Single-step fermentative production of the cholesterol-
lowering drug pravastatin via reprogramming of
Penicillium chrysogenum
Kirsty J. McLeana,1, Marcus Hansb,1, Ben Meijrinkb, Wibo B. van Scheppingenb, Aad Vollebregtb, Kang Lan Teea,
Jan-Metske van der Laanb, David Leysa, Andrew W. Munroa,2, and Marco A. van den Bergb,2
a
Manchester Institute of Biotechnology, Faculty of Life Sciences, The University of Manchester, Manchester M1 7DN, United Kingdom; and bDSM
Biotechnology Center, 2613 AX, Delft, The Netherlands
Edited by Arnold L. Demain, Drew University, Madison, NJ, and approved January 28, 2015 (received for review October 3, 2014)
The cholesterol-lowering blockbuster drug pravastatin can be
produced by stereoselective hydroxylation of the natural product
compactin. We report here the metabolic reprogramming of the
antibiotics producer Penicillium chrysogenum toward an industrial
pravastatin production process. Following the successful introduction
of the compactin pathway into the β-lactam–negative P. chrysoge-
num DS50662, a new cytochrome P450 (P450 or CYP) from Amy-
colatopsis orientalis (CYP105AS1) was isolated to catalyze the final
compactin hydroxylation step. Structural and biochemical charac-
terization of the WT CYP105AS1 reveals that this CYP is an effi-
cient compactin hydroxylase, but that predominant compactin
binding modes lead mainly to the ineffective epimer 6-epi-prava-
statin. To avoid costly fractionation of the epimer, the enzyme
was evolved to invert stereoselectivity, producing the pharma-
cologically active pravastatin form. Crystal structures of the opti-
mized mutant P450Prava bound to compactin demonstrate how the
selected combination of mutations enhance compactin binding
and enable positioning of the substrate for stereo-specific oxida-
tion. Expression of P450Prava fused to a redox partner in compac-
tin-producing P. chrysogenum yielded more than 6 g/L pravastatin
at a pilot production scale, providing an effective new route to
industrial scale production of an important drug.
pravastatin | P450 engineering | fermentation | statin | cholesterol
S tatin drugs inhibit 3β-hydroxy-3-methylglutaryl CoA (HMG-
CoA) reductase, the enzyme catalyzing the rate-limiting step
in cholesterol biosynthesis. Statins reduce “bad” plasma (LDL)
cholesterol levels and are thus effective against hypercholestero-
lemia. Several statins are on the market; the most prominent
being the completely synthetic atorvastatin (Lipitor), the semi-
synthetic simvastatin (Zocor), and pravastatin (Pravachol) (Fig. 1)
(1, 2). Atorvastatin was identified by a Pfizer R&D pharma-
ceutical screening program using entirely unnatural lead struc-
tures (3); simvastatin and pravastatin, discovered by Merck and
Sankyo, respectively, are examples of synthetic variants of the
naturally occurring statins lovastatin and compactin, and have
superior pharmacokinetic properties (4, 5). Lovastatin and
compactin are produced by filamentous fungi such as Aspergillus
terreus and Penicillium citrinum, respectively (Fig. 1A). In recent
years, their biosynthetic gene clusters were identified and com-
prise a set of nine proteins, including a HMG-CoA reductase, a
transporter, and a transcriptional activator. Two large polyketide
synthases produce the nonaketide precursor and methylbutyrate
side chain. Additional oxidoreductase and hydroxylase enzymes
lead to the production of lovastatin and compactin (6, 7). Simva-
statin (Fig. 1B) is a semisynthetic variant of lovastatin wherein
the methylbutyrate side chain is converted into a dimethylbutyrate
moiety. This outcome can be achieved by direct methylation or by
deacylation and subsequent addition of a de novo chemically
synthesized dimethylbutyrate group (8). Pravastatin (Fig. 1B) is
obtained after stereoselective hydroxylation of compactin at the
C-6 position. Industrially, this is achieved by a two-step production
www.pnas.org/cgi/doi/10.1073/pnas.1419028112
process. First, fermentation of P. citrinum produces compactin.
Next, the statins are purified and the lactone ring is opened by
addition of sodium hydroxide. After neutralization, the open
lactone form of compactin is converted to pravastatin in a bio-
transformation step using the bacterium Streptomyces carbophilus.
This organism contains a cytochrome P450 enzyme (P450 or
CYP) that is capable of hydroxylating compactin stereospecifi-
cally at C-6 with an overall yield of 70% (P450sca-2) (9). CYPs
are hemoproteins of biotechnological interest due to their ability
to perform regio- and stereoselective oxygen insertion chemistry.
They are best known for their pivotal roles in human steroid
synthesis and in drug/xenobiotic detoxification and metabolism
(10). In addition, nonhuman CYPs have increasing importance
in production of antibiotics and other agents beneficial to
human health (11). Among the statins, pravastatin has the
lowest potential for drug interactions (e.g., with antibiotics) as
it is not extensively metabolized by human CYPs (12). However,
low P450sca-2 activity during biotransformation, compactin tox-
icity, and low pravastatin yield hamper large-scale manufacturing
Significance
Statins are successful widely used drugs that decrease the risk
of coronary heart disease and strokes by lowering cholesterol
levels. They selectively inhibit the key regulatory enzyme of
the cholesterol synthesis pathway, thus lowering levels of
plasma LDL (bad) cholesterol. Pravastatin is one of the leading
and most effective statins, derived from the natural product
compactin. However, pravastatin production involves a costly
dual-step fermentation and biotransformation process. Here
we present a single-step fermentative method for production
MICROBIOLOGY
of the active drug pravastatin. Reprogramming of the antibiotics-
producing fungus Penicillium chrysogenum, with discovery and
engineering of an enzyme involved in the hydroxylation of com-
pactin, enables high level fermentation of the correct form of
pravastatin to facilitate efficient industrial-scale statin drug
production.
Author contributions: K.J.M., M.H., W.B.v.S., A.W.M., and M.A.v.d.B. designed research;
K.J.M., M.H., B.M., W.B.v.S., A.V., K.L.T., and J.-M.v.d.L. performed research; K.J.M., M.H.,
CHEMISTRY
B.M., W.B.v.S., A.V., K.L.T., J.-M.v.d.L., D.L., A.W.M., and M.A.v.d.B. analyzed data; and
K.J.M., M.H., D.L., A.W.M., and M.A.v.d.B. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Freely available online through the PNAS open access option.
