Structural fine-tuning of a multifunctional cytochrome P450 monooxygenase

Article abstract of DOI:10.1021/ja110146z

AurH is a unique cytochrome P450 monooxygenase catalyzing the stepwise formation of a homochiral oxygen heterocycle, a key structural and pharmacophoric component of the antibiotic aureothin. The exceptional enzymatic reaction involves a tandem oxygenatio

Full text of DOI:10.1021/ja110146z

ARTICLE
pubs.acs.org/JACS
Structural Fine-Tuning of a Multifunctional Cytochrome P450
Monooxygenase
||
Georg Zocher,*,†,‡,# Martin E. A. Richter,§,# Uwe Mueller,‡ and Christian Hertweck*,§,
†Interfakult€ares Institut f€ur Biochemie, Eberhard Karls Universit€at T€ubingen, Hoppe-Seyler-Str. 4, 72074 T€ubingen, Germany
‡Helmholtz Zentrum for Materials and Energy Berlin, Macromolecular Crystallography Group, Albert-Einstein-Str. 15, D-12489 Berlin, Germany
§Leibniz Institute for Natural Product Research and Infection Biology (HKI), Beutenbergstr. 11a, 07745 Jena, Germany
Chair for Natural Product Chemistry, Friedrich Schiller University, Jena, Germany
S Supporting Information
b
ABSTRACT: AurH is a unique cytochrome
P450 monooxygenase catalyzing the step-
wise formation of a homochiral oxygen
heterocycle, a key structural and pharmaco-
phoric component of the antibiotic aure-
othin. The exceptional enzymatic reaction
involves a tandem oxygenation process in-
cluding a regio- and stereospecific hydro-
xylation, followed by heterocyclization. For
the structural and biochemical basis of this
unparalleled sequence, four crystal struc-
tures of AurH variants in different confor-
mational states and in complex with the
P450 inhibitor ancymidol were solved, which represent the first structures of the CYP151A group. Structural data in conjunction
with computational docking, site-directed mutagenesis, and chemical analyses unveiled a switch function when recognizing the two
substrates, deoxyaureothin and the hydroxylated intermediate, thus allowing the second oxygenation-heterocyclization step.
Furthermore, we were able to modify the chemo- and regioselectivity of AurH, yielding mutants that catalyze the regioselective six-
electron transfer of a nonactivated methyl group to a carboxylic acid via hydroxyl and aldehyde intermediates.
’ INTRODUCTION
scarce, and there is an evident gap in knowledge on their
structure and function. The full spectrum of P450-mediated
oxygenation reactions clearly remains to be explored.
Cytochrome P450 monooxygenases (P450s) are heme con-
taining enzymes that play pivotal roles in catabolism and metabolism
in all kingdoms of life.1 In biosynthetic pathways, this group of
enzymes has proven to be highly versatile not only with respect to
the range of substrates, but also regarding the reactions catalyzed.
These include C-C cleavage, heteroatom oxidation, as well as
hydroxylation and epoxidation reactions,2,3 which are often impor-
tant tailoring reactions in terpene, alkaloid, and polyketide biosyn-
thetic pathways. A hallmark of P450s involved in biosynthetic
pathways is their high regio- and stereoselectivity, even when
modifying nonactivated C-H groups.4 Because these reac-
tions are particularly difficult or even impossible to emulate
with the currently available synthetic protocols, P450s have
become highly valuable for biotechnological applications, in
particular as biocatalysts5 and biosensors.6 During the past
decades, a large body of knowledge on enzyme structures and
the mechanism of P450-catalyzed oxygen transfer has been
established.2 However, it is remarkable that to date virtually
all P450-catalyzed oxygen transfers are hydroxylations and
epoxidations.3 Multifunctional P450 monooxygenases are
Recently, we have discovered an unusual multifunctional P450
monooxygenase, AurH from Streptomyces thioluteus, that is
uniquely capable of forming a homochiral tetrahydrofuran ring.7
This O-heterocycle is a vital component of the polyketide
antibiotic aureothin (1, Scheme 1A),8 as it lends rigidity to the
carbon backbone and significantly contributes to the antifungal
activity of the antibiotic.9 In vivo and in vitro studies have revealed
that AurH catalyzes an unprecedented tandem oxygenation
process. First, AurH catalyzes an asymmetric hydroxylation of
deoxyaureothin (3), yielding (7R)-7-hydroxydeoxyaureothin (2)
as an intermediate, thus setting the absolute configuration of the
