Differential stereocontrolled formation of tricyclic triterpenes by mutation of tyrosine 99 of the oxidosqualene-lanosterol cyclase from saccharomyces cerevisiae

Article abstract of DOI:10.1002/ejoc.200900638

The function of the Tyr99 residue from Saccharomyces cerevisiae oxidosqualene-lanosterol cyclase (ERG7) was analyzed by constructing deletion, and site-saturated mutants. Two truncated intermediates, (13αH)- isomalabarica-14Z,17E,21-trien-3βol and. (13αH)

Full text of DOI:10.1002/ejoc.200900638

FULL PAPER
DOI: 10.1002/ejoc.200900638
Differential Stereocontrolled Formation of Tricyclic Triterpenes by Mutation
of Tyrosine 99 of the Oxidosqualene-Lanosterol Cyclase from Saccharomyces
cerevisiae
Tung-Kung Wu,*[a] Wen-Hsuan Li,[a] Cheng-Hsiang Chang,[a] Hao-Yu Wen,[a]
Yuan-Ting Liu,[a] and Yi-Chun Chang[a]
Keywords: Mutagenesis / Cyclization / Enzymes / Reaction mechanisms / Terpenoids
The function of the Tyr99 residue from Saccharomyces cerevis- the functional role of ERG7Y99 is to affect both chair–boat 6–6–
iae oxidosqualene-lanosterol cyclase (ERG7) was analyzed by 5 tricyclic Markovnikov C-14 cation stabilization and the
constructing deletion and site-saturated mutants. Two trunc-
ated intermediates, (13αH)-isomalabarica-14Z,17E,21-trien-3β-
stereochemistry of the protons at the C-15 position for subse-
quent deprotonation.
ol and (13αH)-isomalabarica-14E,17E,21-trien-3β-ol, were iso- (© Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim,
lated from the ERG7Y99X mutants. These results suggest that Germany, 2009)
low the molecular plane of the substrate were observed in
Introduction
the human OSC enzyme.[21] Additionally, the Tyr98 of the
Oxidosqualene-lanosterol cyclases catalyze the conver-
human OSC was thought to be spatially positioned to en-
sion of (3S)-2,3-oxidosqualene (OS, 1) to lanosterol (LA,
2). The same substrate is utilized to synthesize cycloartenol
by oxidosqualene-cycloartenol synthase. An intricate and
yet highly stereo- and regiospecific cationic cyclization/re-
arrangement mechanism that encompasses substrate con-
formational prefolding, oxirane ring protonation and cleav-
age, cation/π interaction-directed consecutive tetracyclic
ring annulation, 1,2-shifted hydride and methyl group mi-
force the energetically unfavorable boat conformation of OS
for lanosterol B-ring formation by pushing the methyl
group at C-8 (lanosterol numbering) below the molecular
plane.[13] The Tyr99 residue of Saccharomyces cerevisiae ox-
idosqualene-lanosterol cyclase (ERG7) corresponds to the
Tyr98 residue in human OSC. S. cerevisiae ERG7 has 40%
sequence identity and 55% similarity with human OSC. We
previously performed site-saturated mutagenesis experi-
gration, and final specific deprotonation, has been pro-
ments on the Trp232, His234, Phe445, Tyr510, and Tyr707
posed for both enzymes.[1,2] Recently, the elucidation of the
residues of ERG7 to determine their catalytic activities and
complexities of the oxidosqualene cyclases, including the
product profiles.[11,14–17,19] The results showed that both the
catalytic mechanism, functional role of specific amino acid
ERG7F445X and ERG7H234X mutant enzymes produced
residues, and mutation-induced product specificity/diversity
truncated tricyclic and altered deprotonation products, indi-
profile, has been greatly facilitated through substrate ana-
cating the catalytic role of the residues in cationic stabiliza-
logues, site-specific mutagenesis coupled with product char-
tion at the C-14 position for the tricyclic product, and/or
acterization, and X-ray crystallographic analyses.[3–20] How-
ever, the important residues involved in stereo- or regios-
the C-8/C-9 position for the final deprotonation product.
In addition, a π-electron-rich pocket of aromatic residues
pecific control of these product profiles remain unclear.
around the Phe445 residue of ERG7 was suggested to either
The catalytic importance of the Tyr98 residue of human
stabilize the electron-deficient cationic intermediate or to be
oxidosqualene-lanosterol cyclase (OSC) was inconspicuous
responsible for the equilibrium shift toward the lanosteryl
until the recent X-ray crystallographic analysis of the pro-
C-8/C-9 cation during the rearrangement process and for
tein structure.[13] Homology modeling between bacterial
the formation of lanosterol. Of those aromatic residues,
squalene-hopene cyclase and human OSC showed that a
Tyr99 was located at a distance of approximately 3.7 Å,
one-residue insertion above and a one-residue deletion be-
within range to interact with the Phe445 residue, and close
to the C/D ring of the substrate.[16] In addition, although
[a] Department of Biological Science and Technology, National
alignment of multiple sequences revealed that the Tyr99 res-
Chiao Tung University,
idue of ERG7 is highly conserved in both oxidosqualene-
lanosterol cyclases and oxidosqualene-cycloartenol syn-
thases, different amino acids were at that position in related
enzymes β-amyrin synthase and lupeol synthase. Also, a
300 Hsin-Chu, Taiwan, Republic of China
Fax: +886-3-5725700
E-mail: tkwmll@mail.nctu.edu.tw
Supporting information for this article is available on the
WWW under http://dx.doi.org/10.1002/ejoc.200900638.