Data deposition: The atomic coordinates and structure factors have been deposited in the
Protein Data Bank, www.pdb.org (PDB ID code 4OQS and 4OQR). The sequence reported
in this paper has been deposited in the GenBank database (accession no. KF751385).
1
K.J.M. and M.H. contributed equally to this work.
2
To whom correspondence may be addressed. Email: andrew.munro@manchester.ac.uk
or marco.berg-van-den@dsm.com.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1419028112/-/DCSupplemental.
PNAS | March 3, 2015 | vol. 112 | no. 9 | 2847–2852

(13). Here, we present a radically different approach: development
of a one-step, fully fermentative pravastatin production process.
The industrial high producing P. chrysogenum strain DS17690 was
genetically reprogrammed by introducing the full compactin bio-
synthetic pathway and an evolved compactin hydroxylase CYP
from Amycolatopsis orientalis, leading to an efficient pravastatin
producing strain.
Results
P. chrysogenum as a Platform for Compactin Production. The β-lactam
antibiotics producer P. chrysogenum is a well-developed industrial
microorganism that underwent numerous rounds of classical strain
improvement, leading to current penicillin production titers of
more than 50 g/L (14). Because these accumulated mutations
contributed to the industrial robustness and the high flux from
glucose to secondary metabolites, we speculated that this highly
adapted organism would be suitable for producing other sec-
ondary metabolites. To produce other pharmaceuticals in this
species, an isolate completely devoid of β-lactam antibiotics was
needed. Deleting all penicillin biosynthetic genes in a high pro-
ducing strain led to such a host (15). To test whether the dele-
tions had any adverse effect on secondary metabolite production
potential, we reintroduced the biosynthetic genes, resulting in
the recovery of penicillin production (Table S1). Next, the
complete compactin (ML-236B) gene cluster from P. citrinum
(7) was subcloned on three plasmids (Fig. S1). All nine mlc genes
were kept under control of their natural P. citrinum promoter
and terminator regions, assuming that these would function simi-
larly in P. chrysogenum. Transformants were selected on acetamide
due to cotransformation of an acetamidase expression cassette.
As DNA integration in P. chrysogenum occurs almost exclusively
via the nonhomologous end joining (NHEJ) pathway, all frag-
ments randomly integrate in the genome. Successful integration
was confirmed by PCR and Southern blot analyses (SI Materials
and Methods). Stable transformants harboring all DNA frag-
ments were cultivated in synthetic medium and analyzed for
statin production. As expected, all three fragments must be
cotransformed in one strain to enable statin formation (Table
S2). Several strains produced relatively high amounts of statins:
up to 624 mg/L in shake flasks (Fig. 2A). This titer is remarkable,
because the natural compactin producer P. citrinum produces only
19 mg/L (Fig. 2A). These results confirm that the P. chrysogenum
β-lactam free platform strain retains its features for high sec-
ondary metabolite productivities, even for heterologous prod-
ucts. Moreover, the P. citrinum promoters are very efficient in
P. chrysogenum and do not underperform compared with known
strong homologous promoters such as pcbC (Table S2).
Deletion of Esterase Activity Improves Compactin Yield in P. chrysogenum.
A surprisingly high ratio of deacylated (ML-236A) vs. acylated
(compactin) statins (Fig. 1) in P. chrysogenum (up to 60%; Fig. 2A)
suggested an imbalance in the introduced pathway or a competing
activity. Initial experiments confirmed in vivo cleavage of the
2848 | www.pnas.org/cgi/doi/10.1073/pnas.1419028112
Fig. 1. Structures of relevant statins and deriva-
tives. (A) Naturally produced statins (i, compactin;
ii, lovastatin). (B) Semisynthetic pharmaceutical
statins [i, pravastatin (Pravachol); ii, simvastatin (Zocor)].
(C) Alternative compactin oxidation products (i, 6-epi-
pravastatin; ii, 3-α-iso-pravastatin). (D) Deacylated
derivatives [i, deacylated compactin (ML-236A);
ii, deacylated pravastatin).
methylbutyrate side chain of compactin (SI Results), most likely
by an inducible esterase or lipase. Due to numerous candidate
genes in the P. chrysogenum genome (>200) (16), a biochemical
approach was used to identify the responsible enzyme. Com-
pactin producing P. chrysogenum strains were cultivated under
statin deacylating (induced by using urea as nitrogen source; SI
Results) and control (i.e., noninducing, by using yeast extract as
nitrogen source) conditions (Fig. 2B). Proteins from cell lysates
were fractionated by hydrophobic interaction chromatography
and fractions with high compactin deacylation activity analyzed
by SDS/PAGE. Relevant bands were characterized by MS
analysis after tryptic digestion. A protein encoded by Pc15g00720
(GenBank accession no. CAP82958.1), a putative esterase, was
only detectable in samples with increased statin deacylation. De-
leting this gene reduced statin deacylation to an absolute minimum,
even after cultivation with urea as sole N source (Fig. S2).
Identification of an A. orientalis Compactin Hydroxylase. The S. carbo-
philus P450sca-2 enzyme (9) functions using NADPH-ferredoxin
oxidoreductase and ferredoxin redox partners. Attempts to express
P450sca-2 directly in our P. chrysogenum compactin-producing
strain did not lead to detectable pravastatin (Table S2). Several
eukaryotic and prokaryotic systems capable of hydroxylating
compactin have been reported (17–22), but none give 100%
conversion nor show real improvements over S. carbophilus.
Moreover, as the natural P450sca-2 redox partners are unknown
and we required compactin hydroxylase functionality in a fungal
host, we set out to isolate a new hydroxylating enzyme that is
functional in eukaryotes. Several microorganisms were screened
for compactin to pravastatin conversion, and the actinomycete
A. orientalis showed the greatest hydroxylation activity (up to 100%;
SI Materials and Methods). To clone the relevant gene, a genome
library was constructed and screened for compactin hydroxylation.
