Aristolochene synthase: Mechanistic analysis of active site residues by site-directed mutagenesis

Article abstract of DOI:10.1021/ja0499593

Incubation of farnesyl diphosphate (1) with Penicillium roqueforti aristolochene synthase yielded (+)-aristolochene (4), accompanied by minor quantities of the proposed intermediate (S)-(-)germacrene A (2) and the side-product (-)-valencene (5) in a 94:4:2 ratio. By contrast, the closely related aristolochene synthase from Aspergillus terreus cyclized farnesyl diphosphate only to (+)-aristolochene (4). Site-directed mutagenesis of amino acid residues in two highly conserved Mg²⁺-binding domains led in most cases to reductions in both k_cat and k_cat/K _m as well as increases in the proportion of (S)-(-)germacrene A (2), with the E252Q mutant of the P. roqueforti aristolochene synthase producing only (-)-2. The P. roqueforti D115N, N244L, and S248A/E252D mutants were inactive, as was the A. terreus mutant E227Q. The P. roqueforti mutant Y92F displayed a 100-fold reduction in k_cat that was offset by a 50-fold decrease in K_m, resulting in a relatively minor 2-fold decrease in catalytic efficiency, k_cat/K_m. The finding that Y92F produced (+)-aristolochene (4) as 81% of the product, accompanied by 7% 5 and 12% 2, rules out Tyr-92 as the active site Lewis acid that is responsible for protonation of the germacrene A intermediate in the formation of aristolochene (4).

Full text of DOI:10.1021/ja0499593

Published on Web 05/21/2004
Aristolochene Synthase: Mechanistic Analysis of Active Site
Residues by Site-Directed Mutagenesis
Brunella Felicetti and David E. Cane*
Contribution from the Department of Chemistry, Box H, Brown UniVersity,
ProVidence, Rhode Island 02912-9108
Received January 3, 2004; E-mail: David_Cane@brown.edu
Abstract: Incubation of farnesyl diphosphate (1) with Penicillium roqueforti aristolochene synthase yielded
(+)-aristolochene (4), accompanied by minor quantities of the proposed intermediate (S)-(-)germacrene
A (2) and the side-product (-)-valencene (5) in a 94:4:2 ratio. By contrast, the closely related aristolochene
synthase from Aspergillus terreus cyclized farnesyl diphosphate only to (+)-aristolochene (4). Site-directed
mutagenesis of amino acid residues in two highly conserved Mg2+-binding domains led in most cases to
reductions in both kcat and kcat/Km as well as increases in the proportion of (S)-(-)germacrene A (2), with
the E252Q mutant of the P. roqueforti aristolochene synthase producing only (-)-2. The P. roqueforti D115N,
N244L, and S248A/E252D mutants were inactive, as was the A. terreus mutant E227Q. The P. roqueforti
mutant Y92F displayed a 100-fold reduction in kcat that was offset by a 50-fold decrease in Km, resulting in
a relatively minor 2-fold decrease in catalytic efficiency, kcat/Km. The finding that Y92F produced
(+)-aristolochene (4) as 81% of the product, accompanied by 7% 5 and 12% 2, rules out Tyr-92 as the
active site Lewis acid that is responsible for protonation of the germacrene A intermediate in the formation
of aristolochene (4).
Sesquiterpenes are among the most widely occurring and
structurally diverse families of natural products. Compounds
within this class display a broad range of physiological
properties, including antibiotic, antitumor, phytotoxic and anti-
fungal activities.1 Sesquiterpenoids are produced by both marine
and terrestrial plants as well as by numerous microorganisms,
including fungi and actinomycetes.2 More than 300 distinct
sesquiterpene carbon skeletons have been identified to date and
thousands of naturally occurring oxidized or otherwise modified
derivatives have been isolated. Remarkably all sesquiterpenes
are formed by cyclization of a common acyclic precursor,
farnesyl diphosphate (FPP, 1),3 in a reaction catalyzed by
enzymes known as sesquiterpene cyclases.4 Each cyclase utilizes
a common mechanism involving binding and ionization of the
allylic diphosphate ester FPP (1) followed by a precise sequence
of intramolecular electrophilic addition reactions and rearrange-
ments. The diversity in product structure and stereochemistry
can be explained in large part by the precise folding of the
substrate FPP (1) at the sesquiterpene synthase active site.4
(+)-Aristolochene synthase is a fungal cyclase that catalyzes
the divalent metal ion-dependent cyclization of farnesyl diphos-
phate (1) to the eremophilane sesquiterpene hydrocarbon (+)-
aristolochene (4). Two distinct but closely related aristolochene
synthases have been isolated to date, one from Aspergillus
terreus5 and the other Penicillium roqueforti.6 On the basis of
structural considerations (+)-aristolochene (4) is considered the
likely parent hydrocarbon of several eremophilene toxins and
bioregulators produced by a variety of filamentous fungi,
including sporogen-AO 1,7a phomenone,7b and bipolaroxin.7c
(+)-Aristolochene (4) is believed to serve as the precursor of
PR-toxin, a mycotoxin produced by P. roqueforti.7d The
enantiomer (-)-4 has been isolated from a variety of plant
sources, including Aristolochia indica8a and Bixa orellana,8b as
well as from the defensive secretions of Syntermes soldier
termites.8c
Aristolochene synthase from P. roqueforti, has been purified
to homogeneity,6 the structural gene has been cloned,9 and the
protein overexpressed in Escherichia coli.10 The recombinant
(4) (a) Cane, D. E. Acc. Chem. Res. 1985, 18, 220-226. (b) Cane, D. E. Chem.
ReV. 1990, 90, 1089-1103. (c) Cane, D. E. In Isoprenoids Including
Carotenoid and Steroids; Cane, D. E., Ed. (Volume 2 in ComprehensiVe
Natural Products Chemistry; Barton, D, Nakanishi, K., Meth-Cohn, O.,
Eds.); Pergamon Press: Oxford, 1999; pp 155-215.
(5) Cane, D. E.; Kang, I. Arch. Biochem. Biophys. 2000, 376, 354-364.
(6) Hohn, T. M.; Plattner, R. D. Arch. Biochem. Biophys. 1989, 272, 137-
143.
(7) (a) Tanaka, S.; Wada, K.; Marumo, S.; Hattori, H. Tetrahedron Lett. 1984,
25, 5907-5910. (b) Riche, C.; Pascard-Billy, C.; Devys, M.; Gaudemer,
A.; Barbier, M.; Bousquet, J. F. Tetrahedron Lett. 1974, 32, 2765-2766.
(c) Sugawara, F.; Strobel, G.; Fisher, L. E.; Van Duyne, G. D.; Clardy, J.
Proc. Natl. Acad. Sci. U.S.A. 1985, 82, 8291-8294. (d) Moreau, S.; Biguet,
J.; Lablache-Combier, A.; Baert, F.; Foulon, M.; Delfosse, C. Tetrahedron
1980, 36, 2989-2997.
(8) (a) Govindachari, T. R.; Mohamed, P. A.; Parthasarathy, P. C. Tetrahedron,
1970, 26, 615. (b) Lawrence, B. M.; Hogg, J. W. Phytochem. 1973, 12,
2995. (c) Baker, R.; Cole, H. R.; Edwards, M.; Evans, D. A.; Howse, P.
E.; Walmsley, S. J. Chem. Ecol. 1981, 7, 135.
(1) Lee, K.-H.; Hall, I. H.; Mar, E.-C.; Starnes, C. O.; El-Gebaly, S.; Waddell,
T. G.; Hadgraft, R. I.; Ruffner, C. G.; Weidner, I. Science 1977, 196, 533-
536. Anke, H. A.; Sterner, O. Planta Med. 1991, 57, 344-346. Habtemar-
iam, S.; Gray, A. I.; Waterman, P. G. J. Nat. Prod. 1993, 56, 140-143.
Kubo, I.; Himejima, M. Experientia 1992, 48, 1162-1164.
(2) Devon, T. K.; Scott, A. I. Handbook of Naturally Occurring Compounds;
Academic Press: New York, 1972. Connolly, J. D.; Hill, R. A. Dictionary
of Terpenoids; Chapman & Hall: London, 1992. Glasby, J. S. Encyclopedia
of Terpenoids; Wiley: Chichester, 1982.
