Double oxidation of the cyclic nonaketide dihydromonacolin L to monacolin J by a single cytochrome P450 monooxygenase, LovA

Article abstract of DOI:10.1021/ja201138v

Lovastatin, a cyclic nonaketide from Aspergillus terreus, is a hypercholesterolemic agent and a precursor to simvastatin, a semi-synthetic cholesterol-lowering drug. The biosynthesis of the lovastatin backbone (dihydromonacolin L) and the final 2-methylbutyryl decoration have been fully characterized. However, it remains unclear how two central reactions are catalyzed, namely, introduction of the 4a,5-double bond and hydroxylation at C-8. A cytochrome P450 gene, lovA, clustered with polyketide synthase lovB, has been a prime candidate for these reactions, but inability to obtain LovA recombinant enzyme has impeded detailed biochemical analyses. The synthetic codon optimization and/or N-terminal peptide replacement of lovA allowed the lovA expression in yeast (Saccharomyces cerevisiae). Both in vivo feeding and in vitro enzyme assays showed that LovA catalyzed the conversion of dihydromonacolin L acid to monacolin L acid and monacolin J acid, two proposed pathway intermediates in the biosynthesis of lovastatin. LovA was demonstrated to catalyze the regio- and stereo-specific hydroxylation of monacolin L acid to yield monacolin J acid. These results demonstrate that LovA is the single enzyme that performs both of the two elusive oxidative reactions in the lovastatin biosynthesis.

Full text of DOI:10.1021/ja201138v

COMMUNICATION
pubs.acs.org/JACS
Double Oxidation of the Cyclic Nonaketide Dihydromonacolin L to
Monacolin J by a Single Cytochrome P450 Monooxygenase, LovA
Jorge Barriuso,†,# Don T. Nguyen,†,# Jesse W.-H Li,§ Joseph N. Roberts,† Gillian MacNevin,†
Jennifer L. Chaytor,§ Sandra L. Marcus,§ John C. Vederas,*,§ and Dae-Kyun Ro*,†
†Department of Biological Sciences, University of Calgary, 2500 University Drive NW, Calgary T2N 1N4, Canada
§Department of Chemistry, University of Alberta, Edmonton T6G 2G2, Canada
S Supporting Information
b
ABSTRACT: Lovastatin, a cyclic nonaketide from Aspergillus
terreus, is a hypercholesterolemic agent and a precursor to
simvastatin, a semi-synthetic cholesterol-lowering drug. The
biosynthesis of the lovastatin backbone (dihydromonacolin L)
and the final 2-methylbutyryl decoration have been fully char-
acterized. However, it remains unclear how two central reactions
are catalyzed, namely, introduction of the 4a,5-double bond and
hydroxylation at C-8. A cytochrome P450 gene, lovA, clustered
with polyketide synthase lovB, has been a prime candidate for
these reactions, but inability to obtain LovA recombinant
enzyme has impeded detailed biochemical analyses. The syn-
thetic codon optimization and/or N-terminal peptide replace-
ment of lovA allowed the lovA expression in yeast (Saccharomyces
cerevisiae). Both in vivo feeding and in vitro enzyme assays
showed that LovA catalyzed the conversion of dihydromonaco-
lin L acid to monacolin L acid and monacolin J acid, two
proposed pathway intermediates in the biosynthesis of lovasta-
tin. LovA was demonstrated to catalyze the regio- and stereo-
specific hydroxylation of monacolin L acid to yield monacolin
J acid. These results demonstrate that LovA is the single enzyme
that performs both of the two elusive oxidative reactions in the
lovastatin biosynthesis.
ovastatin (open form acid 5a, lactone 5b) is a natural
Figure 1. (A) Lovastatin gene cluster. Two cytochrome P450 genes are
L
polyketide product produced from the filamentous fungus,
shown as open triangles. Black arrows are the genes involved in
lovastatin biosynthesis, and the gray arrows are the genes of unknown
functions. (B) Proposed biosynthetic pathway for lovastatin. Com-
pounds are shown as their hydroxy acids a, but corresponding lactone
forms b (e.g., 1b) are frequently isolated. The hydroxy acid forms
(1aꢀ5a) will exist as salts in in vivo and in vitro conditions.