Fig. 2. Statin production levels in Penicillium strains. Red, ML-236A; blue,
compactin. (A) Left bar, P. citrinum NRRL 8082; other bars, P. chrysogenum
strains. (B) Influence of nitrogen source on ratio of deacylated (ML-236A) to
acylated (compactin) statin. (a) Sodium nitrite; (b) yeast extract; (c) arginine;
(d) lysine; (e) urea; (f) acetamide; (g) ammonium sulfate; (h) hydroxylamine;
(i) peptone; (j) ammonium nitrite.
McLean et al.
Two positive clones were isolated, and sequence analysis revealed
a complete ORF with 40–60% amino acid identity to known CYP
enzymes in one of the clones (Fig. S3), containing a conserved heme
binding motif around the cysteine ligand to the CYP heme iron
(FGGGIHQCLG) (5). This ORF was cloned into the Escherichia
coli expression vector pBAD-DEST49 under control of the minimal
arabinose (araBAD) promoter, and activity assays confirmed the
CYP enzyme as a compactin hydroxylase (cmpH; accession no.
KF751385; CYP105AS1; Fig. S3).
Expression of CYP105AS1 in P. chrysogenum Leads to epi-Pravastatin.
One-step fermentative production of pravastatin requires the
A. orientalis CYP105AS1 to be functional in compactin-producing
P. chrysogenum. Various publications describe the use of re-
ductase domains from self-sufficient CYPs, such as the RhF from
Rhodococcus sp. and BM3 from Bacillus megaterium, for acti-
vation of type I CYP enzymes (23–25). Such hybrid fusion
enzymes may catalyze oxidation reactions in absence of other
oxidoreductase/ferredoxin partners. A codon-optimized version
of CYP105AS1 was fused in frame to the RhF reductase (FMN-
and 2Fe-2S cluster-binding) domain, and the resulting hybrid gene
was expressed under the strong pcbC promoter of P. chrysogenum
(Fig. 3A). On introduction to P. chrysogenum compactin-producing
strains, 19 of 192 transformants were identified that produced sig-
nificant amounts of pravastatin (up to 550 mg/L; Fig. 3B).
The conversion activity of CYP105AS1 expressed in P. chrysogenum
was good, but not complete: ∼90% of the secreted statin was
pravastatin (Fig. 3B; WT and WT-CpO). However, detailed
analyses with NMR and liquid chromatography (LC)-MS/MS
revealed that, although the correct compactin C-6 was hydrox-
ylated, the wrong stereoisomer 6-epi-pravastatin (Fig. 1C) was
produced almost exclusively. This isomer was also reported for S.
carbophilus bioconversion, but at around 10% (9). Here, E. coli
and P. chrysogenum strains harboring CYP105AS1 produced 98–
100% 6-epi-pravastatin and maximally only ∼2% pravastatin
(Table S3 and Fig. 3B). Although producing the nondesired
epimer, the transformants demonstrated that functional expres-
sion of the self-sufficient bacterial CYP in P. chrysogenum is
sufficient to produce high amounts of hydroxylated statin.
Evolution of CYP105AS1 Toward an Efficient Pravastatin Synthase.
The CYP105AS1 structure was solved to 2.05 Å using molecular
replacement with the vitamin D hydroxylase CYP105A1 (26)
[Protein Data Bank (PDB) ID code 2ZBY; 44% sequence
identity]. The final structure contains residues Glu10 to Asp400
with the exception of the polypeptide stretch from Ser77 to
Ala85, with an RMSD of 1.38 Å for 355 Ca atoms (Fig. 4), and is
most similar to the filipin I:CYP105P1 complex (27) (PDB ID
code 3ABA; 42% sequence identity). The disordered nature of
the BC-loop stretch and the open position of the FG-loop ren-
ders the heme pocket and adjacent I-helix solvent accessible in
the ligand-free state, as is observed for other CYP105 members
(including CYP105P1). Extensive attempts to obtain a cocrystal
structure of CYP105AS1 with compactin/pravastatin revealed
uninterpretable electron density in the active site (Fig. S4),
suggesting there is ample space for the substrate to occupy
multiple configurations. We hypothesized that minor changes in
protein structure could influence and/or restrict the conforma-
tional landscape of the substrate to achieve different stereo-
chemistry, resulting in increased pravastatin production at the
expense of 6-epi-pravastatin. To this end, the CYP105AS1 gene
was mutated by error-prone PCR so that each clone contained
two to three mutations on average. PCR products were cloned in
pBAD-DEST under control of arabinose induction. Per round,
5,000 clones were analyzed for compactin hydroxylation, focus-
ing on the ratio of pravastatin to 6-epi-pravastatin. After a single
round of error-prone PCR, mutants with significantly improved
pravastatin:6-epi-pravastatin ratios were identified (Table S3).
Although CYP105AS1 (WT) hydroxylates compactin with a
pravastatin:6-epi-pravastatin ratio of 3:97, the best mutant (#6)
had a ratio of 48:52. Sequence analysis of mutants with the
McLean et al.
Fig. 3. Expression and evolution of the A. orientalis CYP105AS1. (A) Sche-
matic representation of the self-sufficient CYP105AS1 expression construct.
(B) Statin production in P. chrysogenum transformants.
highest pravastatin:6-epi-pravastatin ratios revealed that, in
most cases, a single amino acid change altered stereoselectivity.
Interestingly, in addition to active site mutations, outer shell
mutations were also identified. Furthermore, expression mutants
were isolated (#1; I233T) that did not alter the ratio but im-
proved activity fourfold. The 10 best mutants were taken as
templates and site-saturated fusion PCR used to achieve all
possible combinations. Screening these allowed isolation of
mutants with corresponding pravastatin:6-epi-pravastatin ratios
up to 96:4 (Table S3). Thus, with only two rounds of mutagenesis
and three to five amino acid changes, the ratio of the compactin
hydroxylation products pravastatin, 6-epi-pravastatin was com-
pletely inverted, and CYP105AS1 was engineered into a pravastatin
synthase (P450Prava).
