Form of Taxadiene Synthase Involved in Paclitaxel

Archives of Biochemistry and Biophysics Vol. 379, No. 1, July 1, pp. 137–146, 2000 doi:10.1006/abbi.2000.1865, available online at http://www.idealibrary.com on

Heterologous Expression and Characterization of a “Pseudomature” Form of Taxadiene Synthase Involved in Paclitaxel (Taxol) Biosynthesis and Evaluation of a Potential Intermediate and Inhibitors of the Multistep Diterpene Cyclization Reaction1

David C. Williams,* Mark R. Wildung,* Alan Qingwu Jin,† Dolan Dalal,‡ John S. Oliver,‡ Robert M. Coates,† and Rodney Croteau2 *Institute of Biological Chemistry, Washington State University, Pullman, Washington 99164-6340; †Department of Chemistry, University of Illinois, 600 South Matthews Avenue, Urbana, Illinois 61801; and ‡Department of Chemistry, Box H, Brown University, Providence, Rhode Island 02912

Received February 23, 2000, and in revised form April 4, 2000

with kinetics comparable to the native enzyme. In ad- The diterpene cyclase taxadiene synthase from yew dition to the major product, taxa-4(5),11(12)-diene (Taxus) species transforms geranylgeranyl diphos- (94%), this enzyme produces a small amount of the phate to taxa-4(5),11(12)-diene as the first committed isomeric taxa-4(20),11(12)-diene (ϳϳ5%), and a product step in the biosynthesis of the anti-cancer drug Taxol. tentatively identified as verticillene (ϳϳ1%). Isotopi- Taxadiene synthase is translated as a preprotein bear- cally sensitive branching experiments utilizing (4R)- 2 ing an N-terminal targeting sequence for localization [4- H1]geranylgeranyl diphosphate confirmed that the to and processing in the plastids. Overexpression of two taxadiene isomers, and a third (taxa-3(4),11(12)- the full-length preprotein in Escherichia coli and pu- diene), are derived from the same intermediate tax- rification are compromised by host codon usage, inclu- enyl C4-carbocation. These results, along with the fail- sion body formation, and association with host chap- ure of the enzyme to utilize 2,7-cyclogeranylgeranyl erones, and the preprotein is catalytically impaired. diphosphate as an alternate substrate, indicate that Since the transit peptide-mature enzyme cleavage site the reaction proceeds by initial ionization of the could not be determined directly, a series of N-termi- diphosphate ester and macrocyclization to the verti- nally truncated enzymes was created by expression of cillyl intermediate, followed by a secondary cycliza- the corresponding cDNAs from a suitable vector, and tion to the taxenyl cation and deprotonation (i.e., for- each was purified and kinetically evaluated. Deletion mation of the A-ring prior to B/C-ring closure). Two of up to 79 residues yielded functional protein; how- potential mechanism-based inhibitors were tested ever, deletion of 93 or more amino acids resulted in with recombinant taxadiene synthase but neither pro- complete elimination of activity, implying a structural vided time-dependent inactivation nor afforded more or catalytic role for the amino terminus. The than modest competitive inhibition. © 2000 Academic Press pseudomature form of taxadiene synthase having 60 Key Words: Taxol biosynthesis; paclitaxel; taxadiene amino acids deleted from the preprotein was found to synthase; Taxus; yew. be superior with respect to level of expression, ease of purification, solubility, stability, and catalytic activity

Taxadiene synthase from Taxus (yew) species cata- 1 This investigation was supported in part by grants CA-55254 lyzes the ﬁrst committed step in the biosynthesis of the (R.C.), GM-13956 (R.M.C.), and GM-31354 (R.C.) from the National 3 Institutes of Health, by Cytoclonal Pharmaceutics, and by McIntire- anti-cancer drug Taxol and related taxane diterpe- Stennis Project 0967 from the Washington State University Agricul- tural Research Center. 2 To whom correspondence and reprint requests should be ad- 3 Paclitaxel is the generic name for Taxol, which is now a regis- dressed. Fax: (509) 335-7643. E-mail: [email protected]. tered trademark of Bristol-Meyers Squibb. Because of the far greater

FIG. 1. Cyclization of geranylgeranyl diphosphate to taxa-4(5),11(12)-diene and subsequent elaboration to Taxol. Ac, acetyl group; Bz, benzoyl group; OPP, diphosphate moiety. noids by the cyclization of the universal precursor of where proteolytic processing of the transit peptide oc- diterpenes, geranylgeranyl diphosphate, to the olefinic curs and where the mature form of the enzyme func- product (Fig. 1) (1). This enzyme catalyzes a slow, but tions in the first step of Taxol biosynthesis. The pre- apparently not rate limiting, step in the Taxol biosyn- protein forms of the terpene synthases are often suffi- thetic pathway (2). Taxadiene synthase has been iso- ciently active to permit identification of clones by lated from yew saplings (3) and cell cultures (2), and functional heterologous expression from standard clon- the enzyme exhibits properties typical of the few diter- ing vectors (7), and the taxadiene synthase was con- pene synthases from gymnosperms and angiosperms firmed by this means (6). However, in high level ex- thus far defined (4). Initial assessment of the mecha- pression systems, the presence of the transit peptide of nism of taxadiene synthase (5) indicates that the reac- the terpene synthases often contributes to the formation involves the ionization and cyclization of gera- tion of inclusion bodies and can compromise purifica- nylgeranyl diphosphate to a transient verticillyl cation tion of the soluble form(s) by promoting aggregation intermediate (1), with intramolecular transfer of the and/or the tight association with host chaperones and C11 proton to C7 to initiate transannular B/C-ring other proteins (8). Additionally, the preproteins may be closure to the taxenyl cation (2), followed by deproto- catalytically impaired relative to their native counter- nation at C5 to yield the taxa-4(5),11(12)-diene prod- parts. Soluble taxadiene synthase has been overex- uct; however, few other details of the stereochemical pressed recently as a His-tagged, thioredoxin fusion of mechanism or of the structure of this novel enzyme are the preprotein (9). The enzyme is catalytically capable known. when expressed in this form; however, such preprotein A cDNA encoding the taxadiene synthase from fusions are unsuitable for detailed mechanistic and Taxus brevifolia has been obtained by a homology- crystallographic study. based PCR4 cloning method (6). Like the monoterpene Deciphering the transit peptide-mature protein synthases and other diterpene synthases (7), taxadiene cleavage junction of terpene synthase preproteins from synthase is translated as a preprotein bearing an N- sequence information alone is not possible because of terminal targeting sequence for localization to plastids the considerable variation in primary sequence of these targeting peptides (10–13), and studies with the native enzymes to define this feature directly have indicated that most are N-terminally blocked (8, 14). Moreover, familiarity of the word Taxol, we use it in this paper in lieu of paclitaxel. detailed mass spectrometric analysis of a prototype 4 Abbreviations used: DTT, dithiothreitol; EDTA, ethylenediami- terpene synthase, limonene synthase, indicated that notetraacetic acid; FPLC, fast protein liquid chromatography; GC- plastidial proteolytic processing to the native form in MS, gas chromatography-mass spectrometry; Hepes, N-(2-hydroxy- this case was quite imprecise (8). Therefore, to prepare ethyl)piperazine-NЈ-(2-ethanesulfonic acid); LSC, liquid scintillation counting; Mopso, 3-(N-morpholino)-2-hydroxypropanesulfonic acid; a “pseudomature” form of such a terpene synthase PCR, polymerase chain reaction; SDS–PAGE, sodium dodecylsul- requires the empirical construction and testing of a fate-polyacrylamide gel electrophoresis. series of truncated versions of the target protein. In HETEROLOGOUS EXPRESSION OF TAXADIENE SYNTHASE 139

