Studies on taxadiene synthase: Interception of the cyclization cascade at the isocembrene stage with GGPP analogues

Article abstract of DOI:10.1021/jo0517489

The cyclization of GGPP to taxadiene, catalyzed by taxadiene synthase, has been suggested to proceed through a series of monocyclic isocembrenyl- and bicyclic verticillyl-carbocationic intermediary stages. A set of GGPP analogues with abolished or perturbed π-nucleophilicity at the Δ¹⁰ double bond (GGPP numbering) was synthesized and incubated with taxadiene synthase to intercept the cyclization cascade at the monocyclic stage. Each analogue was transformed by taxadiene synthase in vitro to hydrocarbon products in varying yields, and the structures of the major product in each reaction were solved by GCEIMS and one- and two-dimensional (¹H and ¹³C) NMR and found to be 14-membered monocyclic isocembrenyl diterpenes, indicating that the first C-C bond formation catalyzed by taxadiene synthase could be uncoupled from the other subsequent bond formation events by using suitably designed substrate analogues. The formation and isolation of these isocembrenyl diterpene products using taxadiene synthase supports proposals that the isocembrenyl cation is an intermediate in the cyclization of GGPP to taxadiene.

Full text of DOI:10.1021/jo0517489

Studies on Taxadiene Synthase: Interception of the Cyclization
Cascade at the Isocembrene Stage with GGPP Analogues
Siew Yin Chow, Howard J. Williams, Qiulong Huang,† Samik Nanda,‡ and A. Ian Scott*
Department of Chemistry, Texas A&M University, College Station, Texas 77843-3255
scott@mail.chem.tamu.edu
Received August 18, 2005
The cyclization of GGPP to taxadiene, catalyzed by taxadiene synthase, has been suggested to
proceed through a series of monocyclic isocembrenyl- and bicyclic verticillyl-carbocationic intermedi-
ary stages. A set of GGPP analogues with abolished or perturbed π-nucleophilicity at the ∆10 double
bond (GGPP numbering) was synthesized and incubated with taxadiene synthase to intercept the
cyclization cascade at the monocyclic stage. Each analogue was transformed by taxadiene synthase
in vitro to hydrocarbon products in varying yields, and the structures of the major product in each
reaction were solved by GCEIMS and one- and two-dimensional (1H and 13C) NMR and found to be
14-membered monocyclic isocembrenyl diterpenes, indicating that the first C-C bond formation
catalyzed by taxadiene synthase could be uncoupled from the other subsequent bond formation
events by using suitably designed substrate analogues. The formation and isolation of these
isocembrenyl diterpene products using taxadiene synthase supports proposals that the isocembrenyl
cation is an intermediate in the cyclization of GGPP to taxadiene.
Introduction
80% yield from a related compound, 10-desacetyl baccatin
III (2),3 isolated from the renewable needles of Taxus
baccata. Since 2002, Bristol-Myers-Squibb replaced the
semisynthetic route with an even more environmentally
friendly method using cell fermentation. In this process,
plant cells from Taxus chinensis are initially grown on a
petri dish and later transferred to a large-scale fermen-
tor. The development of this method was honored with a
presidential award and is expected to provide a steady
supply of paclitaxel.10
Paclitaxel (1, Figure 1) is an important natural product
belonging to the isoprenoid family.1 Originally isolated
from the bark of the yew tree Taxus brevifolia by Wani
and Wall in 1963, paclitaxel was approved as a chemo-
therapy agent for breast and ovarian cancer in late 1992.2
The clinical effectiveness and the scarcity of paclitaxel
makes this drug very precious, and to date, paclitaxel
has achieved $9 billion in sales worldwide.3 Because
paclitaxel can be isolated only in low yields from the bark
of the yew tree, at one time it was thought that continu-
ous production would eventually cause a supply shortage
of the drug. As such, paclitaxel became a leading target
for total synthesis, but commercial-scale production has
not been possible.4-9 The supply problem was solved in
1995, and paclitaxel is now prepared in approximately
The biosynthesis of paclitaxel begins with the forma-
tion of taxa-4(5),11(12)diene (3), which is then trans-
(5) (a) Nicolaou, K. C. et al. Nature 1994, 367, 630-634. (b) Nicolaou,
K. C. Nantermet, P. G.; Ueno, H.; Guy, R. K.; Coulandouros, E. A.;
Sorensen, E. J. J. Am. Chem. Soc. 1995, 117, 624-633. Nicolaou, K.
C.; Guy, R. K. Angew. Chem., Int. Ed. Engl. 1995, 34, 2079-2090.
(6) (a) Masters, J. J.; Link, J. T.; Snyder, L. B.; Young, W. B.;
Danishefsky, S. J. Angew. Chem., Int. Ed. Engl. 1995, 34, 1723-1726.
(b) Danishefsky, S. J. et al. J. Am. Chem. Soc. 1996, 118, 2843-2859.
(7) (a) Wender, P. A. et al. J. Am. Chem. Soc. 1997, 119, 2755-
2756. (b) Wender, P. A. et al. J. Am. Chem. Soc. 1997, 119, 2757-
2758.
(8) (a) Morihira, K.; Hara, R.; Kawahara, S.; Nishimori, T.; Naka-
mura, N.; Kusama, H.; Kuwajima, I. J. Am. Chem. Soc. 1998, 120,
12980-12981. (b) Kusama, H.; Hara, R.; Kawahara, S.; Nishimori, T.;
Kashima, H.; Nakamura, N.; Morihira, K.; Kuwajima, I. J. Am. Chem.
Soc. 2000, 122, 3811-3820.
