(+)-Pinoresinol/(+)-lariciresinol reductase from Forsythia intermedia. Protein purification, cDNA cloning, heterologous expression and comparison to isoflavone reductase.

Lignans are a widely distributed class of natural products, whose functions and distribution suggest that they are one of the earliest forms of defense to have evolved in vascular plants; some, such as podophyllotoxin and enterodiol, have important roles in cancer chemotherapy and prevention, respectively. Entry into lignan enzymology has been gained by the ∼3000-fold purification of two isoforms of (+)-pinoresinol/(+)-lariciresinol reductase, a pivotal branchpoint enzyme in lignan biosynthesis. Both have comparable (∼34.9 kDa) molecular mass and kinetic (Vmax/Km) properties and catalyze sequential, NADPH-dependent, stereospecific, hydride transfers where the incoming hydride takes up the pro-R position. The gene encoding (+)-pinoresinol/(+)-lariciresinol reductase has been cloned and the recombinant protein heterologously expressed as a functional β-galactosidase fusion protein. Its amino acid sequence reveals a strong homology to isoflavone reductase, a key branchpoint enzyme in isoflavonoid metabolism and primarily found in the Fabaceae (angiosperms). This is of great evolutionary significance since both lignans and isoflavonoids have comparable plant defense properties, as well as similar roles as phytoestrogens. Given that lignans are widespread from primitive plants onwards, whereas the isoflavone reductase-derived isoflavonoids are mainly restricted to the Fabaceae, it is tempting to speculate that this branch of the isoflavonoid pathway arose via evolutionary divergence from that giving the lignans.

Lignans are a large, structurally diverse, class of vascular plant metabolites having a wide range of physiological functions and pharmacologically important properties (1,2). Because of their pronounced biological (antimicrobial, antifungal, and antiviral) (1,3), antioxidant (4,5), and anti-feedant (6) properties, a major role of lignans in vascular plants is to help confer resistance against various opportunistic biological pathogens and predators. They have also been proposed as cytokinins (7) and as intermediates in lignification (8), suggesting a critical role in plant growth and development. It is widely held that elaboration of this biochemical pathway was essential for the successful transition of aquatic plants to their vascular dry land counterparts (9) some 480 million years ago (10).
Based on existing chemotaxonomic data, the lignans are present in "primitive" plants, such as the fern Blechnum orientale (11) and the hornworts, e.g. Dendroceros japonicus and Megaceros flagellaris (12,13), with the latter recently being classified as originating in the Silurian period (14). Interestingly, evolution of both gymnosperms and angiosperms was accompanied by major changes in the structural complexity and oxidative modifications of the lignans (9,15).
In addition to their functions in plants, lignans also have important pharmacological roles. For example, podophyllotoxin, as its etoposide and teniposide derivatives, is one of the very few examples of plant anticancer agents successfully employed (1). Antiviral properties have also been reported, e.g. (Ϫ)-arctigenin (16), (Ϫ)-trachelogenin (16), and nordihydroguaiaretic acid (17) are effective against HIV due to their pronounced reverse transcriptase inhibitory activities. Some lignans, e.g. matairesinol (18), inhibit cAMP-phosphodiesterase, whereas others enhance cardiovascular activity, e.g. syringaresinol ␤-D-glucoside (19). There is also a high correlation between the presence of the "mammalian" lignans or "phytoestrogens," enterolactone and enterodiol, formed following digestion of high fiber diets, and the reduced incidence rates of breast and prostate cancers (so-called chemoprevention) (20).
The biosynthetic pathways to the lignans are only now being defined, although no purification of any enzyme, or the cloning of any gene, in their pathways has hitherto been reported. Based on radiolabeling experiments with crude enzyme extracts from Forsythia intermedia, it was established that entry into the 8,8Ј-linked lignans, which represent the most prevalent dilignol linkage known (21), occurs via stereoselective coupling of two achiral coniferyl alcohol molecules to afford (ϩ)pinoresinol (22,23), a furofuran lignan. In F. intermedia, and presumably other species, (ϩ)-pinoresinol undergoes sequential reduction to generate (ϩ)-lariciresinol and then (Ϫ)-secoisolariciresinol (24,25). (Ϫ)-Matairesinol is subsequently formed via dehydrogenation of (Ϫ)-secoisolariciresinol, further metabolism of which presumably affords lignans such as the antiviral (Ϫ)-trachelogenin in Ipomoea cairica and (Ϫ)-podophyllotoxin in Podophyllum peltatum (Fig. 1). Thus, the reductive steps giving (ϩ)-lariciresinol and (Ϫ)-secoisolariciresinol are pivotal points in lignan metabolism, since they represent entry into the furano, dibenzylbutane, dibenzylbutyrolactone, and aryltetrahydronaphthalene lignan subclasses. This paper describes the purification and characterization of (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase from F. intermedia, the cloning of its cDNA, and the heterologous expression of its recombinant ␤-galactosidase fusion protein in catalytically active form in Escherichia coli.

EXPERIMENTAL PROCEDURES
Plant Materials-F. intermedia plants were either obtained from Bailey's Nursery (var. Lynwood Gold, St. Paul, MN), and maintained in Washington State University greenhouse facilities, or were gifts from the local community.
Materials-All solvents and chemicals used were reagent or HPLC grade. Unlabeled (Ϯ)-pinoresinols and (Ϯ)-lariciresinols were synthesized as described (24). [(4R)- 3 H]NADPH was obtained as previously reported (25) by modification of the procedure of Moran et al. (26), and [(4R)-2 H]NADPH was prepared according to Anderson and Lin (27). Yeast glucose-6-phosphate dehydrogenase (type IX, 22 Oligonucleotide primers for polymerase chain reaction (PCR) 1 and sequencing were synthesized by Life Technologies, Inc. Geneclean II ® kits (Bio-101 Inc.) were used for purification of PCR fragments, with the gel-purified DNA concentrations determined by comparison to a low DNA mass ladder (Life Technologies, Inc.) in 1.5% agarose gels.