FIG. 2. Possible mechanisms for the utilization of substrate analogs by taxadiene synthase. OPP, diphosphate moiety. this paper, we describe the construction, heterologous 65 min). The unstable chlorides, contaminated by about 10% of the methylenecyclohexene elimination product, were characterized by expression and purification of a series of truncated 1 13 forms of the preprotein, report on the kinetic proper- IR, H NMR, and C NMR spectra, and used without purification. Selected NMR data (benzene-d , 400 MHz) for 2,7-cyclogeranylgera- ties, solubility and stability of these “pseudomature” 6 nyl chloride: ␦H 1.04, 1.56, 1.60 (3s, 3 ϫ 3H,3CH3), 3.83 and 3.93 forms of taxadiene synthase, and discuss the mecha- (ABq, 2H, J ϭ 11.4 CH2Cl), 5.24 (m, 2H, vinyl H); ␦c 41.2 (CH2Cl). nistic implications of several new findings revealed by Reaction of the chlorides with (Bu4N)3HP2O7 (1.08 equiv) in CH3CN studies with the recombinant enzyme. (1 ml) containing powdered 3 Å molecular sieve was carried out for 48 h at room temp. according to literature procedures (20, 21). Purification by ion exchange chromatography, selective precipita- MATERIALS AND METHODS tion, and cellulose chromatography afforded 3(16 mg, 26%) and [1,1- 2 Substrates, standards, and inhibitors. (E,E,E)-[1-3H]Gera- H2]3 (7.9 mg, 10%) as white solids, which were characterized by 1 31 nylgeranyl diphosphate (120 Ci/mol) was prepared as described pre- TLC, H, and P NMR spectra, and negative ion FAB-MS. Data for 1 ϩ ϩ viously (15). (Ϯ)-Taxa-(4)5,11(12)-diene and (Ϯ)-taxa-4(20),11(12)- 3: H NMR (D2O 0.06% (w/v) EDTA concentrated NH4OH to pH ␦ ϫ diene (16) were gifts from Robert M. Williams, Colorado State Uni- 9, 400 MHz) 0.88 (s, 3 H, CH3), 1.41 (s, 6 H, 2 CH3), 1.48 (s, 3 H, versity (Fort Collins, CO). All other biochemicals and reagents were CH3), 1.59 (s, 3 H, CH3), 1.04–1.20 (m, 2 H, CH2), 1.21–1.34 (m, 1 H), purchased from Sigma Chemical Co. or Aldrich Chemical Co., unless 1.42–1.54 (m, 1 H), 1.56–1.68 (m, 1 H), 1.68–1.94 (m, 8 H, CH2), 4.25 ϭ otherwise noted. and 4.35 (ABq, 2 H, J 9.4 Hz, CH2O), 4.63 (s, NH4), 4.97 (m, 2 H, 31 ϩ ϩ (Ϯ)-2,7-Cyclogeranylgeranyl diphosphate (3, Fig. 2) and its deute- vinyl H); P NMR (D2O 0.06% (w/v) EDTA concentrated NH4OH 2 ␦ Ϫ ϭ Ϫ ϭ rium labeled form [1,1- H2]3 were prepared from 2,7-cyclogeranylge- to pH 9, 162 MHz) 6.2 (d, J 20.8 Hz), 10.2 (d, J 20.8 Hz); ϩ raniol (17) by conversion to the respective chlorides and displace- Low resolution negative ion FAB-MS: m/z 449.22 M . Spectra of 2 1 ment with pyrophosphate trianion. Reduction of methyl 2,7-cy- [1,1- H2]3 were identical to those of 3 except for the following: H 2 ␦ clogeranylgeranoate with LiAl H4 (17) afforded the corresponding NMR: 4.25 and 4.35 (CH2OPP) absent; negative ion FAB MS: low 1 ϩ labeled alcohol (305 mg, 77%, 99% d2 by H NMR integration) after resolution: m/z 451.2 (M ); high resolution Calcd for C20H33D2O7P2, purification by flash chromatography (18) on silica gel. The 1H and 451.198359; Found, 451.197500. 13 C NMR spectra (CDCl3, 400 MHz) were identical to those of the For preparation of the 13-cyclopropylidene analog of geranylgera- unlabeled alcohol except for the following: ␦H 4.07 and 4.18 (CH2OH) nyl diphosphate (4, Fig. 2), geranylgeraniol (5 mmol) was first con- missing; ␦c 58.0 (quintet, J ϭ 21.7 Hz, CD2OH). Conversions to the verted to the acetate ester (22) (91.5% yield following flash chroma- unlabeled and labeled allylic chlorides (75 mg, 95%; 97 mg, 92%, tography (18)) and then to the 14,15-epoxide by treatment with respectively) were carried out by the Meyers procedure (19), using 6 3-chloroperoxybenzoic acid in CH2Cl2 (23). The epoxide (2 mmol) was equiv CH3SO2Cl, 5 equiv LiCl, and 8.8 equiv of collidine (DMF, 0°C, transformed to 14,15-dihydroxygeranylgeranyl acetate (25% yield) 140 WILLIAMS ET AL.

