Stereochemistry a t C-1 of Geranyl Pyrophosphate and Neryl Pyrophosphate in the Cyclization to (+)- and (-)-Bornyl Pyrophosphate*

(1R)-l-3H-labeled and (1S)-l-3H-labeled geranyl py- rophosphate and neryl pyrophosphate were prepared from the corresponding 1-’H-labeled aldehydes by a combination of enzymatic and synthetic procedures. Following admixture with the corresponding 2-14C- labeled internal standard, each substrate was converted to (+)-bornyl pyrophosphate and (-)-bornyl pyrophosphate by cell-free enzyme preparations from sage (Salvia officinalis) and tansy (Tanaceturn vul-gare), respectively. Each pyrophosphate ester was hy- drolyzed, and the resulting borneol was oxidized to camphor. The stereochemistry of labeling at C-3 of the derived ketone was determined by base-catalyzed exchange, taking advantage of the known selective ex- change of the exo-a-protons. By comparison of such exchange rates to those of product generated from (1RS)-2-’4C,1-3Hz-labeled substrate, it was demonstrated that geranyl pyrophosphate was cyclized to bornyl pyrophosphate with net retention of configuration at C-1 of the acyclic precursor, whereas neryl pyrophosphate was cyclized to product with inversion of configuration at C- 1. The observed stereochemistry is consistent with a reaction mechanism

(1R)-l-3H-labeled and (1S)-l-3H-labeled geranyl pyrophosphate and neryl pyrophosphate were prepared from the corresponding 1-'H-labeled aldehydes by a combination of enzymatic and synthetic procedures. Following admixture with the corresponding 2-14Clabeled internal standard, each substrate was converted to (+)-bornyl pyrophosphate and (-)-bornyl pyrophosphate by cell-free enzyme preparations from sage (Salvia officinalis) and tansy (Tanaceturn vulgare), respectively. Each pyrophosphate ester was hydrolyzed, and the resulting borneol was oxidized to camphor. The stereochemistry of labeling at C-3 of the derived ketone was determined by base-catalyzed exchange, taking advantage of the known selective exchange of the exo-a-protons. By comparison of such exchange rates to those of product generated from (1RS)-2-'4C,1-3Hz-labeled substrate, it was demonstrated that geranyl pyrophosphate was cyclized to bornyl pyrophosphate with net retention of configuration at C-1 of the acyclic precursor, whereas neryl pyrophosphate was cyclized to product with inversion of configuration at C-1. The observed stereochemistry is consistent with a reaction mechanism whereby ger-any1 pyrophosphate is first stereospecifically isomerized to linalyl pyrophosphate which, following rotation about C-2-C-3 to the cisoid conformer, cyclizes from the anti-endo configuration. Neryl pyrophosphate cyclizes either directly or via the linalyl intermediate without the attendant rotation.
It has long been recognized that geranyl pyrophosphate, the ubiquitous C,, intermediate of isoprenoid metabolism, cannot be converted directly to cyclohexanoid monoterpenes because of the topological constraints imposed by the trans-C-2-C-3 double bond of the acyclic prenyl chain. Many proposals have thus focused on the possible intermediacy of neryl pyrophosphate and linalyl pyrophosphate ( Fig. 1) in enzymatic cyclization processes and on possible means by which geranyl pyrophosphate might be isomerized to such a sterically more suitable intermediate (1)(2)(3). Over the last several years, major advances have been made in understanding the formation of the main classes of monocyclic and bicyclic monoterpenes *This investigation was supported in part by Grant GM 31354 from the National Institutes of Health. This is Scientific Paper 7008, Project 0268, College of Agriculture Research Center, Washington State University, Pullman, WA 99164-6340. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
2 To whom correspondence should be addressed.

FIG. 1. Pathway for the conversion of acyclic precursors to
from their acyclic precursors (3,4), and from such investigations, as well as studies on related sesquiterpene cyclizations (5), the general features of this reaction type have begun to emerge (6)(7)(8).
