratrol was proposed after in vivo substrate-feeding experiments (Rupprich et al., 1980). However, in this species, resveratrol and another likely intermediate, resveratrol-3-Oglucoside, appear in high concentrations in the rhizome (Pü ssa et al., 2009). In contrast, in both wild-type P. abies as well as transgenic lines overexpressing the PaSTS1 resveratrol synthase, only minute amounts of resveratrol and its 3-O-glucoside were detected. This could indicate that metabolite channeling (Winkel-Shirley, 1999) of resveratrol to the next pathway enzyme, a stilbene 3#-hydroxylase, occurs in P. abies, as has been shown for flavonoid biosynthesis in Arabidopsis (Arabidopsis thaliana; Burbulis and Winkel-Shirley, 1999). STS enzymes evolved from CHS by gene duplication and neofunctionalization. Therefore, it is not surprising that both enzymes share the same substrate, p-coumaroyl-CoA. Thus, in both stilbene and flavonoid biosynthesis, further oxidation, O-methylation, or other modifications at positions on the aromatic ring derived from the phenylpropanoid-CoA substrate must occur in later steps of the pathway.

Genetic Transformation of Spruce with PaSTS Provided New Insights into Phenylpropanoid Biosynthesis in the Pinaceae
Genetic transformation can produce unintended pleiotropic effects on occasion (Yabor et al., 2006;Abdeen and Miki, 2009) that negatively affect plant phenotype. In tobacco (Nicotiana tabacum), for example, stilbene overexpression led to male sterility (Fischer et al., 1997;Hö fig et al., 2006), probably due to shortages of substrates for flavonoid and sporopollenin production. However, in the PaSTS overexpression lines described here, despite increased stilbene formation, we detected no negative impact on flavonoid biosynthesis. In fact, in one transgenic line (line 5), the transcript levels of putative LAR enzymes (Tanner et al., 2003), responsible for the catalysis of late steps in the flavonoid pathway, were elevated as compared with the vector control. The levels of the flavan-3-ols, catechin and proanthocyanidins, the catalytic products of LAR, were also higher in the transgenic lines than in the vector control, hinting at a positive interaction between flavonoid and stilbene biosynthesis in spruce. However, more transgenic lines need to be evaluated to see if line 5 is representative of a general pleiotropic effect.
Overexpression of STS also provided hints on how the metabolic pathway is regulated. In the bark of fungus-inoculated wild-type P. glauca saplings, increases in STS transcript levels correlated well with the subsequent stilbene accumulation, suggesting regulation by the level of steady-state transcript. However, in the PaSTS-overexpressing transgenic lines, STS transcripts were much higher in stems than in needles and higher in needles than in roots. These patterns did not correlate well with stilbene accumulation, which was highest in roots, intermediate in stems, and lowest in needles. Such discrepancies may be due to differences in substrate availability or posttranscriptional regulation among organs or to stilbene transport among organs.

CONCLUSION
This study demonstrated that spruce STS enzymes, although only making resveratrol as their intermediate Figure 9. Biosynthetic pathway for astringin formation in P. abies depends on the substrate specificity of STS. Both STS1 and STS2 were found to utilize only p-coumaroyl-CoA as substrate, indicating that the most likely pathway is II.
Stilbene Biosynthesis in Spruce product, still contribute to the formation of the major tetrahydroxylated stilbenes in the tree. It could also be shown that fungal infection induced the tree to produce higher levels of STS transcript and tetrahydroxylated stilbene glycosides and that these compounds have antifungal activity. However, additional research is needed to understand what factors limit stilbene accumulation in both healthy and fungally infected spruce, as stilbene formation may be adjusted to different levels depending on the species of pathogen, the degree of infection, and the presence of other biotic and abiotic stresses. In addition, further investigation is required to determine whether any of the other roles proposed for stilbenes in plants (e.g. antiherbivore protection, allelopathy, and resistance to oxidative stress) can be supported. The ability to manipulate stilbene concentrations independently of other factors, as in the transgenic spruce lines described here, will be valuable in pursuing such work.