Characterization of P450Prava Explains Enantiomeric Specificity. To
characterize in vitro the influence of selected mutations on
compactin affinity and CYP structure, CYP105AS1, five in-
dividual point mutants and combinations thereof (Table S4)
were expressed and purified. UV-visible (UV-Vis) spectros-
copy revealed a typical CYP heme spectrum for WT CYP105AS1,
with a Soret maximum at 419 nm, and α/β (Q) bands at 570/537 nm,
respectively. Subtle variations were observed for mutant forms
(Table S4). Both compactin and pravastatin induced Soret shifts
(toward 393 nm) through displacement of the heme iron axial
water ligand, causing a low spin (LS) to high spin (HS) shift of
the ferric heme iron. CYP105AS1 did not fully convert to HS on
saturation with compactin, displaying an ∼47% conversion and a
MICROBIOLOGY
Kd value of 29.3 ± 1.1 μM (Table S4). The compactin-saturated
WT CYP105AS1 spectrum has a prominent HS shoulder at
395 nm, and a LS Soret peak shifted from 419 to 417 nm.
All individual mutations improve both compactin affinity and
the extent of HS shift and had additive effects. P450Prava dis-
plays the best Kd (1.4 ± 0.4 μM), which is a 21-fold improve-
ment over WT CYP105AS1, with near full conversion to the
substrate-bound HS form and a Soret maximum at 393 nm
(Fig. S5).
CHEMISTRY
The cocrystal structure of P450Prava with compactin at 1.8 Å
led to a readily interpretable electron density for the ligand (Fig. 4).
Comparison with the CYP105AS1 open structure shows that
compactin binding to P450Prava reorientates the BC-loop region
(with the Pro82-Ala88 region now disordered) and reorganizes
the FG helices. This closing motion shields the ligand and active
site from bulk solvent, as previously observed in other CYP105
enzymes (26, 27). Compactin displaces the aqua-sixth ligand
from the heme iron and is bound in an active conformation with
the C6 carbon placed within 4.7 Å of the heme iron (Fig. 4 B and
D). Direct protein-ligand interactions are limited to hydrophobic
contacts with residues from the BC-loop (Phe76, Pro80, Trp93,
Thr95), the FG-loop (Met179, Val180, Val182), the I-helix
PNAS | March 3, 2015 | vol. 112 | no. 9 | 2849

Fig. 4. Structure of the WT CYP105AS1 and P450Prava. (A) An overlay of the
WT CYP105AS1 (in blue) and the P450Prava mutant structure (in green). The
positions of the mutations are indicated by spheres. (B) Stereo-and regio-
selectivity of the P450 prava mutant. The compactin substrate (yellow) is
orientated such that the C6 Hpro-S (β-configuration) is positioned directly
adjacent to the heme iron. The heme is shown in red spheres with the heme
iron in orange. The compactin C6 carbon is in cyan. (C) The WT CYP105AS1
open active site, with key residues shown in atom colored sticks. (D) The
corresponding view of the P450Prava:compactin complex (compare with C).
Compactin is shown in atom colored sticks (yellow carbons) with the corre-
sponding 2Fo-Fc electron density contoured at 1σ as a blue mesh.
(lle236, Val239, Ala240, Thr244), the K-β1 loop (Val282), and
the C-terminal region (Ala388, Ala389). Indirect hydrogen bonding
contacts, mediated via water molecules, occur between the com-
pactin ester C8-substituent and residues Trp93 and Thr95, in ad-
dition to indirect contact made between the carbonyl lactone
moiety and Thr286. Three of the five mutations are located
within van der Waals contact of compactin (I95T, A180V, and
L236I), each leading to improved surface complementarity be-
tween ligand and protein (Fig. 4). In addition, the I95T mu-
tation strengthens the polar interaction network made with
compactin. In contrast, Q127R and A265N are distant from the
active site and unlikely to have direct effects on ligand affinity/
conformation. In the WT structure, Gln127 is part of an exten-
sive network of polar contacts made between the EG loop and
the D-helix (Fig. S6). This network changes on active site clo-
sure, due to motion of the G-helix and EG-loop with respect to
the D-helix. We suggest that the Q127R mutation disrupts the
open conformation network in particular, thus lowering the en-
ergetic barrier to the formation of the closed conformation and
favoring ligand binding (similar to gatekeeper mutations recently
described for P450BM3) (28). In contrast, the contribution of the
A265N mutation, located at the protein surface and distant from
regions involved in the open to closed transition, cannot be ra-
tionalized from the structural data available.
Expression of P450Prava in a Compactin Production Strain. P450Prava
with a pravastatin:6-epi-pravastatin ratio of 96:4 was fused to the
Rhf reductase and randomly integrated in the best P. chrysogenum
compactin strain. The resulting transformants produced large
amounts of pravastatin; the highest isolate produced 688 mg/L
(Fig. 3B). Most importantly, the improved stereoselectivity ratio
pravastatin:epi-pravastatin of the mutant P450Prava as observed in
E. coli was also confirmed in all P. chrysogenum transformants.
To assess industrial performance of the P450Prava-P. chrys-
ogenum strain, 10-L fed-batch fermentations were carried out (SI
Materials and Methods). After a 200-h fermentation time, the
strain produced more than 6 g/L pravastatin (Fig. S7), with only
minor amounts (<0.5 g/L) of deacylated statins and compactin.
2850 | www.pnas.org/cgi/doi/10.1073/pnas.1419028112
Discussion
Process improvement of existing pharmaceuticals is mostly in-
cremental, involving classical strain and process improvement
for fermentation processes and optimizing yields for chemical
processes. For the semisynthetic molecule pravastatin, process
improvement was also limited to classical strain improvement
and to the cloning of new hydroxylases. Here, we applied a
radical approach and completely redesigned the pravastatin
production process. We hypothesized that using existing in-
dustrial fermentation strains would shorten lead times and in-
crease opportunities to obtain desired strains and product levels.
We used an antibiotic-producing strain of P. chrysogenum as
a chassis. P. chrysogenum is well known for β-lactam production,
but this is an undesirable feature when producing other phar-
maceuticals (29). Therefore, we used a variant in which the
β-lactam biosynthetic genes are removed (15), essentially putting
into practice the suggestion made by Jami et al. (30) to use
P. chrysogenum for other biotechnological purposes. Reintroduction
of the penicillin biosynthetic genes proved that all capabilities were
retained in this mutant, and introducing the compactin cluster
demonstrated that beneficial mutations are not restricted to a single
compound of interest but have general advantages for production of
secondary metabolites. These results demonstrated that regulatory
mechanisms from the donor species, P. citrinum, were fully
functional in the recipient P. chrysogenum. Although not often
mentioned in the literature, degradation of statins via cleavage of
the methylbutyrate side chain was already reported (31, 32).