Investigations of the enzymology of camphor biosynthesis have provided the most detailed information to date regarding a monoterpene cyclization process. The biosynthesis of (+)camphor in sage (Salvia officinalis) has been shown (9-12) to involve the conversion of geranyl pyrophosphate to (+)-bornyl pyrophosphate which is subsequently hydrolyzed by a distinct pyrophosphatase to (+)-borneol, followed by the NAD-dependent dehydrogenation of the alcohol to the ketone ( Fig.  1). (-)-Camphor in tansy (Tanaceturn uulgare) and rosemary (Rosmarinus officinalis) is derived by a similar sequence of reactions (9, 13, 14), and sage leaf extracts have also been shown to contain a minor, and readily separable (-)-bornyl pyrophosphate cyclase activity. Studies on the biosynthesis of (+)-bornyl pyrophosphate are of particular significance in establishing for the first time that geranyl pyrophosphate is the preferred substrate for cyclization and that neryl pyrophosphate was not a mandatory intermediate (11). Thus, it was shown with partially purified preparations, free of contaminating phosphohydrolases which compete for the prenyl pyrophosphate substrates, that V/KM for geranyl pyrophosphate was 20 times that for the cis-isomer, neryl pyrophosphate. Furthermore, geranyl pyrophosphate and neryl pyrophosphate, as well as the tertiary isomer linalyl pyrophosphate, were converted to the product without detectable interconversion of the three acyclic precursors and without the preliminary formation of any other free intermediate. Additionally, it was demonstrated that [G-'4C,1-3H2]geranyl pyrophosphate was converted to (+)-[G-'4C,3-3H2]bornyl pyrophosphate (as well as to (-)-[G-'4C,3-3H2]bornyl pyrophosphate) without loss of hydrogen from C-1 (14, E), eliminating earlier proposals for trans-to-cis isomerization involving redox processes (1,16  The summary of the evidence clearly indicates that the enzyme is capable not only of the relevant cyclization but also of the required isomerization of geranyl pyrophosphate to a bound intermediate competent to cyclize. The overall conversion is also unusual in that the pyrophosphate moiety of the substrate is retained in the bicyclic product. The bornyl pyrophosphate cyclases have therefore provided a unique opportunity to examine the role of the pyrophosphate moiety in the coupled isomerization-cyclization process (17).
Schemes for the formation of (+)-bornyl pyrophosphate and (-)-bornyl pyrophosphate from geranyl pyrophosphate and neryl pyrophosphate, which are completely consistent with the results of numerous model studies of terpenoid cyclizations (18)(19)(20)(21)(22)(23), are illustrated in Fig. 2. The first step is metal ion-assisted ionization with syn-isomerization to the bound tertiary allylic isomer (or its ion-paired equivalent) now free to rotate about the C-2-C-3 bond. Following this rotation, ionization of the tertiary intermediate, with electrophilic attack by C-1 of the now cisoid, anti-endo conformer at C-6 of the adjacent double bond, generates the cyclohexanoid ring. Subsequent electrophilic addition to the newly created cyclohexene double bond and capture of the resulting cation by the pyrophosphate anion provide bornyl pyrophosphate with the requisite stereochemistry. One consequence of this scheme is that cyclization should proceed with net retention of configuration at C-1 of geranyl pyrophosphate and with inversion of configuration at C-1 of neryl pyrophosphate (the stereochemical consequences are identical whether neryl pyrophosphate cyclizes directly or via preliminary isomerization to linalyl pyrophosphate).
In this communication, we describe the conversion of both (1R)-and (1S)-l-3H-labeled geranyl pyrophosphate and neryl pyrophosphate to both (+)-and (-)-bornyl pyrophosphate and the stereochemical location of 3H in each product by selective exchange of the exo-a-hydrogen of the derived camphor. The utilization of all four stereospecifically labeled acyclic precursors with the enantiomeric cyclizing enzymes has provided an extremely powerful probe for testing and confirming these stereochemical predictions and, thus, the validity of the coupled isomerization-cyclization scheme.