final product (Scheme 1A). Second, AurH mediates another
C-O bond formation that leads to O-heterocyclization.10 AurH
is capable of accepting a variety of pyrones as substrates, allowing
for the first chemoenzymatic total synthesis of aureothin,11 and
Received: November 11, 2010
Published: January 31, 2011
r
2011 American Chemical Society
2292
dx.doi.org/10.1021/ja110146z J. Am. Chem. Soc. 2011, 133, 2292–2302
|

Journal of the American Chemical Society
ARTICLE
Scheme 1. (A) AurH Catalyzes Two Subsequent Oxidation
Steps and Ring Formation from Deoxyaureothin (3) to
Aureothin (1) via the Secondary Alcohol (7R)-7-Hydroxy-
deoxyaureothin (2) and (B) Structure of the P450 Inhibitor
Ancymidol
determined in another crystal form (nterm-AurH-I) with an
unstructured N-terminal region that does not intersect the en-
trance of the active center (see Supporting Information Figure
S1B). The structure of wild-type AurH (wt-AurH) was established
at 2.1 Å resolution after reengineering the expression plasmid to
avoid the N-terminal overhang. Finally, the structure of nterm-
AurH-II in complex with the P450 inhibitor ancymidol (Scheme 1B;
Ancy:AurH), a competitive inhibitor for P450s,15 was determined at
2.3 Å resolution. Initial difference electron density maps clearly
revealed the presence of the ligand in nterm-AurH-II and also
showed that the N-terminal overhang formerly inserted into the cleft
of the active center was removed by the soaking procedure. The
structures of nterm-AurH-I, wt-AurH, and Ancy:AurH were deter-
mined by molecular replacement using nterm-AurH-II. Crystal-
lographic refinement of all structures yielded models that possess
low free R-factors, excellent stereochemistry, and small root-mean-
square deviations (rmsd) from ideal values for bond lengths and
bond angles (Table 2).
Overall Three-Dimensional Architecture of AurH and
Structure Comparison. AurH is a monomeric protein and
exhibits a typical prismlike cytochrome P450 fold with a length
of about 60 Å and a thickness of about 35 Å. An AurH molecule
consists of 15 helices (16 for nterm-AurH-II) and two β-sheets
(Figure 1A). The overall structure and protein topology of AurH
are similar to those of other P450 enzymes. A search for structural
homologues using DALI16 revealed that the closest structural
relatives to AurH are PikC from pikromycin biosynthesis in
Streptomyces venezuelae,17 P450-SU1, whose native function in
Streptomyces griseoleus is unknown,18 Bacillus subtilis P450bioI,
which plays a key role in biotin biosynthesis,19 CalO2 from the
calicheamycin biosynthetic pathway in Micromonospora echinospora,20
StaP from staurosporin biosynthesis in Streptomyces sp. TP-A0274,21
and P450epoK from epothilone biosynthesis in Sorangium cellulosum
(see Supporting Information Table S1).22 A closer inspection of
these enzymes revealed conserved residues and regions crucial for
cofactor binding and oxygen activation. Compared to the structural
homologues, four regions of AurH showed significant CR-rmsd
deviations. These alternating positions are located at the substrate
recognition sites (SRS)23 and are composed of the N-terminal loop
(amino acids 44 and 45), the FG-loop (amino acids 174 to 181), and
the β2-loop (amino acids 389 to 394). Furthermore, compared to the
structural homologues AurH varies at amino acids 66 to 91, which
form a rigid two-helix bundle (helices B2 and B20) and therebycreate
a hydrophobic wall that perfectly shields the active center from
solvent. Other P450 proteins mainly exhibit random coil configura-
tions with solvent-filled channels in this region. These solvent
channels give rise to several possible entrance pathways as observed
for a fatty acid hydroxylase from Bacillus subtilis.24 In contrast, only a
single entrance path exists in AurH because the two-helix bundle
rigidly seals the active center. This substrate channel is classified as
channel 2b based on the common notation of access pathways in
P450 enzymes.25 Besides AurH, such a two-helix bundle is only
observed in CalO2.20 However, compared to CalO2, the helices B2
and B20 are kinked toward the N-terminal domain and thus yield an
extended cavity. In general, the superimposition of the structural
homologues described above revealed that AurH has a single
substrate entrance leading to an elongated and wide cavity, which
can perfectly harbor the substrate deoxyaureothin (3).
