Plant Physiol. (1989) 89, 1351-1357 Received for publication July 1, 1988 0032-0889/0000/1351 /07/$01 .00/0 and in revised form December 2, 1988 Biochemical and Histochemical Localization of Monoterpene Biosynthesis in the Glandular Trichomes of Spearmint (Mentha spicata)" 2

Jonathan Gershenzon, Massimo Maffei3, and Rodney Croteau* Institute of Biological Chemistry, Washington State University, Pullman, Washington 99164-6340

ABSTRACT type diterpenes found in the glandular exudate. The head cells The primary monoterpene accumulated in the glandular tri- oftobacco trichomes are the only leafcells able to incorporate chomes of spearmint (Mentha spicata) is the ketone (-)-carvone radiolabeled acetate into duvane diterpenes; removal ofthese which is formed by cyclization of the C10 isoprenoid intermediate cells severely reduces or eliminates duvane production (22). geranyl pyrophosphate to the olefin (-)-limonene, hydroxylation Thus, it is tempting to speculate that all terpenes found in to (-)-trans-carveol and subsequent dehydrogenation. Selective glandular trichomes are synthesized in situ. However, recent extraction of the contents of the glandular trichomes indicated reports of monoterpene synthesis by undifferentiated cells in that essentially all of the cyclase and hydroxylase activities culture (5, 28), coupled to evidence for monoterpene transport resided in these structures, whereas only about 30% of the in intact plants (7, 31), suggest that the site of synthesis may carveol dehydrogenase was located here with the remainder not necessarily be the same as the site of accumulation. located in the rest of the leaf. This distribution of carveol dehy- drogenase activity was confirmed by histochemical methods. Spearmint (Mentha spicata, Lamiaceae) accumulates large Electrophoretic analysis of the partially purified carveol dehydro- quantities of monoterpenes in glandular trichomes, the major genase from extracts of both the glands and the leaves following constituent of which is (-)-carvone (20). This monocyclic gland removal indicated the presence of a unique carveol dehy- ketone is biosynthesized by a three step pathway (Fig. 1) in drogenase species in the glandular trichomes, suggesting that which the ubiquitous primary intermediate geranyl pyrophos- the other dehydrogenase found throughout the leaf probably phate is cyclized to the olefin (-)-limonene4 (23), which is utilizes carveol only as an adventitious substrate. These results then hydroxylated by a cytochrome P450-dependent mono- demonstrate that carvone biosynthesis takes place exclusively oxygenase to (-)-trans-carveol (F Karp, R Croteau, unpub- in the glandular trichomes in which this natural product accumu- lished data). (-)-trans-Carveol is subsequently dehydrogen- lates. ated to (-)-carvone. This last enzymic step has not been previously demonstrated in spearmint, but has ample prece- dent in monoterpene metabolism in other systems (16, 24). The site of monoterpene biosynthesis in spearmint was studied by investigating the locations of the enzymes catalyz- Many kinds of lipophilic natural products accumulate in ing the three steps in carvone biosynthesis. Procedures for the modified epidermal hairs known as glandular trichomes (17). selective extraction ofenzymes from glandular trichomes were Prominent among these substances are various types of ter- used (12, 18), which allowed comparison of cell-free prepa- penes, including the monoterpenes and sesquiterpenes of the rations from the trichomes with those of whole leaves from essential oils. It has been generally assumed that the terpenes which the glandular trichomes had been removed. Also, since found in glandular trichomes are synthesized there, since the final step in carvone biosynthesis is mediated by a dehy- gland cells display many ultrastructural features indicative of drogenase, histochemical techniques for studying this enzyme active lipid metabolism and secretion (17, 29). However, the type (26, 32) were employed to probe the cellular and sub- difficulties of identifying terpene secretion products micro- cellular location of this activity. scopically have precluded definite proof of this assumption by ultrastructural methods. MATERIALS AND METHODS Direct evidence for the biosynthetic capabilities ofglandular trichomes has come from studies showing that these structures Plant Materials and Reagents can incorporate labeled precursors such as sucrose, acetate, and mevalonate into terpenes (9). In tobacco, glandular tri- Spearmint (Mentha spicata L.) plants were grown from chomes appear to be the sole site for the synthesis ofduvane- stolons under controlled conditions as previously described (1 1). Newly emerged leaves (3-20 mm long) were used in all 'This investigation was supported in part by U.S. Department of experiments. Polystyrene resin (Amberlite XAD4; Rohm and Energy grant (DE-FG06-88ER1 3869) and by Project 0268 from the Haas) was prepared for use by standard procedures (27). All Agricultural Research Center, Washington State University, Pullman, WA 99164. 