Monoterpene Synthases from Common Sage (Salvia officinalis)*

Common sage (Salvia officinalis) produces an extremely broad range of cyclic monoterpenes bearing diverse carbon skeletons, including members of thep-menthane (1,8-cineole), pinane (α- and β-pinene), thujane (isothujone), camphane (camphene), and bornane (camphor) families. An homology-based polymerase chain reaction cloning strategy was developed and used to isolate the cDNAs encoding three multiproduct monoterpene synthases from this species that were functionally expressed in Escherichia coli. The heterologously expressed synthases produce (+)-bornyl diphosphate, 1,8-cineole, and (+)-sabinene, respectively, as their major products from geranyl diphosphate. The bornyl diphosphate synthase also produces significant amounts of (+)-α-pinene, (+)-camphene, and (±)-limonene. The 1,8-cineole synthase produces significant amounts of (+)- and (−)-α-pinene, (+)- and (−)-β-pinene, myrcene and (+)-sabinene, and the (+)-sabinene synthase produces significant quantities of γ-terpinene and terpinolene. All three enzymes appear to be translated as preproteins bearing an amino-terminal plastid targeting sequence, consistent with the plastidial origin of monoterpenes in plants. Deduced sequence analysis and size exclusion chromatography indicate that the recombinant bornyl diphosphate synthase is a homodimer, whereas the other two recombinant enzymes are monomeric, consistent with the size and subunit architecture of their native enzyme counterparts. The distribution and stereochemistry of the products generated by the recombinant (+)-bornyl diphosphate synthase suggest that this enzyme might represent both (+)-bornyl diphosphate synthase and (+)-pinene synthase which were previously assumed to be distinct enzymes.

The cyclization of the universal precursor geranyl diphosphate to form monocyclic and bicyclic monoterpenes is cata-lyzed by a group of enzymes termed monoterpene synthases (or cyclases). The biochemical transformation of geranyl diphosphate to cyclic products has been investigated using enzymes from a variety of plants, including both angiosperms (1) and gymnosperms (2)(3)(4), and a mechanistic paradigm for these transformations (Scheme 1) is well established (1,5). Thus, geranyl diphosphate is initially ionized and isomerized to form either (3R)-or (3S)-linalyl diphosphate, depending on the particular enzyme. This step permits rotation about the C2-C3 single bond of the bound allylic isomer to the cisoid conformer which, upon subsequent ionization, promotes electrophilic attack by C1 on the C6 -C7 double bond, resulting in the formation of the ␣-terpinyl cation as a central intermediate. Further transformations of this reactive intermediate may be effected by additional intramolecular electrophilic additions, hydride shifts, or other rearrangements before termination of the sequence by deprotonation of the final cation or capture by an external nucleophile, such as a hydroxyl ion or the diphosphate group. Although the fate of the substrate has been well characterized in numerous monoterpene cyclization reactions, the molecular mechanisms by which the enzymes effect these transformations is still poorly understood.
In this report, we describe the homology-based cloning, and subsequent sequencing and heterologous expression, of three monoterpene synthase cDNA genes from sage, the recombinant enzymes from which produce three different major types of cyclic monoterpene products, (ϩ)-sabinene (a bicyclic olefin), 1,8-cineole (a bicyclic ether), and (ϩ)-bornyl diphosphate (a bicyclic diphosphate ester), respectively (Scheme 1). Comparison of the sizes, subunit architectures, and product distributions of these multiple-product enzymes clarifies the assignment of specific catalytic functions to defined monoterpene synthases, and the deduced sequences provide information on the relatedness of these enzymes within the species and to other terpenoid synthases of plant origin. Additionally, comparison of the primary structures of these mechanistically different monoterpene synthases from the same plant, the first examples of this type thus far available, allows preliminary assessment of active site interactions and provides the foundation for more detailed study of structure-function relationships.