was also successfully employed in biotransformation reactions.9
NorH, from the neoaureothin (spectinabilin) pathway, is a homo-
logue of AurH and the only other known P450 enzyme catalyzing
such an exceptional heterocyclization reaction.12
Here we report the first crystal structures of the multifunc-
tional cytochrome P450 monooxygenase AurH. The structural
data, in conjunction with inhibition studies, modeling, site-directed
mutagenesis, and structure elucidation of novel metabolites, provide
intriguing insights into the function of this fine-tuned biocatalyst.
Furthermore, through protein engineering we succeeded in gen-
erating variants of AurH with altered function and regioselectivity,
which led to the formation of a carboxylate branch in lieu of the
THF ring.
’ RESULTS
Determination of the Structure of Cytochrome P450
Monooxygenase AurH. AurH was assigned to the cytochrome
P450 superfamily by its P450 signature motif.13 According to the
CYP450 engineering database,14 AurH is a member of the
CYP151A family, which has not yet been structurally char-
acterized. The protein, which consists of 406 amino acids, was
overproduced in E. coli and purified to homogeneity.10
A
sample of pure AurH was crystallized, and its structure (nterm-
AurH-II) was solved to 1.54 Å resolution by the use of multi-
wavelength anomalous diffraction (MAD) at the K-edge of iron
(Table 1). Our initial construct of AurH revealed an N-terminal
overhang of 10 amino acids, which was a remnant from the cloning
and production strategy. This overhang swapped into a cleft above
the entrance of the active center (see Supporting Information
Figure S1A), and its impact on the structure of AurH was
Comparison of the crystal structures of AurH disclosed that
the largest deviations are located in mainly the same regions as
described for the structural homologues. While nterm-AurH-I
exhibits a closed setup, nterm-AurH-II revealed an artificially
2293
dx.doi.org/10.1021/ja110146z |J. Am. Chem. Soc. 2011, 133, 2292–2302

Journal of the American Chemical Society
ARTICLE
2294
dx.doi.org/10.1021/ja110146z |J. Am. Chem. Soc. 2011, 133, 2292–2302

Journal of the American Chemical Society
ARTICLE
Table 2. Refinement Statistics
nterm-AurH-I
30-2.3
nterm-AurH-II
Ancy:AurH
wt-AurH
resolution range (Å)
19.3-1.54
0.156
20-2.30
0.183
20.0-2.1
21.4
Rcryst
Rfree
0.226
0.279
0.188
0.231
26.4
Number of Non-H Atoms
protein
heme
3154/3171
86
3503
86
3193
43
3164/3164
86
ancymidol
chloride
sulfate
19
2
3
3
5
5
5
5
glycerol
water
18
253
12
426
12
189
340
Average Isotropic B-Factor (Å2)
main chain
side chain
27.0/32.2
27.6/32.8
18.7/20.9
29.8
32.4
18.3
31.7
32.0
13.6
26.2
37.6
75.6
23.5
0.006
0.922
27.9/22.7
29.5/24.8
17.9/22.3
heme
ancymidol
chloride
58.6
37.5
40.4
glycerol
57.7
33.4
water
26.2
26.8
28.4
rmsd for bond lengths (Å)
rmsd for bond angles (deg)
0.011
1.319
0.015
1.575
0.013
0.732
Ramachandran Regions
most favorable (%)
allowed (%)
96.3
3.2
98.5
1.5
95.5
3.2
96.9
2.8
outliers (%)
0.5
0.3
0.3
0.3
Figure 1. (A) Representation of Ancy:AurH exhibiting a typical prismlike shape of cytochrome P450 enzymes. The structure can be divided into a
β-rich domain (blue) and an R-helical domain (green). Regions revealing the largest CR-rmsd distances compared to nterm-AurH-I, nterm-AurH-II, and
wt-AurH are highlighted in red. The inhibitor ancymidol, the heme, and several amino acids of the active center are shown in pink, yellow, and gray,
respectively. Amino acids exchanged by site directed mutagenesis are ochre colored. (B) Superimposition of all AurH structures revealed that the
differences are mainly located in the regions of substrate recognition. The flexible regions are emphasized by colors for Ancy:AurH (light green), nterm-
AurH-II (dark green), nterm-AurH-I (light blue), and wt-AurH (violet).