Instrumentation-1 H nuclear magnetic resonance spectra (300 and 500 MHz) were recorded on Brü ker AMX300 and Varian VXR500S spectrometers, respectively, using CDCl 3 as solvent with chemical shifts (␦ ppm) reported downfield from tetramethylsilane (internal standard). UV (including RNA and DNA determinations at A 260 ) and mass spectra were obtained on Lambda 6 UV/Vis and VG 7070E (ionizing voltage 70 eV) spectrophotometers, respectively. High performance liquid chromatography was carried out using either reversedphase (Waters, Nova-pak C 18 , 150 ϫ 3.9-mm inner diameter) or chiral (Daicel, Chiralcel OD or Chiralcel OC, 250 ϫ 4.6-mm inner diameter) columns, with detection at 280 nm (25). Radioactive samples were analyzed in Ecolume (ICN) and measured using a liquid scintillation counter (Packard, Tricarb 2000 CA). A Temptronic II thermocycler (Thermolyne) was used for all PCR amplifications. Purification of DNA for sequencing employed a QIAwell Plus plasmid purification system (QIAGEN) followed by PEG precipitation (28), with DNA sequences determined using an Applied Biosystems model 373A automated sequencer. Amino acid sequences were obtained using an Applied Biosystems protein sequencer with on-line HPLC detection, according to the manufacturer's instructions.
General Procedures for Enzyme Purification-Protein purification procedures were carried out at 4°C with chromatographic eluents monitored at 280 nm, unless otherwise stated. Protein concentrations were determined by the method of Bradford (29) using ␥-globulin as standard. Polyacrylamide gel electrophoresis used gradient (4 -15%, Bio-Rad) gels under denaturing and reducing conditions, these being performed in Laemmli's buffer system (30). Proteins were visualized by silver staining (31).
Preparation of Crude Extracts-F. intermedia stems (20 kg) were harvested, cut into 3-6-cm sections, and stored at Ϫ20°C until needed. Batches of stems (2 kg) were frozen (liquid N 2 ) and pulverized in a Waring Blendor. The resulting powder was homogenized with potassium phosphate buffer (0.1 M, pH 7.0, 4 liters), containing 5 mM dithiothreitol. The homogenate was filtered through four layers of cheesecloth into a beaker containing 10% (w/v) polyvinylpolypyrrolidone. The filtrate was centrifuged (12,000 ϫ g, 15 min). The resulting supernatant was fractionated with (NH 4 ) 2 SO 4 , with proteins precipitating between 40 and 60% saturation recovered by centrifugation (10,000 ϫ g, 1 h). The pellet was next reconstituted in a minimum amount of Tris-HCl buffer (20 mM, pH 8.0), containing 5 mM dithiothreitol (buffer A) and desalted using prepacked PD-10 columns (Sephadex G-25 medium) equilibrated with buffer A.
Affinity (Affi-Blue Gel) Chromatography-The crude enzyme preparation (191 mg in buffer A, 5 nmol h Ϫ1 mg Ϫ1 ) was applied to an Affi-Blue Gel column (2.6 ϫ 70 cm) equilibrated in buffer A. After washing the column with 200 ml of buffer A, pinoresinol/lariciresinol reductase was eluted with a linear NaCl gradient (1.5-5 M in 300 ml) in buffer A at a flow rate of 1 ml min Ϫ1 . Active fractions were stored (Ϫ80°C) until needed.
Hydrophobic Interaction Chromatography (Phenyl-Sepharose)-After thawing, 10 preparations resulting from the previous step (150 mg, 51 nmol h Ϫ1 mg Ϫ1 ) were combined and applied to a phenyl-Sepharose column (1 ϫ 10 cm) equilibrated in buffer A, containing 5 M NaCl. The column was washed with two bed volumes of the same buffer. Pinoresinol/lariciresinol reductase was eluted using a linear gradient of decreasing concentration of NaCl (5-0 M in 40 ml) in buffer A at a flow rate of 1 ml min Ϫ1 . Fractions catalyzing pinoresinol/lariciresinol reduction were combined and pooled.
Hydroxyapatite I Chromatography-Active protein (31 mg, 91 nmol h Ϫ1 mg Ϫ1 ) from the preceding step was applied to a hydroxyapatite column (1.6 ϫ 70 cm) equilibrated in 10 mM potassium phosphate buffer, pH 7.0, containing 5 mM dithiothreitol (buffer B). Pinoresinol/ lariciresinol reductase was eluted with a linear gradient of potassium phosphate buffer, pH 7.0 (0.01-0.4 M in 200 ml) at a flow rate of 1 ml min Ϫ1 . Active fractions were combined. The buffer was then exchanged with buffer A using PD-10 prepacked columns.
Affinity (2Ј,5Ј-ADP-Sepharose) Chromatography-The resulting enzyme solution (6.5 mg, 463 nmol h Ϫ1 mg Ϫ1 ) was next loaded on a 2Ј,5Ј-ADP-Sepharose (1 ϫ 10 cm) column, previously equilibrated in buffer A containing 2.5 mM EDTA (buffer AЈ), and then washed with 25 ml of buffer AЈ. Pinoresinol/lariciresinol reductase was eluted with a step gradient of NADP ϩ (0.3 mM in 10 ml) in buffer AЈ at a flow rate of 0.5 ml min Ϫ1 . (NAD ϩ (up to 3 mM) did not elute pinoresinol/lariciresinol reductase activity.) Because of the interference of the absorbance of the NADP ϩ , it was not possible to directly monitor the eluent at 280 nm. Protein concentrations for each fraction were determined spectrophotometrically according to Bradford (29).
