ftfcoswECHMgscsaoaiec

Archives of Biochemistry and BiophysicsVol. 393, No. 2, September 15, pp. 222–235, 2001doi:10.1006/abbi.2001.2483, available online at http://www.idealibrary.com on

Cloning, Heterologous Expression, and FunctionalCharacterization of 5-epi-Aristolochene-1,3-Dihydroxylasefrom Tobacco (Nicotiana tabacum)1

Lyle Ralston, Soon Tae Kwon,2 Mark Schoenbeck, Jennifer Ralston, David J. Schenk,*Robert M. Coates,* and Joe Chappell3

Department of Agronomy, Plant Physiology, Biochemistry, and Molecular Biology Program, University of Kentucky,Lexington, Kentucky 40546-0091; and *Department of Chemistry, University of Illinois, Urbana, Illinois 61801

Received March 26, 2001, and in revised form June 4, 2001; published online August 24, 2001

irCctsaactRPapemaettdtwawh

Capsidiol is a bicyclic, dihydroxylated sesquiter-pene produced by several solanaceous species in re-sponse to a variety of environmental stimuli. It is theprimary antimicrobial compound produced by Nicoti-ana tabacum in response to fungal elicitation, and it isormed via the isoprenoid pathway from 5-epi-aris-olochene. Much of the biosynthetic pathway for theormation of this compound has been elucidated, ex-ept for the enzyme(s) responsible for the conversionf 5-epi-aristolochene to its dihydroxylated form, cap-idiol. Biochemical evidence from previous studiesith N. tabacum (Whitehead, I. M., Threlfall, D. R., andwing, D. F., 1989, Phytochemistry 28, 775–779) andapsicum annuum Hoshino, T., Yamaura, T., Imaishi,., Chida, M., Yoshizawa, Y., Higashi, K., Ohkawa, H.,izutani, J., 1995, Phytochemistry 38, 609–613. sug-

ested that the oxidation of 5-epi-aristolochene to cap-idiol was mediated by at least one elicitor-inducibleytochrome P450 hydroxylase. In extending these ob-ervations, we developed an in vivo assay for 5-epi-ristolochene hydroxylase activity and used it to dem-nstrate a dose-dependent inhibition of activity byncymidol and ketoconazole, two well characterizednhibitors of cytochrome P450 enzymes. Using degen-rate oligonucleotide primers designed to the wellonserved domains found within most P450 enzymes,

1 The nucleotide sequences reported in this study have been de-posited with GenBank: CYP71D20, AF368376; CYP71D21,AF368377; CYP73A27, AF368378; CYP73A28, AF368379; CYP82E1,AF368380; and CYP92A5, AF368376.

2 Permanant address: Department of Life Science and NaturalResources, Andong National University, Kyungpook, Korea.

3 To whom correspondence should be addressed: Fax: (859) 257-

7125. E-mail: chappell@pop.uky.edu.

222

ncluding the heme binding domain, cDNA fragmentsepresenting four distinct P450 families (CYP71,YP73, CYP82, and CYP92) were amplified from aDNA library prepared against mRNA from elicitor-reated cells using PCR. The PCR fragments were sub-equently used to isolate full-length cDNAs (CYP71D20nd D21, CYP73A27 and A28, CYP82E1 and CYP92A5),nd these in turn were used to demonstrate that theorresponding mRNAs were all induced in elicitor-reated cells, albeit with different induction patterns.epresentative, full-length cDNAs for each of the450s were engineered into a yeast expression system,nd the recombinant yeast assessed for functional ex-ression of P450 protein by measuring the CO differ-nce spectra of the yeast microsomes. Only microso-al preparations from yeast expressing the CYP71D20

nd CYP92A5 cDNAs exhibited significant CO differ-nce absorbance spectra at 450 nm and were thusested for their ability to hydroxylate 5-epi-aris-olochene and 1-deoxycapsidiol, a putative mono-hy-roxylated intermediate in capsidiol biosynthesis. In-erestingly, the CYP71D20-encoded enzyme activityas capable of converting both 5-epi-aristolochenend 1-deoxycapsidiol to capsidiol in vitro, consistentith the notion that this P450 enzyme catalyzes bothydroxylations of its hydrocarbon substrate. © 2001

Academic Press

Key Words: 5-epi-aristolochene hydroxylase; sesquit-erpene cyclase; phytoalexin; cytochrome P450.

Capsidiol is a bicyclic, dihydroxylated sesquiterpeneproduced by many solanaceous species in response to avariety of environmental stimuli, including exposure to

UV light (3) and infection by microorganisms (4–6). It

0003-9861/01 $35.00Copyright © 2001 by Academic Press

All rights of reproduction in any form reserved.

ti

botb

ecshpTacctvc

223CLONING AND CHARACTERIZATION OF 5-epi-ARISTOLOCHENE-1,3-DIHYDROXYLASE

is the primary antibiotic or phytoalexin produced intobacco in response to fungal elicitation, and it is de-rived from the isoprenoid pathway via its hydrocarbonprecursor, 5-epi-aristolochene (Scheme 1). Several ofthe biosynthetic enzymes leading up to 5-epi-aris-tolochene formation have been studied (7), especially5-epi-aristolochene synthase (EAS)4 (8–11). EAS com-mits carbon to sesquiterpene metabolism by catalyzingthe cyclization of FPP to 5-epi-aristolochene. The en-zyme(s) responsible for the conversion of 5-epi-aris-olochene to capsidiol, however, has yet to be fullydentified and characterized.

Biochemical evidence from previous studies in to-acco (1) and green pepper (2) have suggested that thexidation of 5-epi-aristolochene to capsidiol occurs in awo-step process with one of the hydroxylation stepseing constitutive and the other being mediated by an

4 Abbreviations used: EAH, 5-epi-aristolochene hydroxylase; EAS,5-epi-aristolochene synthase; FPP, farnesyl diphosphate; FC, flashchromatography; Mops, 4-morpholinepropanesulfonic acid; SSC,

SCHEME 1. Proposed pathways for the biosynthesis of capsidisynthesized from FPP by the action of a sesquiterpene cyclase, 5-epiand C3 to form capsidiol. A definitive order for these hydroxylatiWhitehead et al. (1) have suggested that the hydroxylation on C3 of 5followed by hydroxylation at C1 catalyzed by a constitutive enzymNADPH and O2 (2).

standard sodium citrate.

licitor-inducible cytochrome P450 (Scheme 1). Be-ause 1-deoxycapsidiol had been isolated from naturalources (12), Whitehead et al. (1) surmised that per-aps the biosynthesis of this intermediate was due toathogen induction of a corresponding hydroxylase.hey therefore prepared synthetic 1-deoxycapsidiolnd reported a modest conversion of this compound toapsidiol when fed to control or unelicited tobacco cellultures. This was further supported by their observa-ion that radiolabeled 5-epi-aristolochene was only con-erted to capsidiol when fed to elicitor-induced cellultures but not to control cultures. Whitehead et al. (1)

therefore concluded that the 3-hydroxylase, responsi-ble for hydroxylation of 5-epi-aristolochene at C3 togenerate 1-deoxycapsidiol, was pathogen/elicitor in-ducible, while the 1-hydroxylase, responsible for hy-droxylating 1-deoxycapsidiol at the C1 to generate cap-sidiol, was constitutive. Hoshino et al. (2) extended theobservations of Whitehead et al. (1) by directly mea-suring 3-hydroxylase activity in microsomal prepara-tions of arachidonic acid-elicited Capsicum annuum

n elicitor-treated Nicotiana tabacum cells. 5-epi-Aristolochene isstolochene synthase (EAS), and is subsequently hydroxylated at C1

has not been determined. However, based on indirect evidence,i-aristolochene occurs first via an inducible P450-mediated enzyme,ctivity. One or both of these hydroxylation reactions may require

ol i-arions-epe a

fruits and seedlings. These assays consisted of incubat-

actcw

ctbCtiswccajca

fooicdlhdoiwdartcscec

tm

rp

aS

224 RALSTON ET AL.

ing 5-epi-aristolochene with microsome preparationsnd subsequently determining the amount of 1-deoxy-apsidiol generated by a combination of TLC separa-ions and GC. Their evidence demonstrated that theonversion of 5-epi-aristolochene to 1-deoxycapsidiolas dependent on both NADPH and O2, and that 1-de-

oxycapsidiol accumulation in vitro was arrested by theP450 antagonists carbon monoxide (13), ancymidol(14), and ketoconazole (15).