wide opened state of the enzyme. The structures of Ancy:AurH
and wt-AurH showed an intermediate conformation of both
extremes but are related to nterm-AurH-II and nterm-AurH-I,
respectively (Figure 1B). Differences between the two extrema,
nterm-AurH-II (open state) and nterm-AurH-I (closed state),
were observed at the two-helix bundle, where the tip of the two
helix bundle (G76) is shifted by about 5.1 Å toward the active
center. Furthermore, the FG-loop moves toward the same
2295
dx.doi.org/10.1021/ja110146z |J. Am. Chem. Soc. 2011, 133, 2292–2302

Journal of the American Chemical Society
ARTICLE
Figure 2. (A) Stereoview of the active center of nterm-AurH with ancymidol (magenta). Residues that were mutated to determine the influence on
the catalytic reactivity are shown in green. The conformations of the Q91 and S66 in the unbound state (nterm-AurH-I in violet, Ancy:AurH in green)
are highlighted by transparent ball-and-stick representation. The (2Fo - Fc)-omit map of ancymidol (gray) shown at an σ-level of 1.2 clearly defines the
position of the inhibitor within the active center. (B, C) Proposed reaction pathway on the bases of in silico docking of the substrates 2 and 3 into the
cavity of (B) nterm-AurH-II and (C) wt-AurH, repspectively, using AUTODOCK4.
direction, and the β2-loop is shifted by about 4 Å at G392 toward
the FG-loop, which results in a weak main chain contact fixing the
β2-loop in the closed setup. In this conformation, the entrance to
the active center in wt-AurH and nterm-AurH-I is restricted.
Investigation of the intermediate conformations of Ancy:AurH
and wt-AurH indicated a subsequent closure of the two-helix
bundle and disclosed the different conformation of the β2-loop,
changing from an opened to a closed conformation.
Inhibitor Induces Conformational Changes at the Active
Site. The active center of AurH is formed by the two-helix
bundle that consists of B2 and B20, the FG-loop, the I-helix,
the loop between helix K and N-terminal β-sheet, and loop β2.
The cavity of AurH is mainly formed by hydrophobic residues
(Figure 2A), which is in agreement with the hydrophobic
nature of the substrates (2, 3). The inhibitor ancymidol is
primarily bound through the interaction of the pyrimidine ring
system to the heme iron and to the conserved residue T247.
Other interactions to the inhibitor are mediated by loose and
hydrophobic contacts (mainly to F89, A243, L175, L290,
T295). This is also supported by the B-factor distribution of
the inhibitor increasing from 20 Å2 at the pyrimidine ring
system to 52 Å2 at the aromatic phenol moiety. The weak
interaction of ancymidol is furthermore displayed by the low
occupancy, which is refined to 75%. These observations reflect
the low inhibitory effect of ancymidol on the in vitro activity of
AurH and can be explained by the loose binding of the small
inhibitor in a cavity, which is designed for the more volumi-
nous substrates.
In comparison to the substrate free crystal forms of AurH, we
found remarkable changes within the structure containing the
2296
dx.doi.org/10.1021/ja110146z |J. Am. Chem. Soc. 2011, 133, 2292–2302

Journal of the American Chemical Society
ARTICLE
inhibitor ancymidol. First of all, binding of ancymidol is accom-
panied by a movement of the two helix bundle toward the closed
state of the enzyme. Furthermore, inhibitor binding results in a
reorientation of side chain Q91 toward the rigid wall of B2 and
B20. In one conformation, the amide function of Q91 is placed
between the two negatively charged carboxy groups of the heme
cofactor, thus presenting the hydrophobic carbon atoms of the
side chain into the active center (Figure 2A). The reorientation of
Q91 demonstrates flexibility of this residue during substrate
binding. We concluded that Q91 plays a key role for catalysis and
is able to adopt different conformations upon inhibitor binding,
which results in presenting either the polar amide head or the
hydrophobic carbons toward the active center. Notably, the
glutamine residue is also present in NorH, which catalyzes
an analogous reaction in the neoaureothin pathway. However, it
is replaced in the structural homologues mainly by a hydro-
phobic residue. The importance of this position for catalysis is
supported by structure comparison and site-directed muta-
genesis. Structural comparisons revealed that hydrophobic
residues, mainly leucine, replaced the polar glutamine in all
homologous proteins depicted in Table S1. Furthermore,
preparation of the Q91L mutant abolished catalysis completely.