Hydroxyapatite II Chromatography-Fractions exhibiting pinoresinol/lariciresinol reductase activity (0.85 mg, 1,051 nmol h Ϫ1 mg Ϫ1 ) were combined and directly applied to a second hydroxyapatite column (1 ϫ 3 cm), equilibrated in buffer B, with the enzyme eluted with a linear gradient of potassium phosphate buffer, pH 7.0 (0.01-0.4 M in 45 ml), at a flow rate of 1 ml min Ϫ1 .
Fast Protein Liquid Chromatography (Superose 12 and Mono Q Chromatography)-Combined fractions from the preceding step having the highest activity (50 g, 10,940 nmol h Ϫ1 mg Ϫ1 ) were pooled and concentrated to 1 ml, using a Centricon 10 microconcentrator (Amicon, Inc.). The enzyme solution was then applied in portions of 200 l to a fast protein liquid chromatography column (Superose 12, HR 10/30). Gel filtration was performed in a buffer containing 20 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 5 mM dithiothreitol at a flow rate of 0.4 ml min Ϫ1 . Pinoresinol/lariciresinol reductase was eluted with 12.8 ml of the mobile phase. The active fractions that coincided with the UV profile (absorbance at 280 nm) were pooled (20 g, 15,300 nmol h Ϫ1 mg Ϫ1 ) and desalted (PD-10 prepacked columns). This was next applied to a Mono Q HR 5/5 column (Pharmacia), equilibrated in buffer A. The column was washed with 10 ml of buffer A and pinoresinol/lariciresinol reductase activity eluted using a linear NaCl gradient (0 -500 mM in 50 ml) in buffer A at a flow rate of 0.5 ml min Ϫ1 . Aliquots (30 l) of the collected fractions were analyzed by SDS-polyacrylamide gel electrophoresis, using a gradient (4 -15% acrylamide) gel. Proteins were visualized by silver staining. Active fractions 34 -37 (27,760 nmol h Ϫ1 mg Ϫ1 ) and 38 -41 (30,790 nmol h Ϫ1 mg Ϫ1 ) were pooled separately and immediately used for characterization.
pH and Temperature Optima-To determine the pH optimum of pinoresinol/lariciresinol reductase, standard assay conditions were employed except that the buffer was replaced with 50 mM Bis-Tris propane buffer in the pH range of 6.3-9.4. The temperature optimum was examined in the range between 4 and 80°C under standard assay conditions.
pI Determination-The isoelectric point of pinoresinol/lariciresinol reductase was estimated by chromatofocusing on a Mono P HR 5/20 FPLC column. For this purpose, active fractions from the Superose 12 gel filtration column were pooled and the buffer exchanged with 25 mM Bis-Tris, pH 7.1, using prepacked PD-10 columns, equilibrated in the same buffer. The preparation so obtained was loaded on the chromatofocusing column, and a pH gradient between 7.1 and 3.9 was formed, using Polybuffer 74 as eluent at a flow rate of 0.5 ml min Ϫ1 . Aliquots (200 l) of each fraction were assayed for pinoresinol/lariciresinol reductase activities. The remainder of the fractions was used to determine the pH gradient.
Kinetic Parameters-Initial velocity studies were performed in triplicate experiments, using 50 mM Bis-Tris propane buffer, pH 7.4, containing 5 mM dithiothreitol, pure enzyme (after Mono Q anion-exchange chromatography), 10 different substrate concentrations (between 8.8 and 160 M) at a constant NADPH concentration (80 M). Incubations were carried out at 30°C for 10 min (within the linear kinetic range). Kinetic parameters were determined from Lineweaver-Burk plots.
Enzymatic Formation of (ϩ)-[(7ЈR)Ϫ 2 H]Lariciresinol-A solution of (Ϯ)-pinoresinols (5.2 mM in MeOH, 4 ml) was added to Tris-HCl buffer (20 mM, pH 8.0, containing 5 mM dithiothreitol, 22 ml) and [(4R)-2 H]NADPH (20 mM in H 2 O, 4 ml), with the whole added to the enzyme preparation (20 ml). After incubation at 30°C for 1 h with shaking, the assay mixture was extracted with EtOAc (2 ϫ 50 ml). The EtOAc solubles were combined, washed with saturated NaCl (50 ml), dried (Na 2 SO 4 ), and evaporated to dryness in vacuo. The resulting extract was reconstituted in a minimum amount of EtOAc, applied to a silica gel column (0.5 ϫ 7 cm), and eluted with EtOAc/hexanes (1:2). Fractions containing the enzymatic product were combined and evaporated to dryness to give (ϩ)-[(7ЈR)-2 H]ariciresinol (1.8 mg, 18%) as an amorphous powder. 1  Pinoresinol/Lariciresinol Reductase Amino Acid Sequencing-The (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase N-terminal amino acid sequence was obtained from each of the purified proteins, and a mixture of both, using an Applied Biosystems protein sequencer with on-line HPLC detection. For trypsin digestion, the purified enzyme (150 pmol) was suspended in 0.1 M Tris-HCl (50 l, pH 8.5), with urea added to give a final concentration of 8 M in 77.5 l. The mixture was incubated for 15 min at 50°C, following which 100 mM iodoacetamide (2.5 l) was added, with the whole kept at room temperature for 15 min. Trypsin (1 g in 20 l) was then added, with the mixture digested for 24 h at 37°C, following which TFA (4 l) was added to stop the enzymatic reaction. The resulting mixture was subjected to reversed-phase HPLC analysis (C-8 column, Applied Biosytems), this being eluted with a linear gradient over 2 h from 0 to 100% acetonitrile (in 0.1% TFA) at a flow rate of 0.2 ml/min with detection at 214 nm. Fractions containing individual oligopeptide peaks were collected manually and directly submitted to amino acid sequencing. Cyanogen bromide digestion was performed by incubation of the reductase (150 pmol) with 0.5 M cyanogen bromide in 70% formic acid for 40 h at 37°C, following which the cyanogen bromide and formic acid were removed by centrifugation under reduced pressure (SpeedVac). The resulting oligopeptide fragments were separated by HPLC and sequenced (see "Results and Discussion").