(9) Proctor, R. H.; Hohn, T. M. J. Biol. Chem. 1993, 268, 4543-4548.
(10) Cane, D. E.; Wu, Z.; Proctor, R. H.; Hohn, T. M. Arch. Biochem. Biophys.
1993, 304, 415-419.
(3) Ruzicka, L. Experientia 1953, 9, 357-396.
9
7212
J. AM. CHEM. SOC. 2004, 126, 7212-7221
10.1021/ja0499593 CCC: $27.50 © 2004 American Chemical Society

Aristolochene Synthase
A R T I C L E S
Scheme 2. Sesquiterpene Hydrocarbons Isolated from P.
Scheme 1. Aristolochene Synthase Mechanism
Roqueforti
Scheme 3. Thermal and Acid-catalyzed Rearrangement of
Germacrene A
protein has a deduced composition of 342 amino acids corre-
sponding to a MD 39,191 Da. The A. terreus aristolochene
synthase has also been purified to homogeneity and the
corresponding structural gene has been cloned and overexpressed
in E. coli.5 The A. terreus protein has a deduced composition
of 320 amino acids and is a monomer of MD 36 480 Da. The P.
roqueforti and A. terreus cyclases have 66% identity at the
nucleic acid level and a 70% identity at the deduced amino acid
level. Although neither of the two genes shows any significant
overall sequence similarity to any protein in the protein- and
DNA-databases, including known terpene synthases of both
microbial and plant origin5 the structure of P. roqueforti
aristolochene synthase11 shares a high level of structural
homology with all other reported microbial and plant sesqui-
terpene synthases.12
intermediate in the cyclization of FPP (1) to a variety of bicyclic
sesquiterpenes.14 On the other hand, the formation of (S)-(-)-
germacrene A (2) by aristolochene synthase had never previ-
ously been directly observed, although investigations with both
mechanism based inhibitors13c and substrate analogues13d have
provided substantial circumstantial evidence for its intermediacy
in the formation of aristolochene. Interestingly, it has very
recently been reported that P. roqueforti produces a set of
sesquiterpenes of which the most abundant are (+)-aristolochene
(4), valencene (5), R-selinene (6) and â-elemene (7), all of
unspecified configuration (Scheme 2).15a Sporulated surface
cultures of P. roqueforti also have been reported to produce
(+)-aristolochene (4), valencene (5) and â-elemene (7) as the
main volatile metabolites.15b Since it is well-known that
â-elemene (7) and R-selinene (6) are readily formed from
germacrene A (2) by thermal or acid-catalyzed rearrangement,
respectively (Scheme 3),16,17 detection of these sesquiterpenes
in cultures of P. roqueforti producing (+)-aristolochene (4) is
plausible but circumstantial evidence for the presence of
germacrene A (2), itself possibly produced by aristolochene
synthase. In like manner, valencene (5) could conceivably be
formed by alternative deprotonation at C-6 of the penultimate
Extensive mechanistic investigations using stereospecifically
labeled derivatives of FPP (1) have provided strong support for
an aristolochene synthase cyclization mechanism initiated by
ionization of the allylic diphosphate ester to the corresponding
allylic cation-pyrophosphate ion pair, followed by electrophilic
attack at C-10 of the distal double bond and removal of a proton
from the cis-C-12 methyl group, resulting in formation of the
intermediate (S)-germacrene A (2) (Scheme 1).13 The cyclization
has been shown to take place with inversion of configuration
at C-1 of FPP, as expected for a direct displacement mech-
anism.13a Once formed, the intermediate germacrene A (2) would
be further cyclized by protonation at C-6 and intramolecular
attack of the resultant carbocation on the C-2,3 bond to form
the eudesmane cation (3). Successive 1,2-methyl migration and
hydride shift followed by stereospecific deprotonation of H-8si
(FPP numbering) will then result in the formation of (+)-
aristolochene (4).13b The observed sequence of anti-migration
and syn-deprotonation is readily explained by invoking a chair-
boat conformation for the cyclizing FPP and intermediate
germacrene A.
(14) (a) Whitehead, I. M.; Atkinson, A. L.; Threlfall, D. R. Planta 1990, 182,
81-88. (b) Back, K.; Chappell, J. J. Biol. Chem. 1995, 270, 7375-7381.
(c) Rising, A. K.; Starks, C. M.; Noel, J. P.; Chappell, J. J. Am. Chem.
Soc. 2000, 122, 1861-1866. (d) Munck, S. L.; Croteau, R. Arch. Biochem.
Biophys. 1990, 282, 58-64. (e) Steele, C. L.; Crock, J.; Bohlmann, J.;
Croteau, R. J. Biol. Chem. 1998, 273, 2078-2089.
(15) (a) Jelen, H. H. J. Agric. Food Chem. 2002, 50, 6569-6574. (b)
Demyttenaere, J. C.; Adams, A.; Van Belleghem, K.; De Kimpe, N.; Konig,
W. A.; Tkachev, A. V. Phytochemistry 2002, 59, 597-602.
(16) (a) de Kraker, J. W.; Franssen, M. C.; de Groot, A.; Konig, W. A.;
Bouwmeester, H. J. Plant Physiol. 1998, 117, 1381-1392. (b) Weinheimer,
A. J.; Youngblood, W. W.; Washecheck, P. H.; Harns, T. K. B.; Ciereszko,
L. S. Tetrahedron Lett. 1970, 7, 497-500. (c) Bowers, W. S.; Nishino, C.;
Montgomery, M. E.; Nault, L. R.; Nielson, M. W. Science 1977, 196, 680-
681.
(17) Teisseire, P. J. In Chemistry of Fragrant Substances; VCH Publishers
Inc.: New York, 1994; pp 193-289. de Kraker, J. W. The Biosynthesis of
Sesquiterpene Lactones in Chicory (Cichorium intybus L.) Roots, Doctoral
thesis, Department of Organic Chemistry; Wageningen University: Wagenin-
gen, 2002; pp 41-62.
The proposed intermediate germacrene A is itself a commonly
occurring sesquiterpene that has also been proposed as an
(11) Caruthers, J. M.; Kang, I.; Rynkiewicz, M. J.; Cane, D. E.; Christianson,
D. W. J. Biol. Chem. 2000, 275, 25533-25539.
(12) (a) Lesburg, C. A.; Zhai, G.; Cane, D. E.; Christianson, D. W. Science
1997, 277, 1820-1824. (b) Rynkiewicz, M. J.; Cane, D. E.; Christianson,
D. W. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 13 543-13 548. (c)
Rynkiewicz, M. J.; Cane, D. E.; Christianson, D. W. Biochemistry 2002,
41, 1732-1741. (d) Starks, C. M.; Back, K.; Chappell, J.; Noel, J. P. Science
1997, 277, 1815-1820.
(13) (a) Cane, D. E.; Prabhakaran, P. C.; Salaski, E. J.; Harrison, P. H.; Noguchi,
N.; Rawlings, B. J. J. Am. Chem. Soc. 1989, 111, 8914-8916. (b) Cane,
D. E.; Prabhakaran, P. C.; Oliver, J. S.; McIlwaine, D. B. J. Am. Chem.
Soc. 1990, 112, 3209-3210. (c) Cane, D. E.; Bryant, C. J. Am. Chem.
Soc. 1994, 116, 12 063-12 064. (d) Cane, D. E.; Tsantrizos, Y. S. J. Am.
Chem. Soc. 1996, 118, 10 037-10 040.
9
J. AM. CHEM. SOC. VOL. 126, NO. 23, 2004 7213

A R T I C L E S
Felicetti and Cane
Scheme 4. Formation of (+)-aristolochene (4) and (-)-valencene
(5) by Alternative Deprotonation of the Common Intermediate 8
eremophilene cation (8) that normally yields (+)-aristolochene
(4) (Scheme 4).