Identification of Putative STS and CHS Genes from Spruce
Protein sequences from various Pinaceae STSs (Pinus strobus [Raiber et al., 1995] and Pinus densiflora [Kodan et al., 2001]) and CHS (Pinus sylvestris [Fliegmann et al., 1992] and Ginkgo biloba [Pang et al., 2004]) were used to query Picea sitchensis and Picea glauca EST collections (Pavy et al., 2005;Ralph et al., 2008) by tBLASTn for candidate cDNA sequences. Open reading frames from candidate sequences were detected using the software package DNA Star version 8.02 (DNASTAR), and the absence of predicted signal peptides at the N terminus was confirmed by SignalP software (http://www.cbs.dtu.dk/ services).

Cloning and Sequencing PaSTS and PaCHS
For RNA extraction, fresh bark tissue from 4-year-old Picea abies saplings that were grown in an outdoor plot (clone 3369-Schongau; Samenklenge und Pflanzgarten, Laufen, Germany) was ground to a fine powder. Total RNA was extracted using the method developed by Kolosova et al. (2004). One microgram of total RNA was converted to cDNA in a 20-mL reverse transcription reaction using SuperScript II reverse transcriptase (Invitrogen) and 50 pmol of poly(T) 12-18 primer (Invitrogen). Gateway-compatible primers were designed for candidate sequences by using the N-and C-terminal sequences of putative P. sitchensis and P. glauca STS and CHS genes as templates (attB primer sequences are provided in Supplemental Table S1).
Pseudomature forms of PaSTS and PaCHS cDNA were PCR amplified with attB primers (Supplemental Table S1) using Platinum Taq high-fidelity DNA polymerase (Invitrogen) and purified with the QIAquick PCR purification kit (Qiagen). Gateway entry clones were made by using BP Clonase II and pDONR207 (Invitrogen) following the manufacturer's protocols.
pDONR207 constructs containing PaSTS and PaCHS genes were sequenced using 10 pmol of the vector-specific primers pDONR F (5#-TCGCGTTAACGC-TAGCATGGATCTC-3#) and pDONR R (5#-GTAACATCAGAGATTTTGAGA-CAC-3#) and the BigDye Terminator version 3.1 Cycle Sequencing Kit on an ABI Prism R 3100 sequencing system (Applied Biosystems). Sequences from each construct were assembled and translated into protein sequence using DNA Star software. Table S2) were aligned with the automatic alignment program MAFFT version 6 (mafft.cbrc.jp/ alignment/server/) using the BLOSUM 62 scoring matrix with 1.53 gapopening penalty and an offset value of 1.

STS and CHS protein sequences (Supplemental
Phylogenetic analyses were conducted using MEGA version 4 (Center for Evolutionary Medicine and Informatics; Tamura et al., 2007) employing the neighbor-joining method. Evolutionary distances were calculated with the Poisson correction. The tree was searched using the close-neighbor interchange algorithm with pairwise elimination of alignment gaps. The statistical likelihood of tree branches was tested with 10,000 bootstrap replicates.