EXPERIMENTAL PROCEDURES
Plant Materials, Substrate, and Reagents-Sage plants (S. officinalis L.) were grown from seed as described previously (19). [1-3 H]Geranyl diphosphate (250 Ci/mol) was prepared by an established method (8). Terpenoid standards were from our own collection. Unless otherwise stated, all reagents were obtained from Sigma or Aldrich Chemical Co. DNA sequences were assembled and analyzed using GCG software (20).
cDNA Library Construction, PCR 1 -based Probe Generation, and Library Screening-Approximately 15 g of emerging sage leaves (shoot 1 The abbreviations used are: PCR, polymerase chain reaction; bp, base pair(s); GC, gas chromatography; Mopso, 3-(N-morpholino)-2-hydroxypropane sulfonic acid; MS, mass spectrum/spectrometry; SBS, sage bornyl diphosphate synthase; SCS, sage 1,8-cineole synthase; SSS, sage sabinene synthase; SUS, sage unknown synthase. SCHEME 1. Conversion of geranyl diphosphate to the monoterpenes of sage. Formation of the monocyclic and bicyclic products requires preliminary isomerization of geranyl diphosphate to linalyl diphosphate. Acyclic products can be formed from either geranyl diphosphate or linalyl diphosphate. OPP denotes the diphosphate moiety. tips) from 3-week-old plants were ground to a fine powder in liquid nitrogen and extracted into buffer composed of 200 mM Tris-HCl (pH 8.5), 300 mM LiCl, 5 mM thiourea, 1 mM aurintricarboxylic acid, 10 mM dithiothreitol, and 10 mM EDTA, and containing 1% (w/v) polyvinylpyrrolidone (M r ϳ40,000). Total RNA thus extracted was prepared by precipitation with isopropyl alcohol, followed by CsCl density gradient centrifugation, as described previously (21). Poly(A) ϩ mRNA was isolated by chromatography on oligo(dT)-cellulose (Qiagen) and 6.3 g of the resulting mRNA was used to construct a ZAPII cDNA library according to the manufacturer's instructions (Stratagene).
A general strategy for the homology-based PCR cloning of terpenoid synthases of higher plant origin has been suggested (22) based upon a comparison of the cDNA sequences of a monoterpene synthase (23), a sesquiterpene synthase (24), and a diterpene synthase (25) of angiosperm origin. Thus, three regions of deduced amino acid sequence (corresponding to residues 180 -188, 197-203, and 380 -387 of limonene synthase from spearmint (23)) were employed to design primers corresponding to 1F, ). Analysis of the PCR reaction products by agarose gel electrophoresis (26) indicated that only the combination of primers 2F and 3R amplified a discrete product of approximately 600 bp, which was ligated into pT7Blue (Novagen), and transformed into Escherichia coli NovaBlue cells. Plasmid DNA was prepared from 32 individual transformants; seven of these had inserts of the predicted size (ϳ600 bp). These inserts were partially sequenced (DyeDeoxy Terminator Cycle Sequencing; Applied Biosystems) to reveal two distinct "terpenoid synthase-like" sequences.
The relative ability of these two potential probes to hybridize with expressed genes was evaluated by DNA-RNA hybridization. Two samples of sage leaf mRNA isolated as above (3 g each) were electrophoresed on 1% (w/v) agarose under denaturing conditions and blotted onto separate polyvinylidene difluoride membranes using standard techniques (26). Each membrane was evaluated with 32 P-labeled probe, generated from one or the other of the 600-bp fragments using random hexamer priming (28), by standard hybridization and washing protocols (26). Autoradiography of the membrane revealed that both probes hybridized to a 2-kilobase transcript, although one probe generated a significantly stronger signal (ϳ10-fold) than the other. This probe was subsequently employed to screen the cDNA library in an attempt to isolate full-length cDNA sequences encoding the corresponding presumptive terpene synthase. UV cross-linked nitrocellulose lifts containing 3-5 ϫ 10 4 primary plaques (plated on E. coli XL1-Blue-MRFЈ), after pre-hybridization (in 1.25 ϫ SSPE, 0.5 ϫ Denhart's reagent, 9% formamide, 0.002% SDS, and 10 g/ml denatured E. coli DNA, for 2 h at 42°C), were hybridized in the same medium with approximately 8 Ci of the 32 P-labeled probe for 48 h. Filters were washed, first at room temperature (in 2 ϫ SSC with 0.1% SDS) then at 55°C (in 1 ϫ SSC with 0.1% SDS), and subsequently exposed to x-ray film at Ϫ70°C (26). Plaques yielding positive signals were purified through two additional rounds of hybridization. A total of 77 purified ZAP clones so isolated were excised in vivo to generate BluescriptII SK(Ϫ) phagemids and transformed into E. coli SOLR cells according to the Stratagene protocol. The size of each cDNA insert was determined by PCR using T3 and T7 promoter primers, and transformed clones containing an insert Ͼ1.6 kilobases were either expressed to assay for monoterpene synthase activity or sequenced at the 5Ј terminus using the T3 promoter primer. Bluescript plasmids expressing monoterpene synthase activity in cellfree extracts of transformed E. coli (see below) were fully sequenced on both DNA strands by primer walking or by the method of nested deletions using exonuclease III and mung bean nuclease (26).