Important insights have come from analysis of the 2.5 Å
crystal structure of recombinant P. roqueforti aristolochene
synthase11 which displays the all-helical class I R-helical
terpenoid cyclase fold12 in which 6 R-helices out of a total of
11 total surround a large, conical active site cleft approximately
15 Å wide by 20 Å deep (Figure 1). Remarkably, despite the
absence of any significant amino acid sequence similarity,
homologous structures are found in three additional sesqui-
terpene synthases: pentalenene synthase from the gram-positive
bacterium Streptomyces UC5319,12a trichodiene synthase from
the filamentous fungus Fusarium sporotrichioides,12b,c and epi-
aristolochene synthase from the tobacco plant Nicotiana
tabacum.12d Interestingly, the same terpenoid cyclase fold is also
observed for avian farnesyl diphosphate synthase18a and for
human squalene synthase.18b In all of these proteins, the sides
and bottom of the active site cavity are predominately hydro-
phobic and contain a high proportion of aromatic residues. These
amino acids confer an overall shape to the cavity, help to
sequester the substrate and isolate it from the solvent, and define
a unique contour that forms a template for the catalytically
productive conformations of the substrate and intermediates. The
upper region of the active site is usually somewhat hydrophilic
in nature and contains polar and charged residues. Near the edge
of the active site cavity of P. roqueforti arisolochene synthase
is the universally conserved aspartate-rich motif4c,18a,19 D115DVLE
and on the opposite wall is the other highly conserved amino
acid sequence5,12b,c V242VNDIYSYDKE, with the residues of
the magnesium-binding triad indicated in bold (Figure 1). The
crystal structures of N. tabacum epi-aristolochene synthase with
a bound FPP analogue12d and of F. sporotrichioides trichodiene
synthase complexed with the coproduct inorganic pyrophos-
phate12b have provided strong evidence that these two highly
conserved motifs are together responsible for the binding of
FPP through the chelation of three Mg2+ ions that in turn are
complexed with the pyrophosphate moiety of the substrate. In
each structure, two of these Mg2+ ions are liganded either
directly or indirectly by two or more carboxylates of the
aspartate-rich motif, whereas the third Mg2+ ion, which is
complexed to the opposite face of the pyrophosphate moiety,
Figure 1. Active site of aristolochene synthase with model of bound FPP,
showing key active site residues (ref 11).
is itself held in place by coordination to the side chains of the
Asp/Asn, Ser, and Glu residues of the Mg2+-binding triad. The
pyrophosphate moiety of FPP (1) also forms hydrogen-bonds
with basic Arg and Lys residues present at the top edge of the
active site. The importance of the two conserved Mg2+-binding
motifs has been previously been probed by site-directed
mutagenesis of several sesquiterpene synthases, including tri-
chodiene synthase,19d pentalenene synthase,20 δ-selinene syn-
thase,21 and γ-humulene synthase.21
We now describe site-directed mutagenesis experiments on
aristolochene synthase from both P. roqueforti and A. terreus
that substantiate the importance of the two conserved Mg2+
-
binding motifs in each protein, provide direct evidence for the
formation of the proposed intermediate (S)-(-)-germacrene A
by both wild-type and mutant aristolochene synthases, and rule
out the previously suggested role of Tyr-92 as an active site
acid in the proton-initiated cyclization of the intermediate
germacrene A.
(18) (a) Tarshis, L. C.; Yan, M. J.; Poulter, C. D.; Sacchettini, J. C. Biochemistry
1994, 33, 10 871-10 877. (b) Pandit, J.; Danley, D. E.; Schulte, G. K.;
Mazzalupo, S.; Pauly, T. A.; Hayward, C. M.; Hamanaka, E. S.; Thompson,
J. F.; Harwood, H. J. J. Biol. Chem. 2000, 275, 30 610-30 617.
Results
(19) (a) Croteau, R. B. In Isoprenoids Including Carotenoids and Steroids; Cane,
D. E., Ed. (Volume 2 in ComprehensiVe Natural Products Chemistry;
Barton, D, Nakanishi, K., Meth-Cohn, O., Eds.); Pergamon Press: Oxford,
1999; pp 97-153. (b) Tarshis, L. C.; Proteau, P. J.; Kellogg, B. A.;
Sacchettini, J. C.; Poulter, C. D. Proc. Natl. Acad. Sci. U.S.A. 1996, 93,
15 018-15 203. (c) Ashby, M. N.; Edwards, P. A. J. Biol. Chem. 1990,
265, 13 157-13 164. (d) Cane, D. E.; Xue, Q.; Fitzsimons, B. C.
Biochemistry 1996, 35, 12 369-12 376. (e) Marrero, P. F.; Poulter, C. D.;
Edwards, P. A. J. Biol. Chem. 1992, 267, 21 873-21 878. (e) Song, L.;
Poulter, C. D. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 3044-3048. (f) Joly,
A.; Edwards, P. A. J. Biol. Chem. 1993, 268, 26 983-26 989. (g) Peters,
R. J.; Croteau, R. B. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 580-584.
Recombinant wild-type aristolochene synthase from both P.
roqueforti and A. terreus were each expressed and purified as
previously described.5,10 The kcat (0.043 s-1) and Km (0.6 µM)
(20) Seemann, M.; Zhai, G.; de Kraker, J. W.; Paschall, C. M.; Christianson,
D. W.; Cane, D. E. J. Am. Chem. Soc. 2002, 124, 7681-7689.
(21) Little, D. B.; Croteau, R. B. Arch. Biochem. Biophys. 2002, 402, 120-
135.
9
7214 J. AM. CHEM. SOC. VOL. 126, NO. 23, 2004

Aristolochene Synthase
A R T I C L E S
2) and (+)-â-elemene ((+)-7) derived from the sample of 2
produced by aristolochene synthase (Scheme 3) had identical
capillary GC-retention times on an Optima 1701 column and
indistinguishable mass spectra. On the other hand the GC
retention time (48.34 min) of (+)-â-elemene ((+)-7) derived
from (S)-(-)-germacrene A ((-)-2) produced by aristolochene
synthase clearly differed from that of (-)-â-elemene ((-)-7)
(48.60 min) when the two compounds were analyzed by chiral
capillary GC on FS-hydrodex-â-6TBDM.
Aspartate-Rich Domain. To investigate the functional role
of the aspartate-rich domain in P. roqueforti aristochene
synthase, appropriate D115E, D115N, D116E, D116N, E119D,
and E119Q mutants were constructed by site-directed mutagen-
esis and expressed in E. coli BL21(DE3). Each recombinant
mutant protein was purified to homogeneity. The P. roqueforti
aristolochene synthase D115N mutant was completely inactive
within the detection limits of the assay, which could have
measured a reduction in kcat of as much as a factor of 104. The
steady-state kinetic parameters kcat and Km were determined for
each purified mutant, and the product mixtures generated by
each active mutant cyclase were analyzed by GC-MS (Table
2). In contrast to the lack of activity for the D115N mutant,
replacement of Asp-115 with Glu (D115E) gave a mutant that
retained substantial aristolochene synthase activity, displaying
only a ∼3-fold decrease in kcat as well as a 4.5-fold increase in
Km, a net 12-fold decrease in catalytic efficiency, kcat/Km,
compared to the wild-type enzyme. Replacement of Asp-115
with a glutamate also resulted in an increase in the proportion
of both side products, (S)-(-)-germacrene A (2) and (-)-
valencene (5), to 19% and 6%, respectively, of the total product
mixture (Table 2).
Figure 2. Concentration dependence of aristolochene synthase activity.