Heterologous Expression of PaSTS Genes in Escherichia coli
Putative PaSTS1 and PaSTS2 pDONR207 constructs were cloned with LR Clonase II (Invitrogen) according to the manufacturer's instructions into the Gateway-compatible expression vector pH9GW (Yu and Liu, 2006), coding for an N-terminal His tag. Arctic Express (DE3) chemically competent E. coli (Stratagene) was transformed with expression constructs. Single colonies were inoculated into 5 mL of Luria-Bertani broth with 1 mg mL 21 kanamycin and grown for 16 h at 24°C. The 5-mL starter cultures were used to inoculate 100 mL of Overnight Express Instant TB Medium (Novagen) supplemented with 1% (v/v) glycerol and 1 mg mL 21 kanamycin.
Bacterial cultures were grown for 3 d at 12°C (220 rpm) and harvested by centrifugation. Bacteria were resuspended in 10 mL of buffer containing 50 mM Bis-Tris (pH 7.2), 10% (v/v) glycerol, 0.5 mM phenylmethylsulfonyl fluoride, and 1 mM dithiothreitol and disrupted by sonification (65% power, 3 min, two cycles) using a Bandelin Sonoplus HD 2070 (Bandelin Electronics). Insoluble cell debris was removed from the lysate by centrifugation at 4°C. Protein expression was confirmed by western blot using an anti-His horseradish peroxidase conjugate antibody (Novagen).
Expressed proteins were purified with a 1-mL His Trap FF column (GE Healthcare) on an AEKTA 900 chromatography system (GE Healthcare). The column was washed with 50 mM Bis-Tris (pH 7.2) and 10% (v/v) glycerol and eluted with wash buffer amended with 220 mM imidazole. The eluted proteins were desalted into an assay buffer (50 mM Bis-Tris, pH 7.2, 10% [v/v] glycerol, and 1 mM dithiothreitol) on DG-10 desalting columns (Bio-Rad). The protein concentration was determined using the Bradford reagent (Bio-Rad).

Functional Characterization of PaSTS
The aromatic CoA esters cinnamoyl-CoA and p-coumaroyl-CoA were synthesized enzymatically using the methods described by Beuerle and Pichersky (2002). Caffeoyl-CoA and feruloyl-CoA were chemically synthesized following methods from Brand et al. (2006). Identification of these compounds was confirmed by LC-MS with electrospray ionization (ESI).
Enzyme activities were assayed in 200-mL reaction volumes containing 150 mM malonyl-CoA (Sigma), 50 mM individual aromatic CoA esters, and 26 mg of purified enzyme. Reaction mixtures were incubated for 3 h at 28°C, stopped by acidification with 50 mL of 0.1 N HCl, and extracted with 3 volumes of ethyl acetate. The ethyl acetate extracts were evaporated under nitrogen gas flow and redissolved in 50 mL of methanol for LC-ESI-MS analysis. Negative control assays were initiated with heat-denatured enzyme preparations.
Enzyme extracts from transgenic P. abies-overexpressing PaSTS1 were prepared by extracting 100 mg of bark tissue using the method of Martin et al. (2002). After pelleting plant debris, the extract was passed over a 1-mL reactive red agarose (Sigma) affinity chromatography column. The column was washed with 5 column volumes of assay buffer and eluted with 2 column volumes of assay buffer amended with 500 mM NaCl. STS activity was determined as above.
Bacterial cultures were grown for 8 h at 22°C followed by 4 d at 12°C before adding p-coumaric or caffeic acid in dimethyl sulfoxide to the culture medium as substrate to a final concentration of 1 mM. Culture medium (1 mL) was harvested 24, 48, and 72 h after the addition of phenolic acids. After centrifuging the culture medium, the supernatant was acidified with 0.1 N HCl and extracted with 1 volume of ethyl acetate. After evaporation, the extract was redissolved in 50 mL of methanol for LC-ESI-MS analysis.

Genetic Transformation of P. abies Callus with PaSTS1
The PaSTS1 pDONR207 was cloned with LR Clonase II (Invitrogen) into the Gateway-compatible binary vector pCAMGW. pCAMGW STS1 or pCAM-BIA 2301 (as vector control) was transformed into the chemically competent disarmed Agrobacterium tumefaciens strain C58/pMP90 (Schmidt et al., 2010). An embryonic P. abies cell culture (line 186/3c VIII) was transformed as described by Schmidt et al. (2010). From the six kanamycin-resistant transgenic lines obtained, line 5 and line 11 were selected for regeneration of seedlings and further experiments.