by treatment with HClO4, and the diol then cleaved with excess Cells were harvested by centrifugation (3000g for 15 min) and then periodate to afford the geranylgeranyl acetate trisnoraldehyde in resuspended in 50 ml of buffer A (25 mM Mopso (pH 6.8), containing quantitative yield (24, 25). Wittig coupling of the trisnoraldehyde 10% (v/v) glycerol and 1 mM DTT) supplemented with 1 mM EDTA (0.3 mmol) with excess cyclopropyltriphenylphosphonium bromide, and 1 mM phenylmethylsulfonyl fluoride. Following one freeze-thaw followed by hydrolysis of the ester, gave 13-cyclopropylidene- cycle, the collected cells were disrupted by sonication (large probe, 14,15,16-trisnor-geranylgeraniol (50 mg, 35%) (22); this alcohol was medium power, 2 ϫ 1 min, with cooling) and the resulting suspension converted to the bromide and diphosphorylated as before (26). The was centrifuged (30,000g for 30 min) to remove debris. After addition resulting 13-cyclopropylidene analog of geranylgeranyl diphosphate of MgCl2 to a final concentration of 10 mM, the clarified supernatant (4) was purified by combination of cation-exchange and reversed was added to a conical centrifuge tube containing 10 g of ceramic 1 phase chromatography. H NMR (D2O, 400 MHz, HOD in D2Oas hydroxyapatite (Bio-Rad, Hercules, CA) equilibrated with buffer A. internal standard (␦ 4.6)). ␦ 5.58 (tt, 1 H, J 1 ϭ 7 Hz, J 1 ϭ 2 Hz), 5.27 The sample was agitated gently (by turning on a Barnstead/Thermo- (t, 1 H, J ϭ 7 Hz), 4.99 (app t, 2 H), 4.15 (t, 2 H, J ϭ 6.4 Hz), 2.12 line Labquake) for 10 min at 4°C, the gel allowed to settle, and the (app. t, 2 H, J ϭ 7 Hz), 1.97–1.83 (m, 10 H), 1.53 (s, 3 H), 1.42 (s, 3 liquid decanted. By successive cycles of this procedure, the matrix 31 H), 0.84 (s, 4 H). P NMR (D2O, 162 MHz). ␦ Ϫ 6.55 (d, J ϭ 22.10 was washed with buffer A then eluted with buffer A containing 30 Hz), Ϫ9.69 (d, J ϭ 22.15 Hz). mM potassium phosphate. For the preparation of the vinyl analog of geranylgeranyl diphos- The resulting, partially purified enzyme was filtered (0.2 ␮m sy- phate (16-methylidene-geranylgeranyl diphosphate, 7 of Fig. 2), ringe filter, Nalgene, Rochester, NY) then loaded onto a MonoQ 10/10 geranylgeranyl acetate (2.4 mmol) was converted to the terminal anion exchange column (Pharmacia FPLC, Piscataway, NJ) previ- alcohol (34% purified yield) with SeO2/t-butyl hydroperoxide (27) ously equilibrated with buffer A. The column was washed with buffer which was oxidized to the corresponding aldehyde with MnO2 (28) in A until the baseline stabilized (UV monitoring at A 280) and then was 91% yield following flash chromatography (18). Coupling of the ter- eluted with a 150 ml linear gradient from 0 to 30% buffer B (buffer minal aldehyde with methyltriphenylphosphonium bromide (29) pro- A plus 1 M NaCl). Taxadiene synthase eluted at ϳ150 mM NaCl. The vided 16-methylidenegeranylgeranyl acetate in 37% yield following enzyme preparation was next applied directly to a Pharmacia FPLC purification, and the ester was hydrolyzed as before (22), converted 10/10 column packed with ceramic hydroxyapatite (Bio-Rad) that to the corresponding chloride (30), and then diphosphorylated with was previously equilibrated with buffer A. The column was washed tris(tetra-n-butylammonium) hydrogen pyrophosphate (26). Purifi- with buffer A as before until the baseline stabilized and then eluted cation by cation-exchange and reversed phase chromatography gave with a 100 ml linear gradient from 0 to 100 mM potassium phosphate 7 mg (41%) of 16-methylidenegeranylgeranyl diphosphate (7). 1H (pH 6.8). The taxadiene synthase eluted from hydroxyapatite (the ␦ ␦ NMR (D2O, 400 MHz, HOD in D2O as internal standard ( 4.6)) different truncations elute at different phosphate concentrations on 6.15 (m, 1 H), 5.25 (t, 1 H), 4.95 (broad m, 3 H), 4.27 (t, 2 H), 2.02 this column) was applied directly to a Pharmacia FPLC 10/10 column (broad m, 2 H), 1.9 (m, 10 H), 1.5 (d, 6 H, J ϭ 13.8 Hz), 1.4 (d, 9 H, packed with DEAE Sepharose (Sigma, St. Louis, MO) that was ϭ 31 ␦ Ϫ ϭ J 5.8 Hz). P NMR (D2O, 162 MHz) 8.58 (d, J 19.77 Hz), previously equilibrated with buffer A. This weak anion-exchange Ϫ9.85 (d, J ϭ 21.12 Hz). column was washed with buffer A until the baseline stabilized and Expression constructs. The original taxadiene synthase cDNA then eluted with a 100 ml linear gradient from 0 to 1 M NaCl in (pTb42.1) was isolated as a pBluescript SK(Ϫ) clone (6). To make this buffer A. Taxadiene synthase eluted at ϳ350 mM NaCl, and was cDNA insert compatible for expression from pSBET (31), a T7 pro- desalted into buffer A using a Centriprep YM50 centrifugal concen- moter-driven vector that additionally encodes the argU gene speci- trator (Millipore, Bedford, MA) prior to the standard assay. fying the tRNA for rare arginine codon usage commonly encountered The course of the purification was monitored throughout by en- in plant sequences, site-directed mutagenesis (Gene Editor, Pro- zyme assay and SDS–PAGE (33). Protein concentration was deter- mega, Madison, WI) was first employed to simultaneously eliminate mined by the Bradford assay using a standard curve calibrated with two internal NdeI sites by silent substitution and to install a 3Ј- purified, recombinant limonene synthase (34). BamHI site downstream of the stop codon. Site-directed mutagenesis Taxadiene synthase assay and enzyme characterization. The (Quick-Change, Stratagene, La Jolla, CA) was then used to incorpo- standard assay was performed in 25 mM Hepes (pH 8.0) containing 3 rate an NdeI site and thus a new starting methionine in the full- 10% (v/v) glycerol in the presence of 1 mM MgCl2 and [1- H]gera- length cDNA (i.e., the preprotein) and in the truncations correspond- nylgeranyl diphosphate as substrate as previously described (3). The ing to the deletion of 60, 79, 93, 113, and 126 residues of the prepro- reaction products were extracted into pentane, and the pentane tein. Each of these constructs in pBluescript was then digested with extract was passed through a short column of MgSO4-silica gel prior NdeI and BamHI, and the resulting fragment was gel purified and to aliquot counting by LSC and GC-MS analysis for verification of then ligated into gel purified pSBET which had been previously taxane olefin production. For convenience in most assays, the digested with NdeI and BamHI. The resulting six pSBET constructs, MgSO4-silica gel column chromatography step was dispensed with designated M1 (preprotein), M60, M79, M93, M113, and M126 based (this step had no influence on the assay), and the pentane extract on the placement of the new initiating methionine, were then indi- was concentrated and analyzed directly by GC-MS via external cal- vidually transformed into E. coli JM109 cells by standard protocols ibration with authentic taxadiene. The latter assay protocol was (32). The transformants were first confirmed for the presence of the employed to evaluate the conversion of the alternate substrate, 2,7- appropriate insert by PCR using T7 promoter and T7 terminator cyclogeranylgeranyl diphosphate (3), and the catalytic turnover of primers, and the inserts of the purified plasmids were then fully the two potential mechanism-based inhibitors (4 and 7). GC-MS sequenced to verify that no unwanted alterations had occurred in the analysis was performed on a Hewlett-Packard 6890–7293 MSD sys- coding region during the mutagenesis procedures. The confirmed tem using a 30 m fused silica capillary 5MS column. Samples were plasmids were then transformed into E. coli BL21(DE3) (32) for separated using cool-on-column injection (40°C) with temperature expression studies. programming to 320° at 20°C/min and He as carrier gas; 70 eV cDNA expression and enzyme purification. Overnight cultures spectra were recorded. (grown from single colonies isolated from fresh transformation Conversion of the substrate analog 2,7-cyclogeranylgeranyl plates) were used to inoculate 2.8-liter Fernbach flasks containing 1 diphosphate (3) was evaluated at a concentration of 50 ␮M (i.e., liter of Luria-Bertani medium supplemented with 50 mg of kanamy- saturation with the normal geranylgeranyl substrate); the two mech- cin. Cultures were grown at 37°C to A 600 ϭ 0.5–0.7, allowed to anism-based inhibitors (4 and 7) were similarly tested. To examine equilibrate to 20°C, and then induced by addition of 1 mM isopropyl- the time-dependent inactivation of taxadiene synthase, the enzyme 1-thio-␤-D-galactopyranoside and grown for another 12–16 h at 20°C. was incubated at 31°C with 50 ␮M concentrations of either the HETEROLOGOUS EXPRESSION OF TAXADIENE SYNTHASE 141