(9) Mukaiyama, T. et al. Chem. Eur. J. 1999, 5, 121-161.
(10) Ritter, S. K. Chem. Eng. News. 2004, 82 (28), 25-30.
† Current address: Ebioscience, 6042 Cornerstone Court West, San
Diego, CA 92121.
‡ Current address: Biotechnology Research Center, Toyama Pre-
fectural Univesity, 5180 Kurokawa Kosugi, Toyama 939-0398, Japan.
(1) Sacchettini, J. C.; Poulter, C. D. Science 1997, 277, 1788-1789.
(2) Suffness, M.; Wall, M. E. In TAXOL Science and Applications.
Suffness, M., Ed.; CRC Press: Boca Raton, FL, 1995; pp 3-25 and
references therein.
(3) Morrissey, S. R. Chem Eng. News 2003, 81 (37), 17-20.
(4) (a) Holton, R. A. et al. J. Am. Chem. Soc. 1994, 116, 1597-1598.
(b) Holton, R. A. et al. J. Am. Chem. Soc. 1994, 116, 1599-1600.
10.1021/jo0517489 CCC: $30.25 © 2005 American Chemical Society
Published on Web 11/02/2005
J. Org. Chem. 2005, 70, 9997-10003
9997

Chow et al.
step in the biosynthetic pathway to all taxanes. The
chemical structure of taxadiene has been verified inde-
pendently by total synthesis,16 and conversion of radio-
labeled taxadiene to radiolabeled paclitaxel with yew
stem sections proved that taxadiene is the progenitor of
paclitaxel.12a Like other terpene-cyclase-catalyzed reac-
tions, this cyclization reaction has been traditionally
formulated to proceed with multiple carbocationic inter-
mediary stages, as shown in eq 1.
FIGURE 1. Structures of taxane diterpenes.
formed to 10-desacetylbaccatin III (2) in several steps and
eventually to paclitaxel. The many steps that link taxa-
diene to 10-desacetylbaccatin III (2) and to paclitaxel
obviously involve monooxygenations of the taxane skel-
eton and the esterification of some of the resulting
hydroxyl groups.11 To date, taxadiene synthase, a few
monooxygenases, and all five acyl/benzoyl transferases
have been found and cloned into common laboratory
microorganisms.12-14 Although the exact sequence of the
transformations is not known, a statistical survey of the
oxygenation pattern of the entire taxoid family suggests
that oxygenation of taxadiene should occur in the order
C5, C10, C2, C9, C1, and C13 en route to paclitaxel.
Similarly, esterifications should occur in the order C5,
C2, C10, and C13.15
The intermediate carbocations 5-7 are structurally
related to diterpenes isocembrene (8) and verticillene (9),
and both compounds have been proposed as possible
intermediates in the cyclization of GGPP to taxadiene.17
However, early experiments designed to test the inter-
mediacy of these putative monocyclic or bicyclic com-
pounds were unsuccessful. Treatment of cembrene (not
isocembrene) or verticillene with acid in biomimetic
reactions did not convert these materials to products
having the desired tricyclic taxane skeleton.18 Further-
more, incubation of yew extracts with verticillene in
inhibition, trapping, and direct conversion experiments
showed that verticillene is not a freely exchangeable
intermediate in the enzyme-catalyzed reaction.19 None-
theless, the existence of carbocations 5-7 as tightly
bound intermediates in the active site of the enzyme
remains plausible. Under normal circumstances, these
intermediates are reactive, with short lifetimes which
The cyclization of geranylgeranyl diphosphate (GGPP,
4) to taxadiene 3 is a remarkable reaction that involves
the formation of three C-C σ bonds, three carbocyclic
rings, and three new stereocenters from the achiral,
acyclic substrate. This reaction represents the committed
(11) Walker, K.; Croteau, R.; Recent Adv. in Phytochem. 1999, 33,
31-50.
(12) (a) Koepp, A. E.; Hezari, M.; Zjicek, J.; Vogel, B. S.; LaFever,
R. E.; Lewis, N. G.; Croteau, R. J. Biol. Chem. 1995, 270 (15), 8686-
8690. (b) Hezari, M.; Lewis, N. G.; Croteau, R. Arch. Biochem. Biophys.
1995, 322 (2), 437-444. (c) Wildung, M. R.; Croteau, R. J. Biol. Chem.
1996, 271 (16), 9201-9204. (d) Huang, K.-X.; Huang, Q.-L.; Wildung,
M. R.; Croteau, R.; Scott, A. I. Protein Express. Purif. 1998, 13, 90-
96. (e) Huang, Q.; Roessner, C. A.; Croteau, R.; Scott, A. I. Bioorg. Med.
Chem. 2001, 9, 2237-2242.
(13) (a) Hefner, J.; Rubenstein, S. M.; Ketchum, R. E. B.; Gibson,
D. M.; Williams, R. M.; Croteau, R. Chem. Biol. 1996, 3, 479-489. (b)
Chau, M.; Walker, K.; Long, R.; Croteau, R. Arch. Biochem. Biophys.
2004, 430, 237-246. (c) Jennewein, S.; Long, R.; Williams, R. M.;
Croteau, R. Chem. Biol. 2004, 11, 379-387. (d) Jennewein, S.; Rithner,
C. D.; Williams, R. M.; Croteau, R. B. Proc. Natl. Acad. Sci. U.S.A.
2001, 98, 13595-13600. (e) Schoendorf, A.; Rithner, C. D.; Williams,
R. M.; Croteau, R. B. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 1501-
1506. (f) Wheeler, A. L.; Long, R. M.; Ketchum, R. E. B.; Rithner, C.