Forsythia Intermedia Stem cDNA Library Synthesis-Total RNA (ϳ300 g/g fresh weight) was obtained (32) from young green stems of greenhouse-grown F. intermedia plants (var. Lynwood Gold). A F. intermedia stem cDNA library was constructed using 5 g of purified poly(A) ϩ mRNA (Oligotex-dT TM Suspension, QIAGEN) with the ZAP-cDNA ® synthesis kit, the Uni-ZAP TM XR vector, and the Gigapack ® II Gold packaging extract (Stratagene), with a titer of 1.2 ϫ 10 6 pfu for the primary library. A portion (30 ml) of the amplified library (1.2 ϫ 10 10 pfu/ml; 158 ml total) (28) was used to obtain pure cDNA library DNA (33) for PCR.
Pinoresinol/Lariciresinol Reductase DNA Probe Synthesis-The Nterminal and internal peptide amino acid sequences were used to construct the degenerate oligonucleotide primers (see "Results and Discussion"). Purified F. intermedia cDNA library DNA (5 ng) was used as the template in 100-l PCR reactions (10 mM Tris-HCl, pH 9.0, 50 mM KCl, 0.1% Triton X-100, 2.5 mM MgCl 2 , 0.2 mM each dNTP, and 2.5 units of Taq DNA polymerase) with primer PLRN5 (100 pmol) and either primer PLRI5R (20 pmol) or primer PLRI4R (20 pmol). PCR amplification was carried out in a thermocycler as follows: 35 cycles of 1 min at 94°C, 2 min at 50°C, and 3 min at 72°C; with 5 min at 72°C and an indefinite hold at 4°C after the final cycle. Single primer, template-only and primer-only reactions were performed as controls. PCR products were resolved in 1.5% agarose gels, where a single band (380-or 400-bp) was observed for each reaction.
To determine the nucleotide sequence of the amplified bands, five 100-l PCR reactions were performed as above with PLRN5 ϩ PLRI5R and PLRN5 ϩ PLRI4R primer pairs. The five reactions from each primer pair were concentrated (Microcon 30, Amicon Inc.) and washed with TE buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA; 2 ϫ 200 l), with the PCR products subsequently recovered in TE buffer (2 ϫ 50 l). These were resolved in preparative 1.5% agarose gels. Each gel-purified PCR product (ϳ0.2 pmol) was then ligated into the pT7Blue T-vector and transformed into competent NovaBlue cells, according to Novagen's instructions. Insert sizes were determined using the rapid boiling lysis and PCR technique (with R20-mer and U19-mer primers) according to the manufacturer's instructions. Restriction analysis was performed to determine whether all inserts for each primer pair (PLRN5 ϩ PLRI5R or PLRN5 ϩ PLRI4R) were the same, as follows: to 20 l each of a 100-l PCR reaction (insert of interest amplified with R20-mer and U19-mer primers) were added 4 units of HaeIII, 1.5 units of Sau3A, or 5 units of TaqI restriction enzyme. Restriction digestions were allowed to proceed for 60 min at 37°C for HaeIII and Sau3A and at 65°C for TaqI reactions. Restriction products were resolved in 1.5% agarose gels giving one restriction group for all inserts tested. Three recombinant plasmids from PLRN5 ϩ PLRI5R (called pT7PLR1 Ϫ pT7PLR3) and two recombinant plasmids from PLRN5 ϩ PLRI4R (called pT7PLR4 and pT7PLR5) PCR products were selected for DNA sequencing; all contained the same open reading frame (ORF). The (ϩ)-pinoresinol/(ϩ)lariciresinol reductase probe was next constructed as follows: five 100-l PCR reactions were performed as above with 10 ng of pT7PLR3 DNA (containing the PLRN5-PLRI5R insert) with primers PLRN5 and PLRI5R. Gel-purified pT7PLR3 insert (50 ng) was used with Pharmacia's T7 QuickPrime ® kit and [␣-32 P]dCTP, according to kit instructions, to produce a radiolabeled probe (in 0.1 ml), which was purified over BioSpin 6 columns (Bio-Rad) and added to carrier DNA (0.5 mg/ml sheared salmon sperm DNA (Sigma), 0.9 ml).
Library Screening-600,000 pfu of F. intermedia amplified cDNA library were plated for primary screening, according to Stratagene's instructions. Plaques were blotted onto Magna Nylon membrane circles (Micron Separations Inc.), which were then allowed to air dry. The membranes were placed between two layers of Whatman ® 3MM Chr paper. cDNA Library phage DNA was fixed to the membranes and denatured in one step by autoclaving for 2 min at 100°C with fast exhaust. The membranes were washed for 30 min at 37°C in 6 ϫ standard saline citrate (SSC) and 0.1% SDS and prehybridized for 5 h with gentle shaking at 57-58°C in preheated 6 ϫ SSC, 0.5% SDS, and 5 ϫ Denhardt's reagent (hybridization solution, 300 ml) in a crystallization dish (190 ϫ 75 mm). The 32 P-radiolabeled probe (see above) was denatured (boiling, 10 min), quickly cooled (ice, 15 min), and added to a preheated fresh hybridization solution (60 ml, 58°C) in a crystallization dish (150 ϫ 75 mm). The prehybridized membranes were next added to this dish, which was then covered with plastic wrap. Hybridization was performed for 18 h at 57-58°C with gentle shaking. The membranes were washed in 4 ϫ SSC and 0.5% SDS for 5 min at room temperature, transferred to 2 ϫ SSC and 0.5% SDS (at room temperature) and incubated at 57-58°C for 20 min with gentle shaking, wrapped with plastic wrap to prevent drying, and finally exposed to Kodak X-OMAT AR film for 24 h at Ϫ80°C with intensifying screens. Twenty positive plaques were purified through two more rounds of screening with hybridization conditions as above, and eight were found to encode the desired enzyme.