Incubation for 10 min at variable protein concentration. (A) Wild-type P.
roqueforti aristolochene synthase. (B) P. roqueforti Y92F mutant.
of the P. roqueforti cyclase were in good agreement with those
previously reported5 as were the kcat (0.0173 s-1) and Km (0.13
µM) of the A. terreus enzyme.10,22
GC-MS analysis of the pentane extract of an incubation of
FPP (1) with A. terreus aristolochene synthase showed a single
product corresponding to (+)-aristolochene (4) (Figure 3). By
contrast, the P. roqueforti enzyme produced two sesquiterpene
hydrocarbons of m/z 204 in addition to the expected (+)-
aristolochene (4), in relative proportions of 94 (4): 2 (5): 4 (2)
(Figure 4). (-)-Valencene (5) was identified by direct GC-MS
comparison with authentic (+)-valencene using a nonchiral
stationary GC phase. When analyzed by chiral capillary GC,
the enzymatically generated (-)-valencene (5) had a retention
time of 51.04 min compared to a retention time of 51.84 min
for (+)-valencene. (S)-(-)-Germacrene A (2) was first identified
by capillary GC-MS comparison with a sample of (R)-(+)-
germacrene A that was generated from FPP (1) by the W308F
mutant of pentalenene synthase.20 The absolute configuration
of the enzymatically generated (S)-(-)-germacrene A (2) was
established as previously described16a by means of its facile
thermal Cope rearrangement to (+)-â-elemene (7) (Scheme
3).16b Using an injection port temperature of 250 °C, (-)-â-
elemene ((-)-7) generated from (R)-(+)-germacrene A ((+)-
The D116E mutant showed a 30-fold decrease in kcat but only
a 2-fold increase in Km, resulting in a net 60-fold decrease in
catalytic efficiency, while showing an 8-fold increase in the
proportion of (S)-(-)-germacrene A (2) compared to the wild-
type enzyme. Similarly the D116N mutant displayed a 20-fold
reduction in kcat and 3-fold increase in Km, with a product
distribution that was essentially identical to that obtained from
the D116E mutant. By contrast, the E119D mutant did not
display significant changes in either steady-state kinetic param-
eters (4-fold decrease in kcat and 2-fold decrease in Km) or in
product distribution compared to the wild-type (Table 2).
Although the replacement of Glu-119 by glutamine (E119Q)
caused a 10-fold decrease in kcat and a 4-fold decrease in Km,
with little net change in catalytic efficiency, the proportion of
germacrene A (2) generated by the E119Q mutant increased to
14% of the total mixture.
Mg2+-Binding Triad. To probe the role of the three putative
Mg2+-binding residues12 in the conserved V242VNDIYSYDKE
domain of P. roqueforti aristolochene synthase, the N244L,
N244D, S248A, E252D, and E252Q mutants, as well as the
S248A/E252Q double mutant were each expressed in E. coli
BL21(DE3) and purified to homogeneity. Both the N244L
mutant and the double mutant S248A/E252D were completely
inactive, within the detection limits of the assay. Aristolochene
synthase activity was retained, however, when Asn-244 was
replaced with aspartate (N244D), albeit with a kcat that was
reduced 350-fold compared to the wild-type and a Km that was
6-fold higher, corresponding to a nearly 2000-fold decrease in
kcat/Km (Table 2). The N244D mutant also exhibited a dramatic
(22) As previously reported, the activity of P. roqueforti aristolochene synthase
is a simple linear function of protein concentration only within a specific
concentration range.10 The activity of wild-type aristolochene synthase was
linearly proportional to protein concentration only up to 20 nM, after which
the measured activity per mg protein decreased, with negligible increase
in total activity above 60 nM protein (Figure 2A, Table 1). Similar behavior
was also observed for the A. terreus aristolochene synthase, which displayed
nonlinear dependence on protein concentration above 27 nM protein. To
allow meaningful comparisons of steady state kinetic parameters, therefore,
we first determined the concentration dependence of aristolochene synthase
activity for all mutants prepared in this study. (Table 1) Kinetic assays of
each mutant were then carried out only at protein concentrations within
the experimentally established linear range.
9
J. AM. CHEM. SOC. VOL. 126, NO. 23, 2004 7215

A R T I C L E S
Felicetti and Cane
Figure 3. (A) GC-MS total ion chromatogram of A. terreus wild-type aristolochene synthase product. (B) Mass spectrum of (+)-aristolochene (4)
Figure 4. (A) GC-MS total ion chromatogram of P. roqueforti wild-type aristolochene synthase product mixture. (B) Mass spectrum of enzymatically
generated (+)-aristolochene (4). (C) enzymatically generated (-)-valencene ((-)-5). (D) (+)-valencene ((+)-5). (E) enzymatically generated (S)-(-)-germacrene
A ((-)-2). (F) (R)-(+)-germacrene A ((+)-2).
Table 1. Aristolochene Synthase, Wild-Type and Mutants:
Concentration Range for Linear Dependence of Activity on Protein
E252D and S248A mutants of P. roqueforti aristolochene
Concentration
mixture, no (-)-valencene (5) being detected (Table 2). The
synthase displayed more modest changes in steady-state kinetic
parameters compared to the N244D mutant. Both showed a ca.
concentration
protein
range (nM)
30-fold decrease in kcat and a 300-fold decrease in kcat/Km
compared to wild-type. On the other hand, the resulting product
mixtures were almost identical to that produced by N244D, with
(S)-(-)-germacrene A (2) the major product of the cyclization.
Interestingly, the E252Q mutant displayed a 200-fold reduction
in kcat and a nearly 3500-fold reduction in kcat/Km. Most
importantly, this mutant generated almost exclusively (S)-(-)-
germacrene A (2) from FPP (1), with <0.1% (+)-aristolochene
(4) and no detectable valencene (5) (See Supporting Informa-
tion).
P. roqueforti WTa,b
Y92Fb
2-20
2-50
D115Ec
2-160
2-410
2-510
2-290
2-550
2-510
2-410
2-410
2-520
0.9-27
0.9-410
0.9-410
D116Ec
D116Nd
E119Dd
E119Qd
N244Dd
S248Ae
E252Dc
E252Qd
A. terreus WT
AT-N219D
AT-E227D
Although the structure of the A. terreus aristolochene synthase
has not yet been solved experimentally, based on the high level
of sequence similarity a working structural model was generated
using the Swiss Protein Modeler and the P. roqueforti structure
as a template. (Figure 5) Although such models are tentative,
they can serve as useful guide to structure comparisons and as
a source of further experiments, especially for proteins of high
sequence identity. This model indicates that most of the amino
a WT, wild-type. b 2 µM FPP, specific activity 44 µCi/µmol. c 10 µM
FPP, specific activity 44 µCi/µmol. d 20 µM FPP, specific activity 44 µCi/
µmol. e 4 µM FPP, specific activity 170 µCi/µmol.
shift in product distribution, with (S)-(-)-germacrene A (2)
constituting 81% of the pentane-extractable products and (+)-
aristolochene (4) making up the remaining 19% of the product
9
7216 J. AM. CHEM. SOC. VOL. 126, NO. 23, 2004

Aristolochene Synthase
A R T I C L E S
Table 2. Aristolochene Synthase Mutants: Kinetic Parameters and Product Ratios
proportion of sesquiterpenes (%)
steady-state kinetic parameters
aristolochene
valencene
(5)
germacrene A
protein
k
cat (10-3 s-1
)
Km (µM)
kcat/Km (s-1 M-1
)
(4)
(2)
P. roqueforti WT
D115E
D115N
D116E
D116N
E119D
E119Q
N244L
N244D
S248A
E252D
E252Q
S248A/E252D
Y92F
A. terreus WT
AT-N219D
AT-E227D
AT-E227Q
43 ( 2
0.6 ( 0.1
2.7 ( 0.2
7.2 × 104
6.0 × 103
inactive
1.2 × 103
1.5 × 103
3.0 × 104
3.2 × 104
inactive
36.4
94
75
2
6
4
19
16.2 ( 0.4
1.30 ( 0.05
2.2 ( 0.1
9.7 ( 0.2
4.8 ( 0.1
1.10 ( 0.13
1.5 ( 0.1
0.32 ( 0.02
0.15 ( 0.02
62
63
94
84
3
2
2
2
35
35
4
14
0.120 ( 0.003
1.30 ( 0.04
2.20 ( 0.09
0.22 ( 0.01
3.3 ( 0.2
5.5 ( 0.4
3.4 ( 0.4
10.1 ( 0.7
19
21
19
ND
ND*
ND
81
79
81
2.4 × 102
6.5 × 102
21.8
ND
ND
100
inactive
3.4 × 104
1.3 × 105
19.5
0.40 ( 0.02
17.3 ( 0.7
0.076 ( 0.004
1.6 ( 0.1
0.012 ( 0.002
0.13 ( 0.01
3.9 ( 0.5
81
100
44
7
12
ND
56
ND
ND
ND
1.4 ( 0.2
1.1 × 103
26
74
inactive
*WT, wild type; ND, not detected.
an aspartate or a glutamine (AT-E227D, AT-E227Q). Each
protein was overexpressed and purified as previously described.5
The AT-E227Q mutant was inactive, whereas the AT-E227D
mutant behaved similarly to the corresponding P. roqueforti
E252D mutant, with (S)-(-)-germacrene A (2) being the major
cyclization product. The AT-N219D mutant also was similar to
the N244D mutant of P. roqueforti aristolochene synthase in
kinetic properties, with a ∼200-fold decrease in kcat and a net
∼6600-fold decrease in kcat/Km, and in product distribution,
giving rise to a 56:44 mixture of (S)-(-)-germacrene A (2) and
(+)-aristolochene (4) (Table 2).