4Nomenclature used is based on the p-menthane system: (-)- 2 Dedicated to the memory of W. R. Nes, colleague and friend. limonene = 4S-p-mentha- 1(2),8(9)-diene; (-)-trans-carveol = 2S,4R- Present address: Istituto di Botanica Speciale, University ofTurin, p-mentha-l(6),8(9)-dien-2-ol; (-)-carvone = 4R-p-mentha-1(6),8(9)- Italy. dien-2-one. 1351 1 352 GERSHENZON ET AL. Plant Physiol. Vol. 89, 1989

at 195,000g for 2 h. The resulting pellets were resuspended in 0/*. 0 the appropriate buffers for assay. 3 Assay for Geranyl Pyrophosphate: (-)-Limonene Cyclase This activity was measured in a 15 mm sodium/potassium Geranyl (-)-Limonene (-)-trans-Carveol (-)-Carvone phosphate buffer, containing 10% (v/v) glycerol, 5 mM ascor- pyrophosphate bic acid, and 1 mM DTT. One-mL aliquots of the various Figure 1. Pathway for the biosynthesis of (-)-carvone from geranyl extracts were added to Teflon-sealed, screw-capped tubes and pyrophosphate in spearmint. The enzymes involved are geranyl py- rophosphate: (--limonene cyclase (1), (--limonene hydroxylase (2), the reaction initiated by addition of 20 mM MgCl2 and 18 jM and (--trans-carveol dehydrogenase (3). [I-3H]geranyl pyrophosphate (90 Ci/mol) which was synthe- sized and purified by literature procedures (15). As a trap for the volatile olefin products, 1 mL of pentane was carefully other reagents were purchased from Aldrich or Sigma Chem- layered on top of the reaction mixture. Following incubation ical Co. unless otherwise noted. at 30C for 1 h with gentle shaking, the reaction was stopped by vigorous mixing. The limonene generated was extracted Enzyme Extracts from the reaction mixture, and the 3H content determined by chromatographic separation and liquid scintillation counting Extracts ofglandular trichomes were prepared by two meth- essentially as previously described (23). The cyclase activity ods. For the first method, leaves were submerged in prechilled was almost completely restricted (98%) to the 195,000g extraction buffer and gently brushed with a soft-bristle tooth- supernatants. brush (12). The extraction buffer was 100 mm sodium/potas- Assay for (-)-Limonene Hydroxylase sium phosphate (pH 6.5), containing 1 M sucrose, 5 mM MgCl2, 10 mm Na2S205, 50 mm ascorbic acid, 4 mM DTT This activity was determined by incubating 1 mL aliquots (Research Organics), 1 mm EDTA, polyvinylpolypyrrolidone of the preparation in 25 mm sodium/potassium phosphate (1 g/g leaf), and XAD-4 resin (2 g/g leaf). This procedure buffer (pH 7.4), containing 30% (v/v) glycerol and 0.5 mM removed approximately 40 to 60% ofthe glandular trichomes DTT, with 2 mm NADPH and 200 nmol (-)-limonene (op- present on the leaf surface as determined microscopically. tical purity > 80%) for 1 h at 30°C. The reaction was stopped For the second method, extracts were prepared by a mech- by addition of 1 mL diethyl ether followed by vigorous anized procedure in which the leaf surfaces were gently shaking to extract the (-)-trans-carveol formed. After addition abraded with small glass beads (18). Briefly, 5 to 15 g batches of 25 nmol camphor as an internal standard, the ether layer of leaves were extracted using a Bead-Beater cell disrupter was removed and the reaction mixture reextracted twice with (Biospec Products) containing 200 g of0.5 mm diameter glass additional 1 mL portions of ether. The combined ether ex- beads (Biospec Products) and prechilled extraction buffer tracts were decolorized with charcoal, washed with 1 mL of (formulated as described above) added to nearly full volume water, passed through a short column of silica gel (type 60A, of the 10 ounce polycarbonate chamber. Extraction was car- Mallinckrodt) overlaid with anhydrous MgSO4 in a Pasteur ried out in twenty 15 s pulses of operation with the rotor pipet, and concentrated to 200 ,L under vacuum (Savant speed controlled by a rheostat set at 110 V. Between pulses, Speed Vac). the polycarbonate chamber was dismounted and cooled on Reaction products were analyzed by GLC (Hewlett Packard ice for at least 15 s. This procedure resulted in removal of 5890A with 