Aspergillus terreus. Lovastatin, its natural analogues such as
compactin, and their semi-synthetic derivatives simvastatin
(Zocor) and pravastatin (Pravachol) are potent and widely
prescribed cholesterol-lowering drugs.1 These compounds are
effective competitive inhibitors of (3S)-hydroxy-3-methylglutaryl-
coenzyme A reductase, a rate-limiting enzyme in the cholesterol
biosynthetic pathway. The gene cluster for lovastatin biosynthesis was
identified from A. terreus (Figure 1A), and subsequent biochemical
studies revealed that the iterative type I polyketide synthase (PKS)
encoded in lovB catalyzes the synthesis of dihydromonacolin L acid
(1a) in concert with an accessory enzyme, LovC (Figure 1B).2
Recently, it has been shown that in vitro reconstitution of purified
LovB and LovC recombinant enzymes could synthesize the enzyme-
conjugated 1a.3 Although this PKS enzymeꢀproduct complex was
not able to off-load the synthesized product, the addition of a fungal
thioesterase (TE) enzyme in trans facilitated the release of 1a from the
LovB and LovC enzyme complex. Therefore, coordinated reactions
of purified LovB, LovC, and trans-TE can synthesize a key lovastatin
intermediate, 1a, in vitro. The final C-8 side-chain modification is also
well studied.2,4 The 2-methylbutyryl side chain is synthesized by the
second PKS, LovF, as an enzyme-bound thioester and is transferred
directly from this enzyme to the C-8 hydroxyl group of monacolin J
acid (4a) by the acyl transferase, LovD.
Significant progress has been made in elucidating lovastatin
biosynthesis at the entry point (LovB and LovC) and at the last
step of side-chain decoration (LovD and LovF). However, the central
oxidative transformation of 1a to 4a has yet to be fully understood.
During the lovastatin purification from A. terreus, 3R-hydroxy-3,
Received: February 5, 2011
Published: April 15, 2011
r
2011 American Chemical Society
8078
dx.doi.org/10.1021/ja201138v J. Am. Chem. Soc. 2011, 133, 8078–8081
|

Journal of the American Chemical Society
COMMUNICATION
Figure 2. Illustration of the N-terminal modification of LovA (A) and
the immunoblot analysis of LovA derivatives and CPR (B). Lettuce P450
(GenBank number GU198171) was used as a control to assess the expression
levels of LovA derivatives, and commercial antibodies (anti-FLAG and anti-
cMyc) were used to detect the tagged recombinant enzymes.
5-dihydromonacolin L (2a) was also isolated, and subsequent
chemical analysis demonstrated that this unstable compound under-
goes dehydration to yield monacolin L acid (3a).5 In addition,
microsomes isolated from A. terreus could catalyze the C-8 hydro-
xylation of 3a to produce 4a in vitro.6 This C-8 hydroxylation reaction
was blocked by carbon monoxide, implying that the C-8 hydroxyla-
tion of 3a is catalyzed by cytochrome P450 monooxygenase (P450).
The gene cluster for lovastatin biosynthesis indeed encodes two P450
genes, lovA and ORF17 (Figure 1A).2 The lovA gene shares its bi-
directional promoter with lovB with only a 482-bp distance, and
therefore it was logical to postulate that lovA is involved in oxidative
modification of 1a. In agreement with this idea, genetic disruption of
lovA in A. terreus resulted in the complete absence of downstream
intermediates after 1a.7 This genetic evidence demonstrated the
critical role of lovA in oxidative transformation of 1a and also suggested
that the first oxidation is catalyzed by P450 enzyme LovA. Collectively,
these results suggested that two consecutive hydroxylations, both
catalyzed by P450(s), are responsible for the conversion of 1a to 4a.
LovA is the prime candidate for at least one of these oxidative reactions.