In Vivo Excision and Sequencing of pPLR1-pPLR8 Phagemids-Purified cDNA clones were rescued from the phage following Stratagene's in vivo excision protocol. Both strands of the eight different cDNAs (PLR1-PLR8) that coded for (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase were completely sequenced using overlapping sequencing primers (see "Results and Discussion").
Sequence Analysis-DNA and amino acid sequence analyses were performed using the Unix-based GCG Wisconsin Package (34,35) and the ExPASy World Wide Web molecular biology server (Geneva University Hospital and University of Geneva, Geneva, Switzerland).
DNA and RNA Gel Blot Analysis with Primers PLRN5 and PLRI5R-The DNA was transferred to charged nylon membranes (Ge-neScreen Plus ® , DuPont NEN), cross-linked to the membrane (Stratalinker from Stratagene), prehybridized, hybridized with a 32 P-end-la-beled (28) degenerate oligonucleotide (primer PLRI4F) as the probe and washed according to the membrane manufacturer's recommendations for aqueous hybridization conditions. The membrane was then exposed to Kodak X-OMAT AR film for 48 h at Ϫ80°C with intensifying screens. For RNA gel blot analysis, total RNA (30 g per lane) from F. intermedia stem tips was separated by size by non-denaturing agarose gel electrophoresis. The RNA was transferred to charged nylon membranes (GeneScreen Plus ® , DuPont NEN) and visualized as above for the DNA gel blot analysis. The probe for RNA gel blot analysis was the same as that used for library screening (see above).
Expression in E. coli-Purified plasmid DNA from pPLR3 (ORF in frame with the ␤-galactosidase gene ␣-complementation particle in pBluescript) was transformed into NovaBlue cells according to Novagen's instructions. Transformed cells (5-ml cultures) were grown at 37°C with shaking (225 rpm) to mid log phase (A 600 ϭ 0.5) in LB medium (28) supplemented with 12.5 g ml Ϫ1 tetracycline and 50 g ml Ϫ1 ampicillin. IPTG (isopropyl ␤-D-thioglucopyranoside) was then added to a final concentration of 10 mM, and the cells were allowed to grow for 2 h. Cells were collected by centrifugation and resuspended in 500 l (per 5-ml culture tube) of buffer (20 mM Tris-HCl, pH 8.0, 5 mM dithiothreitol). Lysozyme (5 l of 0.1 mg ml Ϫ1 , Research Organics, Inc.) was next added, and following incubation for 10 min, the cells were   3 ϫ 15 s). After centrifugation at 14,000 ϫ g at 4°C for 10 min, the supernatant was removed and assayed for (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase activity (210 l supernatant per assay) as described above. Controls included assays of pPLR2 (cDNA out of frame) with all assay components, as well as pPLR3 and pPLR2 with no substrate except [(4R)-3 H]NADPH. Separation of products and chiral identification were performed by HPLC as described previously (25).

RESULTS AND DISCUSSION
Purification and Characterization of (ϩ)-Pinoresinol/(ϩ)-Lariciresinol Reductase-From our previous investigations (24) using crude enzyme extracts from F. intermedia, it was established that soluble, NADPH-dependent, reductases were capable of converting the sequential conversion of (ϩ)-pinoresinol into (ϩ)-lariciresinol and (Ϫ)-secoisolariciresinol, respectively (Fig. 1). While it was unclear whether more than one reductase was required to catalyze the sequential steps, the reductions proceeded via abstraction of the pro-R hydride of NADPH, resulting in an "inversion" of configuration at both the C-7 and C-7Ј positions of the products, (ϩ)-lariciresinol and (Ϫ)-secoisolariciresinol (25).
In the further study of these lignan branch point enzymes, the first objectives were the purification and characterization of the protein(s) involved. As shown in Table I, purification was achieved via an eight-step procedure involving three types of affinity chromatography, as well as hydrophobic interaction, hydroxyapatite, and gel filtration steps. This resulted in a 3060-fold purification of (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase. As for many of the enzymes involved in phenylpropanoid metabolism, the protein was in very low abundance, i.e. 20 kg of F. intermedia stems yielded only ϳ20 g of the purified (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase. Various permutations of this overall protocol were also examined, but none produced better results than that given. Interestingly, in all stages of the purification protocol, (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase activities co-eluted.
Given this observation, it was next essential to unambiguously ascertain whether more than one form of the protein existed. To this end, the 3060-fold purified protein was subjected to chromatofocusing on a Mono P column, using Polybuffer 74 as eluent, where again only a single peak of activity (corresponding to pI 5.7) was noted. Subsequent application of this preparation to a SDS-gradient gel electrophoresis (4 -15% polyacrylamide) revealed the presence of two protein bands of similar apparent molecular weight ( Fig. 2A, lane 2), whose separation was achieved via anion-exchange chromatography on an Mono Q HR 5/5 matrix (Fig. 2B). As can be seen from SDS-PAGE analysis ( Fig. 2A, lanes 3-9), they had apparent molecular masses of ϳ36 and ϳ35 kDa, respectively.