3392A integrator) using a bonded-phase fused- more than 99% of the glandular trichomes as determined silica open-tubular capillary column (30 m x 0.25 mm i.d.) microscopically. coated with a 0.2 ,um film of Superox FA (Ailtech Associates) Leaves recovered from the second gland extraction proce- and operated with H2 (2 mL/min), on-column injection (in- dure were manually homogenized in a Ten-Broeck homoge- jector temperature ambient), temperature programming (45° nizer to obtain extracts of whole leaves from which the for 5 min, then 10°/min to 2200 with 15 min hold), and flame glandular trichomes had been removed. Homogenization was ionization detector (detector temperature 230C). (-)-trans- carried out in the same extraction buffer, without XAD-4, Carveol was identified by comparison of retention time and but with additional polyvinylpolypyrrolidone (0.5 g/g leaf). mass spectrum with an authentic standard and was quanti- Following homogenization, the extract was slurried with tated by comparison of detector response to that of the XAD-4 (1 g/g leaf) for 5 min. internal standard. Control assays extracted immediately after All extracts were filtered by passage through eight layers of substrate addition (i.e. zero time) were used to determine the premoistened cheesecloth and then four layers of80 ,um nylon background ofendogenous volatile substances in each extract. mesh (Small Parts, Inc.). Filtered extracts were centrifuged at The small amounts of trans-carveol usually found in these 195,000g for 2 h. The supernatants obtained were divided control assays were subtracted from those produced in full- into several portions, dialyzed to specific assay conditions as term assays to determine enzymic production. Limonene noted below, and then concentrated by ultrafiltration (Ami- hydroxylase activity was found only in the 195,000g pellet, con YM-30) to a small volume suitable for assay. The and was not evident in boiled preparations ofthis type. 195,000g pellets were resuspended in extraction buffer, man- ually homogenized, stirred with XAD (1 g/g leaf) to ensure Assay for (-)-trans-Carveol Dehydrogenase complete removal of endogenous monoterpenes, filtered One-mL aliquots of the extract in 100 mm sodium/potas- through cheesecloth and nylon mesh, and centrifuged again sium phosphate buffer (pH 8.0), containing 10% (v/v) glycerol LOCALIZATION OF MONOTERPENE BIOSYNTHESIS IN SPEARMINT TRICHOMES 1 353 and 1 mM DTT, were incubated in the presence of 1 mM current of40 mA for 3 h (approximately 1000 V/h). Standard NADP and 200 nmol (-)-carveol (62% trans, 38% cis, SCM electrode buffers (25 mm Tris, 192 mM glycine) according to Specialty Chemicals) for 1 h at 30°C. Reaction products were Laemmli (25) were used, but the cathode buffer also contained extracted with ether and analyzed by GLC as described above 1 mm sodium thioglycolate, 3 mM 3-mercaptopropionic acid, for hydroxylase assays. Only the (-)-trans-carveol isomer was and 1 mM EDTA to aid in persulfate removal (4). enzymically oxidized to (-)-carvone. Carveol dehydrogenase Gel slices were stained for carveol dehydrogenase activity activity was principally confined (about 90%) to the 195,000g by incubation in 100 mm phosphate buffer (pH 8.0), contain- supernatants. ing 0.4 mm nitroblue tetrazolium, 0.065 mM phenazine meth- osulfate, 0.75 mM NAD, and saturating levels of (-)-carveol. Partial Purification of (-)-trans-Carveol Dehydrogenase Incubation was carried out for approximately 2 h in the dark in a sealed glass bottle that had been purged with N2. Stained Carveol dehydrogenase preparations from glandular tri- slices were scanned at 633 nm with an LKB 2202 Ultroscan chome extracts and whole-leaf (minus trichomes) extracts laser densitometer. Control slices assayed without (-)-carveol, were purified separately for electrophoretic comparison. After or from runs without enzyme, showed no detectable bands. extraction, filtration and centrifugation as detailed above, the Protein concentrations were estimated using the BioRad supernatants were dialyzed against 100 mm sodium/potas- protein assay based upon the dye-binding technique of Brad- sium phosphate buffer (pH 7.0), containing 10% (v/v) glyc- ford (3) with BSA as the standard. erol, 2 mm ascorbic acid, 1 mM DTT, and 1 mm EDTA, and then concentrated by ultrafiltration (Amicon YM-30) to vol- Histochemical Localization of (-)-trans-Carveol umes of approximately 10 mL. Powdered (NH4)2SO4 was Dehydrogenase added until 25% saturation was achieved and the precipitated protein was removed by centrifugation at 27,000g for 10 