However, in the P. chrysogenum background, this reaction was
much more pronounced, and efficient pravastatin production
requires strict control via the nitrogen source or by deleting the
relevant esterase, encoded by Pc15g00720. It is tempting to
speculate that non–statin-producing fungi like P. chrysogenum
have evolved this esterase in nature to protect themselves from
the antifungal properties of statins (33), whereas statin-producing
species solved this by an additional gene copy of the target enzyme
HMG-CoA reductase and a specific statin transporter (34); how-
ever, evidence supporting such hypotheses is lacking.
Initial attempts to use the available P450sca-2 to hydroxylate
compactin directly in fungi were unsuccessful, and we attempted
several strategies to improve its function. The S. carbophilus
genome was sequenced and identified a ferredoxin and several
ferredoxin oxidoreductases, which were cointroduced with both
native and codon-optimized P450sca-2 gene variants (under
control of strong P. chrysogenum promoters) into compactin
producing strains, but these approaches did not lead to pravastatin
production (Table S2). Several studies have engineered CYPs by
fusing them to reductase modules in efforts to enhance their cata-
lytic rates (25, 35). However, fusion constructs between P450sca-2
and the P450BM3 reductase or the reductase module of the
Rhodococcus RhF P450-reductase fusion enzyme also did not
result in compactin to pravastatin conversion. Recently, P450sca-2
was successfully mutated and expressed in E. coli (36, 37). Inter-
estingly, the authors did not select a fusion to the Rhf reductase as
the reductase crystal structure is unknown (38), but selected the
Pseudomonas putida reductase/putidaredoxin system (38, 39),
leading to 377.5 mg/L pravastatin from this three-component
redox system.
In contrast, our search for novel CYPs led to the isolation of
A. orientalis CYP105AS1, which readily hydroxylates compactin
in E. coli. A. orientalis was included in our screen as it was
reported to contain broad substrate-range CYPs (40). Although
stand-alone CYP105AS1 showed low activity in P. chrysogenum,
it was successfully expressed fused to RhF reductase, producing
a fully functional enzyme both in E. coli and P. chrysogenum.
The CYP105AS1 structure reveals an open conformation and
suggests multiple compactin binding modes rather than a distinct
enzyme-substrate complex that could guide rational engineering
for improved stereoselectivity. CYP mutations often result in
novel product mixes and therefore new specificities (41), and
we successfully realigned CYP105AS1 stereospecificity in two
mutation rounds, without compromising expression levels. Most
McLean et al.
beneficial mutations were identified in the active site above the
heme (42). The P450Prava/compactin structure reveals how con-
formational freedom of the substrate is restricted to essentially
a single conformation, where substrate is positioned with the
pro-S hydrogen from the C-6 carbon close to the heme iron,
explaining the enantiomeric excess of the mutant. Three muta-
tions, in particular, appear to reshape the active site, enhancing
ligand affinity and redefining the conformational landscape. This
structural change explains the significant difference in regio- and
stereoselectivity compared with CYP105AS1, also outperforming
P450sca-2 (9). P450sca-2 produces 10% 6-epi-pravastatin (9),
whereas the best CYP105AS1 mutants produce only 4%; this is
a decisive difference when considering industrial production at
a several-ton scale. A second unwanted byproduct of hydroxylation,
3α-iso-pravastatin (Fig. 1C; previously reported as 6α-iso-ML-236B)
(9, 31), was hardly observed in our experiments.
The production titers (>6 g/L) obtained by our engineered
strain (Fig. S7) are much better than the classical two-step prava-
statin production, which yields 2–3 g/L pravastatin (20, 43).
Pravastatin secretion was fully operational in P. chrysogenum
when all compactin biosynthetic genes and the P450Prava mutant
were randomly integrated. These findings suggest that the
P. citrinum protein MlcE is also heterologously functional as
a transporter for all statins studied, including pravastatin.
To our knowledge, our results are the first example of har-
nessing the potential of a previously improved industrial pro-
duction strain in a more generic way: the generation of a
platform strain, harboring beneficial mutations, which can be
used for rapid development of novel production strains for
unrelated chemicals.
Materials and Methods
Additional procedures can be found in SI Materials and Methods.
Strains. E. coli DH10B and TOP10 (Invitrogen) were used for gene cloning.
P. chrysogenum strains used were laboratory strain Wisconsin 54-1255 (44),
the high penicillin producer DS17690 (16), the single penicillin gene cluster
derivative DS47274 (15), and the β-lactam free platform strain DS50662 (15).
For compactin reference fermentations and genomic DNA isolation, P. citrinum
NRRL8082 was used (NRRL).
Media. P. chrysogenum was cultivated for 168 h at 25 °C in synthetic medium
(SI Materials and Methods) at 250 (shake flask) or 400–550 rpm (microtiter
plate; MTP). To produce penicillin V, 2 g/L phenoxyacetic acid was added.
P. citrinum spores were harvested from potato dextrose agar (PDA) slants in
0.5% Tween 80 and cultivated as described previously (45).
Construction of a Compactin Producing P. chrysogenum Strain. The P. citrinum
compactin gene cluster was cloned in three parts. The central and 3′-end (14
and 6 kb) were readily PCR amplified and cloned using Gateway technology
(Invitrogen) into entry vectors pDONRP4-P1R and pDONR221, respectively
(Fig. S1). The 18-kb fragment was cloned in two steps. First, the 10- and 8-kb
fragments were cloned separately in pCR2.1 TOPO T/A (Invitrogen) and
combined afterward via compatible restriction ends. The small piece of
remaining and interfering oligonucleotide sequence in mlcA was removed
via exchange of an internal genomic PCR amplified mlcA fragment (4.1-kb
Acc65I fragment). Ligation of all fragments produced the full-length 18-kb
in pDONR41Zeo. All PCR amplified fragments were verified via sequencing.
Linearized plasmids were transformed to P. chrysogenum as described pre-
viously (15).
LC-MS Analysis of Statins. A Waters Acquity ultraperformance liquid chro-
matography (UPLC) coupled to a Micromass ZQ 2000 mass spectrometer and
controlled by Waters MassLynx software was used for LC-MS analysis. LC
separation was performed on a Waters Acquity UPLC ethylene bridged hybrid
Phenyl column with dimensions 50 × 2.1-mm ID (particle size, 1.7 μm) and
with a flow of 0.5 mL/min and temperature of 60 °C. The binary mobile
phase consisted of water with 0.1% formic acid (A) and acetonitrile with
0.1% formic acid (B) (Table 1).