Eur. J. Org. Chem. 2009, 5731–5737
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5731

T.-K. Wu et al.
FULL PAPER
one-residue insertion proximal to the Tyr99 position of mutations. These results indicated a functional role for
ERG7 was observed in oxidosqualene-lanosterol cyclases Tyr99 in catalysis or structure stabilization. Next, the prod-
from S. cerevisiae, Trypanosoma brucei, and T. cruzi, but uct profiles of each mutant were characterized by growing
not in mammalian oxidosqualene-lanosterol cyclases (data the mutant strains, preparing the nonsaponifiable lipid ex-
not shown). The combination of these observations sug- tract (NSL), separating the fractions on a AgNO3-impreg-
gested that Tyr99 of S. cerevisiae ERG7 may have a dif- nated silica gel column, and analyzing the structures of each
ferent functional role than Tyr98 in human OSC. However, product by using gas chromatography–mass spectrometry
no mutation-induced product isolation and characterization (GC–MS) and nuclear magnetic resonance (1H, 13C NMR,
1
were performed to validate the above-mentioned hypothesis DEPT, H–1H COSY, HMQC, HMBC, and NOE).
about the catalytic mechanism.
Product profiles of the ERG7∆Y99 and ERG7Y99X mu-
To further substantiate the functional role of the Tyr99 tants are summarized in Table 1. No products with molecu-
position in the ERG7-catalyzed cyclization/rearrangement lar masses of m/z = 426 were observed for the nonviable
cascade and to investigate the effects of substitutions of mutants, consistent with the genetic results. The viable mu-
other amino acid residues on the catalytic activity and tants produced either lanosterol alone or lanosterol and two
product profile of the cyclase, we generated a mutant ERG7 unidentified compounds with a C30H50O formula. The first
on which the Tyr99 residue was deleted (ERG7∆Y99) and a new compound was identified by NMR spectroscopy as
set of mutants carrying every amino acid substitution at (13αH)-isomalabarica-14Z,17E,21-trien-3β-ol (3), a chair–
that position (site-saturated mutations, ERG7Y99X) and boat (C–B) 6–6–5 tricyclic product with trans–syn–trans
characterized products profiles. Two abortive cyclization stereochemistry and ∆14Z,17E,21 double bonds. Confirm-
products, (13αH)-isomalabarica-14Z,17E,21-trien-3β-ol (3) ation of the structure, including both the bond connectivity
and (13αH)-isomalabarica-14E,17E,21-trien-3β-ol (4), were and stereochemistry, is described in the Experimental Sec-
isolated, for the first time, from several mutants expressing tion. Specifically, the presence of NOE interactions among
ERG7Y99X as the sole oxidosqualene cyclase. In addition, 3-H/Me-29, Me-30/Me-28, Me-29/5-H, 3-H/5-H, Me-28/9-
compounds 3 and 4 were isolated concomitantly from the H, 15-H/Me-26, and Me-27/13-H, as well as the absence of
ERG7Y99X mutants that produced truncated products, but NOE interactions between Me-27/Me-28, Me-27/9-H, 9-H/
at different ratios. These product profiles correspond to 13-H, Me-28/5-H, and 13-H/15-H, were uniquely consistent
truncation of the cyclization/rearrangement cascade at the with the stereochemistry of the C–B 6–6–5 tricyclic nucleus
chair–boat 6–6–5 tricyclic Markovnikov cation, at the C-14 and the Z geometry for a double bond between C-14 and
position, and subsequent abstraction of protons at the C-15 C-15. The results unambiguously establish compound 3 to
position with different stereochemical preferences. A similar be (13αH)-isomalabarica-14Z,17E,21-trien-3β-ol, a chair–
truncated
intermediate,
(13αH)-isomalabarica- boat 6–6–5 tricyclic product with trans–syn–trans stereo-
14(26),17E,21-trien-3β-ol (5), which is derived from the chemistry and ∆14Z,17E,21 double bonds (Figure 1a). The
same C-14 position that directly abstracts the C-26 proton, second new product, migrating on the GC column with a
was previously isolated from both ERG7F445X and the retention time of 0.2 min relative to the first compound (i.e.,
ERG7H234X mutants.[15,16] Conversely, compound 5 could 3), was isolated and identified. The structure of compound
not be isolated from any of the ERG7Y99X mutants. These 4 was also elucidated by NMR spectroscopy, as described
results suggested a catalytic role of the Tyr99 residue in in the Experimental Section. Interestingly, compound 4 is
ERG7 that not only stabilizes the same chair–boat 6–6–5 structurally similar to compound 3 with differences only in
tricyclic Markovnikov cation as that of Phe445 and His234, the stereochemistry of the carbon double bond located at
but also affects different deprotonation positions, perhaps the C-14/C-15 position. The presence of NOE interactions
through different spatial orientations or interactions with between Me-29/3-H, Me-29/5-H, Me-30/Me-28, Me-28/9-
other proximal residues within the active site cavity.