Expression of lovA in Escherichia coli was not successful, and
yeast (Saccharomyces cerevisiae) was pursued as an alternative host. A
P450 enzyme of this type requires a redox partner, such as
cytochrome P450 oxidoreductase (CPR), to receive reducing
equivalents from NADPH.8 It was reported that native CPR activity
is limiting when heterologous P450 is expressed in yeast.9 Therefore,
we first isolated and characterized A. terreus CPR. Primary sequences
of CPR are highly conserved in eukaryotes, and the BLAST search
identified A. terreus cpr gene from the sequence database. In order to
verify CPR activity, cpr cDNA was isolated from A. terreus, cloned
into the high-copy yeast vector pESC-Leu2d, and expressed in yeast
with a C-terminal cMyc tag.10 Immunoblot analysis using anti-cMyc
antibodies clearly detected the recombinant CPR in the yeast micro-
somes (Figure 2B). Using the isolated microsomes, the catalytic
activity of recombinant CPR was measured by monitoring the
reduction of cytochrome c (Cytc). In these assays, microsomes from
cpr-expressing yeast showed 7.3-fold higher Cytc reduction activity
(4430 ( 260 nmol of Cytc minꢀ1 mgꢀ1) than those from the vector
control (610 ( 220 nmol of Cytc minꢀ1 mgꢀ1). The basal CPR
activity from the control was due to endogenous yeast activities, as
previously reported.9 This result substantiated that the A. terreus cpr
expressed in yeast is biochemically active.
Figure 3. Metabolite profiles of in vivo substrate feeding assays using an
HPLC-UV detector. Compounds 4a, 4b, 3a, and 3b were confirmed as
monacolin J acid, monacolin J lactone, monacolin L acid, and monacolin L
lactone, respectively, by comparison to the authentic standards. Dihydromo-
nacolin L (DML, 1b) and its corresponding acid 1a are not detected at this
wavelength (240 nm). IS = internal standard (cinnamic acid, 20 μM).
The lovA gene in C-terminal FLAG-tagged form was cloned in
the same vector coding cpr, and both lovA and cpr were co-expressed
in yeast, using a Gal1 (for cpr) and Gal10 (for lovA) bi-directional
promoter. However, LovA was not detected in the immunoblot
analysis using FLAG antibodies (Figure 2B), although its mRNAs
could be detected by reverse transcriptase PCR analysis. Hence,
inefficient translation and/or sub-cellular targeting were sus-
pected, and three LovA transcript and protein variants were created
to address this problem. First, lovA cDNA was entirely synthesized
with optimized yeast codons (synthetic lovA, S-lovA) by replacing
430 nucleotides in native lovA (Supporting Information). Second, to
ensure its endoplasmic reticulum localization, the N-terminal
58 amino acids of LovA (putative hydrophobic ER-targeting domain)
were replaced with the N-terminal 43 amino acids from lettuce P450
(LsGAO), which previously showed a high level of expression in yeast
(Figure 2A).11 This fusion protein is referred to as hybrid LovA
(H-lovA). Finally, synthetic lovA with lettuce N-terminus was gener-
ated and referred to as hybrid synthetic lovA (HS-lovA). The expres-
sions of these three lovA variants were examined in comparison to the
native lovA by immunoblot (Figure 2B). The immunoblot analysis
clearly showed that both codon optimization and N-terminal en-
gineering of LovA markedly increased the abundance of LovA protein
in the microsomes. The additive effect of the two modifications was
found in HS-lovA. We focused on the characterization of S-LovA and
HS-LovA recombinant enzymes, but H-LovA was not further studied
because it encodes the same amino acids as HS-LovA.
The catalytic activities of S-LovA and HS-LovA were first
evaluated by feeding the substrate 1a (Li salt form) to the culture
medium at 100 μM. In the HPLC diode array detector (DAD)
analysis of culture extract after 8 h of cultivation, four new peaks
were detected at 240 nm in both S-lovA- and HS-lovA-expressing
yeast strains (Figure 3). The retention times and characteristic
UV spectra (unique triple max peaks at ∼240 nm) of these novel
8079
dx.doi.org/10.1021/ja201138v |J. Am. Chem. Soc. 2011, 133, 8078–8081

Journal of the American Chemical Society
COMMUNICATION
products were identical to those of the authentic standards for
3a/b and 4a/b (Figures 3 and S1). However, the catalytic
efficiency of S-LovA for the synthesis of 4a/b was at least
110-fold higher than that of HS-LovA, taking into consideration
that HS-LovA showed 11.9-fold higher expression than S-LovA.