Molecular Weight-Native molecular weights of each reductase were estimated via comparison of their elution behavior to calibrated molecular weight standards obtained by analysis of their elution profiles on Superose 12, Superose 6, and Superdex 75 gel filtration FPLC columns (see "Experimental Procedures"). For each reductase, an apparent native molecular weight of 59,000 was calculated based on its elution volume, in contrast to that of ϳ36,000 and ϳ35,000 by SDS-polyacryl- (ϩ)-Pinoresinol/(ϩ)-Lariciresinol Reductase amide gel electrophoresis ( Fig. 2A). While the discrepancy between molecular weights from gel filtration and SDS-PAGE remains unknown, it can tentatively be proposed that although the native protein likely exists as a dimer, it could also be a monomer of asymmetric shape, thereby altering its effective Stokes radius (36 -39), as reported for human thioredoxin reductase and yeast metalloendopeptidase.
pI Determinations and pH/Temperature Optima-The pI values were determined by chromatofocusing on a Mono P column, using Polybuffer 74 as eluent, over the pH range of 3.9 -7.1. Each reductase form had a pIϳ 5.7. The pH optimum was examined over the range of pH 6.3 to 9.4 and found to be pH 7.4. At optimum pH, the temperature optimum for the reductase activity was established to be ϳ30°C.
Kinetic Properties-It was instructive to ascertain whether the two reductase forms catalyzed distinct reductions, i.e. that of the conversion of (ϩ)-pinoresinol to (ϩ)-lariciresinol or (ϩ)lariciresinol to (Ϫ)-secoisolariciresinol, respectively, or whether either displayed a preference for (ϩ)-pinoresinol or (ϩ)-lariciresinol as substrates. To this end, initial velocity studies were carried out, individually employing both (ϩ)-pinoresinol and (ϩ)-lariciresinol as substrates in concentrations ranging between 8.8 and 160 M, while keeping the NADPH concentration constant at 80 M. Apparent K m values were obtained from Lineweaver-Burk plots. Importantly, they were essentially the same for both forms (i.e. K m for pinoresinol, 27 Ϯ 1.5 and 23 Ϯ 1.3 M; K m for lariciresinol, 121 Ϯ 5.0 and 123 Ϯ 6.0 M). In an analogous manner, apparent maximum velocities (expressed as mol h Ϫ1 mg Ϫ1 of protein) were also essentially identical (i.e. V max for pinoresinol, 16.2 Ϯ 0.4 and 17.3 Ϯ 0.5; for lariciresinol, 25.2 Ϯ 0.7 and 29.9 Ϯ 0.7). Thus, all available evidence suggests that (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase exists as two isoforms, with each capable of catalyzing the reduction of both substrates. How this reduction is carried out, i.e. whether both reductions are done in tandem, in either quinone or furano ring form will await further study using a more abundant protein source.
Stereospecificity of Hydride Transfer-Since the two (ϩ)-pi-noresinol/(ϩ)-lariciresinol reductase isoforms exhibited essentially identical catalytic characteristics, a preparation containing both isoforms was used to examine the stereospecificity of the hydride transfer. The strategy adopted utilized selective deuterium labeling using NADP 2 H as cofactor for the reduction of (ϩ)-pinoresinol, with the enzymatic product, (ϩ)-lariciresinol, being analyzed by 1 H NMR and mass spectroscopy. Thus, (Ϯ)-pinoresinols were incubated with the enzyme preparation in the presence of stereospecifically deutero-labeled [(4R)-2 H]NADPH prepared via the method of Anderson and Lin (27). The enzymatic product was established to be (ϩ)-[(7ЈR)-2 H]lariciresinol, as evidenced by the disappearance of the 7Ј-pro-R proton at ␦ 2.51 ppm due to its replacement by deuterium and by its molecular ion at (m/z) 361 (M ϩ ϩ 1) corresponding to the presence of one deuterium atom at C-7Ј (see "Experimental Procedures"). Thus, hydride transfer from (ϩ)-pinoresinol to (ϩ)-lariciresinol had occurred in a manner whereby only the 7Ј-pro-R hydrogen position of (ϩ)-lariciresinol was deuterated. An analogous situation was observed for the conversion of (ϩ)-lariciresinol into (Ϫ)-secoisolariciresinol, thereby establishing that the overall hydride transfer was completely stereospecific (Fig. 3).

Cloning of the cDNA Encoding (ϩ)-Pinoresinol/(ϩ)-Lariciresinol
Reductase-With the purified (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase on hand, our next objectives were to prepare the corresponding Forsythia cDNA library and to obtain the gene(s) encoding the enzyme of interest. But, initial attempts to isolate functional F. intermedia RNA from fast-growing green stem tissue were unsuccessful, due to difficulties encountered via facile oxidation by its plant phenolic constituents. This problem was, however, successfully overcome by adaptation of an RNA isolation procedure, specifically designed for woody plant tissue, which uses low pH and reducing conditions in the extraction buffer to prevent oxidation (32). Poly(A) ϩ mRNA (5 g) thus obtained was used to construct a ZAP F. intermedia stem cDNA library (total titer of 1.2 ϫ 10 6 pfu), which was subsequently amplified (158 ml, 1.2 ϫ 10 10 pfu/ml). From an aliquot (30 ml) of the amplified library, the purified F. intermedia cDNA library DNA needed for subsequent PCR experimentation was isolated.