H, Me-27/13-H, and 13-H/15-H, as well as the absence of
NOE interactions between Me-30/3-H, Me-28/Me-27, 13-
H/9-H, 13-H/Me-26, and 15-H/Me-26, indicate the α-orien-
tation for 13-H and the E-conformation for a double bond
between C-14 and C-15. Thus, the structure was determined
Results and Discussion
We generated the ERG7∆Y99 and ERG7Y99X mutations to be (13αH)-isomalabarica-14E,17E,21-trien-3β-ol (4), a
by using the QuikChange site-directed mutagenesis kit.[15,17] tricyclic product with trans–syn–trans stereochemistry and
14E,17E,21
Following the confirmation of the amino acid substitutions
∆
double bonds (Figure 1b). No known truncated
at the Tyr99 position, the recombinant plasmids were trans- or altered rearranged products, such as monocyclic achil-
formed into yeast HEM1 ERG7 double-knockout mutant leol A or camelliol C, bicyclic, tricyclic (13αH)-isomalabar-
TKW14, which is only viable when complemented with cy- ica-14(26),17E,21-trien-3β-ol (5), tetracyclic protosta-20,24-
clase activity derived from ERG7Y99X or when supplied dien-3β-ol, truncated rearranged protosta-12,24-dien-3β-ol,
with exogenous ergosterol. The TKW14[pERG7Y99X] strain and altered deprotonated parkeol or 9β-lanosta-7,24-dien-
expressed ERG7Y99X as the only oxidosqualene cyclase. The 3β-ol (∆7-lanosterol), were observed in the GC–MS analy-
genetic selection results showed that all mutations at Tyr99 ses from any of the mutants. These two compounds, in con-
supported ergosterol-independent growth of TKW14, ex- junction with the previously isolated truncated tricyclic
cept for the Tyr99 deletion and the Tyr99Asn and Tyr99His compound 5, which abstracts the C-26 proton after generat-
5732
www.eurjoc.org
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2009, 5731–5737

Differential Stereocontrolled Formation of Tricyclic Triterpenes
Figure 1. Bond connectivity and stereochemistry established by HMQC/HMBC (bold bond, ––) and NOE interaction (curved arrows)
of (a) (13αH)-isomalabarica-14Z,17E,21-trien-3β-ol and (b) (13αH)-isomalabarica14E, 17E,21-trien-3β-ol.
ing a tricyclic C-14 cation, suggested a direct trapping of
the common C–B 6–6–5 tricyclic Markovnikov C-14 cation
but with a different deprotonation position and stereochem-
istry.[16,17] Interestingly, Matsuda and co-workers charac-
terized a baruol synthase (BARS1) from Arabiopsis thali-
ana, which makes baruol and 22 minor products.[22] Among
the monocyclic (I) or bicyclic (II) cationic position. In
ERG7H234X and ERG7F445X site-saturated mutations, and
ERG7Y510F/H site-directed mutations, direct deprotonation
of Me-26 resulted in the formation of 5 as the truncated
product, whereas in ERG7Y99X site-saturated mutations, de-
protonation of different protons at the C-15 position pro-
duced compounds 3 and 4.[15,16,24] Alternatively, an anchim-
eric cyclopentylcarbinyl cyclohexyl carbocation (IV) re-
arrangement occurred between C-13 and C-14 and concert-
edly expanded the C-ring, which was followed by D-ring
annulation to generate the protosteryl C-20 cation (V). Sub-
sequently, a series of hydride and methyl group rearrange-
ments generated the lanosteryl C-8/C-9 cation (VI). Finally,
a highly specific deprotonation abstracted the proton,
which was either originally at C-9 or, after the hydride shift
from the C-9 to the C-8 position, and this deprotonation
yields lanosterol.
the minor products,
a
tricyclic truncated product,
(13S,14E,17E)-malabarica-14,17,21-trien-3β-ol, was iso-
lated. However, no detailed structure–function relationships
of baruol synthase were reported.
Table 1. Product profiles of S. cerevisiae TKW14 expressing the
ERG7∆Y99 deletion and ERG7Y99X site-saturated mutants.[23] The
product profile was determined by the intensities of the signals in
the GC–MS trace, in conjunction with resolved methyl peaks in
the 600 MHz NMR spectrum.
Amino acid
substitution
Products profile
5
4
2
The question of how the OSC catalyzes the course of
C- and D-ring cyclization remains controversial.[5,15,16,25–27]
Different mechanisms based on either experimental evi-
dence or computational studies have been proposed for the
formation of C- and D-ring intermediates. Formation of
rings C and D could proceed by either the 6–6–5 Markovni-
kov tertiary cyclopentylcarbinyl carbocation (III), which
undergoes an anchimeric cyclocarbinyl-cyclohexyl (IV) C-
ring expansion and D-ring annulation, or the asynchronous
concerted cyclization from A/B-bicyclic cyclohexyl cation
(II) to D-ring formation to generate the protosteryl cation
(V). However, many 6–6–5 tricycles occur in nature or arise
from simple substrate analogs, whereas natural 6–6–6 tri-
terpenes from (oxido)squalene are unknown, favorably sup-
porting the intermediary of the 6–6–5 tertiary cyclopen-
tylcarbinyl carbocation.[2,5,28–31] For example, Corey and
co-workers performed substrate–analog experiments on S.