Moreover, reorientation of Q91 is accompanied by a flip of the
carbonyl group of S66 toward the active center. As S66 is
located at the N-terminal kink of the two-helix bundle and this
reorientation of S66 is not observed in nterm-AurH-II nor in
wt-AurH or nterm-AurH-I, we concluded that this movement is
not a result from the removal of the N-terminal overhang and
displays an inhibitor induced fit which occurs during substrate
binding.
Second, a detailed comparison of the active centers of Ancy:
AurH with the unbound states of AurH revealed an inhibitor-
induced incorporation of an additional water molecule into the
cleft of helix I, which is crucial for the catalytic process and proton
transfer. The incorporation of additional water molecules into
the groove of the I-helix during binding of molecular oxygen was
described for P450Cam.26,27 The incorporation of these active
site water molecules is accompanied by a “flip” of the acidic
carbonyl oxygen, which widens the helix. Although this reorga-
nization is not observed in Ancy:AurH, the hydroxy group of
T247 is pushed outward to form a hydrogen bond to the nitrogen
atom of the inhibitor. In comparison to conformational changes
induced by inhibitor binding of ketoconazole as described for
P450eryF,28 the cleft in helix I remains stable and did not collapse
upon inhibitor accumulation. However, incorporation of the
water molecule and the rotation of T247 toward ancymidol
pushed the conserved residues E246 and T247 upward by about
0.5 Å away from the heme cofactor. This results in a rotation of
the adjacent aromatic side chain of H250 by about 50° and is
accompanied by the loss of the hydrogen bond network, which
connects helix I with loop β2 (Figure 2A). As a consequence,
loop β2 swings out by about 4 Å to avoid a steric clash of the rigid
imidazol ring system of H250 with the branched side chain of
V393. This enables tension of the loop β2 and opens the reaction
center. We assume that this tension might be important for
catalysis as our docking calculations revealed that the pressure of
loop β2 is able to bend the substrate and facilitates a geometry
that presses the substrate deeper into the pocket and pushes
atom C-9a toward the reaction center. A comparable hydrogen
bond network connecting loop β2 with helix I is found in
P450SU-1 from S. griseoleus, which bis-hydroxylates vitamin
D3.18 Yet, a substrate-induced fit of this region during substrate
incorporation has not been observed. Nevertheless, it is well
conceivable that the relaxation of loop β2 is essential for THF
ring formation in transferring the tension of this loop to the
intermediate, which bends the rigid unsaturated part of (7R)-7-
hydroxydeoxyaureothin (2) (Figure 2A). Interestingly, the hy-
drogen bond network of D397-H230-E246 is conserved in NorH,
thus pointing to a crucial catalytic function of the flexible β2-loop.
Substrate Modeling Illuminates Stereochemical Course of
Oxygenation and Unveils Switch Function of AurH. As
attempts to incorporate the natural substrates of AurH into the
crystal have not been successful, we performed a computational
docking of the substrates (2) and (3) on structures of Ancy:
AurH, wt-AurH, and nterm-AurH-I. These dockings resulted in
models for both oxidative reactions and elucidate the crucial role
of Q91, which adopts different conformations depending on the
actual substrate. In silico docking of the substrate deoxyaureothin
(3) into the cavity of nterm-AurH-I using the Lamarckian genetic
algorithm implemented in AUTODOCK429 provides a model
for the first step of the oxidative enzymatic reaction and perfectly
rationalizes its regioselectivity and stereospecificity in forming
the hydroxyl intermediate (Figure 2B). The elongated substrate
fits snugly into the active center. Our model revealed contacts
between the substrate and residues F19, F68, Q91, V292, T295,
L290, F316, L391, and G392. The pyrone ring of the substrate
occupies a hydrophobic pocket built by F85, F89, L175, V242,
A243, and T239. By comparison of structurally homologous
P450s in the substrate bound state, we found that this portion of
the cavity is occupied frequently, as has been observed for
P450bioI,19 StaP,21 P450epoK,22 PikC,17 and P450-SU1.18 Ac-
cording to our model, the substrate adopts an orientation where
C-7 and the reactive oxygen species at the heme iron have a distance
of about 4.0 Å. Locked in this conformation, the substrate is attacked
from the Resite, which is in full accord with the observed 7R
configuration of the product. Moreover, Q91 adopts a conformation
in analogy to the scenario observed in Ancy:AurH and presents the
hydrophobic carbon atoms toward the educt while potentially
forming a hydrogen bond to the carboxy group of the cofactor.