min. Immature leaves (5-10 mm) collected before the plants The supernatants were then brought to 60% saturation with flowered were cut in halfand incubated immediately with the (NH4)2S04 and centrifuged again. The resulting pellets con- reaction mixture described below in Teflon-sealed screw- tained over 97% of the carveol dehydrogenase activity origi- capped tubes covered with aluminum foil. Incubation was nally present in the extracts. These preparations were sus- carried out for 45 min at 30°C with gentle agitation. The pended in a minimum volume of 100 mm phosphate buffer composition ofthe reaction mixture, based on previous stud- (pH 8.0), containing 10% (v/v) glycerol, 1 mm ascorbic acid, ies of dehydrogenase histochemistry (26, 32) and extensive 0.5 mM DTT, and 0.5 mM EDTA, and applied to a 2.5 x 110 preliminary trials, was 20 mm sodium cacodylate (pH 7.5), 3 cm column of Sephacryl S-200 (Pharmacia) equilibrated with mM MgCl2, 1% (w/v) sucrose, 2% (w/v) polyvinylpyrrolidone the same buffer. The column was pumped at a flow rate of (Mr 360,000), 0.01% (v/v) Tween 20, 0.6 mm nitroblue tet- 25 mL/h, and 5.3 mL fractions were collected and assayed razolium, 1 mm phenazine methosulfate, 0.5 mM NADP, and for carveol dehydrogenase activity, which eluted at approxi- 0.5 mM (-)-carveol. Reaction conditions were optimized in a mately 1.55 void volumes. Fractions containing the dehydro- series of experiments by varying concentrations of nitroblue genase were combined, concentrated by ultrafiltration and, tetrazolium, phenazine methosulfate, NADP, (-)-carveol, after a buffer change by which 1% (w/v) sorbitol was substi- and incubation time. Tween 20 was added to enhance tissue tuted for the 10% (v/v) glycerol, the preparation was lyophi- penetration of the hydrophobic substrate, while sucrose and lized and stored under N2 at -25°C. As a result ofthese steps, polyvinylpyrrolidone reduced browning of the tissue. A brief the total purification of carveol dehydrogenase was 30- to 40- (10 s) dip of the leaves in pentane prior to incubation signifi- fold and recovery was 60 to 80%. cantly enhanced the apparent reaction rate, presumably by removing cuticle wax and thereby assisting substrate uptake. Separation of Dehydrogenase Activities by PAGE Direct comparisons were made between leaves incubated with the above reaction mixture and controls incubated without A comparison between carveol dehydrogenase activities carveol. To distinguish carveol dehydrogenase from alcohol located in the glandular trichomes and those located in the (ethanol) dehydrogenase activity, experiments were also un- rest of the leaf was made by nondenaturing discontinuous dertaken with 0.5 mm ethanol as a substrate in place of polyacrylamide gel electrophoresis of the corresponding par- carveol, and by including in the reaction mixture 15 mm tially purified preparations. Vertical slab gels (16 cm x 18 cm pyrazole, an inhibitor of alcohol dehydrogenase. x 1 mm) employing the discontinuous system of Laemmli After incubation, the leaves were rinsed with 20 mm sodium (25) were cast with a separating gel of 7.5% acrylamide. Prior cacodylate (pH 7.0) for 20 min and then fixed with 3% (v/v) to loading the sample, the slab gel was subjected to electro- glutaraldehyde in 20 mm sodium cacodylate buffer for 1 h. phoresis at 40 mA constant current to remove the residual Fixation prior to incubation greatly reduced the rate of the persulfate used in polymerization (21). Lyophilized prepara- dehydrogenase reaction. After washing the fixed tissue with tions of carveol dehydrogenase, partially purified as described sodium cacodylate for 30 min, the leaves were postfixed with above, were dissolved in a buffer, formulated to preserve 1% (W/V) OS04 in 10 mm sodium cacodylate (pH 7.0) for 90 catalytic activity, that contained 50 mM diazabicyclooctane min at room temperature, then washed with distilled water (pH 9.0), 25% (w/v) sucrose, 6 mm 3-mercaptopropionic acid, for 45 min, dehydrated in a graded ethanol series ending with 5 mm ascorbic acid, 1 mm DTT, and 20 mM NaCl. Aliquots 100% propylene oxide, and embedded in Spurr's low viscosity containing from 50 to 200 ,ug protein were loaded into 1 cm- resin (30). Silver-gold thin sections were obtained with a wide wells. Two gels at a time were run in a Hofer SE 600 Reichert OM U2 ultramicrotome. Sections were stained with vertical slab gel apparatus with a 4°C cooling bath at a constant lead citrate, uranyl acetate, and lead citrate again, in sequence 1 354 GERSHENZON ET AL. Plant Physiol. Vol. 89, 1989

-,4 a * 5 4 cu" .