Gly
Ala
Val
Leu
Ile
13
17
0
0
17
8
0
0
0
0
13
21
0
0
6
0
28
–
58
70
0
0
33
5
44
0
0
29
13
100
100
50
87
56
100
100
100
18
Asp
Glu
Gln
Lys
Arg
Ser
Thr
Cys
Met
Phe
Trp
Pro
Asn
His
0
69
59
0
0
4
0
51
–
20
100
100
90
100
21
–
–
–
–
The production of truncated cyclization intermediates by cerevisiae ERG7 wild-type enzyme and obtained a discrete
ERG7Y99X mutants suggests that the substitution of Tyr99 6,6,5,4-tetracyclic compound involved in C-ring closure,
effects the formation of truncated tricyclic triterpenes but providing convincing evidence for the 6–6–5 tertiary cy-
does not result in loss of enforcement of the boat conforma- clopentylcarbinyl carbocation formation.[5] In addition, trit-
tion for lanosterol B-ring formation. In ERG7Y99X mutants, erpenoid compounds with a 6–6–5 C–B tricyclic isomalab-
the (13αH)-isomalabarica-trien-3β-ols 3 and 4 were appar- ricane skeleton have been isolated from natural resources.[31]
ently formed from intermediate C–B 6–6–5 Markovnikov Furthermore, Hess showed that the cyclopentylcarbinyl
tricyclic cation through deprotonation of C-15 (Scheme 1). carbocation formed during the lanosterol biosynthesis
Protonation of oxidosqualene initiates carbocation forma- could quickly undergo a concerted C-ring expansion and
tion and subsequent cation–π annulation to a chair–boat– D-ring closure to give the protosteryl cation without viola-
chair (C–B–C) conformation and Markovnikov-favored 6– tion of the Markovnikov rule.[26] It was proposed that the
6–5 tricyclic C-14 cation (III) without disruption at either OSC plays a functional role in “holding” the substrate in
Eur. J. Org. Chem. 2009, 5731–5737
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
5733

T.-K. Wu et al.
FULL PAPER
Scheme 1. Proposed mechanism for the oxidosqualene to lanosterol cyclization/rearrangement cascade in ERG7Y99X-catalyzed site-satu-
rated mutants.
the proper conformation for the completion of the cascade clic cyclopentylcarbinyl carbocation, whereas the C–C–C 6–
leading to the tetracyclic system. We have previously iso- 6–5 tricyclic cyclopentylcarbinyl cation exhibited an exo-
lated a Markovnikov 6–6–5 tricyclic intermediate from both thermic 6.9 kcal/mol of free energy relative to that of the
the S. cerevisiae ERG7H234X and ERG7F445X mutants, dammarenyl cation. In parallel, Matsuda showed that a
which also supports the intermediary of the 6–6–5 tertiary low-energy 6–6–5 pathway of C- and D-ring formation cir-
cyclopentylcarbinyl carbocation for oxidosqualene cyclase- cumvents the 6–6–6 tricycle and that a horizontal variant
catalyzed cyclization reactions.[15,16] In contrast, a computa- of the secondary cyclohexyl carbocation occurs transiently
tional study of the squalene cyclase reaction performed by as a partially bridged structure in D-ring formation, but is
Ramkumar and Gao suggested that the only significant in- neither an intermediate nor transition state along the mini-
termediate in the squalene-hopene cyclization cascade is the mum-energy path.[30] Perhaps the low barrier of the free
A/B-bicyclic cyclohexyl cation, and the six-membered C- energy for the protosteryl cation formation could be com-
ring species is not a minimum on the free energy surface pensated by the enzyme through “holding” of the proper
whose formation is accompanied by an asynchronous con- carbocationic conformation, and the placement of the nu-
certed cyclization from the A/B-bicyclic cyclohexyl cation cleophilic groups from the protein backbone or side chains
to ring D.[27] Nevertheless, the latter mechanism was only and proceed through a concerted or bridged transition state
supported by the results obtained from free-energy simula- in C-ring expansion/D-ring annulation to the protosteryl
tion and the lack of isolation of a 6–6–6 tricyclic aborted cation, as previously suggested by Jenson and Jorgenson
cyclization product. However, it is interesting to note that and Hess.[25,26] Alternatively, we cannot exclude the pos-
following the B-ring formation, a branched free-energy pro- sibility that stabilization or destabilization effects caused by
file for the C–B–C protosteryl cation and the C–C–C dam- the mutation may subsequently alter the C-ring cyclization
marenyl cation formation was observed. For the protosteryl process from asynchronous concerted A/B-bicyclic cyclo-
cation, an endothermic 9.4 kcal/mol free energy relative to hexyl carbocation to 6–6–5 Markovnikov tertiary cyclopen-
the bicyclic cation was obtained for the C–B 6–6–5 tricyclic tylcarbinyl-cyclohexyl rearrangements. However, a mu-
cyclopentylcarbinyl cation. Conversely, for the dammarenyl tation in the enzyme active site may alter the conformation
cation formation, an exothermic 1.1 kcal/mol free energy and positioning of the cyclizing polyolefin chain or disrupt
relative to that of the C–C bicyclic cation was observed for specific interactions between the enzyme and the carbo-
the C–C–C 6–6–5 tricyclic cyclopentylcarbinyl cation. Sub- cationic intermediate that are needed for the delicate bal-
sequently, the protosteryl cation showed an endothermic ance of thermodynamic or kinetic control of various carbo-
2.7 kcal/mol free energy relative to that of the 6–6–5 tricy- cationic intermediates.
5734
www.eurjoc.org
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2009, 5731–5737

Differential Stereocontrolled Formation of Tricyclic Triterpenes
Why the ERG7Y99X mutants exhibit different tricyclic de- tations. Alternatively, the possibility that mutations at
protonation profiles from those of the ERG7H234X and Tyr99 generate a new base that leads to the Markovnikov
ERG7F445X mutants is another interesting question. Both C–B 6–6–5 cation and the subsequent disproportional for-
the ERG7H234X and ERG7F445X mutants yielded compound mation of compounds 3 and 4 cannot be excluded. In con-
5 as the sole tricyclic deprotonation product, whereas the trast, a deletion at this position may lead to a local main
ERG7Y99X mutants resulted in the production of com- chain adjustment in the mutant protein compared with the
pounds 3 and 4, but not 5 from the common C-14 cation. In wild type, incurring obstruction of substrate binding and
addition, most of the ERG7Y99X mutants produced 14(Z) subsequent catalysis such that no product could be ob-
isomer 3 as a major product. Previous results showed that tained from the ERG7∆Y99 mutant.