D.; Williams, R. M.; Croteau, R. Arch. Biochem. Biophys. 2001, 390,
265-278. (g) Chau, M.; Jennewein, S.; Walker, K.; Croteau, R. Chem.
Biol. 2004, 11, 663-672.
(16) Rubenstein, S. M.; Williams, R. M. J. Org. Chem. 1995, 60,
7215-7223.
(17) (a) Harrison, J. W.; Lythgoe, B. J. Chem. Soc. C 1966, 1932-
1945. (b) Gueritte-Voegelein, F.; Guenard, D.; Potier, P. J. Nat. Prod.
1987, 50, 9-18. (c) Floss, H. G.; Mocek, U. In TAXOL Science and
Applications; Suffness, M., Ed.; CRC Press: Boca Raton, 1995; pp 191-
208.
(18) (a) Dauben, W. G.; Hubbell, J. P.; Oberhansli, P.; Thiessen, W.
E. J. Org. Chem. 1979, 44, 669-673. (b) Begley, M. J.; Jackson, C. G.;
Pattenden, G. Tetrahedron 1990, 46, 4907-4924. Note: In Dauben’s
paper (ref 18a), the term cembrene refers to the monocyclic diterpene
with double bonds at ∆2,4Z,7,11, using the numbering system shown in
(14) (a) Walker, K.; Ketchum, R. E. B.; Hezari, M.; Gatfield, D.;
Goleniowski, M.; Barthol, A.; Croteau, R. Arch. Biochem. Biophys.
1999, 364, 273-279. (b) Walker, K.; Schoendorf, A.; Croteau, R. Arch.
Biochem. Biophys. 2000, 374, 371-380. (c) Walker, K.; Croteau, R.
Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 583-587. (d) Walker, K.;
Croteau, R. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 13591-13596. (e)
Walker, K.; Long, R.; Croteau, R. Proc. Natl. Acad. Sci. U.S.A. 2002,
99, 9166-9171. (f) Walker, K.; Fujisaki, S.; Long, R.; Croteau, R. Proc.
Natl. Acad. Sci. U.S.A. 2002, 99, 12715-12720.
Figure 1. Structure 8 shown in this work, with double bonds at ∆3,7,11,15
,
had previously been referred to as “cembrene” or “cembrene A” by
others (see ref 17a-c). We chose the term “isocembrene” because it is
a single word and to distinguish the structure from that used in
Dauben’s work.
(15) Baloglu, E.; Kingston, D. G. I. J. Nat. Prod. 1999, 62, 1448-
1472.
(19) Lin, X.; Hezari, M.; Koepp, A. E.; Floss, H. G.; Croteau, R.
Biochemistry 1996, 35, 2968-2977.
9998 J. Org. Chem., Vol. 70, No. 24, 2005

Studies on Taxadiene Synthase
Results
Synthesis of Substrates. The synthesis of 10,11R-
dihydroGGPP 10 was accomplished in six steps from the
commercially available S-citronellyl bromide 14, as shown
in Scheme 1. Treatment of bromide 14 with sodium iodide
in acetone, followed by sodium p-toluenesulfinate in DMF
at 0 °C, gave sulfone 16 in 89% yield over two steps.
Sulfone 16 was subjected to a Beillmann coupling at -78
°C with the known allylic bromide 17,23 and subsequent
reduction gave the desired known24 alcohol 19 in 32%
yield over two steps. Alcohol 19 was converted to 10,-
11S-dihydroGGPP 10 in 35% by activation as the bro-
mide 20 (PBr3, Et2O, 0 °C), followed by treatment with
(Bu4N)3HOPP in anhydrous acetonitrile using a modified
literature procedure.25 Using the identical procedure,
commercially available R-citronellyl bromide 21 was used
to synthesize 10,11S-dihydroGGPP (11).
Synthesis of 11-desmethylGGPP 12 and 2E,6E,10Z-
GGPP 13, shown in Scheme 2, employed the common
starting aldehyde 23, prepared readily from farnesyl
benzyl ether 22 in 28% yield.26 A Wittig reaction gave
unsaturated ester 24 in 91% yield, having the desired
trans geometry at the newly created double bond. Reduc-
tion of 24 with DiBAL-H gave alcohol 25 in 93% yield.
Conversion of alcohol 25 to bromide 26 was accomplished
in one pot by treatment with methanesulfonyl chloride
followed by lithium bromide. Beillmann coupling of
bromide 26 with sulfone 27, followed by reduction with
lithium/ammonia, furnished 11-desmethyl GGOH 29 in
45% overall yield from 25.23c Alcohol 29 was converted
to 11-desmethylGGPP 12 in 46% yield as described above
for analogue 10.
FIGURE 2. Analogues 10-13 used in this study, which are
dissimilar to GGPP 4 at the ∆10 double bond.
may prohibit isolation and direct observation. To observe
the intermediates in such situations, a common strategy
has been to use substrate analogues that are unable to
undergo the complete enzymatic cascade, thereby termi-
nating at the intermediate stage.