cyclopropylidene analog (4) or the vinyl analog (7). At 30-min inter- nistic and structural studies, there was little choice but vals, aliquots of the preparations were removed, concentrated, re- to construct, express, and test a series of truncations suspended, and reconcentrated twice in buffer A using a Centriprep concentrator to remove residual inhibitor, and then assayed for ac- surrounding the approximate proteolytic processing tivity with [1-3H]geranylgeranyl diphosphate as substrate using the site based upon rather imprecise predictive methods standard protocol. Untreated preparations that were similarly pro- (10–13). cessed served as the controls. For kinetic analysis, standard assays For this purpose, the full-length preprotein and were performed at ten substrate concentrations ranging from 0.5 to truncations corresponding to removal of 60, 79, 93 , 50 ␮M [1-3H]geranylgeranyl diphosphate. After correction for very minor solvolytic background rates (buffer controls), the averaged 113, and 126 residues from the N-terminus (designated values for triplicate analyses at each substrate concentration were M1, M60, M79, M93, M113, and M126, respectively) plotted and evaluated using the Enzyme Kinetics Program (Trinity were prepared in pSBET (31), a vector that has proved Software, Inc.). to be highly suitable for heterologous expression in E. ␮ The stability of the enzyme in buffer A, in the presence of 50 M coli of plant genes that specify arginine by rare codon substrate, was determined at room temperature with purified preparations overlayed with hexane to exclude air and continuously usage. Low temperature, long duration expression extract the olefin product, the formation of which was conveniently from this vector in E. coli BL21(DE3) (20°C, 14 h) monitored by aliquot counting. The solubility of taxadiene synthase yielded functional, soluble protein in the case of M1, was determined in buffer A at room temp. by sequential ultrafiltra- M60, and M79 (confirmed by assay); SDS–PAGE of the tion (using a Centriprep YM50 concentrator) and centrifugation at soluble protein expressed from M93, M113, and M126 13,000g to remove aggregated protein, while monitoring protein verified the presence of the corresponding recombi- concentration by dye binding assay (33) and UV absorbsion at A 280. nant, but inactive, forms of taxadiene synthase (data RESULTS AND DISCUSSION not shown). Each of the functional taxadiene synthases was pu- Truncation of Taxadiene Synthase rified by the combination of hydroxyapatite chromatog- The cDNA for taxadiene synthase encodes a prepro- raphy (two steps) and anion-exchange chromatography tein with a calculated size of 98.3 kDa (862 amino (two steps) to provide essentially homogenous prepa- acids) (6), compared with a value of about 80 kDa rations (Ͼ99% pure) of the M60 and M79 truncations determined for the native enzyme by SDS–PAGE and by SDS–PAGE (data not shown). Both of these proteins gel permeation chromatography (3). Plastidial process- were expressed to levels of 30–40% of total bacterial ing of the preprotein appears to result in removal of the soluble protein, with little formation of inclusion bod- targeting peptide and N-terminal modification which ies. Minor adjustments in the expression and purifica- blocks Edman degradative sequencing (6). Thus, it is tion protocol have now allowed the routine preparation not possible to determine the proteolytic cleavage site of homogeneous forms of both enzymes in yields of and the exact size of the native enzyme by this means, 30–40 mg/L (M60) and 10–20 mg/L (M79) of bacterial in order to construct a recombinant mature protein by culture. Because of the instability of the full-length precise truncation of the cDNA. A similar problem was preprotein (M1) and the tendency to form inclusion recently addressed in the case of the monoterpene cy- bodies and otherwise aggregate, this form of taxadiene clase (Ϫ)-4S-limonene synthase (35) by systematic synthase (expressed to levels of about 10% of total truncation of the cDNA, and expression and kinetic bacterial soluble protein) could be brought to only evaluation of the resulting enzymes (8). A very tracta- about 50% purity by the same fractionation protocol; ble “pseudomature” form of limonene synthase was however, this was sufficient for accurate kinetic char- prepared by truncation immediately upstream of a pair acterization of the enzyme. Kinetic evaluation of each of tandem arginine residues (R58R59), a sequence ele- functional form of taxadiene synthase by several ment that was ultimately shown to participate in the graphical plotting procedures revealed the truncated preliminary substrate isomerization step that is spe- versions, M60 and M79, to be superior to the prepro- cifically and universally required for monoterpene cy- tein M1 and to the previously expressed thioredoxin- clization reactions (8). Alignment of the deduced taxa- fusion of the preprotein (9), and comparable to the diene synthase sequence with those of several mono- native enzyme based on rough estimates of kinetic terpene synthases and other terpenoid synthases (7), constants from previous studies (2, 3) (Table I). The

in an attempt to deﬁne the corresponding cleavage site, observed K m for the recombinant enzymes is somewhat was not fruitful. The taxadiene synthase sequence con- higher than that estimated for the native enzyme (Ta- tains an unusual insertional element of about 200 ble I) but still in the low ␮M range typical for this class amino acid residues (6, 7) that results in placement of of enzymes (4, 14, 36, 37). Given the turnover numbers this corresponding site some 300 amino acid residues of the M60 and M79 enzymes, and the evaluation of downstream of the starting methionine to afford a pro- approximate rates and protein contents of the crude tein of about 65 kDa, far smaller than the native en- bacterial extracts, it is apparent that the “inactive” zyme. Thus, in producing a recombinant “pseudoma- truncations of taxadiene synthase, i.e., M93, M113,

ture” form of taxadiene synthase for detailed mecha- and M126, could not have possessed k cat values in ex- 142 WILLIAMS ET AL.