progressive reaction energy release was primarily observed
during A- and B-ring formation but was much less energetic
for C- and D-ring annulation when neglecting the role of
the enzyme. Nevertheless, the energetics of A-ring forma-
tion could be dramatically affected by the inclusion of the
enzymatic effects.[30] In addition, comparing energies
among various tricyclic geometric isomers showed energy
decrements from compounds 5 to 3 to 4, and although little
influence on the enzymatic effect was observed when the
tricyclic cation was converted into the tetracyclic intermedi-
ate, the substitution of amino acid residues at different spa-
tial positions may alter a kinetically favored compound 5
that produces the thermodynamically favored compounds 3
and 4. Perhaps, the differences in energies of various tricy-
clic geometric isomers could be compensated by the amino
acid residues in the stabilization of the Markovnikov tricy-
clic cation and/or the subsequent alteration of the depro-
tonation position with differential stereochemical control,
Figure 2. Snapshot of the superposition of the S. cerevisiae
ERG7Y99 wild type structure with the ERG7Y99A mutant structure,
which corresponds to the tricyclic C-14 carbocation intermediate,
which was obtained from homology modeling.
with compound 3 in favor of compound 4. Further, the iso-
lation of compounds 3 and 4, but not that of compound 5,
suggested a discrepancy in the effects of spatially different
amino acid residues within the active site cavity on both
cationic stabilization and deprotonation.
In parallel, the homology modeling also provides some
insight into the relationships between enzyme structure and
product specificity (Figure 2). Residues in this region are
involved in substrate folding, which affects the cyclization/
Conclusions
In conclusion, a genetic selection for active ERG7 mu-
rearrangement mechanism.[13] The homology model tant enzymes with substitutions at Tyr99, coupled with
showed that His234 of ERG7 is hydrogen bonded to Tyr510 characterization of the products generated by those mutant
and located spatially near the upper back end of the active enzymes, suggests that Tyr99 plays a role in catalysis, affect-
site cavity. The Phe445 residue of ERG7 is located near the ing both Markovnikov tricyclic C-14 cationic stabilization
B/C ring fusion and is proximal to the C-8 and C-14 posi- and the regio- or stereochemical control in the active site
tions, neighboring the ceiling of the active site cavity. In- of the protons at the C-15 position for subsequent depro-
spection of the model suggests that the Tyr99 residue is po- tonation but not the B-ring boat conformation. Character-
sitioned in the middle side wall of the active site cavity by istically, structural modification or electronic differentiation
interacting with the C-ring, with the phenolic hydroxy side of active site residue substitutions or repositions may have
chain pointed towards the substrate. The C-14 cation was effects on the stabilization of different cation intermediates
found at a distance of approximately 5.0 Å from the ob- or neighboring amino acid residues or motifs, thus directing
served phenolic oxygen atom of Tyr99, and this distance truncated deprotonation from different orientations and
agrees with the dipole–cation interaction for the phenolic subsequent stereochemical predilections. Moreover, incon-
oxygen atom of Tyr99 in the C-14 cation-bound OSC com- sistencies were observed on the functional roles of human
plex. It is conceivable that amino acid changes at Tyr99 OSC structure-based prediction and S. cerevisiae ERG7 ex-
would strongly affect the orientation or electrostatic interac- perimental results, suggesting that minor dissimilarities may
tion of the phenolic oxygen of Tyr99, which is positioned occur between two enzymes and require independent char-
to stabilize the C-14 cation for C-ring expansion and fur- acterization of their structure–function relationships. Fi-
ther D-ring annulation. Perhaps some additional space for nally, the results obtained from this study provide signifi-
the free rotation of the hydrocarbon side chain moiety is cant insight into the stereo- and regiocontrol of oxido-
offered through mutations at this position, resulting in pro- squalene cyclases in general. Further studies should focus
tons being abstracted from different positions and/or orien- on elucidating the precise molecular interactions in con-
Eur. J. Org. Chem. 2009, 5731–5737
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
5735

T.-K. Wu et al.
FULL PAPER
as 126.8, 137.8 ppm), which are characteristics of double bonds at
the exocyclic hydrocarbon side chains. These correlations suggest
the involvement of a tricyclic ring skeleton. The following are fea-
tures of the HMBC spectrum: (1) The δ = 1.657 ppm protons of
the methyl group are coupled by 2J to the carbon atom at δ =
131.3 ppm (C-22), as well as by 3J to the carbon atoms at δ = 124.3
(C-21) and 17.7 ppm (C-24), thus locating the terminal double
bond. (2) The carbon atom at δ = 124.3 ppm is coupled to protons
at δ = 2.038 and 1.967 ppm, as well as at δ = 1.657 and 1.577 ppm.
(3) The carbon atom at δ = 123.3 ppm is coupled to protons at δ
= 2.584–2.627, 1.967, 1.584, 5.186 ppm. and (4) The protons at δ
= 2.584–2.627 ppm are coupled to carbon atoms at δ = 123.3 (C-
17), 126.8 (C-15), 134.8 (C-18), and 137.8 ppm (C-14). These corre-
lations unambiguously establish the connectivity of hydrocarbon
side chains and relative double bond positions. Furthermore, the
HMBC spectrum showed that the δ = 2.615 ppm methine proton
trolling regio- and stereochemical specificity, chair–boat–
chair and chair–chair–chair conformations, and the re-
arrangement process.