To improve the efficiency of functional expression and facilitate subsequent enzyme purification, each of the apparent full-length pBluescript clones that expressed monoterpene synthase activity was subcloned, in-frame, into pGEX vectors (Pharmacia) using a convenient BamHI (SBS and SSS) or EcoRI (SCS) restriction site at the 5Ј-end, and the XhoI restriction site at the 3Ј terminus. Fidelity in subcloning was confirmed by complete sequencing, and these plasmid constructs were expressed in E. coli XL1-Blue-MFRЈ cells.
cDNA Expression in E. coli-The Bluescript plasmids expressed in E. coli strain XL1-Blue were grown in 5 ml of LB medium (26), supplemented with 100 g of ampicillin/ml, to an A 600 ϭ 0.5 at 37°C with constant shaking, then induced with 1 to 3 mM isopropyl-1-thio-␤-Dgalactopyranoside. The cells were allowed an additional 4 h growth at 37°C before harvesting by centrifugation (2000 ϫ g, 10 min) and lysis by sonication (Braun-Sonic 2000 with microprobe at maximum power for 15 s), on ice, in 50 mM Mopso buffer containing 10% glycerol, 10 mM MgCl 2 , and 5 mM dithiothreitol (either pH 6.5 or 7.1, as appropriate for the assays described below). The sonicates were cleared by centrifugation (18,000 ϫ g, 10 min) and the resulting supernatant was used as the enzyme source. The pGEX constructs in E. coli XL1-Blue-MFRЈ cells were similarly grown at 37°C to A 600 ϭ 1.0 to 1.5, then induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside and incubated overnight at 20°C with constant shaking. The cells were then harvested and lysed, and the soluble supernatant prepared as before. Purification of the resulting fusion proteins was attempted using the glutathione-Sepharose affinity column according to the manufacturer's instructions (Pharmacia). Of the three expressed monoterpene synthases (see below), only one (SBS) bound to the matrix but, even in this case, affinity-based purification proved to be unreliable. Therefore, partial purification of the heterologously expressed synthases was achieved by ion-exchange chromatography on O-diethylaminoethyl-cellulose (Whatman DE-52) using a 0 -400 mM NaCl gradient. The partially purified preparations were desalted by repeated ultrafiltration and dilution using an Amicon Centriprep 30 concentrator (30 kDa cutoff) and the appropriate assay buffer. The pGEX-expressed fusion proteins were also subjected to gel permeation chromatography (Pharmacia FPLC system) using a Pharmacia XY 16 ϫ 70 column packed with Superdex S-200 and equilibrated with the appropriate 50 mM Mopso buffer system. The column was developed at a flow rate of 0.3 ml/min and calibrated using the Sigma MW-GF-200 molecular weight marker kit. K av values of the recombinant enzymes were compared with the calibration standards to establish molecular weights (29), which were then corrected for the engineered fusion and transit peptide to estimate the molecular weight of the corresponding native form.
Enzyme Assay-Monoterpene synthase activities were assayed by methods previously described (6,8,12,30). Briefly, an aliquot of the bacterial cell lysate, appropriate column fractions, or partially purified and desalted enzyme preparation, in 0.5 or 1.0 ml of 50 mM Mopso buffer (pH 6.1 for SBS, and 7.1 for SCS and SSS) containing 10 mM MgCl 2 , 5 mM dithiothreitol, and 10% (v/v) glycerol, was transferred to a 7-ml glass Teflon sealed, screw-capped tube, and the mixture was overlaid with 1 ml of pentane to trap volatile products. The reaction was initiated by the addition of 4.5 M [1-3 H]geranyl diphosphate (1.3 Ci), with incubation at 31°C with gentle shaking for 0.5 to 3.0 h. The pentane layer and an additional pentane extract (2 ϫ 1 ml) were passed over a short column of silica gel surmounted by anhydrous MgSO 4 (in a Pasteur pipette) to afford the monoterpene olefin fraction. Subsequent extraction of the remaining aqueous phase with diethyl either (2 ϫ 1 ml), and passage of this extract through the same column, yielded the oxygenated monoterpene fraction. The residual aqueous phase was then treated with excess potato apyrase and wheat germ acid phosphatase to hydrolyze monoterpenol diphosphate esters (6,30). The liberated alcohols were then extracted into diethyl ether (2 ϫ 1 ml) and the combined extract dried over anhydrous MgSO 4 . Radioactivity in the various fractions was determined by liquid scintillation counting of aliquots (Packard 460 CD with external standard quench correction) and the remaining material was concentrated for radio-GC and GC-MS analysis.