SC

t.".,. i I:,

"f;

-t i.,

1..: ./. :., g I'AL_ek LOCALIZATION OF MONOTERPENE BIOSYNTHESIS IN SPEARMINT TRICHOMES 1355

(14) and viewed with a Hitachi H-300 transmission electron Histochemical Localization of Carveol Dehydrogenase microscope. To confirm that carveol dehydrogenase activity is present throughout the cells ofthe leaf, an adaptation of a commonly RESULTS used method for histochemical localization ofdehydrogenases Localization of Carvone Biosynthetic Capability by Cell- was employed. This method is based on the reduction of Free Assays tetrazolium salts to insoluble formazans by NADPH, with phenazine methosulfate as an intermediate electron carrier To determine whether (-)-carvone biosynthesis takes place (26, 32). Leaf sections were incubated with nitroblue tetrazo- specifically in the glandular trichomes ofspearmint, the activ- lium, phenazine methosulfate, NADP, and the substrate (-)- ities ofthe enzymes associated with the relevant pathway were trans-carveol, and then fixed and prepared for transmission measured in cell-free extracts ofthe glandular trichomes. Two electron microscopy. The presence of carveol dehydrogenase types of procedures were employed to extract selectively the was indicated by small dark granules of formazan precipitate. glandular contents. In the first method, leaves were submerged Formazan granules were found in cells of all tissues of the in extraction buffer and then gently brushed with a toothbrush leaf, including the epidermis, the palisade layer, the spongy to obtain a glandular trichome extract containing the material parenchyma, the vascular bundles, and both types of glan- from only 40 to 60% of the glandular trichomes present on dular trichomes, peltate and capitate (Fig. 4). Granules in the the leaf surface, but with essentially no contamination from glandular trichomes were associated with the smooth endo- underlying tissue (12). All three enzymes involved in the plasmic reticulum (Fig. 6), while granules in the cells of the synthesis of carvone from the ubiquitous isoprenoid inter- rest ofthe leafwere often associated with smooth endoplasmic mediate geranyl pyrophosphate (i.e. the cyclase, hydroxylase, reticulum, and also with other membrane systems (Fig. 7). and dehydrogenase) were readily detected in this extract in- Controls incubated without carveol showed only sporadic dicating that the glandular trichomes were capable of carvone reaction zones in some mesophyll cells, with no consistent biosynthesis. subcellular location (Fig. 5). To examine the possibility that To determine the biosynthetic capability of the trichome the putative carveol dehydrogenase activity was due to alcohol extract relative to that ofthe rest of the leaf, a second experi- (ethanol) dehydrogenase, leaf sections were incubated with ment was performed using a more efficient procedure for ethanol as a substrate under otherwise identical conditions. gland removal. Leaf surfaces were abraded with glass beads The reaction attributed to alcohol dehydrogenase activity was (18) to remove over 99% of all the glandular trichomes (Figs. observed only in mesophyll cells and was localized near the 2 and 3) without extensive damage to the underlying leaf plasmalemma (Fig. 8), a pattern clearly distinguishable from tissue. Using this method, an extract of the contents of vir- that produced by carveol dehydrogenase. Pyrazole, an inhib- tually all ofthe glandular trichomes on the leaf was obtained. itor of alcohol dehydrogenase, also inhibited the reaction of The remaining leaf tissue was extracted by homogenization carveol dehydrogenase and was thus not useful as a diagnostic to give a preparation representing the leaf free of gland con- tool. Therefore, the summation of both histochemical and in tents. Comparison of these two extracts showed that two of vitro results were consistent, and indicated that carveol de- the three enzymes ofthe carvone pathway, geranyl pyrophos- hydrogenase activity is present in glandular trichomes and in phate: (-)-limonene cyclase and (-)-limonene hydroxylase, the rest of the leaf. were almost completely restricted to the glandular trichomes (Table I). The traces of these two activities in the remainder Electrophoretic Comparison of Carveol Dehydrogenase of the leaf probably result from incomplete removal of glan- Activities from Glandular Trichomes and the Remaining dular trichomes from this preparation. The third enzyme, (-)- Leaf trans-carveol dehydroganase, was found in substantial amounts in both the glandular trichomes and the rest of the The oxidation ofcarveol is the final step in carvone biosyn- leaf (Table I). thesis and it therefore seems unlikely that the alcohol would