Degenerate primers (underlined in Fig. 4) for PCR amplification of the F. intermedia (ϩ)-pinoresinol/(ϩ)-lariciresinol re- ductase cDNA were next designed from the N-terminal (30 residues) and internal (following trypsin and cyanogen bromide digestion) amino acid sequences. With the N-terminal and reverse primers, two PCR products (380-and 400-bp from PLRN5 ϩ PLRI4R and PLRN5 ϩ PLRI5R primer pairs, respectively) were obtained as shown in Fig. 5. Verification of the authenticity of the amplified DNA was determined in two different ways. First, DNA gel blot analysis, using the 32 P-endlabeled degenerate primer PLRI4R (see Fig. 4) as a probe, was performed with the amplification products of primers PLRN5 ϩ PLRI5R using purified F. intermedia cDNA library DNA as template. This gave a single band in the autoradiograph, corresponding to the PCR-amplified band. Second, additional PCR analysis (using N-terminal and reverse primers PLRN5 ϩ PLRI4R) produced the 380-bp fragment from the 400-bp PCR species (data not shown).
These two PCR products were next cloned individually into a T-vector for restriction analysis and sequencing. Three restriction enzymes (HaeIII, Sau3A, and TaqI) were then each incubated with 12 and 24 individual inserts for the 380-and 400-bp fragments, respectively (see "Experimental Procedures"). This resulted in only one restriction group for each original PCR amplification product, thus showing that each was due to a single DNA species. Subsequent sequencing of five of the inserts (two from the 380-bp and three from the 400-bp fragments, respectively) revealed that both PCR fragments encoded the same ORF, which contained the N terminus and two of the internal amino acid microsequences of the reductase.
The larger (400-bp) fragment was next radiolabeled (see "Experimental Procedures") and used, under moderately stringent conditions, to probe 600,000 pfu of the amplified F. intermedia stem cDNA library. This resulted in more than 350 positive plaques, with 20 (of different signal intensities) being subjected to two additional rounds of screening. After final purification, 8 of the 20 cDNAs encoded the desired enzyme and were subcloned by in vivo excision into pBluescript. These eight cDNAs (called PLR1-PLR8) were sequenced completely with overlapping primers as shown in Fig. 6. All eight cDNAs had the same coding but different 5Ј-untranslated regions. On the other hand, analysis of the 3Ј-untranslated region of each of the eight cDNAs established that all were truncated versions of the longest cDNA's 3Ј-region. Preliminary RNA gel blot analysis (data not shown) with total RNA from greenhouse-grown plant stem tips confirmed a single transcript with a length of approximately 1.2 kilobase pairs.
Heterologous Expression of (ϩ)-Pinoresinol/(ϩ)-Lariciresinol Reductase-With the cDNAs putatively encoding (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase, our next objective was to prove that the sequence was correct by heterologously expressing the enzyme in active form. (Heterologous expression is also necessary in order to obtain sufficient protein to enable the systematic study of its precise biochemical mechanism at a future date.) Examination of the eight putative (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase clones revealed that one, PLR3, was in frame with the ␣-complementation particle of ␤-galactosidase in pBluescript (Fig. 7). This was fortuitous, since it potentially provided a facile means to express the fully functional fusion protein and, hence, to provide proof that the cloned sequence was correct. Consequently, pPLR3 was introduced into Nova-Blue E. coli cells, which were grown until mid-log phase, then induced with IPTG (10 mM final concentration), grown for a further 2 h, and analyzed for expression of (ϩ)-pinoresinol/(ϩ)lariciresinol reductase activity.
Catalytic activity was established by incubating cell-free extracts (see "Experimental Procedures") for 2 h at 30°C with (Ϯ)-pinoresinols (0.4 mM) and [4R-3 H]NADPH (0.8 mM). Following incubation, unlabeled (Ϯ)-lariciresinols and (Ϯ)-secoisolariciresinols were added as radiochemical carriers, with each lignan isolated by reversed-phase HPLC. As shown in Fig. 8, subsequent chiral HPLC analysis revealed that both (ϩ)-lariciresinol and (Ϫ)-secoisolariciresinol, but not the corresponding antipodes, were radiolabeled (total activity, 54 nmol h Ϫ1 mg Ϫ1 ). By contrast, no catalytic activity was detected either in the absence of (Ϯ)-pinoresinols or when control cells were used that contained a plasmid (pPLR2) not in frame with the ␤-galactosidase gene. Thus, the heterologously expressed (ϩ)-pino-resinol/(ϩ)-lariciresinol reductase and the plant protein function in precisely the same enantiospecific manner. Given that the reductase sequence is now correct, future efforts will be directed to obtaining the reductase proper heterologously expressed in order to obtain a sufficient amount for detailed biochemical analyses and crystals for x-ray analysis.
Sequence Analysis-The full-length sequence (see Fig. 7) of the cloned (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase (PLR3) contains all of the peptide sequences (underlined in Fig. 7) determined by Edman degradation of digest fragments (see Fig. 4). The single ORF predicts a polypeptide of 312 amino acids with a calculated molecular mass of 34.9 kDa, in close agreement with the value (ϳ35 or ϳ36 kDa) estimated previously by SDS-PAGE for (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase. An equal number of acidic and basic residues are also present, with a theoretical isoelectric point (pI) of 7.09, in contrast to that experimentally obtained by chromatofocusing (pI ϳ5.7).
The amino acid composition reveals seven methionine residues. Interestingly, the N terminus of the plant-purified enzyme lacks the initial methionine, this being the most common post-translational protein modification known. Consequently, the first methionine (see Fig. 7) in the cDNA can be considered to be the site of translational initiation. The sequence analysis also reveals a possible N-glycosylation site at residue 215 (although no secretory targeting signal is present) and seven possible protein phosphorylation sites at residues 50 and 228 (protein kinase C-type), residues 228, 250, 302, and 303 (casein kinase II-type) and residue 301 (tyrosine kinase type).