LC-MS analyses of the four compounds confirmed that the
masses of these compounds were consistent with those for 3a/b
and 4a/b. In the negative-ion (ꢀ)LC-MS, 3a and 4a showed
[M ꢀ H]ꢀ ions of m/z 321 and 337, respectively. In the positive
ion (þ)LC-MS, the [M þ H]þ ions of 3a and 4a were labile and
underwent dehydration (ꢀ18 Da) to form [M þ H ꢀ H2O]þ
ions whose m/z values corresponded to the predicted values of
positive ions for 3a (m/z = 305) and 4a (m/z = 321). In
addition, [M þ H]þ ions of the lactone compounds (3b and
4b) were consistent with the predicted masses (m/z = 305 for
3b and m/z = 321 for 4b).
It is potentially feasible that both free acid and lactone forms
(1a and 1b) could be used as LovA substrates. In order to test if
1b can be used as LovA substrate, 100 μM 1b was fed to the yeast
expressing S-lovA and cpr, and the culture was incubated for 8 h.
However, no conversion of 1b was detected by HPLC-DAD
(Figure 3). When the same sample was analyzed by a highly
sensitive LC-MS, the four compounds (3a/b and 4a/b) could be
detected, but their abundance was about 250-fold lower than
those converted from 1a. Therefore, it appears that 3a and 4a
were synthesized from 1a and then converted to the correspond-
ing lactones in an acidic yeast culture medium. In order to verify
this, the same feeding experiments were performed in extended
incubation times (24 h) with varying final pH (3.0ꢀ6.8) in the
medium using different buffer strengths. In acidic conditions
(pH 3), almost all of the monacolin L and monacolin J were
present as their lactone forms (3b and 4b), whereas their free
acid forms (3a and 4a) were dominant in the medium with final
pH 6.8 (Figure S2). This result together with the data from the
1b-feeding assay suggested that 1a is the LovA substrate. The 3b
and 4b apparently resulted from non-enzymatic lactonization in
acidic yeast medium.
Using the pH-optimized conditions, yeast feeding assays were
scaled up (1 L), and 3a and 4a were purified by HPLC. The
structure of the final product 4a was verified by spectral
comparison to authentic standard, and standard NMR analyses
were used to confirm the structure of 3a (Supporting In-
formation). By using FT-ICR-MS, the exact m/z of the
[M ꢀ H]ꢀ for 3a was determined to be 321.20700 and for 4a
to be 337.20212. These values were less than 0.4 ppm deviations
from the theoretical masses. To ensure the reactions were
catalyzed by LovA, in vitro enzyme assays were done using
microsomes prepared from yeast expressing S-lovA and cpr.
When 1a was incubated with the microsomes, 3a and 4a were
produced as shown by LC-MS analysis (Figure 4A). Two
additional [M ꢀ H]ꢀ ions displaying m/z 339 were detected.
One of these compounds is likely to be 3R-hydroxy-3,5-dihy-
dromonacolin L acid (2a), a reported intermediate in the
lovastatin biosynthesis, and we propose that the other compound
is its isomer, 4aR-hydroxy-4a,5-dihydromonacolin L acid (Figure
S3, Supporting Information). No conversion was detected when
1b was incubated with the microsomes, consistent with the
in vivo feeding experiment. As many P450 enzymes catalyze
epoxidations, it has been proposed that 3,4-epoxy-dihydromo-
nacolin L could be a LovA reaction intermediate.7 To examine
this possibility, the pure R and β isomers of 3,4-epoxy-dihy-
dromonacolin L (open forms a, m/z 339) were chemically
Figure 4. In vitro LovA enzyme assays. Total negative ion scans were
performed by LC-MS. Selective ions of m/z 323, 339, 321, and 337 were
used to detect the metabolites shown in Figure 1B. Beside the peaks, m/z
values of [M ꢀ H]ꢀ ions are given. (A) 1a, 4a,5-dihydromonacolin L
acid (substrate), was incubated with the microsomes from yeast expres-
sing either cpr only or cpr and S-lovA. Asterisks indicated compounds
displaying negative ions of m/z 339, proposed to be 2a and its isomer
4aR-hydroxy-4a,5-dihydromonacolin L acid (Supporting Information).