The samples were kept at 10 °C and injected with a 2-μL sample loop,
a needle volume of 30 μL, and at a syringe draw rate of 30 μL/min. The full
loop injection was used with an overfill factor of 5.6 and loop offline after
McLean et al.
Table 1. HPLC time table gradient for the separation of statins
Time (min) A% B% Flow (mL/min) Curve
0.00 67 33 0.50 1
0.65 67 33 0.50 6
1.00 10 90 0.50 6
1.45 10 90 0.50 6
1.50 67 33 0.50 6
1.80 67 33 0.50 6
0.1 min for the Load Ahead option. As needle wash, 400 μL acetonitrile/
water (50:50) was used.
The MS was operated with an Electrospray source in the positive mode
(ES+) with the capillary at 3.00 kV and a cone voltage of 35 V. The source
temperature was 120 °C, and the desolvation temperature was 360 °C. The
desolvation gas flow was 600 L/h, and the cone gas flow was 50 L/h. The MS
settings were as follows: extractor, 3 V; RF lens, 0.3 V; LM resolution, 15.0;
HM resolution, 15.0; ion energy, 0.3 V; multiplier, 650 V; dwell time, 10 ms;
interscan delay, 50 ms; interchannel delay, 10 ms; span, 0. Using this
protocol, statins were measured in the selected ion recording mode as the
sodium adducts m/z 447.25 for the pravastatin isomers and at m/z 431.25 for
compactin. The compounds elute at the following retention times: prava-
statin, 0.61 min; 6-epi-pravastatin, 0.66 min; 3α-iso-pravastatin, 0.78 min;
and compactin, 1.25 min.
Expression of CYP105AS1 and Mutants in P. chrysogenum. Expression of
functional CYPs was done using a synthetic version of pDONR-P4-P1r carrying
an expression cassette consisting of the P. chrysogenum pcbC promoter, the
cmpH gene fused in-frame to the Rhodococcus RhF reductase (23) (codon
optimized for expression in P. chrysogenum), and the P. chrysogenum penDE
terminator (accession no. KF754797). NdeI (CATATG including the CYP ATG
start codon) and SpeI (located in the flexible linker region between the CYP
and reductase domains) sites were engineered to allow introduction of any
CYP variant. Linearized plasmids were transformed to P. chrysogenum.
Evolution of CYP105AS1: Construction of an Error-Prone Library. Mutants of
CYP105AS1 were generated using the Diversify PCR Random Mutagenesis Kit
(Clontech) followed by Gateway cloning. Four different concentrations of
manganese salt were evaluated. Libraries were constructed in a 15-μL BP
reaction using 100 ng PCR-amplified CYP, 200 ng pDONR221, 3 μL BP Clonase II
enzyme mix, and Tris-EDTA buffer (pH 8.0) and incubated for 4 h at 25 °C.
Subsequent transfer to the expression vector [10 μL BP mix, 2 μL (150 ng/μL)
pBAD-DEST49, 3 μL LR clonase II] was done for 2 h at 25 °C. The reaction was
stopped by addition of 4 μg proteinase K (10 min, 37 °C). One microliter was
transformed to E. coli TOP10, and 48 individual clones of each library were
picked to determine mutation frequency.
Screening for Improved Pravastatin:6-epi-Pravastatin Conversion Ratio. Five
thousand individual clones from the error-prone library, with an average of
MICROBIOLOGY
two to three nucleotide mutations per insert, were used to inoculate 96-MTP
containing 200 μL/well Luria-Bertani (LB) medium + 100 μg/mL ampicillin and
grown overnight at 37 °C. Subsequently, 40 μL of the overnight culture was
transferred to deep well 96-MTP filled with 500 μL LB + 100 μg/mL ampicillin
and 1 mM δ-aminolevulinate. Plates were grown for 4 h at 30 °C, 300 rpm.
Subsequently, 0.01% arabinose was added, followed by overnight in-
cubation at 22 °C, 300 rpm. Cells were spun down and resuspended in 300 μL
LB medium containing 100 mM phosphate buffer (pH 6.8), 0.2% glucose,
0.2 g/L compactin, 1 mM δ-aminolevulinate, and 0.01% arabinose. After over-
CHEMISTRY
night incubation at 30 °C, 300 rpm, cells were spun down, and supernatant was
extracted with an equal volume of methanol for LC-MS analysis.
Second Evolution Round: Site Saturation and Error-Prone PCR. DNA sequences
encoding improved enzymes were PCR amplified, purified, and pooled in
equimolar concentrations (10 ng/μL each). Using this as template DNA, site-
saturation PCR was done for six positions (Table S5). DNA fragments were
isolated, mixed in equimolar amounts, and used in a fusion PCR under error-
prone PCR conditions. No additional PCR primers were added. The resulting
library was cloned in pBAD-DEST49 as above.
Crystallization and Structural Determination of CYP105AS1 (WT) and the
P450Prava Mutant. Pure WT CYP105AS1 and P450prava mutant proteins (10–
13 mg/mL) were crystallized in 10 mM Tris and 50 mM NaCl, pH 7.5.
PNAS | March 3, 2015 | vol. 112 | no. 9 | 2851
Compactin (in DMSO) was added to the protein to a final concentration of
1 mM with ≤2% (vol/vol) DMSO for cocrystallogenesis. Protein was centri-
fuged (18,800 × g, 10 min, 4 °C) immediately before crystallization to remove
any precipitate. Initial screens were performed using a Mosquito liquid
handling robot (TTP LabTech) and soluble protein 96-well screens (Molecular
Dimensions). Sitting drops (400 nL) were made by mixing equal volumes of
protein solution and mother liquor and incubating at 4 °C. Crystallization
conditions were identified using the Clear Screen Strategy I (CSS-1), Mor-
pheus, and PACT premier screens (Molecular Dimensions). Crystal microseeding
procedures were used to improve crystal forms (46). Microseed stocks were
prepared by using the Seed Bead kit (Hampton Research) with crystals in 50 μL
of mother liquor from the same condition. Sitting drops were set up as before,
including the microseed at a ratio of 4:1 protein to microseed before mixing
with mother liquor. Diffraction quality crystals of the CYP105AS1 WT protein
were obtained with 0.2 M ammonium chloride, 0.1 M Mes, pH 6.0, and 20%
(wt/vol) PEG 6000, or 0.3 M sodium acetate, 0.1 M Tris, pH 8.5, 10% (wt/vol) PEG
8000, and 10% (wt/vol) PEG 1000. Diffraction quality crystals of the P450Prava
compactin-bound mutant were obtained using 100 mM Bis-Tris propane at
1. Weng TC, Yang YHK, Lin SJ, Tai SH (2010) A systematic review and meta-analysis on
the therapeutic equivalence of statins. J Clin Pharm Ther 35(2):139–151.