The availability of taxadiene synthase, overexpressed
in recombinant E. coli, allowed the mechanistic investi-
gation of the cyclization of GGPP to taxadiene in vitro.20
In this paper, we report the synthesis of four GGPP
analogues 10-13 (Figure 2) that are dissimilar to GGPP
4 at the ∆10 double bond and incubation of these
compounds with taxadiene synthase. By design, the
termini of these analogues remain unchanged with
respect to the natural substrate in order to maintain the
ability to form the C1-C14 bond (GGPP numbering). For
analogues 10 and 11, the monocyclic 5-like intermediates,
if formed, would obviously not be able to undergo the
subsequent C-C σ bond formation step because the
requisite ∆10 double bond is absent. Although analogues
12 and 13 both contain the ∆10 double bond, formation
of the bicyclic 6-like intermediate would be be more
difficult because the ∆10 double bond of 12 is only
disubstituted and hence less nucleophilic, while in 13,
the unnatural Z-geometry of the ∆10 double bond may
not present the proper approach vector to effect C-C
bond formation. If not completely unreactive, these
analogues were expected to halt the cyclization cascade
at the monocyclic isocembrene level. A similar strategy
had been used successfully by others to intercept, for
example, the conversion of oxidosqualene to lanosterol
and the conversion of farnesyl diphosphate to aris-
tolochene.21,22 While this work was in progress, Coates
et al. published a related work on the taxadiene synthase-
catalyzed formation of verticillene-like products.20d
For the preparation of 2E,6E,10Z-GGPP 13, aldehyde
23 was subjected to a Still-Gennari olefination to furnish
unsaturated ester 31, with a Z/E ratio of 98:2 (separable
by column chromatography to 100:0 as determined by 1H
NMR).27 A similar set of transformations, as described
for 24 f 12, provided 2E,6E,10Z-GGPP 13 in similar
yields.
(23) Altman, L. J.; Ash, L.; Marson, S. Synthesis 1974, 129-131.
(b) Sharpless, K. B.; Umbreit, M. A. J. Am. Chem. Soc. 1977, 99, 5526-
5528. (c) Coates, R. M.; Ley, D. A.; Cavender, P. L. J. Org. Chem. 1978,
43, 3, 4915-4922.
(24) Kato, T.; Ebihara, S.-I.; Furukawa, T.; Tanahashi, H.; Hoshika-
wa, M. Tetrahedron: Asymmetry 1999, 10, 3691-3700.
(25) (a) Davisson, V. J.; Woodside, A. B.; Neal, T. R.; Stremler, K.
E.; Muehlbacher, M.; Poulter, C. D. J. Org. Chem. 1986, 51, 4768-
4779. (b) Davisson, V. J.; Woodside, A. B.; Poulter, C. D. Methods
Enzymol. 1984, 110, 130-144. (c) We inadvertently made (NH4)(Bu4N)2
or (Bu4N)3 salts of the isoprenoid diphosphates, found to be stable for
months at -20 °C, which gave reasonable yields in the incubation
reactions. We did not compare the stability or reactivity of these salts
with the (NH4)3 salts commonly used by others. The stoichiometry of
the Bu4N group relative to the isoprenoid diphosphate was determined
by solvent-suppressed 1H NMR, comparing the integration of C1-proton
of the Bu4N group at ∼3.2 ppm to the C1-proton of the isoprenoid
moiety at ∼4.5 ppm. See the Experimental Section and Supporting
Information for details.
(20) (a) Williams, D. C.; Carroll, B. J.; Jin, Q.; Rithner, C. D.; Lenger,
S. R.; Floss, H. G.; Coates, R. M.; Williams, R. M.; Croteau, R. Chem.
Biol. 2000, 7, 969-977. (b) Williams, D. C.; Wildung, M. R.; Jin, A.
Q.; Dalal, D.; Oliver, J. S.; Coates, R. M.; Croteau, R. Arch. Biochem.
Biophys. 2000, 379 (1), 137-146. (c) Huang, Q.; Williams, H.; Scott,
A. I. Tetrahedron Lett. 2000, 41, 9701-9704. (d) Jin, Y.; Williams, D.
C.; Croteau, R.; Coates, R. M. J. Am. Chem. Soc. 2005, 127, 7834-
7842. (e) Jin, Q.; Williams, D. C.; Hezari, M.; Croteau, R.; Coates, R.
M. J. Org. Chem. 2005, 70, 4667-4675.
(26) Chen, K.-M.; Semple, J. E.; Joullie, M. M. J. Org. Chem. 1985,
50, 3997-4005. See also ref 23c.
(27) (a) Still, W. C.; Gennari, C. Tetrahedron Lett. 1983, 24, 4405-
4408. (b) We initially tried to make 2E,6E,10Z-GGPP 13 by coupling
of neryl sulfone with bromide 17, followed by reduction, bromination
and phosphorylation. However, the incubation products showed the
presence of taxadiene, along with 40 in ∼1:1 ratio by TLC. Hydrolysis
of the EEZ-GGPP with phosphatase to EEZ-GGOH also showed ∼10%
all-trans GGOH (GC). Here, assuming that the yield of all-trans GGPP
to taxadiene was e23%, the yield of EEZ-GGPP was calculated to be
g3%. In contrast, when 13 was synthesized as shown in Scheme 2,
GC of the incubation product showed no or negligible taxadiene at tR
∼23.9 min (see Figure 2d).
(21) (a) van Tamelen, E. E.; Sharpless, K. B.; Ranzlik, R.; Clayton,
R. B.; Burlingame, A. L.; Wszolek, P. C. J. Am. Chem. Soc. 1967, 89,
7150-7151. (b) Corey, E. J.; Virgil, S. C.; Liu, D. R.; Sarshar, S. J.
Am. Chem. Soc. 1992, 114, 1524-1525. (c) Krief, A.; Schauder, J.-R.;
Guittet, E.; Herve du Penhoat, C.; Lallemand, H.-Y. J. Am. Chem. Soc.
1987, 109, 7910-7911.
(22) Cane, D. E.; Tsantrizos, Y. S. J. Am. Chem. Soc. 1996, 118,
10037-10040.
J. Org. Chem, Vol. 70, No. 24, 2005 9999

Chow et al.