Kinetic analyses were carried out with the partially purified, recombinant pGEX fusion proteins by determination of initial reaction rates at a minimum of 10 substrate concentrations ranging from 0.45 to 45 M [1-3 H]geranyl diphosphate, at saturating levels of the divalent metal ion cofactor. The results were analyzed by non-linear regression of the Michaelis-Menten equation using the curve-fitting capabilities of Sigma-Plot (Jandel Corp.).
Product Identification-Radio-GC was performed on a Gow-Mac 550P gas chromatograph with thermal conductivity detector directly coupled to a Nuclear-Chicago 8731 gas proportional counter (31). An AT-1000 packed column (Alltech) was used with helium as carrier at 30 ml/min and with temperature programming from 70 to 200°C (at 5°C/min) for analysis of monoterpene olefins, and from 100 to 200°C (at 5°C/min) for analysis of oxygenated monoterpenes. Authentic standards (10 -20 g/component) were included with each injection in order to correlate the retention times determined by mass and radioactivity detectors.
GC-MS was performed on a Hewlett-Packard 6890 GC-quadrupole mass selective detector system interfaced with a Hewlett-Packard Chemstation for data analysis. An Alltech AT-1000 fused silica capillary column (30 m ϫ 0.25 mm inner diameter) was employed. Inleting was done by cool, on-column injection at 40°C, with oven programming from 40°C (50°C/min) to 50°C (5 min hold) then to 160°C (10°C/min), under a constant flow of 0.7 ml of helium/min. Full spectra were recorded for major reaction products which were identified by comparison of retention times to authentic standards and by comparison of spectra to those of the NSB75K library using the G1033A NIST probability based matching algorithm. Chiral phase separations were performed on a Hewlett-Packard 5890 GC by split injection (80:1) on a 30-m cyclodex-B capillary column (J & W Scientific) using H 2 as carrier at 0.6 ml/min and temperature programming from 70 to 200°C at 10°C/min with flame ionization detection. Compound identification was based on retention time identity with the authentic standard.

RESULTS AND DISCUSSION
Similarity-based Cloning of Monoterpene Synthases from S. officinalis-Because S. officinalis (culinary or common sage) produces such a broad structural variety of monoterpenes, this species has been utilized extensively for studies on the enzymology, stereochemistry, and mechanism of monoterpene cyclization reactions (1,5). Structural analyses of the responsible monoterpene synthases, and more detailed study of the cyclization mechanisms, require the isolation of cDNA species encoding these target enzymes. Protein purification from sage, as the basis for cDNA isolation, has been of limited success (17) because of the number of synthases present and their similarity in physical properties (32), and thus far has not permitted a reverse genetic approach to cloning of any of the monoterpene synthases from this species.
As a possible alternative to protein-based cloning of terpene synthases, a homology-based PCR strategy was recently proposed (22) that was developed by comparison of the deduced amino acid sequences of cDNAs encoding a monoterpene synthase (23), a sesquiterpene synthase (24), and a diterpene synthase (25) of phylogenetically distant angiosperm species. Three conserved regions of sequence were identified that appeared to be useful for the design of degenerate PCR primers. Two of these primers ultimately amplified a 600-bp fragment using cDNA from a sage leaf library as template. Cloning and sequencing showed the amplified products to correspond to two distinct sequence groups, both of which showed similarity to sequences of cloned terpene synthases, but only one of which hybridized strongly to a 2-kilobase target upon Northern blot analysis of sage leaf mRNA. This more efficient probe was utilized to screen the sage leaf cDNA library, from which 77 positive phagemids were purified. Size selection of the purified and in vivo excised clones yielded a subset of 44 with insert size greater than 1.6 kilobases, and these were expressed in E. coli XL1-Blue cells and the resulting extracts were assayed for functional monoterpene synthase activity by monitoring the conversion of [1-3 H]geranyl diphosphate to monoterpene olefins, oxygenated monoterpenes, and monoterpenyl diphosphate esters.