Recent results suggest that the hydroxylation of5-epi-aristolochene is an important regulated step inapsidiol biosynthesis. In studies to evaluate the effec-iveness of methyl jasmonate as an inducer of capsidioliosynthesis in tobacco cell cultures, Mandujano-havez et al. (16) reported that the modest accumula-

ion of this phytoalexin was accompanied by a strongnduction of EAS. This surprising result implied thatteps before or after the sesquiterpene cyclase reactionere limiting. Using an in vivo assay measuring the

onversion rate of radiolabeled 5-epi-aristolochene toapsidiol, a very limited induction of the hydroxylasectivity (s) was observed in cells treated with methylasmonate relative to that in fungal elicitor-treatedells. This result pointed to the hydroxylase reactionss a potentially limiting step in capsidiol biosynthesis.The current work was undertaken to characterize

urther the activity(s) responsible for the hydroxylationf 5-epi-aristolochene and to investigate the regulationf this activity(s). In particular, we aimed to determinef these reactions are mediated by one or two cyto-hrome P450 enzymes, and if by two enzymes, then toetermine if there is a preferred order to the hydroxy-ations (Scheme 1). Unfortunately, cytochrome P450ydroxylases comprise a very large family of heme-ependent, mixed-function oxidases that are generallyf low abundance and membrane bound, thus frustrat-ng conventional biochemical approaches. Therefore,e first sought corroborating evidence that the hy-roxylation of aristolochene was indeed mediated byn inducible cytochrome P450-type reaction using theecently developed in vivo assay (16). This informationhen supported an experimental strategy to isolateDNAs for inducible P450 enzymes based on the con-erved heme binding domain found within all cyto-hrome P450 enzymes (17, 18), and to characterize thencoded enzyme activities after expression of theDNAs in a yeast expression system (19).

MATERIALS AND METHODS

ChemicalsStandard laboratory reagents were purchased from Becton Dick-

inson Microbiology Systems (Sparks, MD), FisherBiotech (FairLawn, NJ), Sigma Chemical Company (St. Louis, MO), and AldrichChemical Co. (Milwaukee, WI).

1-Deoxycapsidiol and 5-epi-aristolochene were prepared by chem-

ical reductions of capsidiol derivatives by procedures modeled after

literature descriptions cited below. The purity ($90–95%) and iden-tity of the intermediates and products were established by GC and/or1H NMR spectroscopic analyses, and by correspondence of the NMRspectral data with values given in the literature references. Com-pounds were purified by flash chromatography, FC, on silica gel withthe designated eluant (20). Additional information about proceduresand full characterization data are available in the PhD dissertationof DJS (21).

Capsidiol. Elicitations of capsidiol biosynthesis, extractions, andpurification by FC (1:6 hexane-ethyl acetate) were carried out onlarge scale (480 fresh green bell peppers) using procedures adaptedfrom those previously reported (22): yield, 492 mg (1.0 mg/pepper),mp 149.5–151°C (previously reported as 150–152.5°C (23)), GC(100%), 1H NMR (500 MHz, CDCl3)(1).

1-Deoxycapsidiol. Conversion of capsidiol (140 mg) to its diac-etate was effected with excess acetic anhydride (15 mL) and pyridine(24 mL) at room temperature for 24 h (23). Isolation by ether extrac-tion and FC (10:1) hexane-ethyl acetate) gave capsidiol diacetate:yield, 189 mg (ca. 100%), 1H NMR (400 MHz, CDCl3) (24). Reduction(25, 26) of the diacetate (50.1 mg) in THF (2 mL) with Li (18.6 mg) inliquid NH3 (15 ml) at 278°C for 30 min followed by quenching with-butyl alcohol in ether and ether extractions afforded an 11.6:1ixture (GC) of 1-deoxycapsidiol (epi-aristolochene-3a-ol) and 5-epi-

eremophilene-3a-ol. FC (4:1 hexane-ethyl acetate) gave 21.7 mg (63%ecovery, 97% purity by GC) and 0.5 mg (ca. 1.5%) of epieremo-hilene-3a-ol. The 1H NMR data (500 MHz, CDCl3) agreed with the

literature values for both isomers (25).5-epi-Aristolochene. 1-Deoxy-capsidiol (19.8 mg) was converted to

the phosphoramidate in three operations (27): (1) lithiation with 1 Mbutyllithium in hexane (240 ml), N,N,N9,N9-tetramethylethylenedia-mine (0.15 g), and 1,2-dimethoxyethane (0.7 ml) at 0°C for 0.5 h; (2)phosphorylation with N,N-dimethylphosphoramidic dichloride (0.33g) at room temperature for 3 h; (3) reaction with dimethylamine (5mL) initially at 233°C and then at 0°C for 1 h. Ether extractions andFC (2% MeOH-ethyl acetate) gave 26.0 mg (81%) of the phosphor-amidate as an oil: 1H NMR (500 MHz, CDCl3).

The phosphoramidate (11.9 mg) in THF (400 mL) was added to ablue suspension of Li pieces (20 mg) in dry ethylamine (7 mL), whichwas being stirred and cooled at 278°C. The bath temperature wasraised to 242°C and after 5 min t-butyl alcohol (3 mL in 1 mL of THF)nd 3-hexyne were added in succession to discharge the blue color.aturated aqueous NH4Cl (several drops), pentane (3 ml), and water

(2 ml) were added; the mixture was warmed to room temperatureand the product was extracted with pentane (4 3 2 mL). The yield(7.4 mg, 86%) was determined by quantitative GC (cedrene as exter-nal standard). FC and cautious concentration afforded epi-aris-tolochene containing some residual pentane: 1H NMR (500 MHz,CDCl3) (25).

Biological Materials and Induction Treatments

Nicotiana tabacum cv. KY14 plants and cell suspension cultureswere used. Cell suspension cultures were maintained in modifiedMurashige-Skoog as described previously (11). Cultures in theirrapid phase of growth (3 days old) were used for all experiments. Atthe indicated times, cells were collected and separated from media byvacuum filtration and stored at 280°C.

Induction treatments were performed by the addition of the fungalelicitors, cellulase (Trichoderma viride, Type RS, Onozuka) or para-citicein (28), at the indicated concentrations. Paraciticein was puri-fied from E. coli cells overexpressing a recombinant paraciticeinprotein containing a carboxy-terminal histidine purification tag

(Back and Chappell, unpublished).

artaqsg

vMfoc

abedse

iSwd

225CLONING AND CHARACTERIZATION OF 5-epi-ARISTOLOCHENE-1,3-DIHYDROXYLASE

In Vivo 5-epi-Aristolochene Hydroxylase Assay andInhibition Studies5-epi-Aristolochene hydroxylase activity was measured as the in-

corporation of [6-3H]-5-epi-aristolochene into extracellular capsidiolby intact cells. [3H]-5-epi-aristolochene was produced by incubatingn excess of [1-3H]farnesyl diphosphate (1 mM, 20.5 Ci/mmol) withecombinant 5-epi-aristolochene synthase (29, 30). The hexane ex-ractable radioactivity from reactions was treated with a smallmount of silica to remove any farnesol or residual FPP beforeuantifying the yield of radioactive 5-epi-aristolochene by liquidcintillation counting. The hexane solvent was removed under aentle stream of N2 gas, and the dried residue was redissolved in

acetone. Control and elicitor-treated cells were then incubated with[3H]-5-epi-aristolochene (approximately 100,000 dpm at 2.5 nM) for3-h periods at various points during an induction time course beforecollecting the cell and media samples. Detection and quantification ofcapsidiol in the extracellular culture media was performed as re-ported previously (31), and the amount of radioactivity incorporatedinto capsidiol was determined. For these determinations, sampleswere separated by TLC, and the zones corresponding to capsidiolwere scraped from the plate for scintillation counting.

Inhibition studies were performed by the addition of the P450inhibitors ancymidol (2, 14) and ketoconazole (2, 15) directly to thecell cultures or enzyme assay mix. Cell cultures were incubated inthe presence of cellulase (0.5 mg/mL) and indicated concentrations ofancymidol or ketoconazole for 12 h prior to the addition of [3H]-5-epi-aristolochene. After a further 3-h incubation period, the cells andmedia were collected. The amount of radioactivity incorporated intoextracellular capsidiol was determined as described above. To eval-uate secondary effects of these inhibitors, the level of induciblesesquiterpene cyclase activity in the collected cells was determinedaccording to (32), as well as in vitro assays with purified recombinantEAS (29) incubated with the indicated concentrations of ancymidoland ketoconazole.