Figures 2 and 3. Scanning electron micrographs (x90) of spearmint leaf surfaces before (Fig. 2) and after (Fig. 3) selective removal of the glandular trichomes with glass beads (PT, peltate glandular trichomes). More than 99% of these structures were removed by this method. Arrows in Figure 3 show former locations of glandular trichomes. Samples for scanning electron microscopy were air-dried ovemight at room temperature and gold-coated. Specimens were viewed with an ETEC Autoscan U-1 at 30 kV. Figures 4-8. Histochemical localization of carveol dehydrogenase in sections of young spearmint leaves. Leaves were incubated with the reaction mixtures described in "Materials and Methods," then fixed and prepared for transmission electron microscopy. The presence of carveol dehydrogenase is indicated by a precipitate of formazan granules. Figure 4. Peltate glandular trichome and adjacent mesophyll following incubation with carveol showing a precipitate in all tissues (CU, cuticle; SS, subcuticular space; SC, secretory cell; ST, stalk cell; BC, basal cell; E, epidermis; M, mesophyll) (x1370). Figure 5. Control section incubated without carveol, showing only sporadic precipitate in the mesophyll (x1370). Figure 6. Section of a secretory cell illustrating the subcellular location (arrows) of carveol dehydrogenase in smooth endoplasmic reticulum (SER) (x68,400). Figure 7. Mesophyll cell showing precipitate associated with smooth endoplasmic reticulum and other membrane systems (CW, cell wall) (x1 3,700). Figure 8. Subcellular localization of alcohol (ethanol) dehydrogenase (arrows) near the plasmalemma of a mesophyll cell indicating that carveol dehydrogenase and alcohol dehydrogenase are not equivalent (x13,700). 1356 GERSHENZON ET AL. Plant Physiol. Vol. 89, 1989

Table I. Localization of Enzymes of Carvone Biosynthesis in turing PAGE, stained to reveal carveol dehydrogenase activ- Spearmint Extracts ity, displayed significant differences between the two prepa- Glandular trichomes were first removed from the leaves by abra- rations. Only a single staining band was present in the tri- sion with glass beads, and the remaining leaves were then homoge- chome-free leaf extracts, whereas two bands of activity were nized. The preparation and assay of each enzyme is described in present in extracts of the glandular trichomes (Fig. 9). The "Materials and Methods." slower moving and less intense of these two bands had the Activity same mobility as the single band in the trichome-free leaf Enzyme Glandular Remaining trchomesGlandular extract. A plausible interpretation is that the dehydrogenase unique to the glandular trichome extract is the enzyme re- trichomes leaf sponsible for carveol oxidation in the intact plant (such highly nkat/kg tissue % selective monoterpenol dehydrogenases have been previously Geranyl pyrophosphate: 5220 25.0 99.5 reported in other plants [8, 16]), whereas the enzyme present (--limonene cyclase in both extracts represents a dehydrogenase activity found in (-)Limonene hydroxylase 150 5.6 96.4 all tissues while (-)-trans-Carveol dehydro- 8080 18,100 30.9 which, capable of oxidizing carveol under in genase vitro conditions, is probably not associated with monoterpene biosynthesis in vivo. Thus, the enzymes responsible for the entire pathway of carvone biosynthesis in spearmint appear A to be restricted to the glandular trichomes. cv, co DISCUSSION For many years, glandular trichomes have been regarded C.) as the primary site of monoterpene synthesis in mints and related essential oil plants (1, 9, 17). The present results demonstrate that these structures are in fact the sole sites of %us monoterpene biosynthesis in the leaves of spearmint. In con- junction with the work of Keene and Wagner on the diter- a) penes oftobacco (22), it can now be suggested that all terpenes CO accumulated in, or exuded from, glandular trichomes are coU1) probably synthesized in these highly specialized structures. 0) 0 B Since monoterpene alcohols are known to be transported as aL- ID glycosides (7, 31) (i.e. in mature peppermint (+)-neomethyl- f3-D-glucoside is transported from leaves to rhizomes and -c:a) catabolized at this site [10]), it is conceivable that the mono- m terpenes of spearmint are first synthesized elsewhere in the 0 I I leaf and then transported to the glandular trichomes. How- U1) ever, this possibility appears to be eliminated by the present L.. results. Additionally, it is extremely unlikely that monoter- 0o penes are biosynthesized elsewhere in the plant since stem and root extracts are devoid of limonene cyclase activity (J Gershenzon, R Croteau, unpublished data). 