RESULTS AND DISCUSSION
The proposed model for the cyclization of geranyl pyrophosphate and neryl pyrophosphate to bornyl pyrophosphate predicts that the configuration at C-1 of the trans-precursor will be retained in the transformation, while that of the cisprecursor will be inverted (Fig. 2). To examine the stereochemical fate of C-1 of these acyclic precursors, the (1R)-1-3H-and (1S)-l-3H-labeled enantiomers of both geranyl pyrophosphate and neryl pyrophosphate were prepared by combination of enzymatic and chemical methods. Each chiral substrate, as well as (1RS)-[l-3H2]geranyl pyrophosphate and (1RS)-[l-3Hz]neryl pyrophosphate, was admixed with its 2-"C-labeled counterpart to a 3H/14C ratio of 10, and the posi- tion of the tritium was verified to be a minimum of 95% of that specified (Table I). Each substrate was then enzymatically converted to (+)-bornyl pyrophosphate and (-)-bornyl pyrophosphate by partially purified cyclase preparations from S. officinalis (sage) and T. uulgure (tansy), respectively. The purification procedure (Sephadex G-150 chromatography) served to remove some of the competing phosphohydrolases in the crude extracts and to eliminate, from the sage leaf preparation, a low level of (-)-bornyl pyrophosphate cyclase activity (M, -65,000).
Bornyl pyrophosphate generated enzymatically from each substrate was hydrolyzed in situ by the addition of acid phosphatase plus apyrase, and following extraction and pooling of samples from like assays, the borneol was isolated by TLC2 after OsOl treatment to remove unsaturated compounds. The borneol derived from each substrate with the respective cyclase was oxidized (Cr03) to camphor, and this product then diluted to a specific activity of 250 pCi (3H)/mol for further studies. The 'H/14C ratios of each borneol sample and its respective derived camphor were identical, within experimental error, to that of the corresponding prenyl pyrophosphate starting material (Table I). These results confirm earlier observations that hydrogen from C-1 of the acyclic precursor is not lost in the conversion to (+)-or (-)-camphor (14,15). Such verification is crucial since the transformation to camphor allows advantage to be taken of the selectivity of exo-hydrogen exchange of this ketone (43) in locating the tritium.
A sufficient number of assays were run with four or five separate cyclase preparations to ensure the production of a minimum of 0.25 pCi (3H) of (+)-and (-)-camphor from each substrate with each preparation. Individual exchange runs with each 0.25 pCi (3H) (1 mmol) sample of camphor were carried out in 0.5 N NaOH/dioxane at 70 "C, and the tritium loss at corresponding time points for like samples was aver- stereochemical cyclization scheme (Fig. 2). Comparison of exchange rates from the linear portion of the curves (0-10 min) for the product derived from (1s)and (1R)-labeled precursors, respectively, gave a ratio (18.5:l) comparable to the known ratio of exchange rates (21.3:l) for the exo-versus endo-a-hydrogens of camphor (43). Fig. 3B. In this instance, the rate of tritium loss from camphor derived from (1R)-['4C,1-3H]geranyl pyrophosphate was approximately 20 times that of the product derived from (1S)-['4C,1-3H]geranyl pyrophosphate and roughly 10 times that from the racemic 14C,1-3H2-labeled precursor, when compared at 10 min. Thus, the 1-proR hydrogen of geranyl pyrophosphate gave rise to the a-exohydrogen of the derived (-)-camphor, opposite to the location determined for (+)-camphor and in accord with the proposed cyclization scheme. From the summary of these data with the enantiomeric products, it is clear (cf. Fig. 2) that configuration at C-1 of geranyl pyrophosphate is retained in the conversion to bornyl pyrophosphate.

The averaged exchange curves for (-)-camphor (via
Exchange curves for the (+)-camphor and (-)-camphor derived from the three l-3H-labeled neryl pyrophosphates with the respective (+)-and (-)-bornyl pyrophosphate cyclases are provided in Fig. 3, C and D. Based on arguments similar to those above, the results indicate that the 1-proR hydrogen of neryl pyrophosphate gives rise to the exo-ahydrogen of (+)-camphor, whereas the 1-pros hydrogen of this precursor gives rise to the exo-a-hydrogen of the enantiomeric (-)-camphor. It is clear, with reference to Fig. 2, that the configuration at C-1 of neryl pyrophosphate is inverted in the cyclization to bornyl pyrophosphate.