Role of Tyr-92. The identity of the Lewis acid species in
aristolochene synthase and mechanistically related enzymes
responsible for protonation of the intermediate germacrene A
(2) has been the object of considerable speculation and
experimentation. Chappell and Noel have reported that replace-
ment of Tyr-520 of 5-epi-aristolochene synthase by a phenyl-
alanine abolishes formation of epi-aristolochene and results in
the exclusive formation of germacrene A by this mutant, leading
them to propose that Tyr-520 might correspond to the Lewis
acid that protonates this intermediate in the cyclization of FPP
to epi-aristolochene.14c Our previous analysis of the active site
of P. roqueforti aristolochene synthase led also us to suggest
that Tyr-92 might be suitably positioned to protonate the
intermediate (S)-(-)-germacrene A.11 To investigate the func-
tional role of this residue, we therefore constructed the Y92F
mutant of P. roqueforti aristolochene synthase by site-directed
mutagenesis. Although the Y92F mutant exhibited a 100-fold
decrease in kcat, this reduced turnover number was offset by a
compensating 50-fold decrease in Km, resulting in substantially
no net change in observed catalytic efficiency, kcat/Km, compared
to the wild-type aristolochene synthase.23 More significantly,
this mutant produced (+)-aristolochene as the major product
(81%), with only a modest increase in the relative proportions
of both (S)-(-)-germacrene A (2) (12%) and (-)-valencene (5)
(7%) compared to the wild-type cyclase.
Figure 5. Comparison of Chain B of P. roqueforti aristolochene synthase
(red) (PDB 1DI1B) with the structure of A. terreus aristolochene synthase
(white) (Genbank Accession Number AF198360), calculated using the Swiss
Modeler Program and displayed using the Swiss Protein Database Viewer
(v 3.7).32 Superposition of the two protein backbones gave a calculated
RMSD for 1152 backbone atoms of 1.22 Å. Residues labeled in red are for
the P. roqueforti protein and those in white are for the A. terreus enzyme.
Selected residues labeled only in red are identical in both structures (although
they have different residue numbers.) E119 is part of the conserved aspartate-
rich motif DDXXE. E252 is part of the conserved triad NDXXSXXXE.
acids surrounding the active site are highly conserved in the
two proteins, in accord with the fact that they both catalyze the
cyclization of FPP (1) to (+)-aristolochene (4). In particular,
the highly conserved aspartate-rich motifs (D90DLLE for the
A. terreus synthase and D115DVLE for the P. roqueforti cyclase)
and the magnesium-binding triads (N219DIYSYEKE for the A.
terreus synthase and N244DIYSYDKE for the P. roqueforti
synthase) are present at analogous positions in each cyclase, at
the upper edges of the active site pocket. We therefore con-
structed a limited set of mutants of the A. terreus aristolochene
synthase, targeting two of the residues of the magnesium-binding
triad N219DIYSYEKE. Accordingly, Asn-219 was replaced by
an aspartate (AT-N219D) and Glu-227 was replaced by either
(23) The activity of the Y92F mutant was found to be a simple linear function
of protein concentration only up to 50 nM (Figure 2B, Table 1), with only
minor increases in total cyclase activity above 80 nM mutant protein.
9
J. AM. CHEM. SOC. VOL. 126, NO. 23, 2004 7217

A R T I C L E S
Felicetti and Cane
synthase release a greater proportion of the cyclization inter-
mediate (S)-(-)-germacrene A (2) (14%-35%) compared to the
wild-type (4%). By contrast, the fraction of the alternative
deprotonation product, (-)-valencene (5) is generally un-
changed, increasing from 2% to 6% only for the D115E mu-
tant. More dramatic shifts in product distribution were observed
for mutations in the residues of the Mg2+-binding triads,
N244DIYSYDKE and N219DIYSYEKE, of the P. roqueforti and
A. terreus synthases, respectively, with the proportion of (-)-
germacrene A (2) increasing to 55%-75% for the A. terreus
N219D and E227D mutants, 80% for the three P. roqueforti
N244D, S248A, and E252D mutants, and reaching 100% of
total sesquiterpene for the P. roqueforti E252Q mutant. These
increases in the percentage of prematurely released germacrene
A (2) are associated, however, with substantial kinetic penalties,
with degradations in the observed kcat/Km by factors from 100
to as much as 6500.
Figure 6. Product distribution of P. roqueforti aristolochene synthase
mutants.
Discussion
Formation of (-)-Germacrene A (2) and (-)-Valencene
(5). Cultures of P. roqueforti have been reported to accumulate,
in addition to aristolochene, small quantities of valencene, of
unspecified configuration, and a variety of sesquiterpenes readily
derived from germacrene A.15 Allemann has recently reported
that wild-type recombinant P. roqueforti aristolochene synthase
generates small quantities of both valencene (5) and germacrene
(2), again of unspecified configuration.24 We have now inde-
pendently established that P. roqueforti aristolochene synthase
converts FPP to a 94:2:4 mixture of (+)-aristolochene (4), (-)-
valencene (5), and (S)-(-)-germacrene A (2) and rigorously
determined the absolute configuration of each product. Interest-
ingly, the very closely related A. terreus enzyme generates (+)-
aristolochene as the sole sesquiterpene hydrocarbon product.
Although the formation of product mixtures is unusual for
microbial terpene synthases, it is very common for plant terpene
synthases. Two particularly striking examples are those of
γ-humulene synthase and δ-selinene synthase, which cyclize
FPP to 52 and 34 different sesquiterpenes, respectively.14e
The absolute configuration of the (S)-(-)-germacrene A (2)
produced by P. roqueforti aristolochene synthase is consistent
with the well-supported absolute sense of folding of FPP (1) in
the formation of (+)-aristolochene (4)4b,13a,b and therefore
supports the intermediacy of 2 previously inferred from experi-
ments with substrate analogues13d and mechanism-based
inhibitors.13c The observed formation of a small proportion of
(S)-(-)-germacrene A (2) from FPP is apparently due to its
premature release from the active site before it can be repro-
tonated and further converted to the eudesmane cation. The
absolute configuration of the enzymatically generated (-)-
valencene ((-)-5) is also consistent with the configuration of
the major product, (+)-aristolochene (4). This side product arises
by deprotonation at C-6 instead of C-8 in the penultimate cation
intermediate 8 (Scheme 4). In principle, the same active site
base could remove a proton from either position, resulting in
the formation of minor amounts of (-)-valencene (5) along with
the favored product, (+)-aristolochene (4).
The formation of aberrant product mixtures resulting from
mutations in the Mg2+-binding domains is believed to be due
to perturbations in the precise binding of the pyrophosphate
moiety of the substrate FPP, with consequent mispositioning
of the farnesyl chain as well as the derived carbocationic and
neutral intermediates. Although there is no direct evidence that
the essential Mg2+ cations are bound by aristolochene synthase
at the two universally conserved Mg2+-binding motifs, the
crystal structure of the SmCl3-derivative of P. roqueforti
synthase shows a single Sm3+ ion bound to Asp-115.11
Consistent with this model are the results of site-directed
mutagenesis experiments with trichodiene synthase, in which
replacement of Asp-100 of the aspartate-rich motif with
glutamate resulted in a 22-fold reduction in observed kcat and
the formation of a mixture of five aberrant cyclization products,
in addition to trichodiene.19d Similar results were also obtained
for the D101E and D104E mutants of trichodiene synthase. The
crystal structure of the D100E mutant of trichodiene synthase
with bound inorganic pyrophosphate showed that only two of
the usual three Mg2+ ions remains bound to the inorganic
pyrophosphate, with one of the two Mg2+-ions normally bound
to the aspartate-rich domain of the wild-type cyclase no longer
present, resulting in a shift in the position of the bound inorganic
pyrophosphate.12c Most importantly the conformational changes
associated with pyrophosphate binding by the wild-type tri-
chodiene synthase were attenuated in the D100E mutant,
resulting in a 12% increase in the included active site volume
and a calculated decrease in the packing density of the substrate
FPP (1) from 78% to 67%.12b,c The concomitant increases in
spatial and conformational degrees of freedom for the mutant-
bound substrate and derived reactive intermediates are thought
to permit premature deprotonation of intermediates, resulting
in formation of the observed aberrant products.