Nine functionally active clones were identified by this means, four types of which showed unique product profiles. Two cDNA clones, of which the clone designated 3C6 was most active in expression, yielded an enzyme in the corresponding bacterial extracts that produced principally bornyl diphosphate from geranyl diphosphate. This recombinant enzyme, designated SBS (sage bornyl diphosphate synthase), was presumed to represent the native (ϩ)-bornyl diphosphate synthase of sage, one of the prominent enzymes of oil gland extracts of this species (6) that produces the first dedicated intermediate in (ϩ)-camphor biosynthesis (7,33,34). Four clones, of which clone 3B5 yielded the highest activity, expressed a synthase in bacterial extracts that converted geranyl diphosphate to 1,8cineole as the major product. This acquisition was designated SCS (sage 1,8-cineole synthase) and considered to represent the native 1,8-cineole synthase, an enzyme for which the mechanism of cyclization has been studied in detail (8,35). Two additional clones, of which clone 3F25 was seemingly the most active in expression, yielded E. coli extracts capable of transforming geranyl diphosphate to sabinene as the dominant olefin product. This acquisition was named SSS (sage sabinene synthase), with correspondence assigned to the native (ϩ)sabinene synthase which catalyzes the cyclization to the bicyclic olefin precursor of (Ϫ)-isothujone (9,10,36). The last functional clone (3M13) yielded an expressed activity that catalyzed the conversion of geranyl diphosphate to the monocyclic olefin limonene as essentially the only product.
Clones 3C6, 3B5, 3F25, and 3M13 were fully sequenced (see Fig. 1 for deduced amino acid alignments), whereupon it was shown that the insert of 3M13 was a 5Ј-truncated version of 3F25 (data not shown), resulting in the translation of a shortened fusion protein starting at residue 106 of the sabinene synthase (SSS) that apparently leads to derailment of the normal bicyclization to instead yield the monocyclic product limonene. The remaining 35 positive clones, that were inactive in expressing a functional monoterpene synthase activity, were partially sequenced from the 5Ј terminus to search for additional truncated versions or inserts that were out of frame for proper expression. This approach revealed three truncated and/or out of frame versions of SBS, 20 additional versions of SCS, and 11 versions of SSS, while unveiling one new clone (number 3F5 and designated SUS). This clone showed extensive sequence homology to 3C6, 3B5, and 3F25 but encoded a premature stop codon (Fig. 1) that would result, because of this cloning artifact, in translational truncation toward the carboxyl terminus. A functional form of the SUS protein has not yet been obtained.
Sequence Analysis-Alignment of the deduced amino acid sequences (Fig. 1)  there are several regions of conservation between these eight monoterpene synthases of diverse origin. Comparison of these sequences using the GCG GAP program (20) (Table I) revealed the monoterpene synthases from sage to resemble each other and the limonene synthases from related members of the Lamiaceae (50 -70% identity, 70 -85% similarity) more closely than the monoterpene synthases of the gymnosperm grand fir (ϳ32% identity), or the linalool synthase from Clarkia breweri (44)(ϳ25% identity).
Monoterpene biosynthesis is compartmentalized in plastids (45)(46)(47). Thus, the monoterpene synthases are encoded as preproteins bearing an amino-terminal transit peptide for import of these nuclear gene products into plastids (leucoplasts of the oil gland cells in the present instance), where they are proteolytically processed to the mature forms. In vitro translation and plastidial processing of the gene product has been demonstrated directly with the limonene synthase cDNA from spearmint. 2 In all of the monoterpene synthases (Fig. 1), the 50 -60 amino-terminal residues are characterized by a low degree of similarity, typical of targeting sequences, yet they all share common features of transit peptides in being rich in serine, threonine, and small hydrophobic residues but with few acidic residues (48,49). All native monoterpene synthases thus far examined (2,17,22,23) appear to be NH 2 -terminally blocked, preventing direct determination (by sequencing) of the transit peptide-mature protein cleavage junction. Significantly, a tandem pair of arginine residues (e.g. Arg 55 -Arg 56 of SBS) are strictly conserved in the deduced sequences of all of the monoterpene syntheses (Fig. 1) and they define the most NH 2 -terminal region of obvious homology, suggesting a possible cleavage site. Recent studies 3 have shown that truncation of the recombinant spearmint limonene synthase preprotein immedi-ately upstream of these tandem arginines yields a fully functional "pseudomature" form of the enzyme, whereas truncation downstream from this element severely impairs activity.