TABLE I says. In this case, the use of MgSO4-silica gel was avoided Kinetic Constants for the Native Enzyme, Full-Length and cool-on-column injection was employed to minimize Preprotein, Thioredoxin Fusion, and Truncated the possibility of acid-catalyzed or thermal rearrange- Versions of Taxadiene Synthase ment of the product(s). Each of the functional enzymes, M1, M60 and M79, was shown to generate a qualitatively a ␮ b Ϫ1 Enzyme K m ( M) k cat (s ) and quantitatively identical product set (illustrated for the M60 enzyme in Fig. 3). The major product (retention Native enzyme (ref. 3) 3.0 0.0106 Thioredoxin fusion (ref. 9) 3.1 0.0025 time 10.71 min of Fig. 3A) was confirmed as taxa- M1 preprotein 10.5 0.0024 4(5),11(12)-diene (94.0%), as expected, by comparison of M60 15.9 0.0102 retention time and mass spectrum (Fig. 3B) to those of M79 14.0 0.0080 the authentic standard prepared by total synthesis (16). Ͻ M93 0.00001 One minor product (retention time 10.50 minutes of Fig. M113 Ͻ0.00001 M126 Ͻ0.00001 3A) was similarly confirmed as the isomeric taxa- 4(20),11(12)-diene (4.8%) by comparison of retention time a No entry indicates a K m value could not be determined because and mass spectrum (Fig. 3C) to the authentic standard the activity was negligible. (16). The second, very minor product (retention time b Kinetic constants are averages of triplicate analyses with SE less 10.08 of Fig. 3A) was tentatively identified as verticillene than 10% of the indicated value. (1.2%, presumably the 3(4),7(8),11(12)-isomer derived from intermediate 1 of Fig. 1) based on a retention time cess of 0.0001/s when measured at presumed satura- consistent with this macrocyclic olefin and a mass spec- tion (50 ␮M geranylgeranyl diphosphate). The essen- trum (Fig. 3D) that closely resembles a low resolution tially complete loss of activity by truncation beyond spectrum of the authentic compound (spectrum kindly functional recombinant enzyme M79 (i.e., M93) indi- provided by Professor G. Pattenden, University of Not- cates that the size of the mature taxadiene synthase is tingham, UK). in the 87- to 89-kDa range, somewhat larger than the Since it was independently demonstrated that both the 80-kDa size estimated for the native enzyme from its 4(5),11(12)- and 4(20),11(12)-isomers of taxadiene were natural source (2, 3). entirely stable under the conditions of the analysis (5), it Both the M60 and M79 truncated versions of taxa- can be assumed that the minor olefins are, in fact, au- diene synthase were subsequently evaluated for stabil- thentic products of the taxadiene synthase reaction ity and solubility at room temperature as a prelude to formed by alternative deprotonation steps. Thus, small crystallographic studies. The M60 form, at 10 mg/ml in amounts of verticillene could arise by direct deprotona- 25 mM Mopso buffer (pH 8.0), was stable at room tion of the verticillyl C12-carbocation (1 of Fig. 1), which temperature for at least 10 days (with less than 30% consequently aborts the normal intramolecular hydrogen loss of activity) when maintained in the presence of transfer from C11 to C7 to initiate B/C-ring closure to the 2ϩ taxane skeleton (Fig. 1). The formation of only traces of substrate (50 ␮M) and Mg cofactor (1 mM) and with verticillene is consistent with preliminary mechanistic the continuous removal of product by the use of a studies with taxadiene synthase (5) that have demon- hexane overlay. Under these conditions, the M79 ver- strated the internal transfer of the hydrogen atom to sion was much less stable, losing about 70% activity occur with high efficiency (Ͼ90%) and, thus, to imply over a 2-day period at room temperature. The M60 tight binding of the verticillyl intermediate. In the course form of taxadiene synthase, at room temperature in a of the principal reaction route, the intramolecular migra- buffer consisting of 25 mM Mopso (pH 6.8), containing tion and ring closure steps provide the taxenyl C4-carbo- 10% (v/v) glycerol and 1 mM DTT, could be concen- cation (2), which offers three possibilities for deprotona- trated to at least 20 mg/ml without precipitation. The tion to terminate the cyclization cascade. In approxi- M79 form of the enzyme underwent visible aggregation mately 94% of the reaction cycles, proton loss occurs from under these conditions when the protein concentration C5 to yield the major product taxa-4(5),11(12)-diene. In- exceeded 5 mg/ml. Thus, based on expression yield, frequently, deprotonation occurs from the C20 methyl ease of purification, kinetic constants, stability, and group (ϳ5%) to give the 4(20)-isomer. Proton loss from solubility, the M60 truncation provided a highly useful the bridgehead methine (C3) to give the 3(4)-isomer was “pseudomature” form of taxadiene synthase. not observed (i.e., trace levels of product). The minor diterpene olefin products of taxadiene syn- Product Characterization thase documented here were not observed previously in Each construct which yielded a functional cyclase (M1, studies with the native enzyme from Taxus stems (3) or M60, M79) was evaluated for product formation (olefins cell cultures (2) because too little material was available and oxygenated diterpenes) by coupled GC-MS analysis for detailed analysis compared to the amounts now ac- of pentane extracts of the corresponding large-scale as- cessible via the recombinant forms. The analysis of large- HETEROLOGOUS EXPRESSION OF TAXADIENE SYNTHASE 143