SCHEME 1
SCHEME 2
with GCMS analyses (see Figure 3), indicated that each
analogue gave predominantly one major product com-
prising approximately 50-60% of the overall hydrocarbon
mixture. The molecular weight of the major products
from 10 to 13 were shown to be 274, 274, 258, and 272,
respectively, consistent with the loss of a proton and the
diphosphate group. Large-scale incubations were con-
ducted to obtain more product, and the results are
summarized in Table 1.
NMR Analyses of the Incubation Products. After
purification of the crude hydrocarbon products by pre-
parative GC, we obtained sufficient amounts (∼1 mg) of
the major products for rigorous structural elucidation by
NMR (1H, 13C, DEPT, COSY, NOESY, HSQC, HMBC,
and TOCSY). NMR analysis in all cases used the follow-
ing procedures. The 1H spectrum was first used to count
Enzymatic Cyclization of GGPP Analogues to
Isocembrenoid Diterpenes. E. coli strain BL21(DE3)-
plysS/pTS79H, described previously,12e was used to over-
express taxadiene synthase. After optimization, the crude
enzyme derived from 8 L of cell culture could effect
conversion of 89 µmol of all-trans GGPP to 5.7 mg (21
µmol) of the hydrocarbon product, corresponding to 23%
isolated yield.28 TLC of products from small-scale incuba-
tions of analogues 10-13 with taxadiene synthase, along
FIGURE 3. GCMS analyses, displaying the total ion current
of the hydrocarbon products from the incubation of taxadiene
synthase with analogues 10 (panel a), 11 (panel b), 12 (panel
c), and 13 (panel d). GCMS analyses were performed using a
30 M, 0.25 mm i.d., 0.25 µM DB-5MS column, programmed
120 °C for 2.00 min, to 280 °C at 5 °C/min, final time 15 min.
In control experiments, incubation of analogues 10-13 with
BL21/DE3/plysS (which does not encode taxadiene synthase)
failed to give any hydrocarbon products on TLC.
(28) We opted to use soluble crude lysate instead of purified enzyme
because we felt that the crude lysate could contain pyrophosphatase
activity, which would hydrolyze inorganic phosphate from the reaction,
thereby eliminating product inhibition by diphosphate. A previous
similar experience with the enzyme SAM-synthetase showed that the
soluble crude lysate functioned better than the purified enzyme,
probably for the same reason. See: Park, J.; Tai, J.; Roessner, C. A.;
Scott, A. I. Bioorg. Med. Chem. 1996, 4 (12), 2179-2185.
10000 J. Org. Chem., Vol. 70, No. 24, 2005

Studies on Taxadiene Synthase
1
TABLE 1. Large-Scale Incubation of Taxadiene
the minor products was shown by GCMS and H NMR
analysis to be 10,11-dihydrogeranylmyrcene 41.30m
Synthase with Analogues 10-13
substrate
enzyme
hydrocarbon products
amount
(µmol)
cell
(L)
lysate
(mL)
mass
(mg)
yield
(%)
analogue
10a
377
300
770
n.d.
89
30
32
39
8
750
800
975
200
200
5.9c
6
19
<1
∼3
23
11a
15.7c
∼1-2d
n.d.e
12b
13b
GGPPb
8
5.7c
a [(Bu4N)2NH4 salt], see ref 25c. b [(Bu4N)3 salt], see ref 25c.
c Isolated yield. d Estimated yield. e In some runs, substrate 13,
contaminated with ∼10% all-trans GGPP, gave ∼1:1 monocyclic
products/taxadiene. Given that the yield of GGPP f taxadiene
was 23%, the yield of products from 13 was estimated at ∼3%.
See ref 27b for more details.
vinyl protons and methyl groups and determine methyl
group multiplicities. 13C and DEPT experiments then
allowed double bond and attached hydrogen numbers to
1
be determined. HSQC associated H chemical shifts to
appropriate carbons. In HMBC analysis, correlations
between each methyl 1H peak and 13C of nearby carbons
revealed in all cases the directly attached carbon and at
least two others, usually saturated methylene and two
vinyl carbons. Vinyl 1H correlations to adjacent methyl-
enes completed individual isoprene unit assignments.
TOCSY spectra then supplied connectivity between groups.
The major products from the respective reactions were
assigned structures corresponding to 1S,12R-11,12-di-
hydroisocembrene 37, 1S,12S-11,12-dihydroisocembrene
38, 1S-12-desmethylisocembrene 39, and 1S-11,12-Z-
isocembrene 40 as shown in eqs 2-5 on the basis of these
NMR experiments.29 In every case, the stereochemistry
at C1 was assumed to be the same as in C1 of taxadiene,
i.e., an “S” configuration. The 13C and 1H chemical shifts
of 37-40 are summarized in Tables 2 and 3, and
additional NMR data are shown in the Supporting
Information. Due to limited amounts of material, we were
unable to characterize the minor products in each reac-
tion, except for one case. For the incubation of 11, one of
Discussion
The cyclization of GGPP to taxadiene had been a
subject of continued interest since 1966.17 The availability
of taxadiene synthase, overexpressed in recombinant E.