Downstream of the tandem arginines are several regions of homology, including the highly conserved (I, L, or V)DDXXD motif (e.g. residues Ile 350 -Asp 355 of SBS) found in virtually all deduced sequences for enzymes that utilize prenyl diphosphate substrates (50,51). This aspartate-rich element is now generally recognized as a binding site for the metal ion chelated diphosphate ester substrate (50,(52)(53)(54)(55). Several other highly conserved regions are also apparent, including Arg 299 -Trp-Trp 301 , Arg 372 -Trp-Glu/Gln 374 , Tyr 384 -Met-Gln/Lys 386 , and Cys 516 -Tyr-Met-X-Glu/Asp 520 (amino acid placement referred to SBS). The active site peptide LQLYEASFLL, previously isolated from the co-purified (ϩ)-pinene synthase and (ϩ)-bornyl diphosphate synthase of sage (17), was located at residues 195-204 of SBS and also at residues 187-196 of SSS. Very similar sequences in the same location were found in SCS and in the two limonene synthase sequences from M. spicata and P. frutescens (Fig. 1).
An intriguing aspect of both prenyltransferase and terpenoid cyclization catalysis (1,56,57) is the electrophilic nature of these reaction mechanisms, and the means by which these enzymes stabilize and guide highly reactive carbocationic intermediates without alkylation at the active site. Dougherty (58) has suggested -cation interactions as a means of stabilizing such reactive intermediates in enzyme-catalyzed reactions, and the concept has received some support from both sequence analysis and x-ray crystallographic studies (59 -61). While far from conclusive, the number of conserved aromatic residues displayed among the monoterpene synthases does suggest the possibility of such a role for aromatic side chains in these electrophilic cyclizations.
The (ϩ)-bornyl diphosphate synthase from sage has previously been shown to be inhibited by the "active serine"-directed reagent diisopropyl fluorophosphate, a characteristic not shared by other monoterpene synthases (6). Because of the unique utilization of the substrate diphosphate moiety as the terminating nucleophile by this enzyme (62,63), it was hypothesized that a serine residue may be involved in binding and transfer of the diphosphate function in the course of the reaction. Sequence comparison of SBS with the other two monoterpene synthases of sage reveals five unique serine residues at positions 67, 255, 302, 454, and 469 (the serine at position 454 is conserved in monoterpene synthases from other species). Three of these serines (at positions 67, 255, and 302) are within otherwise highly conserved regions and are, therefore, obvious  targets for selective covalent modification with radiolabeled diisopropyl fluorophosphate and directed mutagenesis studies. Properties of Recombinant Synthases-Calibrated gel permeation chromatography of the pGEX fusion form of SBS revealed a single peak of activity at an elution volume corresponding to an M r ϳ200,000, indicating that the expressed fusion preprotein (corresponding to a molecular weight of about 2 ϫ 96,300) was a functional dimer. Treatment of the SBS protein with thrombin to remove the glutathione S-transferase fusion tag, followed by re-chromatography, indicated a decrease in molecular weight to approximately 135,000, consistent with the loss of the 27-kDa transferase peptide from each subunit at a calculated molecular weigth of 69,300 for the preprotein. Further correction of the molecular mass to account for the transit peptide would yield a dimer of about 120 kDa which corresponds roughly to the native dimer molecular mass of both (ϩ)-bornyl diphosphate synthase and (ϩ)-pinene synthase from sage (6,12), two enzymes which have never been satisfactorily resolved as distinct species. Although a dimeric quaternary structure is not unique to these two synthases, the vast majority of the monoterpene synthases characterized to date are monomeric (1,5). The product profile of SBS clone 3C6 is qualitatively similar to the combination of both of these enzymes (i.e. (ϩ)-bornyl diphosphate and the (ϩ)-series of ␣pinene and related olefins) (6,12), although the quantitative distributions do not exactly match, and the stereochemistry of the olefin products is anomalous. Thus, as indicated previously, (ϩ)-bornyl diphosphate, (ϩ)-␣-pinene, (ϩ)-camphene, and (ϩ)limonene arise via the same overall cyclization stereochemistry, and these enantiomers are produced exclusively from geranyl diphosphate by the native (ϩ)-bornyl diphosphate and (ϩ)-pinene synthase activities (18, 38 -41). The small amounts of (Ϫ)-limonene and (Ϫ)-camphene formed by the recombinant enzyme are attributed to antipodal cyclizations via abnormal, extended substrate conformations, as the phenomenon has been described previously, especially when using neryl diphosphate (the cis-analog of geranyl diphosphate) as an alternate substrate (40,64). The geranyl substrate, however, was verified as Ͼ99% pure, thereby eliminating this possibility in the present instance and suggesting that loss of stereochemical fidelity (to the extent of 5% of the total product mixture) may be attributed to the presence of the glutathione S-transferase fusion peptide plus transit peptide which may alter substrate binding directly, or indirectly by compromising subunit assembly. To address this question, as well as the correct identity of the SBS protein, will require the detailed assessment of truncated enzymes that more closely resemble the native form.