All experiments were replicated in several independent trials.While the absolute values presented may have varied between ex-periments by as much as 50%, the trends and time courses wereconsistent throughout.

Construction of an Elicitor-Induced cDNA LibraryCell cultures were incubated with fungal elicitor (0.5 mg cellulase/

mL) for 6 h before collecting the cells by filtration. The cells werekept frozen at 280°C until total RNA was extracted from them usingTrizol (Life Technologies, Rockville, MD) according to the manufac-turer’s instructions. Poly(A)1 RNA was purified by two rounds ofoligo(dT) cellulose column chromatography (Life Technologies, Rock-ville, MD). cDNA synthesis and library construction were subse-quently carried out using the Uni-ZAP XR library kit (Stratagene, LaJolla, CA), according to manufacturer’s instructions.

PCR Cloning StrategyCytochrome P450 cDNA fragments were amplified from the elici-

tor-induced cDNA library using various combinations of degenerateforward and reverse primers with the vector-specific T3 and T7primers. The template DNA was prepared from a 500 mL aliquot ofthe elicitor-induced cDNA library (3 3 106 PFU/mL) by heat dena-turation at 70°C for 10 min, followed by phenol/chloroform extrac-tion, ethanol precipitation, and resuspension in 500 mL of sterile,deionized water. Amplification reactions were performed in 50-mLolumes containing 50 mM KCl, 10 mM Tris–HCl, pH 8.8, 1.5 mMgCl2, 200 mM of each dNTP, 2 mL template DNA, 20 pmol each of

orward and reverse primer, and 1 U Taq Polymerase (Life Technol-gies). Reactions were preheated at 94°C for 2 min, followed by 35

ycles of denaturing at 94°C for 1 min, annealing at 50°C for 1.5 min,

nd polymerization at 72°C for 2 min. The reactions were completedy a 10-min extension at 72°C. Aliquots of the reaction products werexamined for DNA products by agarose gel fractionation and ligatedirectly into the pGEM-T Easy vector (Promega, Madison, WI). Re-ulting recombinant plasmids containing insert DNAs within thexpected size range were sequenced using T7 and Sp6 primers.

DNA SequencingAll the DNA sequencing reactions were performed using the Big-

Dye Terminator Cycle sequencing kit (Perkin-Elmer, Wellesley,MA), with the sequences being read on an automated ABI Prism 310Genetic Analyzer (Applied Biosystems, Foster City, CA). Computerassessment of the DNA sequence information was performed usingthe MacVector (Oxford Molecular, Madison, WI) software package.

cDNA Library ScreeningThe cDNA library was screened with digoxigenin labeled probes. A

258-bp DNA fragment amplified from the pGEM-deg6.4 clone usinggene-specific forward (59-GGCGGAGAATTTGTCCTGGAATGT-CATTTGGTTTAG-39) and reverse (59-GTACAATAGTGAGGTTGA-CAATG-39) primers was used to screen for CYP71Ds; and a 374-bpDNA fragment amplified from the pBKS-CYPB3.843 clone with spe-cific forward (59-GGTGGTTGTGAATGCATG-39) and reverse (59-TTATGCAGCAATAGGCTTGAAGACA-39) primers. The probes werelabeled with digoxigenin-11–dUTP using the PCR DIG Labeling Mix(Roche Molecular Biochemicals, Indianapolis, IN), hybridized toplaque lifts of the cDNA library plated at approximately 10,000 PFUper 150-mm plate, and hybridization detected with the DIG detectionsystem according to the manufacturer’s instructions (Roche Molecu-lar Biochemicals). Plaques exhibiting strong hybridization wereplaque purified, auto-subcloned to their plasmid forms according tothe manufacturer’s recommendations (Stratagene, La Jolla, CA),and subjected to DNA sequencing as described above.

RNA AnalysisRNA gel blot analysis was carried out using 10-mg aliquots of total

RNA. RNA samples were heat-denatured at 70°C for 15 min insample buffer (13 Mops, 50% formamide, 16% formaldehyde, 30%glycerol, and 3% ethidium bromide) and then size fractionated on a1.2% agarose gel containing 13 Mops and 18.1% formaldehyde (mod-ified from (33)). Uniformity of sample loading was determined byvisual inspection of the gel for rRNA bands. The RNAs were thentransferred to a Zeta Probe nylon membrane (Bio-Rad Laboratories,Hercules, CA) and hybridized according to the manufacturer’s rec-ommendations. Full-length cDNAs probes were labeled with[32P]dCTP (Prime-it Kit, Stratagene, La Jolla , CA) prior to hybrid-zation. After hybridization, the membranes were washed in 23SC/0.1% SDS once at room temperature followed by sequentialashes in 0.23 SSC/0.1% SDS at 42 and 65°C. Hybridization wasetected with a Phosphoimager (Molecular Dynamics, Model 445 SI).

Construction of Yeast Expression VectorsThe coding regions of the P450 cDNAs were cloned into the

pYeDP60 expression vector (19, 34). Appropriate BamHI, EcoRI, andSstI restriction sites (underlined) were introduced via PCR primerscontaining these sequences either upstream of the translation startsite (ATG) or downstream of the stop codon (TAA or TGA). Theprimers used to amplify the CYP71D20 cDNA were 59-GGGGGATC-CATGCAATTCTTCAGCTTGGTTTCC-39 and 59-GGGGAATTCT-TACTCTCGAGAAGGTTGATAAGG-39; for the CYP82E1 cDNA59-CCCGGATCCATGTATCATCTTCTTTCTCCC-39 and 59-GGGGA-ATTCTCAATATTGATAAAGCGTAGGAGG-39; and for the CYP92A3

cDNA 59-CCCGGATCCATGCAATCCTTCAGCTTGGTTTCC-39 and

TC

G

t

UF(Plym

r

2Atayaaf

b2abvg1Eggaumsoa

wrmftmmsawotfsTc

uprd4patm

226 RALSTON ET AL.

59-GGGGAGCTCTCACTCGCAAGAAGATTGATAAGG-39. Two long,overlapping (italicized) primers 59-GCCATTATCGGCGCAA-

ACTAATCTCCAAACTCCGCGGTAAAAAATTCAAGCTCCCAC-TGGTCCAACAGCAGTC-39 and 59-GGGGGATCCATGGACCTC-

CTCCTCATAGAAAAAACCCTCGTCGCCTTATTCGCCGCCATTAT-CGGCGCAATACTA-39 coding for the N-terminal sequence of CYP73A1(GenEMBL Z17369) up to the hinge region were used for the modifica-tion of the membrane anchoring segment of CYP73A27 to avoid possibleproblems with intracellular targeting due to the unusual N-terminus(35); the reverse primer used for both amplifications was 59-

GGGAGCTCTTATGCAGCAATAGGCTTGAAGAC-39. CYP71D20and CYP73A27 were amplified using full-length cDNA templates,whereas CYP82E1 and CYP92A5 were amplified directly from thecDNA library template. Amplifications were performed in 50-mL reac-ions containing 13 Pfx amplification buffer, 1 mM MgSO4, 300 mM of

each dNTP, 10 ng template DNA, 20 pmol each of forward and reverseprimer, and 1.25 U Platinum Pfx Polymerase (Life Technologies). Re-actions were preheated at 94°C for 2 min, followed by 35 cycles ofdenaturing at 94°C for 15 s, annealing at 55°C for 30 s, and elongatingat 68°C for 1.5 min. PCR products were ligated into the pGEM-T Easyvector (Promega) and subcloned into the pYeDP60 vector; and theresulting constructs were validated by a combination of PCR and DNAsequencing.