(B) 3a, monacolin L acid, was used as substrate in the same experimental
conditions as described for panel A.
synthesized from 1b and also independently incubated with
the microsomes, but these were not transformed further (data
not shown). In addition, MS/MS analysis suggested that the two
m/z 339 compounds from the in vitro assays are not 3,4-epoxy-
dihydromonacolin L (Figure S4, Supporting Information). On
the basis of these results, we propose that 2a is synthesized by a
hydrogen (1a C-4a hydrogen) abstraction and subsequent oxy-
gen re-bound (i.e., C-3 hydroxyl group) onto the allylic radical
(Figure S3). Using scaled-up yeast cultures (1 L), we attempted
to purify the two compounds with m/z 339 after feeding 1a, but
the low abundance of these two compounds did not allow us to
acquire sufficient amounts for NMR analyses.
In the assays described thus far using 1a, it cannot be excluded
that the second reaction (the conversion of 3a to 4a) is catalyzed
by an unknown yeast enzyme. Also, 3a could, in principle, be a
reaction shunt product that is released from LovA but cannot be
re-introduced into the LovA biosynthetic pathway. To address
these questions, the purified 3a was used as a substrate for in vitro
LovA assays. In these assays, clear conversion of 3a to 4a was
observed, with no trace of catalytic conversion in the control
microsomes (Figure 4B). The Km value of LovA for 3a was
determined to be 6.2 ( 1.1 μM, and the microsomes showed
Vmax = 9.1 ( 0.5 pmol minꢀ1 mgꢀ1. The sufficiently low Km
value supported the physiological relevance of LovA activity in
A. terreus. These results demonstrate that 3a is a true intermedi-
ate in the lovastatin biosynthetic pathway.
8080
dx.doi.org/10.1021/ja201138v |J. Am. Chem. Soc. 2011, 133, 8078–8081

Journal of the American Chemical Society
COMMUNICATION
In summary, detailed in vivo and in vitro characterizations of
LovA recombinant enzyme demonstrated that LovA is the
missing link in the lovastatin biosynthesis, catalyzing the two
central oxidative reactions from 1a to 4a. Recently, other double
oxidations of bacterial polyketides by single P450 tailoring
enzymes during biosynthesis have been reported.12 These LovA
studies have completed the molecular characterizations of an
entire set of genes required for the lovastatin biosynthesis and
hence provide an opportunity to synthesize lovastatin by means
of metabolic engineering.
(7) Sorensen, J. L.; Auclair, K.; Kennedy, J.; Hutchinson, C. R.;
Vederas, J. C. Org. Biomol. Chem. 2003, 1, 50–59.
(8) (a) Coon, M. J. Annu. Rev. Pharmacol. Toxicol. 2005, 45, 1–25.
(b) Ortiz de Montellano, P. R.; De Voss, J. J. Nat. Prod. Rep. 2002,
19, 477–493.
(9) (a) Ro, D. K.; Ehlting, J.; Douglas, C. J. Plant Physiol. 2002,
130, 1837–1851. (b) Pompon, D.; Louerat, B.; Bronine, A.; Urban, P.
Methods Enzymol. 1996, 272, 51–64.
(10) Ro, D. K.; Ouellet, M.; Paradise, E. M.; Burd, H.; Eng, D.;
Paddon, C. J.; Newman, J. D.; Keasling, J. D. BMC Biotechnol. 2008,
8, 83.
(11) Nguyen, D. T.; Gopfert, J. C.; Ikezawa, N.; Macnevin, G.;
Kathiresan, M.; Conrad, J.; Spring, O.; Ro, D. K. J. Biol. Chem. 2010,
285, 16588–16598.