2. Lamon-Fava S (2013) Statins and lipid metabolism: An update. Curr Opin Lipidol 24(3):
221–226.
3. Roth BD (2002) The discovery and development of atorvastatin, a potent novel hy-
polipidemic agent. Progress in Medicinal Chemistry, eds King FD, Oxford AW (Elsevier
Science BV, Amsterdam), Vol 40, pp 1–22.
4. Yoshino G, et al. (1986) Effect of CS-514, an inhibitor of 3-hydroxy-3-methylglutaryl
coenzyme A reductase, on lipoprotein and apolipoprotein in plasma of hypercholestero-
lemic diabetics. Diabetes Res Clin Pract 2(3):179–181.
5. Serizawa N (1996) Biochemical and molecular approaches for production of pravastatin,
a potent cholesterol-lowering drug. Biotechnol Annu Rev 2:373–389.
6. Kennedy J, et al. (1999) Modulation of polyketide synthase activity by accessory
proteins during lovastatin biosynthesis. Science 284(5418):1368–1372.
7. Abe Y, et al. (2002) Molecular cloning and characterization of an ML-236B (compactin)
biosynthetic gene cluster in Penicillium citrinum. Mol Genet Genomics 267(5):636–646.
8. Thaper RK, Kumar Y, Kumar SMD, Misra S, Khanna JM (1999) A cost-efficient synthesis
of simvastatin via high-conversion methylation of an alkoxide ester enolate. Org
Process Res Dev 3(6):476–479.
9. Matsuoka T, et al. (1989) Purification and characterization of cytochrome P-450sca
from Streptomyces carbophilus. ML-236B (compactin) induces a cytochrome P-450sca
in Streptomyces carbophilus that hydroxylates ML-236B to pravastatin sodium
(CS-514), a tissue-selective inhibitor of 3-hydroxy-3-methylglutaryl-coenzyme-A
reductase. Eur J Biochem 184(3):707–713.
10. Guengerich FP (2005) Human cytochrome P450 enzymes. Cytochrome P450: Structure,
Mechanism and Biochemstry, 3rd Ed), ed Ortiz de Montellano PR (Kluwer Academic/
Plenum Publishers, New York), pp 377–530.
11. Podust LM, Sherman DH (2012) Diversity of P450 enzymes in the biosynthesis of
natural products. Nat Prod Rep 29(10):1251–1266.
12. Neuvonen PJ (2010) Drug interactions with HMG-CoA reductase inhibitors (statins):
The importance of CYP enzymes, transporters and pharmacogenetics. Curr Opin In-
vestig Drugs 11(3):323–332.
13. Hosobuchi M, Kurosawa K, Yoshikawa H (1993) Application of computer to moni-
toring and control of fermentation process: Microbial conversion of ML-236B Na to
pravastatin. Biotechnol Bioeng 42(7):815–820.
14. van den Berg MA (2011) Impact of the Penicillium chrysogenum genome on industrial
production of metabolites. Appl Microbiol Biotechnol 92(1):45–53.
15. Harris DM, et al. (2009) Exploring and dissecting genome-wide gene expression re-
sponses of Penicillium chrysogenum to phenylacetic acid consumption and penicillinG
production. BMC Genomics 10:75.
16. van den Berg MA, et al. (2008) Genome sequencing and analysis of the filamentous
fungus Penicillium chrysogenum. Nat Biotechnol 26(10):1161–1168.
17. Chen C-H, Hu H-Y, Cho Y-C, Hsu W-H (2006) Screening of compactin-resistant mi-
croorganisms capable of converting compactin to pravastatin. Curr Microbiol 53(2):
108–112.
18. Fujii Y, et al. (2011) Construction of a novel expression vector in Pseudonocardia
autotrophica and its application to efficient biotransformation of compactin to prava-
statin, a specific HMG-CoA reductase inhibitor. Biochem Biophys Res Commun
404(1):511–516.
19. Lin C-L, Tang Y-L, Lin S-M (2011) Efficient bioconversion of compactin to pravastatin
by the quinoline-degrading microorganism Pseudonocardia carboxydivorans isolated
from petroleum-contaminated soil. Bioresour Technol 102(22):10187–10193.
20. Park JW, et al. (2003) Bioconversion of compactin into pravastatin by Streptomyces sp.
Biotechnol Lett 25(21):1827–1831.
21. Peng Y, Yashphe J, Demain AL (1997) Biotransformation of compactin to pravastatin
by Actinomadura sp. 2966. J Antibiot (Tokyo) 50(12):1032–1035.
22. Serizawa N, et al. (1983) Microbial hydroxylation of ML-236B (compactin). Studies on
microorganisms capable of 3 beta-hydroxylation of ML-236B. J Antibiot (Tokyo) 36(7):
887–891.
23. Nodate M, Kubota M, Misawa N (2006) Functional expression system for cytochrome
P450 genes using the reductase domain of self-sufficient P450RhF from Rhodococcus
sp. NCIMB 9784. Appl Microbiol Biotechnol 71(4):455–462.
2852 | www.pnas.org/cgi/doi/10.1073/pnas.1419028112
pH 6.5, 200 mM sodium bromide, and 22.5% (wt/vol) PEG 3350. Single
crystals were soaked in mother liquor supplemented with 10% (vol/vol) PEG
200 as cryoprotectant and flash-cooled by plunge freezing in liquid nitrogen.