Experimental Section
Generation and Analysis of Mutant Extracts: The pTKERG7RSWT
plasmid, a pRS314 derivative containing the S. cerevisiae ERG7
gene, was used as a template for Y99 deletion and site-saturated
mutagenesis. The following oligonucleotide primers used (with sub-
stitutions in bold and silent mutation italicized):
ERG7Y99X-Degenerate1:
5Ј-d(CCGTGTCAANNNAAAGGGCCCATGTTCATG)-3Ј
ERG7Y99-Degenerate2:
2
is coupled by J to carbon atoms at δ = 137.8 (C-14), 44.2 (C-8),
5Ј-d(CATGAACATGGGCCCTTTNNNTTGACACGG)-3Ј
ERG7Y99-Deletion1:
and 25.3 (C-12), as well as 3J to carbon atoms at δ = 126.8 (C-
15), 54.4 (C-9), 21.2 (C-11), 22.2 (C-26), and 29.9 ppm (C-27), thus
establishing the adjacent connectivity between the double bond and
the tricyclic nucleus. Chemical shifts were referenced to Si(CH3)4
and are generally accurate to +0.01 ppm. 1H NMR (600 MHz,
CDCl3): δ = 5.186 (t, J = 7.3 Hz, 1 H, 15-H), 5.043 (t, J = 7.6 Hz,
1 H, 17-H), 5.076 (t, J = 6.6 Hz, 1 H, 21-H), 3.237 (dd, J = 11.6,
5.1 Hz, 1 H, 3α-H), 2.585–2.627 (m, 2 H for 16-H, 1 H for 13α-H),
2.045 (m, 2 H, 20-H), 1.960 (m, 2 H, 19-H), 1.860 (m, 1 H, 12α-
H), 1.708–1.786 (m, 3 H, 2α-H, 7-H), 1.657 (s, 3 H, 23-H), 1.626
(m, 1 H, 2β-H), 1.600 (s, 3 H, 26-H), 1.584 (s, 3 H, 25-H), 1.577
(s, 3 H, 24-H), 1.563 (m, 2 H, 9β-H, 11β-H), 1.547 (m, 1 H, 12β-
H), 1.532 (m, 1 H, 6α-H), 1.507 (m, 1 H, 6β-H), 1.485 (m, 1 H,
5α-H), 1.457 (m, 1 H, 1β-H), 1.450 (m, 1 H, 11β-H), 1.389 (m, 1
H, 1α-H), 1.239 (m, 1 H, 6β-H), 1.072 (s, 3 H, 27-H), 0.964 (s, 3
H, 29-H), 0.918 (s, 3 H, 28-H), 0.763 (s, 3 H, 30-H) ppm. 13C NMR
(150 MHz, CDCl3): δ = 33.8 (C-1), 29.2 (C-2), 79.5 (C-3), 39.1 (C-
4), 47.6 (C-5), 18.8 (C-6), 32.5 (C-7), 44.2 (C-8), 54.4 (C-9), 35.6
(C-10), 21.2 (C-11), 25.3 (C-12), 50.8 (C-13), 137.8 (C-14), 126.8
(C-15), 26.9 (C-16), 123.3 (C-17), 134.8 (C-18), 39.7 (C-19), 26.7
(C-20), 124.3 (C-21), 131.3 (C-22), 17.7 (C-23), 25.7 (C-24), 16.1
(C-25), 22.2 (C-26), 29.9 (C-27), 23.1 (C-28), 29.1 (C-29), 15.9 (C-
30) ppm.
5Ј-d(CCGTGTCAAAAAGGGCCCATGTTCATG)-3Ј
ERG7Y99-Deletion2:
5Ј-d(CATGAACATGGGCCCTTTTTGACACGG)-3Ј
The mutagenesis experiments were performed by using the
QuikChange site-directed mutagenesis kit (Stratagene, La Jolla,
CA).[15] Following confirmation of the mutations by DNA se-
quencing, recombinant plasmids were electroporated into the
yeast strain TKW14 and selected for growth on
SD+Ade+Lys+His+Met+Ura+hemin+G418+Erg plates (Calbi-
ochem, La Jolla, CA, USA). The plasmids were then selected on
SD+Ade+Lys+His+Met+Ura+hemin+G418+5-FOA plates for
complementation of cyclase activity as described previously.[11,14]
Transformants were grown in SD+Ade+Lys+His+Met+Ura+
hemin+G418 +Erg medium for nonsaponifiable lipid (NSL) extrac-
tion and column chromatography. The NSL extract was fraction-
ated by silica gel column chromatography and assayed by gas
chromatography–mass spectrometry (GC–MS) to examine triter-
penoid products with a molecular mass of m/z = 426. The mixture
of 2, 3, and 4 (47–65 mg for different mutants, 1.4%–1.9% yield
from the NSL extract) was obtained with n-hexane/ethyl acetate
(19:1) as eluent. Relative ratios of each mutant were determined by
the intensities of the signals in the GC–MS trace in conjunction
with the resolved methyl peaks in the NMR spectrum, as described
previously.[15]
Structure Characterization of (13αH)-Isomalabarica-14E,17E,21-
trien-3β-ol (4): The EI low-resolution mass spectrum exhibited sim-
ilar parent peak and fragment peak patterns to compound 3, sug-
gesting an analogous nucleus skeleton of incomplete cyclization.