Kinetic Properties of Aristolochene Synthase Mutants.
Four of the aristolochene synthase mutants were completely
inactive, including one with a mutation in the aspartate-rich
domain (D115N) and three involving the three residues of the
Mg2+-binding triad: N244L, the double mutant S248A/E252D
and the A. terreus mutant E227Q. The homologous N219L and
N219A mutants of pentalenene synthase have also been found
to be inactive.20 Analysis of the crystal structure of the N219L
mutant of pentalenene synthase revealed no significant changes
in the shape of the active site or in protein structure compared
Mutations in the two Mg2+-binding domains of each of the
two aristolochene synthases substantially affected the observed
distribution of products (Table 2, Figure 6). The various mutants
in the aspartate-rich region, D115DVLE, of the P. roqueforti
(24) (a) Calvert, M. J.; Ashton, P. R.; Allemann, R. K. J. Am. Chem. Soc. 2002,
124, 11 636-11 641. (b) Deligeorgopoulou, A.; Allemann, R. K. Biochem-
istry 2003, 42, 7741-7747. (c) Calvert, M. J.; Taylor, S. E.; Allemann, R.
K. Chem. Commun. 2002, 2384-2385.
9
7218 J. AM. CHEM. SOC. VOL. 126, NO. 23, 2004

Aristolochene Synthase
A R T I C L E S
to the wild-type enzyme, suggesting that the observed loss of
cyclase activity involves disruption of a key interaction with
the substrate.20 Interestingly, the substitution of an alanine for
Ser-248 of aristolochene synthase, a residue that is believed to
coordinate to the same Mg2+ ion as Asn-244, did not abolish
the aristolochene synthase activity, resulting instead in a 300-
fold reduction in kcat/Km and a substantial increase in the
proportion of germacrene A. Similarly, the replacement of Glu-
252 of the P. roqueforti aristolochene synthase with an aspartate
or a glutamine also did not cause complete loss of activity of
the enzyme. Only the P. roqueforti double mutant, S248A/
E252D, in which two Mg2+ coordination sites were perturbed,
was found to be inactive. Although the corresponding E227D
mutant of the A. terreus enzyme was also found to be inactive,
these apparent differences may simply reflect the lower limits
for the detection of reduced cyclase activity. In general,
mutations in the Mg2+-binding triad appeared to have more
profound effects on both catalytic efficiency and product
distribution than variants in the aspartate-rich motif, with
changes in D119 having the smallest net effect.
protonation of the germacrene A intermediate. Interestingly, the
Y520F mutant of epi-aristolochene synthase was also found to
produce both epi-aristolochene and germacrene A, but the
transition to exclusive production of germacrene A occurs at
pH 7, lower than for the wild-type. The identity of the active
site acids in the cyclization of FPP (1) to both (+)-aristolochene
(4) and epi-aristolochene therefore still remains an open
question. One attractive possibility is that a protonated water
molecule might serve as the requisite Lewis acid but there is as
yet no direct data yet available to support this possibility. The
recent observation that the active site of bornyl diphosphate
synthase contains a bound molecule of water is at least consistent
with this hypothesis.28 The identification of the active site acid
of aristolochene synthase will no doubt require analysis of high-
resolution crystal structures of protein with bound substrate or
intermediates analogues.
Conclusions. Site-directed mutagenesis of aristolochene
synthase from both P. roqueforti and A. terreus has provided
strong support for the proposed intermediacy of (S)-(-)-
germacrene A (2) in the conversion of FPP (1) to (+)-
aristolochene (4). The effect of mutations in the two conserved
Mg2+-binding domains has reinforced earlier observations on
the importance of residues in these two regions in binding and
ionization of the pyrophosphate moiety of the FPP. The fact
that the Tyr92Phe mutant of P. roqueforti aristolochene synthase
produces aristolochene as the major (>80%) product, albeit at
reduced rate, definitively rules out this residue as the active
site Lewis acid directly responsible for the protonation and
further cyclization of the germacrene A intermediate. The
identity of the requisite active site bases and acids responsible
for the various deprotonation and protonation steps in the
cyclization cascade catalyzed by aristolochene synthase and
other terpene cyclases remains to be established and is the
subject of further investigation.
Role of Tyr-92. The observation that (+)-aristolochene (4)
is still the major product (81%) of the cyclization reaction
catalyzed by the Y92F mutant of P. roqueforti aristolochene
synthase, and that it is produced with basically the same kcat/
Km, although at substantially lower kcat, compared to the wild-
type enzyme, definitively rules out Tyr-92 as the active site
Lewis acid responsible for the protonation of the intermediate
germacrene A (2).25 Recent reports from other laboratories have
also shed doubt on the original suggestion that Tyr-520 of epi-
aristolochene synthase serves as the active site acid in the
protonation-cyclization of germacrene A. For example, Tyr-
520 and its hydrogen-bonded Asp-444 and Asp-525 are
conserved in chicory germacrene A synthase, as well as in most
of the other known germacrene A synthases.26 The presence of
these residues therefore cannot alone account for the formation
of either germacrene A or epi-aristolochene as the end-product
of their respective synthases. Furthermore, it has recently been
reported that germacrene A is released from the active site of
wild-type epi-aristolochene synthase when cyclizations are
carried out at elevated pH.27 Indeed, at pH > 8.8, germacrene
A becomes the major product of cyclization of FPP (1) by epi-
aristolochene synthase, indicating that the changes in the
protonation state of the active site acid can suppress the normal
Experimental Section
Materials. Recombinant P. roqueforti A. terreus aristolochene
synthases were obtained from E. coli XL1 Blue/pZWO4, E. coli XL1
Blue/pET11rASa, E. coli BL21(DE3)/pZWO4 and E. coli BL21(DE3)-
pLysS/pET11rASa as previously described.5,10 The QuikChange site-
directed mutagenesis kit was purchased from Stratagene. Mutagenic
and sequencing primers were purchased from Integrated DNA Technol-
ogy, Inc. Competent cells of E. coli BL21(DE3), E. coli BL21(DE3)-
pLysS and E. coli XL1-Blue were purchased from Stratagene. The
QIAprep spin Miniprep plasmid purification kit was purchased from
Qiagen. Culture media were obtained from Sigma. Isopropyl-thio-D-
galactopyranoside (IPTG) was purchased from Gibco and â-mercapto-
ethanol was obtained from Sigma. Pre-swollen DE52 anion-exchange
resin was from Whatman. The pre-packed methyl hydrophobic interac-
tion column (methyl-HIC, 5 mL) was obtained from BioRad. The
Resource Q anion exchange column was purchased from Pharmacia
Biotech. Corp. Protein size marker was purchased from Bio-Rad. The
Amicon-YM30 membranes and Centriprep YM-30 disposable ultra-
filtration units were from Millipore. The Bradford protein assay reagent
and the concentrated electrophoresis buffer were from Bio-Rad.
[1-3H]FPP (16.1 Ci/mmol) was purchased from New England Nuclear
(NEN), DuPont. Unlabeled FPP (4) was synthesized following the
procedure described in the literature.29 Authentic (+)-valencene was
(25) Allemann et al.24 have used the p ZWO4 aristolochene synthase expression
plasmid obtained from our laboratory to generate the corresponding Y92F,
Y92V, Y92C, and Y92A mutants. For the Y92F mutant, Allemann reported
a 13-fold reduction in kcat and a ∼80-fold increase in Km, resulting in a
reported net ∼1100-fold decrease in kcat/Km compared to the wild-type.24a
The reported steady-state kinetic parameters were unfortunately determined
at concentrations well above what turns out to be the maximum 50 nM
protein concentration for linear dependence of Y92F activity (cf Table 1),
thereby compromising their quantitative significance. On the basis of an
observed increase in the proportion of germacrene A (28%), of unspecified
configuration, the implausible suggestion was made that Tyr-92 must be
the normal active site Lewis acid responsible for the cyclization of the
germacrene A intermediate, although the major product isolated was still
aristolochene. These workers also reported the isolation of the coproducts
R-selinene (6), â-selinene, and selina-4,11-diene, none of which are
detectable in our analysis of the same Y92F mutant (Table 2). The latter
compounds are well-known artifacts of the acid-catalyzed rearrangement
of germacrene A.16,17,20 Rearrangement of germacrene A is easily avoided
by chromatography over aluminum oxide instead of silica gel and the use
of sodium sulfate in place of magnesium sulfate as drying agent.