In terms of identifying the regions of the polypeptide chain involved in NAD(P)H binding (40 -42), there is a limited number of invariant amino acids in the sequences of different reductases that are viewed to be diagnostic. These include three conserved glycine residues with the sequence GXGXXG, where X is any residue, and six conserved hydrophobic residues. The glycine-rich region is viewed to play a central role in positioning the NAD(P)H in its correct conformation. In this regard, a comparison of the N-terminal region for (ϩ)-pinoresinol/(ϩ)lariciresinol reductase to that of the conserved regions of Drosophila melanogaster alcohol dehydrogenase (41), Pinus taeda cinnamyl alcohol dehydrogenase (43), dogfish muscle lactate dehydrogenase (41), and human erythrocyte glutathione reductase (41) revealed some interesting parallels. As can be seen in Fig. 9, the invariant glycine residues are aligned in every case, as are four of the six hydrophobic residues required for the correct packaging in the formation of the domain. Hence, the NADPH-binding site of (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase isoforms is localized close to the N terminus (and is circled in Fig. 7).
Homology Search: Comparison to Isoflavone Reductase-A BLAST search (44) was next conducted with the translated amino acid sequence of (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase against the nonredundant peptide data base at the National Center for Biotechnology Information. Significant homology was noted for (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase with various isoflavone reductases from the legumes, Cicer arietinum (45) (63.5% similarity, 44.4% identity), Medicago sativa (46) (62.6% similarity, 42.0% identity), and Pisum sativum (47) (61.6% similarity, 41.3% identity). This is of considerable interest since isoflavonoids are formed via a related branch of phenylpropanoid-acetate pathway metabolism. Specifically, isoflavone reductases catalyze the reduction of ␣,␤unsaturated ketones during isoflavonoid formation. For example, in M. sativa, its isoflavone reductase catalyzes the stereospecific conversion of 2Ј-hydroxyformononetin to (3R)vestitone in the biosynthesis of the phytoalexin, (Ϫ)-medi- carpin (46). This sequence similarity may be more than a minor coincidence, given that both lignans and isoflavonoids are offshoots of general phenylpropanoid metabolism, with comparable plant defense functions and pharmacological roles, e.g. as phytoestrogens. Consequently, since both reductases catalyze very similar reactions, it is tempting to speculate that the isoflavone reductases may have evolved from (ϩ)-pinoresinol/ (ϩ)-lariciresinol reductase. This is considered likely since the lignans are present in the pteridophytes, hornworts, gymnosperms, and angiosperms; hence, their pathways apparently evolved prior to the isoflavonoids.
Thus, sequence analysis establishes significant homology between (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase, isoflavone reductases, and putative isoflavone reductase homologs that do not possess isoflavone reductase activity. Indeed, it is also interesting to note that during the purification of the eight (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase cDNAs, two other cDNAs were purified from F. intermedia, which possess even higher levels of homology (75-83% similarity and 59 -74% identity) to the putative isoflavone reductase homologs. Clearly, some other biological activity for these homologs awaits discovery, and future attention will be directed toward elucidating their function in Forsythia.
Last, with both the gene encoding (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase and the corresponding protein now accessible, regulation of the lignan pathway and its parallels to the isoflavonoid pathway can be investigated. In this context, regulation of isoflavonoid biosynthesis is of much interest, since they are synthesized de novo in economically important crops upon exposure to elicitors from fungal pathogens. While research into the transcriptional control of isoflavone reductase has made some significant progress (see Dixon et al. (55) for a pertinent review), the mechanism of regulation of isoflavone reductase at the post-transcriptional level is unknown. It therefore seems significant that both (ϩ)-pinoresinol/(ϩ)-lariciresinol and isoflavone reductases contain five conserved possible phosphorylation sites, including Thr-302 (casein kinase II-type protein phosphorylase site), which are conserved in the homologs as well. Hence, it can be proposed at this point that the activities of both (ϩ)-pinoresinol/(ϩ)-lariciresinol and isoflavone reductases (and their transcript levels) are regulated by protein kinase cascades. This is highly likely, given the pivotal (e.g. branch-point) positions the enzymes play in their respective biosynthetic pathways, their roles in the physiology of the whole organism, and the ubiquity of protein kinase cascades in the regulation of biosynthetic pathways.
Concluding Remarks-Two isofunctional forms of (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase were purified (Ͼ3000-fold) to apparent electrophoretic homogeneity. Both catalyze the sequential reduction of (ϩ)-pinoresinol 3 (ϩ)-lariciresinol 3 (Ϫ)-secoisolariciresinol and have similar kinetic parameters (V max , K m ) and molecular weights. Both reductive steps proceed stereospecifically, with the incoming hydride occupying the pro-R-position at C-7 and C-7Ј in both (ϩ)-lariciresinol and (Ϫ)-secoisolariciresinol. This protein, the first isolated in the lignan pathway, is of particular interest since it catalyzes entry into the furano and dibenzylbutane lignans and is presumed to represent a regulatory point in lignan biosynthesis.
The cDNA encoding (ϩ)-pinoresinol/(ϩ)-lariciresinol reductase from F. intermedia has been cloned, and the recombinant protein heterologously expressed in E. coli as a catalytically active fusion protein. Its high sequence homology to isoflavone reductase may be of evolutionary significance in plant defense. Additionally, the conservation of putative protein phosphorylation sites in both (ϩ)-pinoresinol/(ϩ)-lariciresinol and isoflavone reductases suggests that a similar mode of regulation for both pathways may be in place. Together, these similarities suggest that although the plant defense mechanisms have been maintained, evolutionary divergence of the plant biochemical pathways has occurred. Last, since secoisolariciresinol is considered to be a major source of phytoestrogens in various high-fiber diets, it is now clear that the systematic modulation of levels of such substances can now be investigated.