’ ASSOCIATED CONTENT
(12) (a) Kudo, F.; Motegi, A.; Mizoue, K.; Eguchi, T. Chembiochem
2010, 11, 1574–1582. (b) Wilson, M. C.; Gulder, T. A. M.; Mahmud, T.;
Moore, B. S. J. Am. Chem. Soc. 2010, 132, 12757–12765. (c) Anzai, Y.; Li,
S.; Chaulagain, M. R.; Kinoshita, K.; Kato, F.; Montgomery, J.; Sherman,
D. H. Chem. Biol. 2008, 15, 950–959. (d) Richter, M. E.; Traitcheva, N.;
Knupfer, U.; Hertweck, C. Angew. Chem. Int. Ed. 2008, 47, 8872–8875.
S
Supporting Information. Experimental details, synthetic
b
gene sequences, and NMR and HPLC data. This material is
available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Author
daekyun.ro@ucalgary.ca; john.vederas@ualberta.ca
Author Contributions
#These authors contributed equally.
’ ACKNOWLEDGMENT
We thank Dr. J€urgen Schmidt (Leibniz-Institute of Plant
Biochemistry) for the FT-ICR-MS analysis of 3a and 4a,
Dr. Randy Whittal (University of Alberta) for MS/MS analysis
of compounds with m/z 339, and Zhizeng Gao (University of
Alberta) for assistance with NMR analysis. This work was
supported by the Natural Sciences and Engineering Research
Council of Canada (NSERC), the Canada Research Chair
(CRC) program, and the Canada Foundation for Innovation
(CFI). We also acknowledge the Josꢀe Castillejo Fellowship from
the Spanish Ministry of Education and Science awarded to J.B.
and the Markin Undergraduate Student Research Program
(USRP) in Health and Wellness at University of Calgary awarded
to J.N.R.
’ REFERENCES
(1) (a) Campbell, C. D.; Vederas, J. C. Biopolymers 2010,
93, 755–763. (b) Sadowitz, B.; Seymour, K.; Costanza, M. J.; Gahtan,
V. Vasc. Endovasc. Surg. 2010, 44, 421–433.
(2) (a) Hutchinson, C. R.; Kennedy, J.; Park, C.; Auclair, K.;
Vederas, J. C. In Handbook of Industrial Mycology; An, Z., Ed.; Marcel
Dekker Inc.: New York, 2005; pp 479ꢀ492. (b) Kennedy, J.; Auclair, K.;
Kendrew, S. G.; Park, C.; Vederas, J. C.; Hutchinson, C. R. Science 1999,
284, 1368–1372.
(3) Ma, S. M.; Li, J. W.; Choi, J. W.; Zhou, H.; Lee, K. K.; Moorthie,
V. A.; Xie, X.; Kealey, J. T.; Da Silva, N. A.; Vederas, J. C.; Tang, Y. Science
2009, 326, 589–592.
(4) (a) Gao, X.; Xie, X.; Pashkov, I.; Sawaya, M. R.; Laidman, J.;
Zhang, W.; Cacho, R.; Yeates, T. O.; Tang, Y. Chem. Biol. 2009,
16, 1064–1074. (b) Xie, X.; Meehan, M. J.; Xu, W.; Dorrestein, P. C.;
Tang, Y. J. Am. Chem. Soc. 2009, 131, 8388–8389. (c) Xie, X.; Watanabe,
K.; Wojcicki, W. A.; Wang, C. C. C.; Tang, Y. Chem. Biol. 2006,
13, 1161–1169.
(5) (a) Nakamura, T.; Komagata, D.; Murakawa, S.; Sakai, K.; Endo,
A. J. Antibiot. 1990, 43, 1597–1600. (b) Treiber, L. R.; Reamer, R. A.;
Rooney, C. S.; Ramjit, H. G. J. Antibiot. 1989, 42, 30–36.
(6) Komagata, D.; Shimada, H.; Murakawa, S.; Endo, A. J. Antibiot.
1989, 42, 407–412.
8081
dx.doi.org/10.1021/ja201138v |J. Am. Chem. Soc. 2011, 133, 8078–8081