scale, crude preparations of taxadiene synthase from induced T. cuspidata suspension cell cultures (38) has since verified the biosynthesis of the identical olefin product set by this native enzyme. Analysis of the oxygenated diterpenes produced from geranylgeranyl diphosphate by the recombinant, purified forms of taxadiene synthase revealed the presence of only traces of geranyllinalool, gera- nylnerol and geranylgeraniol produced solvolytically in the course of the assay. To examine in greater detail the alternative depro- tonations catalyzed by taxadiene synthase, we employed the deuterium-labeled substrate (4R)-[4- 2 5 H1]geranylgeranyl diphosphate (84 mol %) to exploit the phenomenon of isotopically sensitive branching (39, 40). Isotopically sensitive branching experiments have been utilized extensively to evaluate monoterpene cyclization reactions (41–45). In the present case, the deuterium-dependent slowing of the C5-deprotonation of a common intermediate (the taxenyl cation 2) should promote the alternative means of terminating the reaction. Comparison of the product mixture derived from the deuterated substrate (Fig. 4A) relative to that generated from the normal substrate (Fig. 3A) revealed a decreased proportion of taxa-4(5),11(12)- diene (77.6%), with an increase in the formation of the 4(20),11(12)-isomer (11.1%) and the production of the tentatively identified 3(4),11(12)-isomer (Fig. 4B, 9.2%) (which is produced at only trace levels with the normal substrate) but without change in the formation of verticillene (1.1%). GC-MS analysis of the products was entirely consistent with the assumed mechanism of taxadiene synthase (1, 5), which predicts loss of the deuterium in formation of taxa-4(5),11(12)-diene but retention of the deuterium in the other olefins. These results confirm that all of the taxadiene isomers are derived from the taxenyl C4-carbocation intermediate

(2) and establish that the 5␤-proton (Hre) is eliminated in forming the 4,5-double bond of the major product. That the rate of formation of verticillene is not altered by the deuterium substitution is expected, since verticillene is formed prior to the isotopically sensitive branch that comprises the terminating deprotonation step of the reaction. A rough estimate of the kinetic

isotope effect for the C5-deprotonation gives a k H/k D of

5 The synthesis of this stereospeciﬁcally deuterium-labeled compound will be described elsewhere in the context of additional mechanistic studies with this enzyme.

from geranylgeranyl diphosphate are illustrated for taxa-4(5),11(12)- diene (B), taxa-4(20),11(12)-diene (C), and a product tentatively identified as verticillene (D). Analytical conditions are described under FIG. 3. Capillary GC-MS analysis of the diterpene olefin products Materials and Methods. Identification of the taxadiene isomers is of the truncated (M60) recombinant taxadiene synthase. The GC based on comparison of retention times and mass spectra to the profile (A, total ion current) and mass spectra of the olefins generated authentic standards. 144 WILLIAMS ET AL.

terminus of the mature enzyme appears to play a functional role. The structural and/or mechanistic rationale for the abrupt loss of activity at truncation M93 is presently unclear.