(26) Bouwmeester, H. J.; Kodde, J.; Verstappen, F. W.; Altug, I. G.; de Kraker,
J. W.; Wallaart, T. E. Plant Physiol. 2002, 129, 134-144.
(28) Whittington, D. A.; Wise, M. L.; Urbansky, M.; Coates, R. M.; Croteau,
R. B.; Christianson, D. W. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 15 375-
15 380.
(29) Cane, D. E.; Sohng, J. K.; Lamberson, C. R.; Rudnicki, S. M.; Wu, Z.;
Lloyd, M. D.; Oliver, J. S.; Hubbard, B. R. Biochemistry 1994, 33, 5846-
5857.
(27) O’Maille, P.; Greenhagen, B.; Chappell, J.; Zhao, Y.; Coates, R. M.; Noel,
J. Poster Presentation: TEAS: A Closer Look at the Cyclization Cascade;
Terpnet 2003, Lexington, Kentucky, 2003.
9
J. AM. CHEM. SOC. VOL. 126, NO. 23, 2004 7219

A R T I C L E S
Felicetti and Cane
obtained from Fluka. (R)-(+)-germacrene A was prepared by incubation
of FPP (1) with the W308F mutant of pentalenene synthase (plasmid
pMSO4), as previously described.20 All other reagents and buffer
components used for protein purification and enzyme assay were of
the highest quality commercially available.
expression, as judged by SDS-PAGE, was then used for kinetic analysis
and preparative scale incubations.
P. roqueforti Aristolochene Synthase. Individual strains of E. coli
BL21(DE3) harboring mutant plasmids for P. roqueforti aristolochene
synthase were each grown overnight in 8 mL of LB medium containing
100 µg/mL ampicillin at 37 °C. The resulting culture (8 mL) was used
to inoculate pre-warmed LB medium (800 mL, 100 µg/mL ampicillin)
and the cells were grown at 30 °C to an OD600 of 1.0. Protein expression
was then induced with 1 mM IPTG. Following 6 h of incubation at 30
°C (250 rpm) the cells were harvested by centrifugation (6000 × g, 25
min, 4 °C). The cell pellet was resuspended in 50 mL of cell lysis
buffer (5 mM EDTA, 5 mM â-mercaptoethanol, 20 mM Tris pH 7.5)
and recentrifuged (6000 × g, 25 min, 4 °C) to remove any remaining
LB medium. The cells were then suspended in 50 mL of lysis buffer
and broken by sonication (three cycles of 10 min each, 30% power
range, 50% duty cycle) in an ice water bath. The pellet was then
resuspended in the same buffer and recentrifuged to remove traces of
soluble protein. The subsequent solubilization and purification of the
mutant proteins were performed as previously described for wild-type
P. roqueforti aristolochene synthase.11 Fractions containing pure enzyme
were identified by SDS-PAGE, combined, and concentrated to a final
volume of 2.5 mL using an Amicon ultrafiltration apparatus equipped
with a YM-30 filter. The final enzyme purity was greater than 90% as
judged by SDS-PAGE analysis. The concentrated protein was dialyzed
against Buffer T (20 mM Tris pH 7.5, 15% glycerol, 5 mM MgCl2, 5
mM â-mercaptoethanol) using a PD-10 desalting column (SephadexTM
G-25, Pharmacia) and the protein was stored at -80 °C.
A. terreus Aristolochene Synthase. Individual strains of E. coli
BL21(DE3)pLysS/pET11rASa and derivatives harboring the wild-type
and mutant plasmids of A. terreus aristolochene synthase were grown
overnight in 10 mL of LB media containing 100 µg/mL carbenicillin
at 37 °C. The overnight seed culture was used to inoculate 1 L of pre-
warmed (30 °C) LB-carbenicillin (100 µg/mL) and the cells were grown
until the OD600 reached 1.0. Protein expression was then induced with
1 mM IPTG. After 6 h of incubation the cells were harvested by
centrifugation (6000 × g, 30 min, 4 °C). The pellet was then
re-suspended in 50 mL of MEG buffer (20 mM Mes pH 6.5, 2 mM
EDTA, 10% glycerol, 5 mM 2-mercaptoethanol) and recentrifuged
(6000 × g, 30 min, 4 °C) to remove any remaining LB medium. The
cells were resuspended in 50 mL of MEG buffer and broken by
sonication (three cycles of 10 min each, 50% duty cycle and 30% power
range) in an ice water bath. The crude cell lysate was centrifuged
(17 000 × g, 30 min, 4 °C) and the supernatant, containing soluble
aristolochene synthase was collected. The subsequent purification of
the A. terreus aristolochene synthase was based on the previously
published procedure for the wild-type protein.5 Fractions containing
pure enzyme were located by SDS-PAGE and combined. The solution
was concentrated to 0.5 mL using an Amicon ultrafiltration apparatus
equipped with an YM-10 filter, the buffer was changed to protein
storage buffer (20 mM Hepes pH 8.0, 5 mM MgCl2 and 10% glycerol)
using a PD-10 desalting column and the protein was stored at -80 °C.
Using the above expression procedure, the A. terreus mutant AT-N219D
was obtained exclusively as inclusion bodies despite the use of lower
induction temperatures, lower IPTG concentrations and longer induction
time. Cells were therefore suspended in 50 mL of cell lysis buffer (MEG
buffer) and sonicated. The lysate was centrifuged (6000 × g for 30
min at 4 °C) and the pellet, containing the insoluble inclusion bodies,
was resuspended in 200 mL of MEG buffer. The inclusion bodies were
solubilized and the protein was purified following the same method
used for the P. roqueforti enzymes11 except that MEG buffer was
utilized for the solubilization and purification. The fractions containing
the protein were combined and concentrated using a centriprep YM-
30 centrifugal filter to a final volume of 2.5 mL, the buffer was changed
to protein storage buffer using a PD-10 desalting column, and the
protein was stored at -80 °C.
General Methods. Standard recombinant DNA and protein ma-
nipulations were carried out according to published procedures.30 PCR
and restriction digestions were run in a MiniCycler thermocycler from
MJ Research, equipped with a hot-bonnet. Aristolochene synthase
mutants were generated with the QuikChange site-directed mutagenesis
kit using the manufacturer’s protocols. DNA sequencing was performed
by the HHMI Biopolymer/Keck Foundation Biotechnology Resource
Laboratory at the Yale University School of Medicine, New Haven
(CT), using the dideoxy dye terminator method and automated
fluorescence sequencing. Protein concentrations were determined by
the Bradford method using commercial reagents (Bio-Rad) and bovine
serum albumin (Sigma) as a calibration standard.31 Liquid scintillation
was performed on a Beckman Model LS-6500 liquid scintillation
counter using Packard OmniFluor scintillation cocktail. Analysis of
DNA and protein sequences as well as PCR primer design utilized the
suite of programs in the GCG Sequence Analysis Package, version 10.0
(Unix), from Accelrys. The Insight II software package (Accelrys) was
used to visualize and manipulate Protein Data Bank (PDB) files
corresponding to the crystal structures of aristolochene synthase
(accession code 1DI1) and that of aristolochene synthase com-
plexed with FPP (1) (accession code 1F1P). A structural model of
the A. terreus aristolochene synthase (Genbank accession number
AF198360) was generated through the Swiss-Model website (http://
swissmodel.expasy.org) using the structure of chain B of the P.
roqueforti enzyme (1DI1B) as a template.32 GC-MS analysis of
enzymatic reaction products was performed using a HP-GCD series II
GC-MS system equipped with either an Optima 1701 capillary GC
column (14% cyanopropyl-phenyl/86% dimethyl polysiloxane) or a
chiral capillary GC column (FS-hydrodex-â-6TBDM), both purchased
from Macherey-Nagel. Both columns were 30 m in length with an
internal diameter of 0.25 mm and 0.25 µm phase thickness.