Diffraction data were collected at Diamond synchrotron beamlines. Data
were reduced/scaled with the XDS suite (47) and processed using the CCP4 suite
(48). The WT CYP105AS1 structure was solved using molecular replacement with
the CYP105A1 structure as a search model (PDB ID code 2ZBY) (26), and the
P450Prava mutant was solved using the WT CYP105AS1 structure. Structures were
refined using Refmac5 (48) and COOT (49). Data and final refinement statistics
are shown in Table S6.
ACKNOWLEDGMENTS. We acknowledge access to Diamond beamlines and
Colin Levy for assistance with synchrotron X-ray data collection. We thank
Marina Golovanova, Rolf Poldermans, Erwin Talens, Paul Klaassen, and
Adriana Carvalho-de Souza for technical support. These studies were funded
by UK Biotechnology and Biological Sciences Research Council Industrial
Partnership Award Research Grant BB/G014329/1 with DSM (to A.W.M., D.L.,
and K.J.M.). This work was sponsored by DSM Anti-Infectives (presently DSM
Sinochem Pharmaceuticals).
24. Fuziwara S, et al. (2002) Catalytically functional flavocytochrome chimeras of P450
BM3 and nitric oxide synthase. J Inorg Biochem 91(4):515–526.
25. Sieber V, Martinez CA, Arnold FH (2001) Libraries of hybrid proteins from distantly
related sequences. Nat Biotechnol 19(5):456–460.
26. Sugimoto H, et al. (2008) Crystal structure of CYP105A1 (P450SU-1) in complex with
1alpha,25-dihydroxyvitamin D3. Biochemistry 47(13):4017–4027.
27. Xu L-H, et al. (2010) Regio- and stereospecificity of filipin hydroxylation sites revealed
by crystal structures of cytochrome P450 105P1 and 105D6 from Streptomyces aver-
mitilis. J Biol Chem 285(22):16844–16853.
28. Butler CF, et al. (2013) Key mutations alter the cytochrome P450 BM3 conformational
landscape and remove inherent substrate bias. J Biol Chem 288(35):25387–25399.
29. Smith RG (2008) Penicillin and cephalosporin drug allergies: A paradigm shift. J Am
Podiatr Med Assoc 98(6):479–488.
30. Jami MS, et al. (2010) The Penicillium chrysogenum extracellular proteome. Conver-
sion from a food-rotting strain to a versatile cell factory for white biotechnology. Mol
Cell Proteomics 9(12):2729–2744.
31. Serizawa N, et al. (1983) 6 alpha-Hydroxy-iso-ML-236B (6 alpha-hydroxy-iso-compactin)
and ML-236A, microbial transformation products of ML-236B. J Antibiot (Tokyo)
36(7):918–920.
32. Komagata D, Yamashita H, Endo A (1986) Microbial conversion of compactin (ML-236B)
to ML-236A. J Antibiot (Tokyo) 39(11):1574–1577.
33. Galgóczy L, Nyilasi I, Papp T, Vágvölgyi C (2011) Statins as antifungal agents. World J
Clin Infect Dis 1(1):4–10.
34. Abe Y, et al. (2002) Effect of increased dosage of the ML-236B (compactin) bio-
synthetic gene cluster on ML-236B production in Penicillium citrinum. Mol Genet
Genomics 268(1):130–137.
35. Dodhia VR, Fantuzzi A, Gilardi G (2006) Engineering human cytochrome P450 en-
zymes into catalytically self-sufficient chimeras using molecular Lego. J Biol Inorg
Chem 11(7):903–916.
36. Li P, Guan H, Li J, Lin Z (2009) Heterologous expression, purification, and character-
ization of cytochrome P450sca-2 and mutants with improved solubility in Escherichia
coli. Protein Expr Purif 65(2):196–203.
37. Fujii T, Fujii Y, Machida K, Ochiai A, Ito M (2009) Efficient biotransformations using
Escherichia coli with tolC acrAB mutations expressing cytochrome P450 genes. Biosci
Biotechnol Biochem 73(4):805–810.
38. Ba L, Li P, Zhang H, Duan Y, Lin Z (2013) Semi-rational engineering of cytochrome
P450sca-2 in a hybrid system for enhanced catalytic activity: Insights into the impor-
tant role of electron transfer. Biotechnol Bioeng 110(11):2815–2825.
39. Ba L, Li P, Zhang H, Duan Y, Lin Z (2013) Engineering of a hybrid biotransformation
system for cytochrome P450sca-2 in Escherichia coli. Biotechnol J 8(7):785–793.
40. Basch J, Chiang S-J (2007) Cloning and expression of a cytochrome P450 hydroxylase
gene from Amycolatopsis orientalis: Hydroxylation of epothilone B for the pro-
duction of epothilone F. J Ind Microbiol Biotechnol 34(2):171–176.
41. Jung ST, Lauchli R, Arnold FH (2011) Cytochrome P450: Taming a wild type enzyme.
Curr Opin Biotechnol 22(6):809–817.
42. Trefzer A, et al. (2007) Biocatalytic conversion of avermectin to 4″-oxo-avermectin:
Improvement of cytochrome p450 monooxygenase specificity by directed evolution.
Appl Environ Microbiol 73(13):4317–4325.
43. Manzoni M, Rollini M (2002) Biosynthesis and biotechnological production of statins
by filamentous fungi and application of these cholesterol-lowering drugs. Appl
Microbiol Biotechnol 58(5):555–564.
44. Elander RP (2002) University of Wisconsin contributions to the early development of
penicillin and cephalosporin antibiotics. SIM News 52:270–278.
45. Benedetti A, Manzoni M, Nichele M, Rollini M (2002) Process for the production of
pravastatin and lovastatin. Patent EP1266967.
46. D’Arcy A, Villard F, Marsh M (2007) An automated microseed matrix-screening
method for protein crystallization. Acta Crystallogr D Biol Crystallogr 63(Pt 4):550–554.
47. Kabsch W (2010) XDS. Acta Crystallogr D Biol Crystallogr 66(Pt 2):125–132.
48. Winn MD, et al. (2011) Overview of the CCP4 suite and current developments. Acta
Crystallogr D Biol Crystallogr 67(Pt 4):235–242.
49. Emsley P, Cowtan K (2004) Coot: Model-building tools for molecular graphics. Acta
Crystallogr D Biol Crystallogr 60(Pt 12 Pt 1):2126–2132.
McLean et al.