As described above, this might trigger an educt-induced fit of
the active center. In docking experiments using wt-AurH,
which exhibits a wider cavity, (7R)-7-hydroxydeoxyaureothin
(2) is able to adopt a conformation that allows THF ring formation
during the second oxidative reaction step (Figure 2C). This model
reveals a hydrogen bond between the secondary alcohol and Q91,
the side chain of which retains its conformation and exerts steric
pressure on the intermediate toward the iron. As a result, the
calculated distance of C-9a to the distal heme water, which is
approximately in the position of the activated oxygen species,
is about 4.7 Å. Formation of the THF ring becomes possible
because the 7-hydroxyl group is oriented in a geometry
enabling an intramolecular back-side attack of an activated
C-9a. The cavity of AurH remains in a closed conformation as
steric pressure is required to bend the rigid unsaturated intermediate.
Nevertheless, compared to nterm-AurH-I, the cavity of wt-AurH
is still wider due to the relaxed state of the two-helix bundle. This is a
reasonable observation because the substrate is bulkier when bent
and explains why such a conformation is not found by in silico
modeling into the tight cavity of nterm-AurH-I. Furthermore, steric
pressure exerted by loop β2 seems to be essential in both oxidative
steps, as docking calculations in Ancy:AurH reveal a degree of
flexibility of the substrate in the cavity, which is not observed in
nterm-AurH-I and wt-AurH. However, most of these conformations
can be neglected as they are inconclusive with the observed
2297
dx.doi.org/10.1021/ja110146z |J. Am. Chem. Soc. 2011, 133, 2292–2302

Journal of the American Chemical Society
ARTICLE
Figure 3. Relative activity of AurH mutants, comprising contents of the
two sequential products, (2) (7R)-7-hydroxydeoxyaureothin and (1)
aureothin.
stereoselectivity and the regioselectivity of the enzyme and are
not supported by our mutagenesis experiments.
Finally, our docking calculations are supported by the finding
that a conversion of the R-pyrone derivative of 3, iso-deoxyaure-
othin, into iso-aureothin is not observed, whereas iso-deoxyaure-
othin is hydroxylated at C-7 and formation of the THF ring does
not take place.9 This is reasonable as steric collision of iso-
deoxyaureothin with the pyrone binding pocket (residues T239
and F89) can easily be avoided by a 180° rotation of the pyrone
ring system compared to deoxyaureothin during the initial hydro-
xylation step. Such a rotation excludes the geometry required for
THF ring formation, as this orientation would force an intramole-
cular clash of the atoms C-5a and C-10 of iso-deoxyaureothin when
adopting the calculated conformation needed for the second
oxidation. Furthermore, we calculated a model for NorH using
MODELLER30 on the basis of wt-AurH and compared regions
exhibiting sequence differences around the active center.
Apart from the catalytic residue E246, which is replaced by a
glutamine residue in NorH, the models only differ in regions
involved in binding of the hydrophobic tail and the aromatic
system of neoaureothin. This clearly supports a correct dock-
ing as differences in the protein sequence are observed in
regions where our substrate models differ from neoaureothin.
Point mutations F19I, L71V, L290T, P291T, L391T, and
F316W would rebuild the cavity of AurH to the wider cavity
of NorH, which fits the extended backbone of neoaureothin.