Somatic Embryogenesis and Plant Regeneration
Transgenic embryonic tissue was maintained at 24°C in the dark and subcultured every 14 d. For plant regeneration, transgenic embryonic tissue was placed on semisolid EMM1 medium (Walter et al., 1999) amended with 6 g L 21 Gelrite (Duchefa), 30 g L 21 Suc, and 15 mg L 21 abscisic acid for 2 weeks. The culture was then placed on EMM2 medium (4.5 g L 21 Gelrite; Walter et al., 1999) for 5 weeks. Embryos (3-5 mm in size) were placed on a sterile nylon mesh in three wells of a six-well cell cluster. The other three wells were filled with sterile water to maintain humidity. Embryos were stored at 4°C for 7 d in the dark. For germination, developed embryos were transferred to modified Litvay medium (Klimaszewska et al., 2005)  Procedures from Klimaszewska et al. (2001) were used to generate the untransformed P. glauca line Pg653.

Quantitative Real-Time PCR
Total RNA for all quantitative real-time PCR experiments was isolated with the Invitrap Spin Plant RNA Mini Kit (Invitek). RNA was quantified spectrophotometrically.
Reverse transcription of 1 mg of RNA into cDNA was accomplished using SuperScript II reverse transcriptase (Invitrogen) and 50 pmol of poly(T) 12-18 primer (Invitrogen). After cDNA was diluted to 10% (v/v) with deionized water, 1 mL of diluted cDNA was used as template for quantitative real-time PCR. PCR was performed with Brilliant SYBR Green QPCR Master Mix (Stratagene) with 10 pmol of the forward primer 5#-GTGGCGAGCAGAA-CACAGACTTC-3# and 10 pmol of the reverse primer 5#-CAGCGATGG-TACCTCCATGAACG-3#, designed to amplify 140 bp of both STS1 and STS2 simultaneously. Primer sequences for PaLAR1, PaLAR2, PaLAR3, and PaPAL are given in Supplemental Table S3. PCR was performed using a Stratagene MX3000P thermocycler: 5 min at 95°C, followed by 40 cycles of 30 s at 95°C, 30 s at 55°C, and 30 s at 72°C, followed by a melting curve analysis from 55°C to 95°C. Reaction controls included nontemplate controls and non-reversetranscribed RNA. Transcript abundance was normalized to the transcript abundance of the ubiquitin (Schmidt et al., 2010; GB:EF681766.1) gene (Supplemental Table S3) and was calculated from three technical replicates each of at least four biological replicates. Relative transcript abundance for the transgenic PaSTS1 lines was calibrated against the transcript abundance of one biological replicate of the vector control. Relative transcript abundance of PgSTS in P. glauca was calibrated against a nonwounded control sample.

Extraction of Phenolic Compounds from Spruce
Spruce tissue was ground to a fine powder in liquid nitrogen and lyophilized using an Alpha 1-4 LDplus freeze dryer (Martin Christ). Approximately 5 mg of dried tissue was extracted with 2 mL of methanol for 4 h at 4°C and then centrifuged at 2,000g, and the supernatant was recovered. Insoluble material was reextracted with 1 mL of methanol for 12 h. The supernatants were combined and evaporated to dryness under a stream of nitrogen and redissolved in 1 mL of methanol containing 50 mg mL 21 chlorogenic acid (Sigma) as an internal standard. For LC-ESI-MS, samples were diluted to 20% (v/v) with methanol.

LC-ESI-MS
Compounds were separated on a Nucleodur Sphinx RP18ec column (250 3 4.6 mm; particle size of 5 mm; Macherey-Nagel) using an Agilent 1100 series HPLC system (Agilent Technologies) with a flow rate of 1.0 mL min 21 at 25°C. Compound detection and quantification were accomplished with an Esquire 6000 ESI ion-trap mass spectrometer (Brucker Daltronics) after diverting column flow-through in a ratio of 4:1. The mass spectrometer was operated as follows: skimmer voltage, 60 V; capillary voltage, 4,200 V; nebulizer pressure, 35 pounds per square inch; drying gas, 11 L min 21 ; gas temperature, 330°C.
Phenolic compounds from spruce and enzyme assay products were separated using 0.2% (v/v) formic acid and acetonitrile as mobile phases A and B, respectively, with the following elution profile: 0 to 1 min, 100% A; 1 to 15 min, 0% to 65% B; 15 to 18 min, 100% B; 18 to 22 min, 100% A. The ESI-MS apparatus was operated in negative mode scanning mass-to-charge ratio (m/z) between 50 and 1,600 with an optimal target mass of 405 m/z. Capillary exit potential was kept at 2121 V.
Aromatic CoA esters to be used as enzyme substrates were separated using 20 mM ammonium acetate (pH 4) and acetonitrile as mobile phases A and B, respectively, with the following elution profile: 0 to 20 min, 5% to 45% B; 20 to 21 min, 45% to 90% B; 21 to 23 min, 90% B; 23 to 28 min, 5% B. The ESI-MS apparatus was operated in alternating mode scanning between m/z 50 and 1,600 with an optimal target mass of 250 m/z. Capillary exit potential was kept at 2109.8 V.