40 50 60 70 Because monoterpene synthesis is restricted to the glandular Mobility (mm) trichomes, it is evident that the timing ofgland ontogeny will influence the concentration ofmonoterpenes in a leaf Figure 9. PAGE of partially purified carveol dehydrogenase prepa- directly rations from glandular trichomes (A) and from the remainder of the and how this varies throughout leaf development. Typically, leaf following gland removal (B). Gels were run, stained, and scanned most glandular trichomes are initiated very early in leaf as described under "Materials and Methods." The glandular trichome development and begin to accumulate monoterpenes before extract contains two species of carveol dehydrogenase activity, the leaves are 5 mm long (1). Therefore, monoterpene con- whereas the extract of the remainder of the leaf contains only the centrations (per gram tissue) are relatively high in young slower migrating species. leaves and decline with further development (13). Ultrastructural studies have concluded that terpene synthe- be transported out of the glandular trichomes for conversion sis within the glandular trichome occurs principally in the to the ketone and then transported back into the trichomes secretory cells rather than the stalk cell or the basal cell (see for storage. Rather, it seems more probable that carveol Fig. 4) (17, 29); the present results support this generalization. oxidation would occur within the trichomes, perhaps cata- Carveol dehydrogenase activity, as indicated by the presence lyzed by a unique species of dehydrogenase. To evaluate this of formazan granules, was always much more prominent in possibility, the carveol dehydrogenase activities from both the secretory cells than in other cell types for both peltate and glandular trichome extract and the remaining leaf (free of capitate trichomes. At the subcellular level, carveol dehydro- trichomes) extract were separately purified by combination of genase activity in spearmint secretory cells was associated (NH4)2S04 precipitation and gel filtration. The activities from with smooth endoplasmic reticulum. This result is somewhat both extracts coeluted on gel filtration. However nondena- surprising in light ofthe observation that the carveol dehydro- LOCALIZATION OF MONOTERPENE BIOSYNTHESIS IN SPEARMINT TRICHOMES 1357 genase was operationally soluble (>90% ofthe activity resided 10. Croteau R, Martinkus C (1979) Metabolism of monoterpenes: in the 195,000g supernatant) and suggests that the Demonstration of (+)-neomenthyl-fl-D-glucoside as a major enzyme is metabolite of (-)-menthone in peppermint (Mentha piperita). released from membranes during tissue extraction (2). How- Plant Physiol 64: 169-175 ever, the subcellular localization of carveol dehydrogenase 11. Croteau R, Venkatachalam KV (1986) Metabolism of monoter- must be viewed with some caution since, in the histochemical penes: Demonstration that (+)-cis-isopulegone, not piperiten- procedures used in this study, NADPH and phenazine meth- one, is the key intermediate in the conversion of (-)-isopiper- itenone to (+)-pulegone in peppermint (Menthapiperita). Arch osulfate are freely diffusible and so may produce artefacts Biochem Biophys 249: 306-315 (26). Smooth endoplasmic reticulum has been frequently 12. Croteau R, Winters JN (1982) Demonstration ofthe intercellular observed in cells assumed to produce terpene secretions (17, compartmentation of l-menthone metabolism in peppermint 19, 29). Yet, the lack of suitable methods for identifying (Mentha piperita). Plant Physiol 69: 975-977 under 13. Croteau R, Felton M, Karp F, Kjonaas R (1981) Relationship of terpenoids the electron microscope has prevented any camphor biosynthesis to leaf development in sage (Salvia ultrastructural demonstration of the direct involvement of officinalis). Plant Physiol 67: 820-824 smooth endoplasmic reticulum in terpene biogenesis. Re- 14. Daddow LYM (1983) A double lead stain method for enhancing cently, however, the synthesis of sesquiterpene olefins in contrast of ultrathin sections in electron microscopy: a modi- mitis fruits was shown to be associated with a fied multiple staining technique. J Microsc 129: 147-153 Citrofortunella 15. Davisson VJ, Woodside AB, Poulter CD (1985) Synthesis of membrane fraction containing endoplasmic reticulum (2). allylic and homoallylic isoprenoid pyrophosphates. Methods Another type of organelle often linked with monoterpene Enzymol 110: 130-144 synthesis is the leucoplast, a plastid ofcomplex shape without 16. Dehal