Yeast Expression Studies

Verified pYeDP60- P450 cDNA constructs were introduced into theyeast WAT11 line, a derivative of the W303-1B strain (MAT a; ade2-1; his 3-11; leu 2-3, 2112; ura 3-1; canR; cyr1), provided by Dr. P.

rban (Centre de Genetique Moleculaire, CNRS, Gif-sur-Yvette,rance). The endogenous NADPH-cytochrome P450 reductase

CPR1) locus has been replaced with ATR1, a NADPH-cytochrome450 reductase from Arabidopsis thaliana (19, 36), in the WAT11

ine. Yeast were grown overnight in a 30°C shaker in YPAD (1 g/Least extract; 1 g/L peptone; 20 g/L glucose; 200 mg/L adenine) liquidedia. Cultures were harvested at an A 600 between 0.5 and 1.5. Cells

were collected by centrifugation at 2500g for 5 min at 4°C, andesuspended in ice-cold, sterile dH2O. Cells were pelleted again as

above and resuspended in 1 M sorbitol. Forty microliters of yeastsuspension was mixed with 0.5 to 1 mg plasmid DNA (in ,5 mLdH2O) in a prechilled 0.5-mL tube and transferred to a chilled cu-vette with a 0.2-cm electrode gap. One pulse at 1.5 kV, 25 mF, and00 Ohms was applied by an Eppendorf electroporator (Model 2510).mixture of 500 mL of YPAD/1 M sorbitol was immediately added to

he electroporated cells. Cells were allowed to recover at 30°C for 1 hnd then spread onto SGI plates (1 g/L bactocasamino acids, 7 g/Least nitrogen base, 20 g/L glucose, 20 mg/L tryptophan, and 20 g/Lgar). Transformed colonies appeared after 3 to 6 days of incubationt 30°C. Recombinant plasmids were confirmed by PCR assays per-ormed directly on randomly selected yeast colonies.

For expression studies, one colony was added to SGI media (1 g/Lactocasamino acids, 7 g/L yeast nitrogen base, 20 g/L glucose, and0 mg/L tryptophan) and grown at 30°C for approximately 24 h. Anliquot of this culture was diluted 1:50 into 250 ml of YPGE (10 g/Lactopeptone, 10 g/L yeast extract, 5 g/L glucose, and 3% ethanol byolume) and was grown until all glucose was consumed. Absence oflucose was determined by placing a 200-mL aliquot of culture into a.5-mL tube, inserting a Diastix urinalysis reagent strip (Bayer,lkhart, IN) for 30 s, and observing colorimetric changes indicatinglucose levels. Induction was initiated by the addition of 5 g ofalactose (final concentration of 2%). The cultures were maintainedt 30°C for an additional 16 h before collecting the cells by centrif-gation at 7000g for 10 min. The pelleted cells were washed with 100L of TES buffer (50 mM Tris–HCl, pH, 7.5, 1 mM EDTA, 0.6 M

orbitol). The cells were centrifuged as above, resuspended in 100 mLf TES-M (TES supplemented with 10 mM 2-mercaptoethanol), and

llowed to incubate at room temperature for 10 min. The yeast cells

ere centrifuged again at 7000g for 10 min, and the pellet wasesuspended in 2.5 mL of extraction buffer (1% bovine serum albu-in, fraction V, 2 mM 2-mercaptoethanol, 1 mM phenylmethylsul-

onyl fluoride, all dissolved in TES). Glass beads (0.5 mm in diame-er, Biospec Products, Inc., Bartlesville, OK) were added until skim-ing the surface of the cell suspension. Cell walls were disruptedanually by hand shaking in a cold room for 10 min at 30-s intervals

eparated by 30-s intervals on ice. Cell extracts were transferred to50-mL centrifuge tube, the glass beads were washed three timesith 5 mL of extraction buffer, and the washes were pooled with theriginal cell extracts. Microsomes were prepared by differential cen-rifugation at 10,000g for 10 min at 4°C to remove cellular debrisollowed by centrifugation at 100,000g for 70 min at 4°C, and micro-omal pellets were resuspended in 1.5 mL of TEG-M buffer (50 mMris–HCl, pH 7.5, 1 mM EDTA, 20% glycerol, and 1.5 mM 2-mer-aptoethanol) and stored frozen at 280°C until further assayed.

CO Difference SpectraFe21 ● CO vs Fe21 difference spectroscopy (13) was performed

sing 0.4 mL of microsomes suspended in 1.6 mL of 50 mM Tris–HCl,H 7.5, 1 mM EDTA, and 20% glycerol. A small amount of theeducing agent, sodium dithionite, was added, and the mixture wasistributed between two cuvettes. A baseline was recorded between00 and 500 nm on a Perkin Elmer Lambda 18 UV/visible spectro-hotometer. CO was then bubbled into the sample cuvette for 1 min,nd the difference spectrum recorded again. The amount of func-ional P450 was estimated based on an absorbance coefficient of 91M21 cm21.

5-epi-Aristolochene–1,3-Dihydroxylase AssaysAssays were performed in 0.5-mL polyethylene tubes in 100-mL

volumes. 5-epi-Aristolochene or 1-deoxycapsidiol dissolved in hexanewas added to the tube, and the organic solvent was removed byincubation of the open tube at 30°C. 5-epi-Aristolochene and 1-de-oxycapsidiol were dissolved in 2 mL of dimethyl sulfoxide beforeadding the reaction mixture. Reactions were carried out in 100 mMTris–HCl, pH 7.5, to which microsomal protein was added to a finalconcentration of 1 mg/mL. Reactions were initiated by the addition of2 mM NADPH. The final concentration of 5-epi-aristolochene and1-deoxycapsidiol in these assays varied from 20 to 50 mM. Afterincubations for variable lengths of time at 30°C, the reactions mix-tures were extracted with two volumes of ethyl acetate. The organicextracts were concentrated and evaluated by GC and GC-MS alongwith standards of 5-epi-aristolochene (1, 30), 1-deoxycapsidiol (37),and capsidiol (38, 39). GC analysis was routinely performed with anHP5890 GC equipped with an Hewelett-Packard HP-5 capillary col-umn (30 m 3 0.25 mm, 0.25 mm phase thickness) and FID asdescribed previously (30). GC-MS analysis was performed at theUniversity of Kentucky Mass Spectrometry Facility using a Varian3400 GC and a Finnigan INCOS 50 quadrupole mass selective de-tector. The GC was equipped with a J&W DB-5ms capillary column(15 m 3 0.25 mm, 0.25 mm phase thickness) and run with He as thecarrier gas (10 p.s.i.). Splitless injections were done at an injectionport temperature of 280°C. The column temperature was maintainedat 40°C for 1 min and then increased to 280°C at 10°C per min.Following separation by the GC column, samples were introduceddirectly into the electron impact ionization source. Mass spectrawere acquired at 70 eV, scanning from 40 to 440 Da in 1 s.

RESULTS

Inhibition of the 5-epi-Aristolochene to CapsidiolConversion by P450 Antagonists

In earlier work by Mandujano-Chavez et al. (16), an

in vivo assay for measuring 5-epi-aristolochene hy-

5dncrniswEcpwca

Pb5

tdiiadaw

lct

qow

227CLONING AND CHARACTERIZATION OF 5-epi-ARISTOLOCHENE-1,3-DIHYDROXYLASE

droxylating (5EAH) activity was described. This assaymeasures the incorporation rate of exogenously sup-plied [3H]-5-epi-aristolochene into extracellular cap-sidiol by cell cultures over a 3-h incubation period, andour intent was to use this assay to validate and opti-mize various aspects of the molecular strategy de-scribed below. Using this indirect assay, a detailedinduction time course of 5EAH activity in elicitor-in-duced cell cultures was determined relative to that ofEAS activity (Fig. 1), the well-characterized sesquiter-pene cyclase activity that catalyzes the formation of5-epi-aristolochene from FPP (Scheme I). Because theEAH assay is an indirect assay, it was not possible toetermine the enzyme activity quantitatively (i.e.,mol product produced/min ● mg protein). Hence, theomparison between EAS and 5EAH is reported as theelative change in the respective enzyme activity. Asoted in our previous reports (11, 32, 40), EAS activity

s not detectable in control cell cultures, but is inducedignificantly within 3 h and reaches its maximal levelithin 15 to 18 h of elicitor treatment. Similar to theAS enzyme activity, 5EAH activity was negligible in

ontrol cell cultures. However, after an apparent laghase of 8 h, a rapid induction of hydroxylase activityas observed 10 to 15 h after elicitor addition to the

ell cultures, reaching a maximum by 18 h followed byrather gradual decline of 10 to 20% over the next 8 h.If the 5EAH activity measured in this assay is a

450-mediated reaction, then P450 antagonists woulde expected to hinder the incorporation of radiolabeled

FIG. 1. Induction time course for sesquiterpene cyclase enzymeactivity and sesquiterpene hydroxylase activity in cellulase-treatedcell cultures. Sesquiterpene cyclase (5-epi-aristolochene synthase,EAS) enzyme activity was determined in extracts prepared fromcontrol (open squares) and elicitor-treated (closed squares) cells col-lected at the indicated time points. Sesquiterpene hydroxylase activ-ity was determined using an indirect assay for control (open circles)and elicitor-treated (closed circles) cells. Cell cultures were incu-bated with [3H]-5-epi-aristolochene for 3 h, ending at the indicatedtime points before quantifying the incorporation of radioactivity intoextracellular capsidiol, a dihydroxylated form of aristolochene (16).