The 1H NMR spectrum showed four distinct vinylic methyl signals
(δ = 1.662, 1.605, 1.582, 1.551 ppm), four methyl singlets (δ = 1.044,
0.956, 0.912, 0.756 ppm), and three double bond protons (δ =
4.997, 5.081, 5.094 ppm), suggesting a tricyclic ring skeleton. Next,
Structure Characterization of (13αH)-Isomalabarica-14Z,17E,21-
trien-3β-ol (3): The EI mass spectrum showed a molecular ion at
m/z = 426 and fragment peaks at m/z = 357, 339, and 247, corre-
sponding to C30H50O [M]+, [M – C5H9]+, [M – C5H9 – H2O]+, and
[M – C13H21 – H2]+, respectively, and suggesting a mass spectral a combination of the apparent HMQC and HMBC correlations
fragmentation of incomplete cyclization. 1H NMR spectroscopy
showed distinct chemical shifts with four vinylic methyl signals (δ
= 1.657, 1.600, 1.584, 1.577 ppm), four methyl singlets (δ = 1.072,
0.964, 0.918, 0.763 ppm), and three sets of triplet double bond pro-
tons (δ = 5.186, 5.076, 5.043 ppm). The HMQC spectrum showed
that the olefinic protons at δ = 5.186, 5.076, and 5.043 ppm are
attached to the carbon atoms at δ = 126.8 (C-15), 124.3 (C-21),
and 123.3 (C-17) ppm, respectively, and that the methine proton at
δ = 3.232 ppm is attached to the carbon atom at δ = 79.5 ppm (C-
3). The HMQC spectrum also showed that the peak at δ = 2.585–
2.627 ppm contained three protons, which are attached to carbon
atoms at δ = 50.8 (C-13, 1 H) and 26.9 ppm (C-16, 2 H). The 13C
NMR spectrum revealed the presence of three tertiary-quaternary
substituted double bonds (δ = 123.3, 131.3 and 124.3, 134.8, as well
showed the following features: (1) The vinyl proton at δ =
4.997 ppm (δC = 125.0 ppm, C-15) is coupled to carbon atoms at
27.0 (C-16), 123.3 (C-17), 59.5 (C-13), and 18.1 ppm (C-26).
(2) The vinyl proton at δ = 5.094 ppm (δC = 123.3 ppm, C-17) is
coupled to carbon atoms at δ = 27.0 (C-16), 39.7 (C-19), and
16.1 ppm (C-25). (3) The vinyl proton at δ = 5.081 ppm (δC
124.32 ppm, C-21) is coupled to carbon atoms at δ = 25.7 (C-23)
and 17.7 ppm (C-24). (4) The proton at δ = 2.684 ppm (δC
=
=
27.0 ppm, C-16) is coupled to carbon atoms at δ = 139.2 (C-14),
125.0 (C-15), 123.3 (C-17), and 135.0 ppm (C-18). (5) The proton
at δ = 2.048 ppm (δC = 59.5 ppm, C-13) is coupled to carbon atoms
at δ = 139.2 (C-14), 125.0 (C-15), 44.5 (C-8), 52.9 (C-9), and
29.8 ppm (C-27). These correlations unequivocally established the
bond connectivity between the tricyclic nucleus skeleton and the
5736
www.eurjoc.org
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2009, 5731–5737

Differential Stereocontrolled Formation of Tricyclic Triterpenes
[7]
E. A. Hart, L. Hua, L. B. Darr, W. K. Wilson, J. Pang, S. P. T.
Matsuda, J. Am. Chem. Soc. 1999, 121, 9887–9888.
B. M. Joubert, L. Hua, S. P. T. Matsuda, Org. Lett. 2000, 2,
339–341.
S. Lodeiro, M. J. Segura, M. Stahl, T. Schulz-Gasch, S. P. T.
Matsuda, ChemBioChem 2004, 5, 1581–1585.
M. M. Meyer, M. J. Segura, W. K. Wilson, S. P. T. Matsuda,
Angew. Chem. Int. Ed. 2000, 39, 4090–4092.
T. K. Wu, J. H. Griffin, Biochemistry 2002, 41, 8238–8244.
T. K. Wu, C. H. Chang, ChemBioChem 2004, 5, 1712–1715.
R. Thoma, T. Schulz-Gasch, B. D’Arcy, J. Benz, J. Aebi, H.
Dehmlow, M. Henning, M. Stihle, A. Ruf, Nature 2004, 432,
118–122.
T. K. Wu, Y. T. Liu, C. H. Chang, ChemBioChem 2005, 6,
1177–1181.
T. K. Wu, Y. T. Liu, C. H. Chang, M. T. Yu, H. J. Wang, J. Am.
Chem. Soc. 2006, 128, 6414–6419.
T. K. Wu, Y. T. Liu, F. H. Chiu, C. H. Chang, Org. Lett. 2006,
8, 4691–4694.
T. K. Wu, M. T. Yu, Y. T. Liu, C. H. Chang, H. J. Wang,
E. W. G. Diau, Org. Lett. 2006, 8, 1319–1322.
T. K. Wu, C. H. Chang, Y. T. Liu, T. T. Wang, Chem. Rec.
2008, 8, 302–325.
T. K. Wu, H. Y. Wen, C. H. Chang, Y. T. Liu, Org. Lett. 2008,
10, 2529–2532.
T. K. Wu, T. T. Wang, C. H. Chang, Y. T. Liu, Org. Lett. 2008,
10, 4959–4962.
T. Schulz-Gasch, M. Stahl, J. Comput. Chem. 2003, 24, 741–
753.
S. Lodeiro, Q. Xiong, W. K. Wilson, M. D. Kolesnikova, C. S.
Onak, S. P. T. Matsuda, J. Am. Chem. Soc. 2007, 129, 11213–
11222.