Compound Identification and Quantification
Several substances were identified by direct comparison with authentic standards, including resveratrol (purchased from Merck) and piceatannol (from Alexis Biosciences). An astringin (10) standard was extracted and purified from spruce bark following a previous protocol (Li et al., 2008), and the identity and purity of this compound were verified by NMR spectroscopy (Supplemental Fig. S3). Standards were unavailable for the other compounds detected, (E)-and (Z)-isorhapontin (11), (E)-and (Z)-6-(3,4-dihydroxystyryl)-4-hydroxy-2-pyrone (13), 6-(3 methoxy,4-hydroxystyryl)-4-hydroxy-2-pyrone (14), as well as (E)-and (Z)-4-hydroxy-6-(4-hydroxystyryl)-2-pyrone (12), and these were identified by mass spectra obtained by negative ionization comparing second-and thirdorder losses with those displayed by the available standards or previously described in the literature. For example, isorhapontin (11) [retention times in Supplemental Table S4] was characterized by a neutral loss of 162, just as seen for astringin (10), yielding a fragment with the same m/z value as the respective deprotonated aglucone (Supplemental Table S4; Supplemental Figs. S4 and S5). Fragmentation of resveratrol (6), piceatannol (7), as well as isorhapontin (11) treated with b-glucosidase prior to analysis was characterized by the appearance of two dominant product ions that represented the sequential neutral loss of two 42-D molecules (Supplemental Table S4; Supplemental Fig. S4). Similar fragmentation spectra have been previously described for resveratrol and piceatannol (Buiarelli et al., 2007;Lo et al., 2007), and the two 42-D molecules lost during fragmentation of these compounds have been identified as ketene molecules involving positions 3, 4, 5, and 6 on the stilbene skeleton (Supplemental Fig. S5). Fragmentation of the styrylpyrones [(12), (13), and (14)], first described by Yamaguchi et al. (1999) as derailment products in PKS enzyme assays, were characterized by the neutral loss of two fragments of 44 and 42 D. Fragments arising from styrylpyrones have the same m/z as those observed during the fragmentation of stilbenes (Supplemental Fig. S4) and therefore represent the sequential loss of a carbon dioxide and a ketene molecule involving positions 2, 3, and 4 on the pyrone ring (Supplemental Fig. S5).
For quantification, compounds were detected in full-scan mode using negative ionization. Ions were extracted from total ion chromatograms using Brucker Daltronics Quant Analysis version 3.4 software employing a standard smoothing width of 3 and Peak Detection Algorithm version 2. Linearity in ionization efficiencies was verified by analyzing serial dilutions of randomly selected samples. External calibration curves for catechin (Sigma) and astringin were created by linear regression. Flavan-3-ol concentrations were determined relative to the catechin calibration curve and stilbene glycosides relative to astringin. All results were normalized relative to the internal standard.