-epi-aristolochene into capsidiol by the elicitor-treated

obacco cell cultures. P450 inhibitors such as ancymi-ol and ketoconazole, however, have been reported tonhibit several enzymatic reactions (14, 15). Therefore,n addition to measuring the effects of these putativentagonists on 5EAH activity, their effects on the in-uction of EAS activity in elicitor-treated cells as wells their direct effects on in vitro assays of EAS activityere determined.Tobacco cell suspension cultures treated with cellu-

ase plus varying concentrations of ancymidol or keto-onazole were preincubated for 12 h before measuringhe cells’ abilities to convert exogenous supplied [3H]-

5-epi-aristolochene to [3H]capsidiol during a subse-uent 3-h incubation period (Fig. 2). Apparent activityf 5EAH was inhibited in a dose-dependent mannerith approximately 50% inhibition by either 25 mM

ancymidol or ketoconazole, and more than 80% by 75mM ancymidol and 95% by 100 mM ketoconazole (Figs.2A and 2B). Importantly, neither the in vitro activity ofrecombinant EAS nor the induction of EAS in the elic-

FIG. 2. Dose-dependent inhibition of 5-epi-aristolochene hydroxy-lase activity by ancymidol and ketoconazole. Cell cultures were in-cubated in the presence of cellulase (0.5 mg/mL) plus the indicatedconcentrations of ancymidol (A) or ketoconazole (B) for 12 h prior tomeasuring the in vivo 5-epi-aristolochene hydroxylase activity in thecell suspension cultures (squares) or the EAS enzyme activity inextracts prepared from the collected cells (triangles). The in vitroactivity of a purified EAS preparation (57) was also measured at theindicated inhibitor concentrations as an additional test for nonspe-

cific effects of these inhibitors (circles).

228 RALSTON ET AL.

itor-treated cell cultures was significantly affected byancymidol at concentrations as high as 100 mM (Fig.2A). Ketoconazole also does not appear to affect the invitro activity of EAS. However, the inducibility of cy-clase activity in elicitor-treated cell extracts was inhib-ited by ketoconazole at concentrations above 50 mM(Fig. 2B). Therefore, the specificity of ketoconazole asan inhibitor of P450 type reactions should be assessedonly at or below a concentration of 50 mM under theseexperimental conditions.

Isolation of Elicitor-Inducible Cytochrome P450cDNAs

The above results corroborated earlier reports (1, 2)suggesting that the EAH activity was mediated by aP450-type enzyme. However, further attempts to char-acterize this activity biochemically in microsomal frac-tions of tobacco cells were frustrated by difficulties inaccurately measuring this activity in cell-free extracts.We therefore pursued a rather common molecular ap-proach of trying to enrich for particular P450 activitiesby overexpressing the corresponding cDNA in a heter-ologous host (18, 41). Isolation of candidate P450cDNAs required a two-step approach. A PCR strategywas first employed using a directional cDNA libraryprepared against mRNA isolated from elicitor-inducedcells as the template and degenerate PCR primers (Fig.3). Sequence alignments of cytochrome P450s frommultiple families across kingdoms were used to iden-tify conserved regions to which a series of degenerateprimers were prepared (Figs. 3A and 3B). While theintent was to design primers capable of amplifyingP450 cDNAs from all families, this was not possible

FIG. 3. A schematic diagram of the primary structure of a generalizprimers highlighted (A). Degenerate P450-specific primers (B) weamplification of cytochrome P450 cDNA fragments. Ethidium bromidirectional cDNA library prepared with mRNA isolated from elicivector-specific primers.

because of incomplete conservation within the chosen

regions across all P450 families. Hence, due consider-ation was given to limiting the degree of degeneracy inthe PCR primers. Primer pairs for the PCR reactionsconsisted of a P450-specific primer in either orientation(forward or reverse) and a corresponding vector-spe-cific primer (T7 or T3). Multiple reaction products weregenerally observed within the size range expected forthe particular primer pair combination. For example,450- to 550-bp products were expected from reactionsutilizing the primer prepared to the heme-binding do-main (GRRXCP(A/G)) and the T7 vector primer (Fig.3C). The mixtures of reaction products were shotguncloned, and approximately 100 of the cloned PCR frag-ments were sequenced. About half of the sequencedDNAs contained signature sequences typical of P450enzymes as revealed by BlastX database searches, andthese corresponded to typical plant P450 family mem-bers of the CYP71, CYP73, CYP92, and CYP82 classes(Table I). Each of these PCR fragments was isolatedmultiple times in separate experiments.

The cloned fragments were used in a second step toisolate full-length clones from the cDNA library.Screening the cDNA library by hybridization with theCYP71 and CYP73 gene fragments yielded four full-length cDNAs, two CYP71Ds, and two CYP73As. Theformer clones were designated CYP71D20 andCYP71D21, and the latter were designated CYP73A27and CYP73A28. The other two cDNA fragments corre-sponded to tobacco cDNAs already found in the Gen-Bank database, CYP82E1 and CYP92A3. These twocDNAs were cloned using specific primers designedwith the help of the available sequence information toamplify the full-length cDNA. All six full-length

cytochrome P450 with conserved domains used for the design of PCRused in various combinations with vector specific primers in thestained agarose gel (C) showing the PCR products amplified from a-treated cells using the degenerate GRRXCP(A/G)-for and the T7

edrede-tor

cDNAs were screened for homology with other se-

an

I

tvpmcylctfwoit9mecCEim

F

cavWametiutnpciaw

1

229CLONING AND CHARACTERIZATION OF 5-epi-ARISTOLOCHENE-1,3-DIHYDROXYLASE

quences in the GenBank database. CYP71D20 andCYP71D21 were most similar to CYP71D7 from Chacopotato (Solanum chacoense); CYP73A27 andCYP73A28 were most closely related to CYP73A15from kidney bean (Phaseolus vulgaris); CYP82E1 hadlready been cloned from tobacco; and CYP92A5 wasearly identical to CYP92A3 from tobacco (Table I).

nduction of Cytochrome P450 mRNAsin Elicitor-Treated Cells

Because it was not possible with the above informa-ion to ascribe a particular biochemical function to thearious cytochrome P450 cDNAs, we rationalized thaterhaps the particular P450 involved in sesquiterpeneetabolism would have an mRNA induction pattern

omparable to the EAS mRNA (16, 42). RNA blot anal-ses were therefore used to determine the steady-stateevels of the mRNAs coding for all four of the cyto-hrome P450 clones and EAS in control and elicitor-reated cells (Fig. 4). Unexpectedly, the mRNAs for allour of the P450s were rapidly and transiently inducedith slightly different time courses relative to one an-ther and to the EAS mRNA. CYP73A27 mRNA, fornstance, displayed an induction pattern similar tohat of EAS with the maximum mRNA level occurring

to 12 h after elicitation. However, while the EASRNA remained high throughout the duration of the

xperiment, the CYP73A27 mRNA was negligible inells 24 h after elicitor-treatment. In contrast, theYP71D mRNA was more rapidly induced than theAS mRNA, reached its maximum 6 to 9 h after elic-

tation, and was declining by 12 h when the EAS

TABLE I

Deduced Amino Acid Sequence Comparisons betweenthe Open-Reading Frames for the Isolated CYP cDNAs

Indicated to the Nearest Family Member Reportedin the GenBank Database

Cytochrome P450cDNA clone

Nearest relative/Accession No.

%Identity

%Similarity

CYP71D20 CYP71D7 (S. chacoense)Gen EMBL U48435

76.5 88.8

CYP71D21 CYP71D7 (S. chacoense)Gen EMBL U48435

76.3 88.8

CYP73A27 CYP73A15 (P. vulgaris)Gen EMBL Y09447

79.4 92.6

CYP73A28 CYP73A15 (P. vulgaris)Gen EMBL Y09447

79.2 92.4

CYP82E1 CYP82E1 (N. tabacum)Gen EMBL AB015762

100.0 100.0

CYP92A5 CYP92A3 (N. tabacum)Gen EMBL X96784

95.5 98.6

Note. Percentage similarities and identities were determined us-ing the align algorithm of the MacVector software suite.