Site-Directed Mutagenesis Supports the Model for Sub-
strate Binding and Reaction Mechanism. To provide experi-
mental support for the calculated models, we performed site-
directed mutagenesis of several amino acids predicted to be
involved in substrate binding. Next, we determined the activity
of the mutated enzymes compared to the wild type by adding
deoxyaureothin (3) to cultures of Streptomyces albus strains
expressing mutated aurH variants (Figure 3). The catalytic
importance of Q91 is emphasized by site-directed mutagenesis
to leucine, which completely abolishes enzymatic activity for the
first reaction step. This was expected because the initial oxygena-
tion reaction is hindered by the branching methyl group at the
γ-carbon atom of leucine. Furthermore, the educt-induced fit of
the rigid two-helix bundle is hampered because leucine is not able
to adopt the conformation toward the carboxy groups of the
cofactor observed in Ancy:AurH. The conversion of intermediate
(2) by the Q91L mutant to aureothin (1) is also abolished (data not
shown), which supports our docking model. The more voluminous
Figure 4. Selected in vivo biotransformations of deoxyaureothin (3) by
(A) AurH wild type, (B) F89W mutant, and (C) T239F mutant. Time
course (front to back): 15, 45 min, 2, 4, 24 h. (D) Biotransformation of
9a-hydroxydeoxyaureothin (4) to the carboxylic acid 6 by AurH(F89W)
over 24 h.
side chains introduced by the mutants L175W and L179W
decreased the catalytic activity for both oxygenation reactions,
which is an effect of steric collision. As expected, the mutation
of L290 to alanine led to a decrease in substrate conversion
because L290 forms a tight hydrophobic contact in both
oxidative reaction models. The loss in activity of F316A clearly
indicates the participation in substrate binding. The decreased
turnover of F316A is expected from our docking model as the
nitro entity of the substrate would interact with the aromatic
system of F316 to yield a close hydrophobic contact. To corroborate
the correct modeling of the pyrone ring portion of the substrate, we
exchanged F89 to tryptophane and T239 to alanine and phenyla-
lanine, respectively. All three mutations significantly decreased
enzymatic performance. Although a minimal conversion to the
intermediate was observed, no ring formation occurred when T239
was altered. While the activity of T239A clearly supports participa-
tion of T239 in a hydrogen bond to the intermediate during
catalysis, an exchange to phenylalanine avoids this stabilizing effect
and furthermore decreases the transformation rate by steric pres-
sure. Interestingly, the ring formation of the mutant F89W is almost
completely lost, while C-7 hydroxylation is still observed at reason-
able rates. As the phenyl moiety of F89 directly contacts the pyrone
entity (Figure 2B, C) one may conclude that the turnover of the
intermediate is impaired by steric collision. Furthermore, the
increased size of W89 pushes the pyrone ring system toward loop
β2 and does not allow the intermediate to adopt the geometry
that is required for a backside attack of the C-7 hydroxyl to C-9a.
As a consequence, THF ring formation has become highly
unfavorable.
Engineering of an AurH Variant Catalyzing a Six-Electron
Oxidation. Among the various engineered mutants, we noted
that two AurH variants containing altered residues that interact
with the pyrone ring transformed 3 into new aureothin derivatives
(Figure 4A-C). From chromatographic profiles, UV absorption,
2298
dx.doi.org/10.1021/ja110146z |J. Am. Chem. Soc. 2011, 133, 2292–2302

Journal of the American Chemical Society
ARTICLE
Scheme 2. AurH Mutants Catalyze the Sequential C-9a
Oxygenation of 3 with Structures of New Aureothin
Analogues 4-6
’ DISCUSSION
In this study, we have elucidated the structure and function of
the unique multifunctional P450 monooxygenase AurH that
sequentially installs two C-O bonds into the aureothin polyke-
tide backbone to yield a homochiral THF ring. We obtained
AurH in four distinct crystallized forms, with one containing the
bound inhibitor ancymidol. These isoforms illustrate the transi-
tion from “open” to “closed” enzyme conformations. Overall, the
AurH structure exhibits a typical P450 fold. However, the two-
helix-bundle (B2, B20), which shields the active center from
solvent, is uncommon and was only reported for CalO2.20 When
AurH changes from “open” to “closed” conformation, the two-
helix-bundle and the FG-loop move toward the active center, and
the β2-loop approaches the FG-loop. Altogether, this seems to
generate steric pressure not only to push the substrate into the
correct position, but also to bend the carbon backbone, thus
facilitating THF-ring formation. The residue at the N-terminal
kink of the two-helix-bundle (S66) moves upon inhibitor binding
and Q91 turns toward the heme group. A plausible scenario
would involve an induced-fit mechanism triggered by substrate
binding. Disintegration of protein crystals in soaking experiments
with substrates implies substantial movements upon substrate
binding and catalysis.