Aristolochene Synthase Mutants. Aristolochene synthase mutants
were prepared by PCR mutagenesis with the QuikChange site-directed
mutagenesis kit according to the manufacturer’s protocols, using as
template plasmid DNA from pZW04, harboring the P. roqueoforti
aristolochene synthase10 or pET11rASa, harboring the A. terreus
aristolochene synthase.5 For the P. roqueforti double mutant, S248A/
E252Q, the E252Q mutation was introduced using the plasmid pZWO4/
S248A as the template for PCR. Plasmid pZWO4 was extracted from
an overnight culture of E. coli XL1Blue/pZWO4 grown in Luria Bertani
(LB) ampicillin (100 µg/mL) medium, and purified using the Miniprep
plasmid purification kit. For each mutation the forward and the reverse
mutagenenic primers consisted of two complementary oligonucleotides
(25-45 nt) containing the desired mutation flanked by unmodified
nucleotide sequences. Each primer was designed and purified according
to the vendor’s recommendations. Mutant plasmids isolated from
overnight cultures of three separate colonies were purified by plasmid
miniprep and the incorporation of the desired mutation was verified
by DNA sequencing. One of the sequenced plasmids was then utilized
to transform competent cells of the expression host strain E. coli
BL21(DE3). From the resultant transformants, three single colonies
were chosen and assayed by SDS-PAGE for aristolochene synthase
production after IPTG induction. All of the colonies were found to
overexpress the enzyme. The colony giving rise to the highest level of
(30) Sambrook, J.; Fritsch, E. F.; Maniatis, T. Molecular Cloning: a Laboratory
Manual; Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY,
1989. Current Protocols in Molecular Biology, CD-ROM; Wiley: New
York, 1999.
(31) Bradford, M. Anal. Biochem. 1976, 72, 248-254.
(32) Schwede, T.; Kopp, J.; Guex, N.; Peitsch, M. C. Nucleic Acids Res. 2003,
31, 3381-3385. Guex, N.; Peitsch, M. C. Electrophoresis 1997, 18, 2714-
2723. Peitsch, M. C. Bio/Technology 1995, 13, 658-660.
9
7220 J. AM. CHEM. SOC. VOL. 126, NO. 23, 2004

Aristolochene Synthase
A R T I C L E S
Table 3. Preparative-Scale Incubation Conditions
Aristolochene Synthase Assay and Kinetics. The activity of the
P. roqueforti mutants was assayed by a modification of the procedure
previously described for wild-type aristolochene synthase.5 To determine
the concentration dependence of activity for each mutant, increasing
concentrations of protein (2 nM to 510 nM) were incubated with FPP
(1) at constant concentrations (2-20 µM) of FPP (1) in Buffer T. The
specific activity of the FPP was adjusted by mixing unlabeled FPP
with [1-3H]FPP (16 Ci/mmol) to a value of 40-170 µCi/µmol.
Incubations were initiated by addition of cyclase (30 µL containing 40
ng to 10 µg of protein) to 500 µL of pre-warmed (30 °C) Buffer H (20
mM Hepes pH 8.0, 5 mM MgCl2, 5 mM 2-mercaptoethanol) containing
FPP. After 10 min the reaction was quenched by addition of 200 µL
of 100 mM EDTA (pH 7.5) and the mixture was transferred to an ice
bath. Hexane (1 mL) was added, the mixture was vortexed for 20 s,
and the organic extract was withdrawn and passed through a 2-cm silica
gel column, directly into a scintillation vial containing 10 mL of
scintillation cocktail. Another extraction was performed with 500 µL
of hexane, followed by a final wash of 900 µL hexane. The hexane
elutable activity was measured by scintillation counting. A blank control
was run without enzyme to correct for the nonenzymatic solvolysis of
FPP. Each assay was conducted in triplicate under the same experi-
mental conditions.
Kinetic assays of aristolochene synthase mutants were carried out
in the same manner. Each assay used a constant protein concentration
(2.6-400 nM) and varying concentrations of [1-3H]FPP (1) (specific
activity 20-800 µCi/µmol). The concentrations of [1-3H]FPP were
chosen so as to bracket the experimental Km for each mutant. The
enzyme concentration was set within the range in which the activity
of the mutant had been found to be a linear function of protein
concentration and such that the percentage conversion of FPP to
products was less than 10% over the 10 min assay period. Steady-state
kinetic parameters were determined by direct fitting of the data to the
Michaelis-Menten equation by nonlinear least squares regression using
the commercial Kaleidagraph package (Synergy Software).
protein
buffer T (ml)
protein (nM)
FPP (µM)
P. roqueforti WTa
D115E
D116E
D116N
E119D
E119Q
N244D
S248A
E252D
E252Q
Y92F
A. terreus WT
AT-N219D
AT-E227D
20
20
20
20
20
20
160
20
80
160
410
510
260
510
510
410
410
510
50
25
25
25
25
25
25
50
50
50
50
25
25
25
25
20
160
170
60b
170b
150b
100
410
410
a WT, wild-type. b Buffer H.
for the maximum concentration of protein so as to produce enough
product for detection by GC-MS (Table 3). The aqueous layer was
overlaid with HPLC-grade pentane (10% by vol). The hydrocarbon
products were extracted twice with pentane (10% by vol). The extracts
were combined, concentrated to 5 mL under reduced pressure at 0 °C,
and passed through an alumina column (3 cm, packed in a Pasteur
pipet) overlaid with sodium sulfate. The organic layer was then
concentrated to 200 µL under reduced pressure at 0 °C. The concentrate
was analyzed by GC-MS using either an Optima 1701 or chiral FS-
hydrodex-â-6TBDM capillary GC column. With the Optima 1701
column, the injection port temperature was either at 150 or 250 °C
with an initial column temperature 45 °C for 4 min, followed by a
gradient of 6 °C/min to a maximum of 220 °C. Under these conditions
(+)-aristolochene (4), (-)-valencene (5) and (-)-germacrene (2) and
(+)-â-elemene (7) eluted typically at 15.47, 15.61, 15.71 and 13.51
min, respectively. For the FS-hydrodex â-6TBDM column, splitless
injections were carried out using an injection port temperature of 150
or 250 °C with an initial column temperature 45 °C for 4 min, followed
by a gradient of 2 °C/min to a maximum of 180 °C.16 Under these
conditions (+)-aristolochene (4), (-)-valencene (5), (S)-(-)-germacrene
A (2) and (+)-â-elemene (7) typically eluted at 51.57, 51.04, 53.33,
and 45.41 min, respectively. Mass spectra data were compared to those
in the NIST98 mass spectral database and to those obtained from
authentic standards.
For wild-type and mutant A. terreus aristolochene synthase, activity
assays were conducted in Buffer H following the procedure described
above, using 2-10 µM FPP (1) (specific activity 50-150 µCi/µmol)
and fixed concentrations of 0.9-410 nM protein. Kinetic assays of A.
terreus aristolochene synthase mutants were carried out as described
above. Each assay was conducted in Buffer H and used a constant
protein concentration (1-100 nM) and varying concentrations of
[1-3H]FPP (1) (specific activity 150-500 µCi/µmol).
Product Analysis. For preparative-scale incubations, purified wild-
type or mutant P. roqueforti aristolochene synthase (80-510 nM) was
incubated with FPP (1) (25-50 µM) in Buffer T at 30 °C for 3 h.
Preparative-scale incubations of A. terreus wild-type and mutant
aristolochene synthases were carried out in the same manner using
Buffer H. The concentration of each mutant was the maximum value
tested at which each protein retained concentration-independent specific
activity. The volume of buffer used in each incubation was adjusted
Acknowledgment. This research was supported by an NIH
Merit Award (GM30301) to D.E.C.
Supporting Information Available: GC-MS analysis of
E252Q product. This material is available free of charge via
the Internet at http://pubs.acs.org.
JA0499593
9
J. AM. CHEM. SOC. VOL. 126, NO. 23, 2004 7221