RNA level was still very high.

unctional Identification of CYP71D20 as 5-epi-Aristolochene Hydroxylase

To ascribe functional identity to the various P450DNAs, full-length cDNAs for CYP71D20, CYP82E1nd CYP92A5 were inserted into the yeast expressionector pYeDP60 (19, 34) and the expression of each inAT11, a yeast line containing an integrated A. thali-

na cytochrome reductase gene (19, 36), was deter-ined. Engineering the CYP73A27 cDNA required an

xtra modification because of an unusually long N-erminus with several hydrophilic residues that couldnterfere with proper intracellular targeting (35). Thisnusual leader sequence was therefore replaced withhe membrane anchoring sequence of CYP73A1, a cin-amate 4-hydroxylase previously demonstrated to ex-ress well in yeast (19, 43). Expression of all theDNAs was under the control of the glucose-repress-ble, galactose-inducible GAL10-CYC1 promoter (44),nd expression was compared to yeast transformedith the parent pYeDP60 vector (control) alone.After induction with galactose for approximately

6 h, control cells and cells containing the various P450

FIG. 4. Induction time course for CYP71D, CYP73A, CYP82E,CYP92A, and EAS transcript accumulation in elicitor-treated cells.Total RNA was extracted from tobacco suspension cells incubatedwith the cellulase elicitor for the indicated times, size fractionated byagarose gel electrophoresis under denaturing conditions, and trans-ferred to a nylon membrane before probing with the respective full-length cDNAs. Uniformity of sample loading was verified by

ethidium bromide staining of ribosomal RNA (Loading control).

1siaGfaLtswMsNewrrdttstt(t

sattcfi

tda1

230 RALSTON ET AL.

constructs were collected, and microsomes preparedfrom each were analyzed for general P450 expressionby CO-difference spectroscopy (13). Preparations fromcells containing the CYP73A and CYP82E constructs,or the vector-only controls did not display any COdifference absorbance peaks (data not shown). How-ever, microsomes prepared from cells containing theCYP71D20 (Fig. 5B) and CYP92A5 (Fig. 5A) constructsboth showed characteristic CO difference spectra withpeaks at 450 nm, indicating that the encoded proteinswere assembling properly with their heme cofactor.Using the extinction coefficient of 91 mM21 cm21 forheme binding proteins (13), it was determined thatapproximately 107 pmol of CYP71D20 and 268 pmol ofCYP92A5 were expressed in the yeast cells per mg oftotal yeast protein.

Because earlier work (1, 2) suggested that the twohydroxylations of capsidiol might be mediated by sep-arate reactions, microsomes from control cells(pYeDP60 vector only) or from cells expressing eitherthe CYP71D20 and CYP92A5 cDNAs were tested for

FIG. 5. CO difference spectra of the microsomal fraction isolatedfrom yeast expressing the CYP92A5 (A) and CYP71D20 (B) cDNAs.Expression of the respective plasmid constructs engineered into theyeast (WAT11) cells was induced by a galactose treatment, followedby isolation of microsomal preparations. The difference adsorptionspectra of microsomes incubated in the presence (solid lines) andabsence (broken lines) of carbon monoxide was determined.

their ability to metabolize 5-epi-aristolochene and

-deoxycapsidiol (Fig. 6). Microsomes were incubatedeparately with 5-epi-aristolochene or 1-deoxycapsidioln the presence or absence of NADPH, and the ethylcetate-extractable reaction products were analyzed byC. Whether or not NADPH was present, microsomes

rom control yeast cells did not metabolize either 5-epi-ristolochene (Fig. 6B) or 1-deoxycapsidiol (Fig. 6D).ikewise, no discernable metabolism of 5-epi-aris-olochene or 1-deoxycapsidiol was observed with micro-omes from cells expressing the CYP92A5 cDNAhether NADPH was present or not (data not shown).icrosomes from yeast expressing the CYP71D20 con-

truct, however, metabolized both substrates in anADPH-dependent manner (Figs. 6A and 6B). Inter-

stingly, both 5-epi-aristolochene and 1-deoxycapsidiolere metabolized to only one product with the same

etention time as capsidiol. Obvious by its absence, noeaction product having a retention time similar toeoxycapsidiol was detectable in the 5-epi-aris-olochene incubations (Fig. 6A). Coinjection of authen-ic capsidiol with the respective reaction products re-ulted in a single GC peak having a 16.2-min retentionime, identical to capsidiol. Mass spectra patterns forhe separate reaction products (EI-MS m/z (%), 22112), 157 (31), 93 (63), 55 (60), 41 (100)) were identicalo those for the capsidiol standard.

DISCUSSION

The current results confirm and extend previous ob-servations. Using an in vivo assay similar to that de-scribed here, Whitehead et al. (1) determined thatwhile little radiolabeled 5-epi-aristolochene was incor-porated into capsidiol by control tobacco callus cul-tures, much more incorporation occurred when the la-bel was fed to elicitor-treated callus tissues. They alsoreported that 1-deoxycapsidiol (5-epi-aristolochene hy-droxylated at C3) was converted to capsidiol when fedto control cultures. These results, the authors sug-gested, were consistent with a simple two-step path-way for the conversion of 5-epi-aristolochene to cap-idiol (Scheme I). Because the enzyme responsible fordding a hydroxyl to C1 of 1-deoxycapsidiol appearedo be constitutive, Whitehead et al. (1) concluded thathe inducible C3 hydroxylase was a committed step inapsidiol biosynthesis. Hoshino et al. (2) provided therst in vitro characterization of the C3 hydroxylase

using microsomal preparations from elicitor-inducedpepper. The assay developed by these investigatorsmeasured the amount of 1-deoxycapsidiol generatedupon incubation of pepper microsomes with 5-epi-aris-olochene by GC analysis. This assay was then used toocument the NADPH dependence of the hydroxylasectivity, an inhibition of hydroxylase activity by CO,00 mM ancymidol, and 100 mM ketoconazole, and an

approximate 20-fold induction of enzyme activity

ch1rotPFtbrr

htzu

eslaiimpo(so(ytcabt5t

231CLONING AND CHARACTERIZATION OF 5-epi-ARISTOLOCHENE-1,3-DIHYDROXYLASE

within 18–24 h of elicitor treatment. Interestingly,Hoshino et al. (2) reported never observing 3-deoxycap-sidiol as a product, nor apparently the dihydroxylatedproduct, capsidiol. Together, these two reports sug-gested that the conversion of 5-epi-aristolochene toapsidiol occurred in two independent steps. The firstydroxylation at C3 of 5-epi-aristolochene to generate-deoxycapsidiol being directed by a P450 mediatedeaction that was elicitor-inducible, followed by a sec-nd hydroxylation at C1 being mediated by a constitu-ive activity that may or may not be mediated by a450-like enzyme. The in vivo assay data presented inigs. 1 and 2 of the current work are consistent withhe conversion of 5-epi-aristolochene being catalyzedy at least one inducible cytochrome P450 mediatedeaction. An important caveat to this conclusion is thatesults with concentrations of ketoconazole above 50

mM should be interpreted with caution, since higherconcentrations clearly had secondary effects, as notedfor the induction of sesquiterpene cyclase activity inthe elicitor-treated cells (Fig. 2). This in vivo assay,

owever, could not determine whether the hydroxyla-ions were being catalyzed by one or two P450 en-ymes, and whether one activity or both might be reg-lated by elicitation of the cells.Biochemical characterization of the enzyme activity

ncoded by the CYP71D20 cDNA expressed in yeastuggests that this enzyme is, in fact, capable of cata-yzing both the C1 and C3 hydroxylations of 5-epi-ristolochene in a sequential manner (Fig. 6). The ev-dence for this is, however, inferred from the biochem-cal analyses performed with yeast microsomes, a

ixture of many proteins and possibly nonspecific ter-ene hydroxylases, rather than a purified preparationf only the CYP71D20 protein. Control microsomesmicrosomes from yeast containing vector control con-tructs) do not metabolize 5-epi-aristolochene or 1-de-xycapsidiol in the presence or absence of NADPHFigs. 6B and 6D). Moreover, only microsomes fromeast expressing the tobacco CYP71D20 cDNA are ableo convert 5-epi-aristolochene or 1-deoxycapsidiol toapsidiol in an NADPH-dependent manner (Figs. 6And 6C), demonstrating that this one enzyme catalyzesoth hydroxylations. An alternative interpretation ishat the CYP71D20 encoded enzyme is converting-epi-aristolochene to either 1- or 3-deoxycapsidiol,hen releasing this intermediate reaction product

FIG. 6. Gas chromatograms of the reaction products formed uponincubation of microsomes isolated from WAT11 yeast cells contain-ing the CYP71D20 expression construct (A and C) or vector controlDNA (B and D) with sesquiterpene substrates. Microsomes isolatedfrom the indicated yeast lines were incubated with 5-epi-aris-tolochene (A and B) or 1-deoxycapsidiol (C and D) in the presence(solid lines) or absence (dashed lines) of NADPH. Identity of 5-epi-

aristolochene,1-deoxycapsidiol and capsidiol were verified by MS.