Gel permeation chromatography of SCS revealed a single peak of activity at an elution volume corresponding to an M r of 72,000, whereas SSS gave two peaks of activity, an aggregated form eluting in the void volume and a second corresponding to an M r of 60,000. Both of these molecular weights are significantly lower than those predicted from pGEX expression-based fusion of the glutathione S-transferase (27 kDa) with the respective preproteins (both SCS and SSS ϳ96 kDa). Thrombin treatment was without influence on the gel permeation chromatographic behavior of these enzymes, indicating the absence of the glutathione S-transferase peptide tag and rationalizing the previously observed inability of the recombinant SCS and SSS enzymes to bind to the glutathione-affinity column. Inspection of the 5Ј-sequences of the corresponding pGEX constructs showed both to be free of in-frame stop codons that might have permitted polycistronic translation of the preprotein devoid of the glutathione S-transferase peptide. The apparent truncation was therefore attributed to proteolytic processing of the recombinant SCS and SSS in the E. coli host to proteins that more closely resemble the preprotein forms of the native, monomeric 1,8-cineole synthase (8,35) and (ϩ)-sabinene synthase (9, 10) of sage. Similar proteolytic processing of a recombinant limonene synthase preprotein from spearmint has been observed previously in this E. coli host (23).
1,8-Cineole synthase has never been satisfactorily separated from the aforementioned (Ϫ)-pinene synthase from sage but, in this instance, the product distribution of SCS does not match the product distribution of (Ϫ)-pinene synthase, either quantitatively, qualitatively, or in stereochemical terms, since the reactions catalyzed are of the opposite antipodal series (8,11,18). However, the product distribution of SCS shows some parallels with that of the recently described cyclase III which produces (ϩ)-␣-pinene and (ϩ)-␤-pinene (13,14). Even here, the match is not perfect and the production of anomalous products of the antipodal (Ϫ)-series (Ͻ6% of total) again suggests that substrate binding interactions may be compromised. To assess the latter possibility, the K m values for SCS (7. Although the combination of products generated by the three recombinant enzymes described in this paper represent many of the monoterpenes synthesized by extracts of sage oil gland cells, several lines of evidence suggest that more monoterpene synthases remain to be acquired from this species. Thus, clone 3F5 (SUS) has not yet been functionally expressed, one of the two probes generated by PCR has not yet located the corresponding cDNA (one probe matches SUS), and an active site peptide (17) has not yet been matched to a cDNA sequence. The lack of the full set of synthases, coupled to the production of multiple products, including abnormal enantiomers, by the extant recombinant forms, has obscured the assignment of cDNA sequences to their native monoterpene synthase counterparts. This is most vividly illustrated by clone 3C6 which is clearly the recombinant form of (ϩ)-bornyl diphosphate synthase and may also represent the enzyme responsible for the activity previously assigned to (ϩ)-pinene synthase. Nevertheless, it is the great mechanistic diversity of this group of enzymes, originating from a single plant species yet producing a cyclic ether, diphosphate ester, and numerous olefins, that makes this set of catalysts unique and so biochemically appealing. Although the enzymatic generation of multiple products is an unusual and complex phenomenon, it does provide a most convenient and very powerful reporting device for communicating the influence of directed mutagenic change on the cyclization reaction. Defining how the similarities and differences of these enzymes relate to the control of catalytic channeling to the different products can be expected to provide the foundation for a detailed, molecular level understanding of structurefunction relationships in the monoterpene synthases.