The in vivo TKW14[pERG7Y99X] product profiles represent
metabolic equilibrium of lanosterol and related triterpenes un-
der physiological conditions, which may vary from that of the
in vitro assay.
S. Lodeiro, W. K. Wilson, H. Shan, S. P. T. Matsuda, Org. Lett.
2006, 8, 439–442.
C. Jenson, W. L. Jorgenson, J. Am. Chem. Soc. 1997, 119,
10846–10854.
B. A. Hess Jr, J. Am. Chem. Soc. 2002, 124, 10286–10287.
R. Rajamani, J. Gao, J. Am. Chem. Soc. 2003, 125, 12768–
12781.
T. Hoshino, M. Kouda, T. Abe, S. Ohashi, Biosci. Biotechnol.
Biochem. 1999, 63, 2038–2041.
R. Xu, G. C. Fazio, S. P. T. Matsuda, Phytochemistry 2004, 65,
261–291.
S. P. T. Matsuda, W. K. Wilson, Q. Xiong, Org. Biomol. Chem.
2006, 4, 530–543.
L. McCabe, J. Clardy, Tetrahedron Lett. 1982, 23, 3307–3310.
Received: June 11, 2009
exocyclic hydrocarbon side chain, as well as the double bond posi-
tions. Chemical shifts were referenced to Si(CH3)4 and are generally
accurate to +0.01 ppm. 1H NMR (600 MHz, CDCl3): δ = 5.094 (t,
J = 7.1 Hz, 1 H, 17-H), 5.069 (t, J = 7.3 Hz, 1 H, 21-H), 4.997 (t,
J = 7.0 Hz, 1 H, 15-H), 3.231 (dd, J = 11.7, 5.0 Hz, 1 H, 3α-H),
2.687 (m, 2 H, 16-H), 2.078–2.047 (m, 3 H, 2 H for 20-H, 1 H for
13α-H), 1.964 (m, 2 H, 19-H), 1.874–1.928 (m, 1 H, 12α-H), 1.711–
1.739 (m, 1 H, 2α-H), 1.662 (s, 3 H, 23-H), 1.622 (m, 2 H, 1 H for
2β-H, 1 H for 7α-H), 1.605 (s, 3 H, 25-H), 1.582 (s, 3 H, 24-H),
1.551 (s, 3 H, 26-H), 1.537–1.511 (m, 4 H, 1 H for 6α-H, 1 H for
9β-H, 1 H for 11β-H, 1 H for 12β-H), 1.470–1.420 (m, 1 H, 5α-H),
1.387 (m, 3 H, 2 H for 1-H, 1 H for 11α-H), 1.222–1.202 (m, 2 H,
1 H for 6β-H, 1 H for 7β-H), 1.044 (s, 3 H, 27-H), 0.956 (s, 3 H,
29-H), 0.911 (s, 3 H, 28-H), 0.756 (s, 3 H, 30-H) ppm. 13C NMR
(150 MHz, CDCl3): δ = 34.1 (C-1), 29.1 (C-2), 79.5 (C-3), 39.0 (C-
4), 47.0 (C-5), 18.6 (C-6), 32.1 (C-7), 44.5 (C-8), 52.9 (C-9), 35.4
(C-10), 20.9 (C-11), 26.9 (C-12), 59.5 (C-13), 139.2 (C-14), 125.0
(C-15), 27.0 (C-16), 123.3 (C-17), 135.0 (C-18), 39.7 (C-19), 26.7
(C-20), 124.3 (C-21), 131.3 (C-22), 25.7 (C-23), 17.7 (C-24), 16.1
(C-25), 18.1 (C-26), 29.8 (C-27), 23.1 (C-28), 29.1 (C-29), 15.9 (C-
30) ppm.
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
Supporting Information (see footnote on the first page of this arti-
cle): MS, 1H NMR, 13C NMR, HMQC, HMBC spectra of com-
pounds 3 and 4.
Acknowledgments
We are grateful to Dr. John H. Griffin and Prof. Tahsin J. Chow for
helpful advice. We thank the Ministry of Education, Aiming for
Top University Plan (MOE ATU Plan), and the National Chiao
Tung University for financial support. This work was also sup-
ported in part by the National Science Council of the Republic of
China (contract no. NSC-94-2113-M-009-011 and NSC-96-2627-
M-009-003). We also thank the National Center for High-Perform-
ance Computing for running Gaussian jobs.
[23]
[24]
[25]
[1] I. Abe, M. Rohmer, G. D. Prestwich, Chem. Rev. 1993, 93,
2189–2206.
[2] K. U. Wendt, G. E. Schulz, E. J. Corey, D. R. Liu, Angew.
Chem. Int. Ed. 2000, 39, 2812–2833.
[3] E. J. Corey, H. Cheng, C. H. Baker, S. P. T. Matsuda, D. Li, X.
Song, J. Am. Chem. Soc. 1997, 119, 1289–1296.
[4] E. J. Corey, S. C. Virgil, J. Am. Chem. Soc. 1991, 113, 4025–
4026.
[5] E. J. Corey, S. C. Virgil, H. Cheng, C. H. Baker, S. P. T. Mat-
suda, V. Singh, S. Sarahar, J. Am. Chem. Soc. 1995, 117, 11819–
11820.
[26]
[27]
[28]
[29]
[30]
[31]
[6] K. U. Wendt, K. Poralla, G. E. Schulz, Science 1997, 277,
1811–1815.
Published Online: September 30, 2009
Eur. J. Org. Chem. 2009, 5731–5737
© 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
5737