232 RALSTON ET AL.

lirigi

mht(cdftSibPmlwlP

eflo

rdcfC3cCh5dalStpattCdewtsitro

233CLONING AND CHARACTERIZATION OF 5-epi-ARISTOLOCHENE-1,3-DIHYDROXYLASE

which is rapidly converted to capsidiol by some endog-enous activity within the yeast microsomes. Thisclearly cannot be the case for 1-deoxycapsidiol, sincethe control microsomes cannot metabolize this com-pound (Fig. 6D). We formally cannot exclude at thistime the possibility that the CYP71D20 enzyme cata-lyzes the synthesis and release of 3-deoxycapsidiol asan intermediate. However, if this were the case, thenaccumulation of this mono-hydroxylated intermediatein reactions initiated with 5-epi-aristolochene might beanticipated, and yeast must possess a hydroxylasewith exquisite specificity for a mono-hydroxylatedplant terpene. Accumulation of 3-deoxycapsidiol hasnot been observed under a variety of reaction condi-tions (Fig. 6A), and we consider it unlikely that yeastpossess an hydroxylase that can so completely andefficiently differentiate between 1-deoxycapsidiol and3-deoxycapsidiol. Contrary to the conclusions ofHoshino et al. (2), the simplest interpretation of ourdata is therefore that the CYP73D20 enzyme catalyzessequential hydroxylations at the C1 and C3 positions of5-epi-aristolochene without release of mono-hydroxy-ated intermediates. Because a preferred order for thenitial hydroxylation cannot be inferred from the cur-ent information, we can indicate only that the enzymes capable of first hydroxylating at the C3 position toenerate 1-deoxycapsidiol and then hydroxylating thisntermediate at C1 to produce capsidiol.

Cytochrome P450 enzymes capable of catalyzingultiple reactions including sequential hydroxylationsave been described previously. In recent studies onhe catabolism of vitamin D, Akiyoshi-Shibata et al.45) and Beckman et al. (46) reported that mammalianytochrome P450cc24 catalyzes two independent hy-roxylations on the side-chain of vitamin D3, plus aurther oxidation of one hydroxyl group to a ketonehat leads to a C–C cleavage and side-chain removal.equential hydroxylations at C20 and C22 and an ox-

dative cleavage of the cholesterol side-chain have alsoeen ascribed to a functionally related cytochrome450 enzyme involved in pregnenolone biosynthesis inammals (47). Catalytically similar to these mamma-

ian enzymes is the bacterial cytochrome P450BM-3,hich introduces 3 hydroxyls into palmitic acid fol-

owed by the oxidation of one of the hydroxyls (48).lants also contain multifunctional cytochrome P450

FIG. 7. Amino acid sequence alignment of the N. tabacum 5-epAccession No. AF368376) and NtCYP71D21 (AF368377) with othermonohydroxylation of nerol and geraniol, linear monoterpenes, whi(52). MsCYP71D18 (AAD44150) and MpCYP71D15 (AAD44152) cmonoterpene, respectively (51). AtCYP701A3 (AAC39505) encodes foa hydroxylation followed by two oxidations of a diterpene (50). TcCYinto a diterpene taxane skeleton (55). The six putative substrate(corresponding sequences are boxed), as well as the position corre

hydroxylase from spearmint has been attributed to subtle changes at t

nzymes. Using a complementation assay to restoreower color in petunia mutants, Holton et al. (49) dem-nstrated that flavonoid-39,59-hydroxylases (CYP75s)

catalyze the stepwise hydroxylation at the equivalentmeta-positions of the B ring of flavanones, intermedi-ates in the anthocyanin biosynthetic pathway. Al-though several terpene hydroxylating activities inplants have been reported in the literature (50, 51),only ent-kaurene oxidase, a member of the CYP701family of P450 enzymes, has been demonstrated to bemultifunctional. Kaurene oxidase catalyzes the intro-duction of a hydroxyl group into ent-kaurene (a precur-sor to the growth regulator gibberellic acid) and itssubsequent oxidation to ent-kaurenoic acid (50).

The identification of CYP71D20 as a multifunctionalP450 enzyme that inserts 2 hydroxyls into a sesquiter-pene skeleton is interesting in comparison to ent-kau-ene oxidase (a diterpene oxidase) (50), geraniol hy-roxylase (52) and several other recently characterizedytochromes P450 that introduce a single hydroxylunction into monoterpene and diterpene skeletons.YP71D13 from peppermint encodes for limonene-hydroxylase which catalyzes a regio- and stereo-spe-ific hydroxylation of 4S-limonene at C3, whileYP71D18 from spearmint encodes for an activity thatydroxylates 4S-limonene only at the C6 position (51,3). Interestingly, the regio-specificity of the C6 hy-roxylase can be altered to a C3 specificity by a singlemino acid substitution of phenylalanine 361 with iso-eucine (54). Also important for comparitive purpose,choendorf et al. (55) recently reported on the charac-erization of a diterpene hydroxylase from Taxus cus-idata. This enzyme introduces an hydroxyl functiont the 10b-position of taxa-4(20),11(12)-dien-5a-yl ace-ate. Unfortunately, primary structure alignment ofhese hydroxylases relative to the CYP71D20 andYP71D21 described here does not readily point toomains that could be important for substrate prefer-nce (sesquiterpene versus mono- or diterpene) norhere catalytic specificity for sequential hydroxyla-

ions might lie (Fig. 7). Instead, this alignment is con-picuous because of the extensive sequence similarity/dentity dispersed throughout the aligned proteins andhe correspondence between the six putative substrateecognition sequences (56). Identifying those domainsf CYP71D20 important for substrate specificity and

istolochene (sesquiterpene) hydroxylases NtCYP71D20 (GenBanknt terpene hydroxylases. NrCYP71A5v1 (CAA70575) catalyzes theaCYP71A1 (A35867) catalyzes the epoxidation of these substrates

lyze the mono-hydroxylation at C6 and C3 of limonene, a cyclicnt-kaurene oxidase, which catalyzes a three-step reaction, including(AF318211) catalyzes the introduction of a single hydroxyl functionognition sequences, SRS, described by Gotoh (56) are highlightednding to F361 (*) in CYP71D18. Regio-specificity of 4S-limonene

i-arplale Patar eP

recspo

his site (54).

1

11

11

1

1

1

11

2

2

2

2

2

2

22

2

2

3

3

3

3

3

3

3

3

3

3

44

4

4

4

4

4

234 RALSTON ET AL.

multiple hydroxylations might be better accomplishedby a comparison to a more closely related enzyme suchas vetispiradiene hydroxylase from Hyoscyamus muti-cus. H. muticus is a solanaceous plant like tobacco, butmakes solavetivone as its primary phytoalexin in re-sponse to pathogen or elicitor challenge (57). Solav-etivone is derived from vetispiradiene, a bicyclic ses-quiterpene having one six-carbon membered ring andone five-carbon spirane, via a single hydroxylation atthe equivalent C2 position of the six-carbon ring. Se-quence alignment of 5-epi-aristolochene hydroxylase tovetispiradiene hydroxylase must await the isolationand functional characterization of the correspondingH. muticus cDNA.

ACKNOWLEDGMENTS

This work was supported by the Kentucky Agricultural Experi-ment Station and grants from NSF (J.C.) and NIH (R.M.C.). Wethank Drs. Michael Barrett and Daniel Werck-Reichhart for helpfulsuggestions during the course of this work, Dr. Jack Goodman for hisvaluable assistance with the GC–MS analysis, and Dr. David Nelsonfor assigning the cytochrome nomenclature for the enzymes de-scribed here.