The summary of the results with all the l-3H-labeled precursors and both enantiomeric cyclases is entirely consistent with, and thus supports, the cyclization scheme (Fig. 2), which is itself based on numerous model reactions and related biosynthetic studies. Specifically, geranyl derivatives have been shown to cyclize in solution via preliminary conversion to the tertiary allylic, linalyl system (20), this cationic isomerization being favored by the presence of a large counterion such as phosphate (21). The allylic isomerization of geranyl pyrophosphate to linalyl pyrophosphate is considered to occur with overall syn stereochemistry based on an examination of the origin of linalool (46), and the transposition of trans,trunsfarnesyl pyrophosphate to nerolidyl pyrophosphate (the CIS analogs of geranyl pyrophosphate and linalyl pyrophosphate,  (1:l) at 70 "C. Aliquots were removed at the intervals indicated, the camphor purified by TLC, and the 3H/'4C ratio determined, from which tritium retention was calculated with reference to the starting material. Tritium retentions for each time period were averaged from four or five independent exchange runs. The apparent rapid deceleration of exo-exchange is attributed to an ~x o -~H isotope effect promoting inversion at C-3 via endo-'H loss and exo-'H return. respectively) by an enzyme system from Gibberella fujikuroi has been demonstrated to occur by a net suprafacial process (7). Linalyl derivatives afford the monocyclic product a-terpineol by solvolytic cyclization predominantly from the antiendo transition state (19), as indicated in Fig. 2, and neryl derivatives similarly cyclize in a stereospecific manner with inversion at C-1 of this acyclic precursor (22). (The stereochemical consequences are identical whether neryl pyrophosphate cyclizes directly or via prior isomerization to linalyl pyrophosphate as indicated in Fig. 2.) The conversion of acyclic prenyl pyrophosphates to the monocyclic a-terpenyl intermediate is a central feature of essentially all proposals for the biosynthesis of cyclohexanoid monoterpenes (3,6,8). The model studies described and the present results from examination of two distinct and enantiomer-generating enzyme systems provide solid experimental support for the stereochemistry of this crucial monoterpene cyclization step. It is also important to note that the enzymatic conversion of trans, trans-farnesyl pyrophosphate to the sesquiterpene trichodiene, a reaction considered to occur by an analogous coupled isomerization-cyclization sequence (8), proceeds with net retention of configuration at C-1 of the trans-acyclic precursor (47), and retention of configuration at C-1 of trans,trans-farnesyl pyrophosphate has been inferred in the cyclization to ybisabolene (48). Similar ring-generat-ing allylic displacements in the diterpene series also occur with overall anti-stereochemistry (49-55). Thus, a consistent pattern for terpenoid cyclizations is beginning to emerge.
The absolute configuration of the end products of the bornyl pyrophosphate cyclizations permits prediction of the configuration of the proposed linalyl pyrophosphate intermediates (i.e. (-)-(3R)and (+)-(3s)-linalyl pyrophosphates should give rise, respectively, to (+)-and to (-)-bornyl pyrophosphate, respectively (Fig. 2)); such insight is not possible in the case of symmetrical cyclization products such as y-terpinene (56) and 1,8-cineole (57). Since the bornyl pyrophosphate cyclases have been shown to utilize (3RS)-[l-Hz]linalyl pyrophosphate as an acyclic precursor, these predictions can be tested directly with the appropriately labeled (3R)-and (38)-enantiomers. These experiments are now underway.    (Table I) The 'Fi:14C ratio of each product was redetermined after pyrophosphorylation (a@ the diphenylurethane of the enzymatic hydrolysis prvduct), and we8 shown to be wirhin t 4 T of the scarring meterial in all caees ( Table 1)