SS, Croteau R (1987) Metabolism of monoterpenes: spec- thylakoids (6). Plastids of this description were present in the ificity of the dehydrogenases responsible for the biosynthesis of camphor, 3-thujone and 3-isothujone. Arch Biochem Bio- secretory cells of spearmint glandular trichomes. However, phys 258: 287-291 no histochemical evidence was found for the association of 17. Fahn A (1979) Secretory Tissues in Plants. Academic Press, New carveol dehydrogenase activity with these organelles. Leuco- York plasts in spearmint glands might serve as compartments for 18. Gershenzon J, Duffy MA, Karp F, Croteau R (1987) Mechanized techniques for the selective extraction of enzymes from plant some of the earlier steps of monoterpene synthesis. Detailed epidermal glands. Anal Biochem 163: 159-164 evaluation of the localization of monoterpene biogenesis at 19. Gleizes M, Carde J-P, Pauly G, Bernard-Dagan C (1980) In vivo the subcellular level must await the development of immu- formation of sesquiterpene hydrocarbons in the endoplasmic nochemically based techniques. reticulum of pine. Plant Sci Lett 20: 79-90 20. Guenther E (1974) The Essential Oils, Vol III (reprinted). Krieger, Huntington, NY, p 681 ACKNOWLEDGMENTS 21. Heeb MJ, Gabriel 0 (1984) Enzyme localization in gels. Meth- ods Enzymol 104: 416-439 We thank Greg Wichelns for raising the plants, Nancy Madsen for 22. Keene CK, Wagner GJ (1985) Direct demonstration of duvatri- typing the manuscript, and the staff of the Electron Microscopy enediol biosynthesis in glandular heads of tobacco trichomes. Center, Washington State University, for assistance. Plant Physiol 79: 1026-1032 23. Kjonaas R, Croteau R (1983) Demonstration that limonene is the first cyclic intermediate in the biosynthesis of oxygenated LITERATURE CITED p-menthane monoterpenes in Mentha piperita and other Men- tha species. Arch Biochem Biophys 220: 79-89 1. Amelunxen F (1965) Elektronenmikroskopische Untersuchungen 24. KJonaas R, Venkatachalam KV, Croteau R (1985) Metabolism an den Druisenschuppen von Mentha piperita L. Planta Med of monoterpenes: Oxidation of isopiperitenol to isopiperiten- 13: 457-473 one, and subsequent isomerization to piperitenone by soluble 2. Belingheri L, Pauly G, Gleizes M, Marpeau A (1988) Isolation enzyme preparations from peppermint (Mentha piperita) by an aqueous two-polymer phase system and identification of leaves. Arch Biochem Biophys 238: 49-60 endomembranes from Citrofortunella mitis fruits for sesquiter- 25. Laemmli UK (1970) Cleavage of structural proteins during the pene hydrocarbon synthesis. J Plant Physiol 132: 80-85 assembly of the head of bacteriophage T4. Nature 227: 680- 3. Bradford MM (1976) A rapid and sensitive method for the 685 quantitation of microgram quantities of protein utilizing the 26. Lewis PR (1977) Other cytochemical methods for enzymes. In principle of protein-dye binding. Anal Biochem 72: 248-254 AM Glauert, ed, Practical Methods in Electron Microscopy, 4. Brewer JM (1967) Artifact produced in disc electrophoresis by Vol 5. North-Holland, Amsterdam, pp 225-287 ammonium persulfate. Science 156: 256-257 27. Loomis WD, Lile JD, Sandstrom RP, Burbott AJ (1979) Adsor- 5. Charlwood BV, Charlwood KA (1983) The biosynthesis ofmono- bent polystyrene as an aid in plant enzyme isolation. Phyto- and sesquiterpenes in tissue culture. Biochem Soc Trans 11: chemistry 18: 1049-1054 592-593 28. Nabeta K, Ohnishi Y, Hirose T, Sugisawa H (1983) Monoterpene 6. Cheniclet C, Carde J-P (1985) Presence ofleucoplasts in secretory biosynthesis by callus tissues and suspension cells from Perilla cells and of monoterpenes in the essential oil: a correlative species. Phytochemistry 22: 423-425 study. Isr J Bot 34: 219-238 29. Schnepf E (1974) Gland cells. In AW Robards, ed, Dynamic 7. Croteau R (1987) Biosynthesis and catabolism of monoterpen- Aspects of Plant Ultrastructure. McGraw-Hill, London, pp oids. Chem Rev 87: 929-954 331-357 8. Croteau R, Felton M (1980) Substrate specificity ofmonoterpenol 30. Spurr AR (1969) A low-viscosity epoxy resin embedding medium dehydrogenases from Foeniculum vulgare and Tanacetum vul- for electron microscopy. J Ultrastruct Res 26: 31-43 gare. Phytochemistry 19: 1343-1347 31. Stahl-Biskup E (1987) Monoterpene glycosides, state-of-the-art. 9. Croteau R, Johnson MA (1984) Biosynthesis of terpenoids in Flavour Fragr J 2: 75-82 glandular trichomes. In E Rodriguez, PL Healy, I Mehta, eds, 32. Van Noorden CJF (1984) Histochemistry and Cytochemistry Biology and Chemistry of Plant Trichomes. Plenum, New of Glucose-6-Phosphate Dehydrogenase. Gustav Fischer, York, pp 133-185 Stuttgart