coli, allowed the mechanistic investigation of the cycliza-
tion of GGPP to taxadiene in vitro.19,20 For example, by
the use of appropriately deuterated substrates, the
TABLE 2. 1H (500 MHz) and 13C NMR (125 MHz) Data for Compounds 37 and 38 in CDCl3
37
38
positiona
δC
δH (mult, J, int)
δC
δH (mult, J, int)
1
2
49.2
33.4
1.91 (m, 1H)
45.9
33.9
1.98 (m, 1H)
1.96, 2.04b (m, 2H total)
5.17 (dd 9.3, 5.4 Hz, 1H)
-
1.97 (m, 2H)
3
125.0
135.3
39.4
124.7
135.1
39.4
5.18 (br t, 7.2 Hz, 1H)
-
2.17 (m, 2H)
2.25 (m, 2H)
4
5
2.14, 2.23b (m, 2H total)
2.24 (m, 2H)
6
24.6
24.9
7
123.4
134.7
37.5
5.07 (t, 6.4 Hz, 1H)
-
124.6
134.1
38.3
5.02 (t, 6.5 Hz, 1H)
-
8
9
1.89, 2.06b (m, 2H total)
1.350, 1.44b (m, 2H total)
1.16, 1.31b (m, 2H total)
1.349 (m, 1H)
1.93, 2.04b (m, 2H)
1.411 (m, 2H)
0.92, 1.412b (m, 2H)
1.49 (m, 1H)
10
11
12
13
14
15
16
17
18
19
20
21.6
21.9
32.1
31.0
32.6
29.0
34.2
0.76, 1.29b (m, 2H total)
1.09, 1.53b (m, 2H total)
-
32.1
1.05, 1.27b (m, 2H)
1.14, 1.51b (m, 2H)
-
28.0
25.3
149.4
109.7
19.3
149.0
110.1
18.8
4.72, 4.73 (m, 2H total)
1.69 (s, 3H)
0.86 (d, 6.7 Hz, 3H)
1.63 (s, 3H)
1.59 (s, 3H)
4.71, 4.73 (each br s, 1H)
1.67 (s, 3H)
0.83 (d, 6.7 Hz, 3H)
1.61 (s, 3H)
1.60 (s, 3H)
21.0
20.6
17.5
16.6
15.1
15.4
a See eq 2 for the representative numbering system. b The assignments of diastereotopic protons as the R or â configurations are not
possible because of the flexibility of the skeleton.
J. Org. Chem, Vol. 70, No. 24, 2005 10001

Chow et al.
TABLE 3. 1H (500 MHz) and 13C NMR (125 MHz) Data for Compounds 39 and 40 in CDCl3
39
40
positiona
δC
δH (mult, J, int)
δC
δH (mult, J, int)
1
2
47.1
33.0
124.9
135.0
39.3
24.9
124.3
135.5
39.0
26.8
131.7
128.8
24.9
31.3
149.0
110.0
19.2
-
2.052 (m, 1H)
48.4
33.1
2.01 (m, 1H)
1.97, 2.049b (m, 2H)
5.10 (br t, 7.9 Hz, 1H)
-
1.99, 2.079b (m, 2H)
5.09 (m, 1H)
-
3
124.8
135.3
39.6
4
5
2.08, 2.19b (m, 2H)
2.09, 2.28b (m, 2H)
5.03 (t, 7.0 Hz, 1H)
-
2.081 (m, 2H)
2.071, 2.28b (m, 2H)
5.04 (t, 7.14 Hz, 1H)
-
6
24.8
7
124.6
136.6
39.7
8
9
2.055, 2.16b (m, 2H total)
2.18, 2.22b (m, 2H total)
5.35 (m, 1H)
1.934, 2.21b (m, 2H)
2.12 (m, 2H)
5.10 (m, 1H)
-
10
11
12
13
14
15
16
17
18
19
20
29.3
126.0
135.5
30.6
5.51 (m, 1H)
1.81, 1.91b (m, 2H total)
1.30, 1.45b (m, 2H total)
-
1.65, 1.93b (m, 2H)
1.29, 1.44b (m, 2H)
-
29.8
148.8
110.1
19.3
4.74, 4.75 (m, 2H total)
1.70 (s, 3H)
4.75, 4.76 (each br. s, 1H)
1.71 (s, 3H)
-
24.3
1.68 (s, 3H)
17.5
15.6
1.62 (s, 3H)
17.6
1.60 (s, 3H)
1.59 (s, 3H)
15.5
1.58 (s, 3H)
a See eq 2 for the representative numbering system. b The assignments of diastereotopic protons as the R or â configurations are not
possible because of the flexibility of the skeleton.
stereochemical course of the cyclization had been eluci-
dated in detail.19,20a,e Also, alkylation of the enzyme by
potential suicide inhibitors had been attempted, unfor-
tunately, with little success.20b Using a fluorinated
analogue, Coates et al. recently observed verticillene
synthase activity by taxadiene synthase directly,20d which
supports the early biogenetic postulates.17 In this pub-
lication, we refine the cyclization mechanism of taxadiene
synthase by demonstrating isocembrene synthase activity
by taxadiene synthase through the isolation of isocem-
brenoid diterpene analogues 37-40.
In the active site of taxadiene synthase, the terminal
double bond of GGPP is brought into close contact with
C1 to initiate the cyclization reaction. It appears that a
similar conformation can be adopted by analogues 10, 11,
and 13 despite the increased rotational freedom or
altered geometry relative to all-trans GGPP at the ∆10
double bond, indicating that the enzyme allows a certain
amount of steric flexibility for binding of these analogues.
The total hydrocarbon yield from 10 (6%) was signifi-
cantly lower than the yield from 11 (19%), suggesting
that the chirality of the binding pocket fits 11 better than
10.
brenyl cation is quenched by capture of water to give a
desmethyl cembrenol, or (d) the substrate is irreversibly
attacked by a nucleophilic site in the enzyme, giving rise
to inhibition. Possibility (a), that analogue 12 is unreac-
tive, seems unlikely due to the significantly higher yields
seen with the other analogues. Possibilities (b), (c), and
(d) have precedents in other terpene-cyclase systems, but
remain unexplored for this reaction.31 In our reactions,
because the incubations were performed using soluble
crude lysate containing many polar endogenous metabo-
lites, we did not attempt to locate any potentially
unreacted substrate or alcoholic products from the incu-
bation of 12.28 We are currently reinvestigating this
reaction to explore these possibilities.