Inhibitors and Alternative Substrates The conversion of geranylgeranyl diphosphate to the taxane skeleton by initial macrocyclization and bridging to form a verticillyl intermediate followed by proton-induced closure of the B/C-rings (Fig. 1) was first proposed in 1966 (50). The evidence, thus far, based largely on the cyclization to taxa-4(5),11(12)-diene catalyzed by taxadiene synthase, supports this mechanistic proposal (1–3). However, exogenous (Ϯ)-verticillene is not detectably incorporated into taxadiene by the enzyme (5), nor do verticillene and its various epoxide derivatives undergo such transannular cyclizations in chemical model reactions (51). Nevertheless, the apparent production of trace levels of verticillene in the enzymatic reaction is consistent with the proposed mechanism, and with the tight binding of such a transient verticillyl intermediate required to enforce a con- formation conducive to the subsequent cyclization to the B/C-ring system and to effect the efficient intramolecular hydrogen transfer step (5) (Fig. 1). Another plausible mechanism for taxadiene forma- FIG. 4. Capillary GC-MS analysis of the diterpene olefin products of the truncated (M60) recombinant taxadiene synthase generated tion could involve formation of the C-ring first, with the 2 from (4R)-[4- H1]geranylgeranyl diphosphate. The new product generation of 2,7-cyclogeranylgeranyl diphosphate (3, (compare to Fig. 3A) at retention time 10.29 min of the GC profile (A, Fig. 2) as a transient intermediate instead of the ver- total ion current) was tentatively identified as taxa-3(4),11(12)-diene ticillyl structure. Completion of the cyclization could based on retention behavior and mass spectrum (B). then occur by similar intramolecular alkylation of the terminal double bond in the sidechain followed by about 2. The formation of multiple products is a com- bridging and intramolecular C11 to C3 proton transfer mon feature of terpenoid cyclization reactions (46, 47). (based on available deuterium labeling studies and MS For example, some sesquiterpene synthases produce in analysis of the derived taxadiene product (5), this sce- excess of 30 different olefin products (48). The fidelity nario is indistinguishable from the previously proposed of taxadiene synthase in the production of largely a hydrogen transfer from C11 to C7), and final proton single product (taxa-4(5),11(12)-diene at 94%) is, in elimination from C5 to form taxa-4(5),11(12)-diene. fact, notable. Structural precedent for the proposed cyclogera- The truncation studies indicate that the N-terminus nylgeranyl-type intermediate (3) in this alternative of taxadiene synthase is essential for catalysis (i.e., route is provided by the natural occurrence of ␣- and there is complete loss of function by truncation be- ␤-cyclogeraniols (52), picrocrocin (53), and safranal tween aa79 and aa93 (see Table I)), but does not sig- and related monoterpenes (53–55) in various plant nificantly influence product outcome in the distribution sources. Furthermore, similar proton-initiated cycliza- of olefins (at least for M1, and truncations M60 and tions of geranylgeranyl diphosphate in the olefinic M79). The M60, M79, M93, M113, and M126 versions chain distal to the diphosphate ester group are well- of taxadiene synthase were intended to represent established in the biosynthesis of the diterpenoid resin forms of the preenzyme from which the transit peptide acids and gibberellins (4, 56). 2 was deleted and for which, contrary to observation, the (Ϯ)-2,7-Cyclo[1- H2]geranylgeranyl diphosphate was minimum mechanistic consequences were anticipated. therefore prepared and tested at saturating concentra- Structural studies with the monoterpene cyclase (Ϫ)- tions (50 ␮M,with1mMMg2ϩ) as a substrate for limonene synthase (8) and the sesquiterpene cyclase taxadiene formation with the purified M60 version of 5-epi-aristolochene synthase (49) indicate the active the synthase under established assay conditions. No sites of these enzymes to be largely derived from the detectable taxadiene, or other diterpene olefin, was C-terminal domain; however, in both cases, the N- produced in the incubation of cyclogeranylgeranyl HETEROLOGOUS EXPRESSION OF TAXADIENE SYNTHASE 145 diphosphate with the synthase under conditions which conditions (at 1 mM Mg2ϩ) with the M60 version of the efficiently yielded ample product (Ͼ100 pmol) with the enzyme. However, when tested as substrates under normal geranylgeranyl substrate; these results argue standard assay conditions, the cyclopropylidene analog against the alternative pathway to taxadiene involving (4) did yield an olefinic product (with parent ion at m/z 2,7-cyclogeranylgeranyl diphosphate as an intermedi- 270 and base peak at m/z 122 (5) consistent with for- ate. However, negative results regarding the interven- mation of the cyclopropylidene analog of taxadiene (6)) tion and conversion of potential intermediates of the at a rate corresponding to about 30% of that observed reaction (e.g., both verticillene and cyclogeranylgera- for total olefin formation with the normal geranylgeranyl diphosphate) do not eliminate the possible involve- nyl substrate. The vinyl homolog (7) was not utilized, ment of the corresponding, tightly bound, transient yielding no detectable olefinic (e.g., 9) or oxygenated intermediate that is not accessible to, or exchangeable diterpenoid products when incubated with taxadiene with, exogenous material. Thus, on binding the gera- synthase. Thus, it appeared that the enzyme escaped nylgeranyl diphosphate substrate, taxadiene synthase the possibility of alkylation by the cyclopropylidene could undergo conformational alteration as a means of analog (4) by conduct of the normal reaction, albeit at shielding subsequently formed carbocation intermedi- a somewhat compromised rate, to produce the corre- ates from water. Such conformational change could as sponding 16,17-cyclotaxadiene derivative (6). In the a consequence also prevent the egress of any stable case of the vinyl-substituted homolog (7), it appeared intermediate from the active site, and disallow the that this substrate, modified by the additional meth- productive binding from solution of such an exogenous ylidene, did not undergo normal, productive binding to intermediate (to either conformational form). Similar initiate the ionization and macrocyclization steps re- stable, but tightly bound, intermediates, especially ole- quired to unmask the novel, potentially reactive, car- finic intermediates, are thought to participate in re- bocation (8 of Fig. 2). A preliminary evaluation of all of lated cyclizations in the monoterpene (46, 57), sesqui- the substrate analogs (3, 4, and 7) as presumptive terpene (37, 58) and diterpene (15, 47) series, yet direct competitive inhibitors of the normal cyclization reac- evidence for the participation of such neutral interme- tion of [1-3H]geranylgeranyl diphosphate (at 30 ␮M) diates of the reaction cycle has been similarly difficult revealed all three to be relatively weak inhibitors of to obtain (59). Conversely, the participation of certain taxadiene synthase (M60 version), with approximate diphosphorylated intermediates in terpenoid cycliza- I 50 values of 90 ␮M(3), 120 ␮M(4) and 100 ␮M(7), tion reactions (linalyl diphosphate in the monoterpene respectively. These results suggest that none of the series (36, 46), nerolidyl diphosphase in the sesquiter- substrate analogs is readily recognized by taxadiene pene series (58), and copalyl diphosphate in the diter- synthase, which exhibits a relatively low K m value of pene series (4, 47)) has been amply demonstrated by 15.0 ␮M for the normal geranylgeranyl substrate that direct testing of these “alternate” substrates. These the proposed reaction mechanism suggests must be latter results support the argument that the failure of precisely aligned at the active site for the cyclization to taxadiene synthase to utilize cyclogeranylgeranyl occur (5). Further studies to understand active site diphosphate indicates that this cyclic isomer of the structure and the mechanism of action of this unusual normal substrate is not an intermediate of the cycliza- catalyst are in progress.5 tion reaction. In a final set of experiments, two potential, mecha- ACKNOWLEDGMENTS nism-based inhibitors (4 and 7) of the taxadiene cy- We thank C. Sanchez, D. Pouchnik, and G. Munske for nucleotide clization reaction were tested with the purified M60 sequencing and primer synthesis, B. Long for assistance with the version of the synthase. These substrate analogs are protein expression, and Joyce Tamura for typing the manuscript. designed to reposition and/or stabilize positive charge in intermediate(s) generated in the course of the nor- REFERENCES mal cyclization reaction (Fig. 2), with the anticipation 1. Koepp, A. E., Hezari, M., Zajicek, J., Stofer Vogel, B., LaFever, that alteration in charge configuration of the carboca- R. E., Lewis, N. G., and Croteau, R. (1995) J. Biol. Chem. 270, tion will result in alkylation of a proximate basic resi- 8686–8690. due of the synthase. This general approach has been 2. Hezari, M., Ketchum, R. E. B., Gibson, D. M., and Croteau, R. utilized to inactivate (and specifically label) certain (1997) Arch. Biochem. Biophys. 337, 185–190. monoterpene synthases (60, 61), and has been widely, 3. Hezari, M., Lewis, N. G., and Croteau, R. (1995) Arch. Biochem. and successfully, employed to probe the structure and Biophys. 322, 437–444. mechanism of triterpene synthases (62). 4. MacMillan, J., and Beale, M. H. (1999) in Comprehensive Nat- ural Products Chemistry: Isoprenoids Including Carotenoids and In the present instance, neither the cyclopropylidene Steroids (D. E. Cane, Ed.), Vol. 2, pp. 217–243, Elsevier Science, analog (4) nor the vinyl homolog (7) at a concentration Oxford, UK. of 100 ␮M exhibited time-dependent inactivation of 5. Lin, X., Hezari, M., Koepp, A. E., Floss, H. G., and Croteau, R. taxadiene synthase when incubated under standard (1996) Biochemistry 35, 2968–2977. 146 WILLIAMS ET AL.