REFERENCES

1. Whitehead, I. M., Threlfall, D. R., and Ewing, D. F. (1989)Phytochemistry 28, 775–779.

2. Hoshino, T., Yamaura, T., Imaishi, H., Chida, M., Yoshizawa, Y.,Higashi, K., Ohkawa, H., and Mizutani, J. (1995) Phytochemistry38, 609–613.

3. Back, K., He, S., Kim, K. U., and Shin, D. H. (1998) Plant CellPhysiol. 39, 899–904.

4. Keller, H., Czernic, P., Ponchet, M., Ducrot, P. H., Back, K.,Chappell, J., Ricci, P., and Marco, Y. (1998) Planta 205, 467–476.

5. Molot, P. M., Mas, P., Conus, M., Ferriere, H., and Ricci, P.(1981) Physiol. Plant Pathol. 18, 379–389.

6. Stolle, K., Zook, M., Shain, L., Hebard, F., and Kuc, J. (1988)Phytopathology 78, 1193–1197.

7. Chappell, J. (1995) Annu. Rev. Plant Physiol. Plant Mol. Biol. 46,521–547.

8. Back, K., and Chappell, J. (1996) Proc. Natl. Acad. Sci. USA 93,6841–6845.

9. Mathis, J. R., Back, K., Starks, C., Noel, J., Poulter, C. D., andChappell, J. (1997) Biochemistry 36, 8340–8348.

0. Starks, C. M., Back, K., Chappell, J., and Noel, J. P. (1997)Science 277, 1815–1820.

1. Vogeli, U., and Chappell, J. (1988) Plant Physiol. 88, 1291–1296.2. Watson, D. G., Baker, F. C., and Brooks, C. J. W. (1983) Biochem.

Soc. Trans. 11, 589.3. Omura, T., and Sato, R. (1964) J. Biol. Chem. 239, 2370–2378.4. Coolbaugh, R. C., Hirano, S. S., and West, C. A. (1978) Plant

Physiol. 62, 571–576.5. Rademacher, W. (2000) Annu. Rev. Plant Physiol. Plant Mol.

Biol. 51, 501–531.6. Mandujano-Chavez, A., Schoenbeck, M. A., Ralston, L. F., Lo-

zoya-Gloria, E., and Chappell, J. (2000) Arch. Biochem. Biophys.381, 285–294.

7. Chapple, C. (1998) Annu. Rev. Plant Physiol. Plant Mol. Biol. 49,

311–343.

8. Schuler, M. A. (1996) Crit. Rev. Plant Sci. 15, 235–284.9. Pompon, D., Louerat, B., Bronine, A., and Urban, P. (1996)

Methods Enzymol. 272, 51–64.0. Still, C. W., Kahn, M., and Mitra, A. (1978) J. Org. Chem. 43,

2923–2925.1. Schenk, D. J. (2000) Ph.D. dissertation, Univ. of Illinois,

Urbana–Champaign, IL.2. Whitehead, I. M., Threlfall, D. R., and Ewing, D. F. (1987)

Phytochemistry 26, 1367–1369.3. Hoyano, Y., Stoessl, A., and Stothers, J. B. (1980) Can. J. Chem.

58, 1894–1896.4. Gordon, M., Stoessl, A., and Stothers, J. B. (1973) Can. J. Chem.

51, 748–752.5. Whitehead, I. M., Ewing, D. F., Threlfall, D. R., Cane, D. E., and

Prabhakaran, P. C. (1990) Phytochemistry 29, 479–482.6. Barton, D. H. R., and Laws, G. F. (1954) J. Chem. Soc. 54, 52–63.7. Das, J., Valenta, Z., Liu, H-J., and Ngooi, T. K. (1984) Can.

J. Chem. 62, 481–483.8. O’Donohue, M. J., Gousseau, H., Huet, J. C., Tepfer, D., and

Pernollet, J. C. (1995) Plant Mol. Biol. 27, 577–586.9. Back, K., Yin, S., and Chappell, J. (1994) Arch. Biochem. Bio-

phys. 315, 527–532.0. Rising, K. A., Starks, C. M., Noel, J. P., and Chappell, J. (2000)

J. Am. Chem. Soc. 122, 1861–1866.1. Chappell, J., Nable, R., Fleming, P., Andersen, R. A., and Bur-

ton, H. R. (1987) Phytochemistry 26, 2259–2260.2. Vogeli, U., Freeman, J. W., and Chappell, J. (1990) Plant

Physiol. 93, 182–187.3. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular

Cloning: A Laboratory Manual, Cold Spring Harbor LaboratoryPress, Cold Spring Harbor, NY.

4. Urban, P., Cullin, C., and Pompon, D. (1990) Biochimie 72,463–472.

5. Nedelkina, S., Jupe, S. C., Blee, K. A., Schalk, M., Werck-Reich-hart, D., and Bolwell, G. P. (1999) Plant Mol. Biol. 39, 1079–1090.

6. Urban, P., Mignotte, C., Kazmaier, M., Delorme, F., and Pom-pon, D. (1997) J. Biol. Chem. 272, 19176–19186.

7. Whitehead, I. M., Ewing, D. F., Threlfall, D. R., Cane, D. E., andPrabhakaran, P. C. (1990) Phytochemistry 29, 479–482.

8. Milat, M. L., Dijon, F., Ricci, P., Bonnet, P., and Blein, J. P.(1991) Phytochemistry 30, 2171–2173.

9. Whitehead, I. M., Threlfall, D. R., and Ewing, D. F. (1987)Phytochemistry 26, 1367–1369.

0. Chappell, J., and Nable, R. (1987) Plant Physiol. 85, 469–473.1. Gonzalez, F. J., and Korzekwa, K. R. (1995) Annu. Rev. Phar-

macol. Toxicol. 35, 369–390.2. Facchini, P. J., and Chappell, J. (1992) Proc. Natl. Acad. Sci.

USA 89, 11088–11092.3. Fahrendorf, T., and Dixon, R. A. (1993) Arch. Biochem. Biophys.

305, 509–515.4. Guarente, L., Yocum, R. R., and Gifford, P. (1982) Proc. Natl.

Acad. Sci. USA 79, 7410–7414.5. Akiyoshi-Shibata, M., Sakaki, T., Ohyama, Y., Noshiro, M.,

Okuda, K., and Yabusaki, Y. (1994) Eur. J. Biochem. 224, 335–343.

6. Beckman, M. J., Tadikonda, P., Werner, E., Prahl, J.,Yamada, S., and Deluca, H. F. (1996) Biochemistry 35, 8465–

8472.

4

4

5

5

235CLONING AND CHARACTERIZATION OF 5-epi-ARISTOLOCHENE-1,3-DIHYDROXYLASE

47. Miller, W. L. (1987) J. Steroid. Biochem. 27, 759–766.8. Boddupalli, S. S., Pramanik, B. C., Slaughter, C. A., Estabrook,

R. W., and Peterson, J. A. (1992) Arch. Biochem. Biophys. 292,20–28.

9. Holton, T. A., Brugliera, F., Lester, D. R., Tanaka, Y., Hyland,C. D., Menting, J. G., Lu, C. Y., Farcy, E., Stevenson, T. W., andCornish, E. C. (1993) Nature 366, 276–279.

0. Helliwell, C. A., Poole, A., Peacock, W. J., and Dennis, E. S.(1999) Plant Physiol. 119, 507–510.

1. Lupien, S., Karp, F., Wildung, M., and Croteau, R. (1999) Arch.

Biochem. Biophys. 368, 181–192.

52. Hallahan, D. L., Lau, S. M., Harder, P. A., Smiley, D. W., Daw-son, G. W., Pickett, J. A., Christoffersen, R. E., and O’Keefe, D. P.(1994) Biochim. Biophys. Acta. 1201, 94–100.

53. Wust, M., Little, D. B., Schalk, M., and Croteau, R. (2001) Arch.Biochem. Biophys. 387, 125–136.

54. Schalk, M., and Croteau, R. (2000) Proc. Natl. Acad. Sci. USA 97,11948–11953.

55. Schoendorf, A., Rithner, C. D., Williams, R. M., and Croteau,R. B. (2001) Proc. Natl. Acad. Sci. USA 98, 1501–1506.

56. Gotoh, O. (1992) J. Biol. Chem. 267, 83–90.

57. Back, K., and Chappell, J. (1995) J. Biol. Chem. 270, 7375–7381.