In this work, the analogues 10-13 were designed
without any modification to the ∆14 double bond in order
to maintain the ability to form the C1-C14 bond (GGPP
numbering). It appears that the ∆14 double bond is the
preferred site of reaction despite the presence of ∆6 (and
10
∆
for 12 and 13) double bond in all these analogues to
potentially compete for ring closure, and the observed
regioselectivity is undoubtedly imposed by the enzyme.
When farnesyl diphosphate (FPP) was incubated with
taxadiene synthase, the absence of the ∆14 double bond
allowed the otherwise disfavored formation of six-
membered bisabolyl sesquiterpenes.20c Finally, we note
that the major products from each analogue are terminal
olefins that resulted from deprotonation at C16, rather
than at C1, which would lead to the more stable Zaitsev-
like product, suggesting that an enzymatic base could be
responsible for regioselective deprotonation.
In contrast, 11-desmethyl GGPP analogue 12, which
is able to fit into the binding pocket as readily as the
natural substrate, gave a very low (<1%) yield. There
are a few possibilities that would explain the low
hydrocarbon yield: (a) the analogue remains dormant in
the active site without reacting, (b) the analogue is
hydrolyzed back to the corresponding alcohol 29, either
by taxadiene synthase or by other hydrolases, (c) cycliza-
tion does take place, but the resulting desmethyl isocem-
Conclusion
(29) Compounds 37-39 are new compounds. Compound 40 had been
previously synthesized using a nonenzymatic method. The 1H NMR
of 40 showed good agreement with the literature data. See: Kato, T.;
Suzuki, M.; Kobayashi, T.; Moore, B. P. J. Org. Chem. 1980, 45, 1126-
1130.
The results presented in this publication indicate that
the first C-C bond formation catalyzed by taxadiene
synthase can be uncoupled from the other subsequent
bond formation events by using suitably designed sub-
strate analogues. The formation and isolation of isocem-
brenyl diterpene products using taxadiene synthase
(30) The olefinic region in the 1H NMR of our product was similar
to that of geranylmyrcene, less a triplet. See: Baeckstrom, P.; Li, L.
Tetrahedron 1991, 47, 6533-6538. See also the Supporting Information
for more details.
10002 J. Org. Chem., Vol. 70, No. 24, 2005

Studies on Taxadiene Synthase
onto the silica column, saving a small amount for TLC.
Typically, approximately 250 mL of silica gel [equilibrated with
solvent C (2:1 v/v 2-propanol/concd NH4OH)] was used for
every 3 mL of sample solution. The column was eluted with
solvent C, and the eluents were collected in fractions. A TLC
analysis (in solvent C, anisaldehyde stain) of the crude mixture
showed the desired product at the baseline Rf 0.0-0.1, along
with impurities with Rf ∼ 0.5-1.0. When all the impurities
had eluted from the column, solvent D [6:3:11 (v/v/v) 2-pro-
panol/concd NH4OH/doubly distilled H2O (ddH2O)] was used
to elute the desired product from the column. The fractions
containing the product were pooled and concentrated in vacuo
using a rotary evaporator (bath temperature e 40 °C) to
remove bulk solvent. Rotary evaporation was discontinued
when the product solution started to bubble, and the cloudy
product solution was lyophilized to give a colorless gum. The
product was taken up in ddH2O and centrifuged at 11000g for
10 min. The supernatant was collected, transferred to a tared
flask, and lyophilized to give 10,11R-dihydroGGPP 10 [0.71
g, 0.74 mmol, 35%, (Bu4N)2NH4 salt by 1H NMR]. The product
was kept as a frozen aqueous solution (25 mg/mL) at -20 °C
in plastic centrifuge tubes: solvent suppressed 1H NMR (D2O,
500 MHz) 5.39 (br t, J ) 6.5 Hz, 1H), 5.13 (br t, J ) 6.8 Hz,
1H), 5.08 (br t, J ) 6.8 Hz, 1H), 4.45 (br, 2H), 3.17 (m, 16H),
2.14-1.87 (8H), 1.76-1.51 (m, 28H), 1.47-1.24 (br, 4H and
sextet, J ) 7.4 Hz, 16H) 1.19-1.00 (m, 3H), 0.94 (t, J ) 7.5
Hz, 24H), 0.87 (t, J ) 6.35 Hz, 3H); 13C NMR (D2O, 125 MHz)
δ 143.1, 137.3, 132.9, 127.5, 127.0, 123.3 (d, J ) 8.5 Hz), 65.0
(d, J ) 5.5 Hz), 60.8 (Bu4N), 42.7, 42.4, 39.9, 39.4, 34.9, 29.3,
28.20, 28.17, 28.10, 25.9 (Bu4N), 22.1, 21.8 (Bu4N), 20.0, 18.7,
18.3, 15.6 (Bu4N); 31P NMR (D2O, 202 MHz) δ -9.1 (d, J )
20.0 Hz), -10.8 (d, J ) 18.7 Hz); ESIHRMS calcd for
C20H37O7P2 451.2015, found 451.2031.