6. Wildung, M. R., and Croteau, R. (1996) J. Biol. Chem. 271, 36. Croteau, R. (1987) Chem. Rev. 87, 929–954. 9201–9204. 37. Cane, D. E. (1990) Chem. Rev. 90, 1089–1103. 7. Bohlmann, J., Meyer-Gauen, G., and Croteau, R. (1998) Proc. 38. Ketchum, R. E. B., Gibson, D. M., Croteau, R. B., and Shuler, Natl. Acad. Sci. USA 95, 4126–4133. M. L. (1999) Biotechnol. Bioeng. 62, 97–105. 8. Williams, D. C., McGarvey, D. J., Katahira, E. J., and Croteau, 39. Jones, J. P., Korzekwa, K. R., Rettie, A. E., and Trager, W. F. R. (1998) Biochemistry 37, 12213–12220. (1986) J. Am. Chem. Soc. 108, 7074–7078. 9. Huang, K.-X., Huang, Q.-L., Wildung, M. R., Croteau, R., and 40. Harada, N., Miwa, G. T., Walsh, J. S., and Lu, A. Y. H. (1984) Scott, A. I. (1998) Protein Express. Purif. 13, 90–96. J. Biol. Chem. 259, 3005–3010. 10. von Heijne, G., Steppuhn, J., and Herrmann, R. G. (1989) Eur. 41. Croteau, R., Wheeler, C. J., Cane, D. E., Ebert, R., and Ha, H.-J. J. Biochem. 180, 535–545. (1987) Biochemistry 26, 5383–5389. 11. Gavel, Y., and von Heijne, G. (1990) FEBS Lett. 261, 455–458. 42. Wagschal, K., Savage, T. J., and Croteau, R. (1991) Tetrahedron 12. Keegstra, K., Olsen, J. J., and Theg, S. M. (1989) Annu. Rev. 47, 5933–5944. Plant Physiol. Plant Mol. Biol. 40, 471–501. 43. Rajaonarivony, J. I. M., Gershenzon, J., and Croteau, R. (1992) 13. Kouranov, A., and Schnell, D. J. (1996) J. Biol. Chem. 271, Arch. Biochem. Biophys. 296, 49–57. 31009–31012. 44. Pyun, H.-J., Coates, R. M., Wagschal, K. C., McGeady, P., and 14. Alonso, W. R., and Croteau, R. (1993) Methods Plant Biochem. 9, Croteau, R. (1993) J. Org. Chem. 58, 3998–4009. 239–260. 45. Wagschal, K. C., Pyun, H.-J., Coates, R. M., and Croteau, R. 15. LaFever, R. E., Stofer Vogel, B., and Croteau, R. (1994) Arch. (1994) Arch. Biochem. Biophys. 308, 477–487. Biochem. Biophys. 313, 139–149. 46. Wise, M. L., and Croteau, R. (1999) in Comprehensive Natural 16. Rubenstein, S. M., and Williams, R. M. (1995) J. Org. Chem. 60, Products Chemistry: Isoprenoids Including Carotenoids and Ste- 7215–7223. roids (D. E. Cane, Ed.), Vol. 2, pp. 97–153, Elsevier Science, 17. Coates, R. M., and Jin, A. Q. (1997) J. Org. Chem. 62, 7475–7478. Oxford. 18. Still, W. C., Kahn, M., and Mitra, A. (1978) J. Org. Chem. 43, 47. Davis, E. M., and Croteau, R. (2000) Top. Curr. Chem. 209, 2923–2925. 53–95. 19. Collington, E. W., and Meyers, A. I. (1971) J. Org. Chem. 36, 48. Steele, C. L., Bohlmann, J., Crock, J., and Croteau, R. (1998) 3044–3045. J. Biol. Chem. 273, 2078–2089. 20. Woodside, A. B., Huang, Z., and Poulter, C. D. (1988) Org. Syn. 49. Starks, C. M., Back, K., Chappell, J., and Noel, J. P. (1997) Coll. Vol. VIII, 616–620. Science 277, 1815–1820. 21. Ravn, M. M., Jin, A. Q., and Coates, R. M. (2000) Eur. J. Org. Chem. 2000, in press. 50. Harrison, J. W., Scrowton, R. M., and Lythgoe, B. (1966) J. Chem. Soc. C 1933–1945. 22. McMurry, J. E., and Erion, M. D. (1985) J. Am. Chem. Soc. 107, 2712–2720. 51. Begley, M. J., Jackson, C. B., and Pattenden, G. (1990) Tetrahe- dron 96, 4907–4927. 23. van Tamelen, E. E., and Sharpless, K. B. (1967) Tetrahedron Lett. 28, 2655–2659. 52. Erman, W. F. (1985) in Chemistry of the Monoterpenoids, Part A, pp. 480–505, Marcel Dekker, New York. 24. Pappo, R., Allen, D. S., Jr., Lemieux, R. U., and Johnson, W. S. (1956) J. Org. Chem. 21, 478–479. 53. Simonsen, J. L. (1953) in The Terpenes, Vol. 1, pp. 114–122, Cambridge Univ. Press, Cambridge, UK. 25. Cane, D. E., Iyengar, R., and Shiao, M.-S. (1981) J. Am. Chem. Soc. 103, 914–931. 54. Zarghami, N. S., and Heinz, D. E. (1971) Phytochemistry 10, 26. Davisson, V. J., Woodside, A. B., Neal, T. R., Stremler, K. E., 2755–2761. Muehlbacher, M., and Poulter, C. D. (1986) J. Org. Chem. 51, 55. Dev, S., Narula, A. P. S., and Yadev, J. S. (1982) in CRC Hand- 4768–4779. book of Terpenoids. Monoterpenoids (S. Dev, Ed.), Vol. II, pp. 27. Umbreit, M. A., and Sharpless, K. B. (1977) J. Am. Chem. Soc. 321–329, CRC Press, Boca Raton, FL. 99, 5526–5528. 56. West, C. A. (1981) in Biosynthesis of Isoprenoid Compounds 28. Attenburrow, J., Cameron, A. F. B., Chapman, J. H., Evans, (J. W. Porter and S. L. Spurgeon, Eds.), Vol. 1, pp. 375–411, R. M., Hems, B. A., Jansen, A. B. A., and Walker, T. (1952) Wiley, New York. J. Chem. Soc. 1094–1111. 57. Croteau, R., Alonso, W. R., Koepp, A. E., and Johnson, M. A. 29. Greenwald, R., Chaykovsky, M., and Corey, E. J. (1963) J. Org. (1994) Arch. Biochem. Biophys. 309, 184–192. Chem. 28, 1128–1129. 58. Cane, D. E. (1999) in Comprehensive Natural Products Chem- 30. Cane, D. E., Oliver, J. S., Harrison, P. H. M., Abell, C., Hubbard, istry: Isoprenoids Including Carotenoids and Steroids (D. E. B. R., Kane, C. T., and Lattman, R. (1990) J. Am. Chem. Soc. Cane, Ed.), Vol. 2, pp. 155–200, Elsevier Science, Oxford, UK. 112, 4513–4524. 59. Rising, K. A., Starks, C. M., Noel, J. P., and Chappell, J. (2000) 31. Schenk, P. M., Baumann, S., Mattes, R., and Steinbiss, H. (1995) J. Am. Chem. Soc. 122, 1861–1866. BioTechniques 19, 196–200. 60. Croteau, R., Alonso, W. R., Koepp, A. E., Shim, J.-H., and Cane, 32. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) in Molecu- D. E. (1993) Arch. Biochem. Biophys. 307, 397–404. lar Cloning: A Laboratory Manual, 2nd. ed., pp. 7.20–7.43, Cold 61. McGeady, P., and Croteau, R. (1995) Arch. Biochem. Biophys. Spring Harbor Laboratory, Plainview, NY. 317, 149–155. 33. Laemmli, U. K. (1970) Nature (London) 227, 680–685. 62. Abe, I., and Prestwich, G. D. (1999) in Comprehensive Natural 34. Bradford, M. M. (1976) Anal. Biochem. 72, 248–254. Products Chemistry: Isoprenoids Including Carotenoids and Ste- 35. Colby, S. M., Alonso, W. R., Katahira, E. J., McGarvey, D. J., and roids (D. E. Cane, Ed.), Vol. 2, pp. 267–298, Elsevier Science, Croteau, R. (1993) J. Biol. Chem. 268, 23016–23024. Oxford, UK.