Treatment of P. glauca Saplings with
Ceratocystis polonica A C. polonica isolate (CMW 7749) provided by the culture collection of the Forestry and Agricultural Biotechnology Institute (University of Pretoria) was grown on 3% (w/v) malt extract agar (Carl Roth) for 12 d at 25°C. P. glauca saplings originating from embryonic tissue (Pg653) were grown in a walk-in growth chamber for 2 years in 2-L pots under a light/temperature regime alternating between 4 months of exposure to 16 h of light at 25°C and 4 months of exposure to 8 h of light at 16°C. Inoculations with C. polonica were performed after saplings were acclimatized for 6 weeks to the higher light/ temperature regime and had completed their "spring" flush.
A bark plug, 4 mm in diameter, was removed from the lower part of the sapling between the second and third branch whorl with a cork borer. A 4-mm plug from the C. polonica culture was placed into the wound with the mycelium oriented toward the wood surface and sealed with Parafilm. For the wounded control treatment, plugs of sterile malt extract agar were inserted.
Bark tissue from inoculated and wounded saplings was harvested 5, 15, and 25 d after the onset of the experiment. Four saplings were sacrificed per time point for each treatment. Bark material was flash frozen in liquid nitrogen and stored at 280°C.

Determination of in Vitro Fungal Growth on Stilbene-Containing Medium
To obtain sufficient stilbene-containing extract for fungal bioassays, extractions were performed on the organs containing the highest stilbene content, the roots. Root tissue from the STS-overexpressing line 5 and the empty vector control was extracted with water (25 mL g 21 tissue) at 4°C overnight and centrifuged. The supernatant was retained, and the insoluble debris was reextracted for 5 h with water (10 mL g 21 tissue). Extracts were combined and sterilized by filtration through a membrane (pore size of 0.2 mm). Solid fungal growth medium was prepared by steam sterilizing 5% (w/ v) agar in water (Carl Roth) and adding root extracts at a ratio of 1:1 to the hot medium. Medium (15 mL) was dispensed in petri dishes (8.4 cm in diameter).
For assays testing fungal growth on astringin, growth medium was prepared by steam sterilizing water agar (4%, w/v) amended with carrot (Daucus carota) juice (2% volume 21 ). Solutions of 4 and 0.4 mg mL 21 astringin in water were prepared and sterilized by filtration through a membrane (pore size of 0.2 mm). Astringin solutions and the agar mixture were mixed at a ratio of 1:2, and 8 mL was dispensed in petri dishes (5.2 cm in diameter).
After the medium solidified, an agar plug (4 mm in diameter) from 14-dold C. polonica stationary culture was placed in the middle of each petri dish, sealed with Parafilm, and incubated at 26°C in the dark. Fungal growth was measured every 24 h until growth reached the margins of the petri dish.

Statistical Analysis
Graphical representations of results are presented as means 6 SE. Statistical significance of differences in STS-transformed lines was determined using a one-way ANOVA on log-transformed data. Significance of differences in the C. polonica inoculation trial was determined using a two-way ANOVA. Differences in means were calculated using Tukey's posthoc pairwise comparisons test at a 95% confidence level. Differences in fungal in vitro growth were calculated using linear models. Analyses were conducted using the open source software R version 2.81 (www.r-project.org) and the LAERCIO package for pairwise comparisons.
Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers JN 400047 to JN 400078.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S2. Relative PaSTS gene copy number in transgenic PaSTS1 lines and vector control.
Supplemental Figure S3. The NMR spectrum measured for astringin isolated from P. abies.
Supplemental Figure S4. Representative mass fragmentation spectra of tetrahydroxylated stilbene compounds reported in this study.
Supplemental Figure S5. Hypothetical mass fragmentation reactions of stilbenes leading to the fragmentation spectra used in identification.
Supplemental Table S1. Forward and reverse attB primers for amplifying and cloning P. abies STS and CHS into pDONR207.
Supplemental Table S2. National Center for Biotechnology Information accession numbers of CHS and STS sequences used for phylogenetic analysis.
Supplemental Table S3. Forward and reverse primers for quantitative real-time PCR of P. abies.
Supplemental Table S4. Analytical data for stilbenes and derailment products.
Supplemental Materials and Methods S1. Quantitative genomic PCR for copy number determination.