Incubation of 10,11R-DihydroGGPP with Taxadiene
Synthase. Cell pellets from an 8 L culture of E. coli strain
BL21(DE3)plysS/pTS79H12e were resuspended and lysed in
200 mL of buffer consisting of 30 mM NaHEPES, 5 mM sodium
metabisulfite, 2.5 mM L-ascorbic acid, 10 mM KF, 2 mM
dithiothreitol, 2 mM â-cyclodextrin hydrate, and 1 mM MgCl2
at pH 8.4. Substrate analogue 10 (96 mg) was added, and the
mixture was shaken gently for 16-24 h at room temperature.
The reaction was extracted with 2 × 600 mL of hexane (HPLC
grade); the organic layer was dried with MgSO4 and evapo-
rated. The residue was redissolved in 1 mL of HPLC hexane
and was loaded onto a 10 mL silica column. The column was
eluted with 20 mL of HPLC hexane to give 1.6 mg (5.7 µmol,
6%) of crude product after evaporation of solvent. The major
product 37 was isolated by preparative GC; see Table 2 for 1H
and 13C NMR: GCEIMS m/z 274.
support the proposals that the isocembrenyl cation is
possibly an intermediate in the cyclization of GGPP to
taxadiene.
Experimental Section
10,11R-Dihydrogeranylgeranyl Bromide (20). To a
solution of 19 (0.62 g, 2.1 mmol) in Et2O (21 mL) at 0 °C was
added PBr3 (0.7 mmol, 68 µL) using a plastic delivery pipet.
The mixture was warmed to room temperature and stirred for
30 min, and additional (68 µL) PBr3 was added if TLC showed
any 19 that remained unreacted. The reaction was diluted with
Et2O (21 mL) and quenched with saturated aqueous NaCl (42
mL). The organic layer was separated, dried with MgSO4,
filtered, transferred into a cold dry flask, and concentrated in
vacuo to give 10,11R-dihydroGGBr 20 (0.63 g, 1.8 mmol, 85%),
which was used immediately in the next step.
10,11R-Dihydrogeranylgeranyl Diphosphate, 10. To
bromide 20 (0.64 g, 1.8 mmol) was added anhydrous CH3CN
(10 mL), followed by (Bu4N)3HOPP (3.25 g, 3.6 mmol). The
reaction was stirred at room temperature for 2 h and concen-
trated using a rotary evaporator (bath temperature e 40 °C).
The clear syrup obtained was dissolved in minimal amount of
solvent A [1:49 (v/v) 2-propanol/25 mM aq NH4HCO3], loaded
onto a column containing 80 mL of Sigma Dowex 50WX8-200
resin (pretreated with concd aq NH4OH,32 washed with excess
distilled water, and then preequilibrated with 60 mL of solvent
A) and eluted with 160 mL of solvent A. Typically, the product
would elute immediately, and fractions containing the product
were cloudy and/or yellow colored, as followed by a silica gel
TLC with anisaldehyde visualization. The desired fractions
were pooled and concentrated in vacuo using a rotary evapora-
tor (e40 °C), followed by lyophilization to dryness to give either
a gum or a thick liquid.
Solid-liquid extraction of the crude product was performed
by following the literature procedure.25b The lyophilized prod-
uct was dissolved in a minimal amount of 0.1 M aq NH4HCO3
and treated with solvent B [1:1 (v/v) 2-propanol/acetonitrile].
The mixture was centrifuged at 11000g for 10 min, and the
clear supernatant was collected and saved. After two identical
treatments, the supernatants were combined and concentrated
in vacuo using a rotary evaporator (bath temperature e 40
°C) to give a thick yellow liquid, which was either stored frozen
at -20 °C or used directly in the next chromatography step.
1H and 13C NMR in D2O at this stage showed significant
presence of the Bu4N group, indicating that the initial cation-
exchange step was not completely successful.32 As such,
cellulose chromatography was not attempted, and the product
was purified by silica chromatography instead.33
The thick yellow liquid from the solid-liquid extraction was
dissolved in a minimal amount of 0.1 M NH4HCO3 and loaded
Acknowledgment. We thank the National Insti-
tutes of Health, MERIT Award DK32034, the Robert
A. Welch Foundation, and the Texas Advanced Technol-
ogy and Research Program (TATRP) for financial sup-
port. We also thank Dr. Charles Roessner for advice on
gene expression and Dr. Jim Pennington for helpful
discussions.
(31) (a) Cane, D. E.; Pawlak, J. L.; Horak, R. M.; Hohn, T. M.
Biochemistry 1990, 29, 5476-5490. (b) Croteau, R.; Felton, M.; Ronald,
R. C. Arch. Biochem. Biophys. 1980, 200, 524-533. (c) Cane, D. E.;
Bryant, C. J. Am. Chem. Soc. 1994, 116, 12063-12064. (d) Croteau,
R.; Alonso, W. R.; Koepp, A. E.; Shim, J.-H.; Cane, D. E. Arch. Biochem.
Biophys. 1995, 317, 149-155.
(32) We initially presumed that used resin (Bu4N form) would be
regenerated to the NH4 form by treatement with concd NH4OH. A
closer inspection of the literature (ref 25b) suggested that the resin
(Bu4N form) should first be converted to the H+ form before conversion
to the NH4 form.
Supporting Information Available: Complete reference
information, synthesis procedures, and characterization data
for compounds not shown in the Experimental Section. This
material is available free of charge via the Internet at
http://pubs.acs.org.
(33) Purification of isoprenoid diphosphates by silica chromatogra-
phy has been previously used by others; see: Ohnuma, S.; Ito, M.;
Koyama, T.; Ogura, K. Tetrahedron 1989, 45, 6145-60. Keller, R. K.;
Thompson, R. J. Chromatogr. 1993, 645, 161-167.
JO0517489
J. Org. Chem, Vol. 70, No. 24, 2005 10003