Another unique feature of AurH is the S66-Q91 pair, which is
not found in similar P450s, except for the homologue NorH from
the neoaureothin pathway. Docking studies imply a switch
function in recognizing the two substrates, deoxyaureothin (2)
and the hydroxylated intermediate (3). The abolished activity in
the Q91L mutant points to an important role of Q91 in enzyme
function and/or structure.
and MS of the products, these mutants obviously catalyzed aberrant
oxygenation reactions. To elucidate these, a full structural determi-
nation of the new derivatives was needed. Since the amounts
obtained in the in vitro assay were too low, we complemented a
Streptomyces albus host harboring the aur biosynthesis gene cluster
except for aurH (knockout mutant) with aurH constructs coding for
the AurH(F89W) and AurH(T239F) variants, respectively. Culti-
vation of these S. albus strains afforded sufficient material for a full
characterization by 1D and 2D NMR spectroscopy. Analyses of the
NMR data revealed the structures of the new metabolites (4-6).
Interestingly, all new compounds represent deoxyaureothin deriva-
tives oxygenated at C-9a, differing only in the degree of oxidation
(Scheme 2). This fact is well depicted by the proton and carbon
NMR signals at this position. While the signals corresponding to the
methylene group at C-7, the place of hydroxylation of native AurH,
remain almost constant with ascending degree of C-9a oxygenation
(see Supporting Information Table S2), the singlet at 1.96 ppm,
which corresponds to three methyl protons at C-9a in deoxyaure-
othin (3) changes to a singlet at 4.22 ppm (relative intensity of two
protons) in 9a-hydroxydeoxyaureothin (4), a singlet at 9.48 ppm
(one proton) in deoxyaureothin-9a-aldehyde (5), andabroadsignal
at 12.9 ppm (one proton) in deoxyaureothin-9-carboxylic acid (6),
respectively. Carbon NMR signals support the identities of the
functional groups (18.1, 64.9, 194.9, and 170.3 ppm for methyl,
hydroxyl, aldehyde, and carboxylic acid group, respectively).
Interestingly, while the mutated enzymes were still able to
hydroxylate C-7, no derivatives resulting from oxygenation at
both C-7 and C-9a were detected. Apparently, products that are
oxygenated at C-9a cannot bind to the engineered enzyme in a
manner that would allow for C-7-oxygenation and vice versa.
Finally, through the successful biotransformation of pure 4 into
6, we proved that the oxygenation of the alcohol to the acid is
indeed catalyzed by AurH variants (Figure 4D).
Our in silico docking calculations were complemented and
fully supported by a series of site-directed mutagenesis experi-
ments. The activities of the mutants were determined in vivo and
compared to the wild type. Although the reaction was remarkably
slower in vivo (50% substrate conversion not reached after 24 h)
than in vitro using wild type enzyme (50% conversion in about
eight hours10), this setup allowed for a quick and parallel investiga-
tion of substrate conversion. This is not surprising as one should
take into account that the in vivo enzyme concentration does not
reach such high values as 5 μmol L-1 in vitro. However, results from
our comparative studies with wild type and mutants clearly support
the hypothesis that narrowing the cavity for pyrone binding prevents
the substrate from adopting the correct steric conformation for THF
ring formation, although the initial hydroxylation is still possible to
some degree. This is reasonable, because bending the substrate to
obtain correct geometry for THF ring formation results in a bulkier
reactant, and formation of the heterocycle is excessively sensitive
because of this steric pressure.
Interestingly, the two successive substrate binding modes
observed for AurH are clearly distinct from other known multi-
functional P450s. For PikC, two different desosamine anchoring
sites have been identified that allow for the recognition of two
different substrates.17 P450BioI positions the fatty acid substrate
in a U-shaped conformation that brings both carbon atoms in
position for consecutive hydroxylation and oxidative C-C-bond
cleavage.19 In contrast to AurH, a conformational change of
the substrate between the first and the second reaction has not
been observed.
The crystal structures and mutagenesis studies indicate that
the substrate is directed to the active site mostly by hydrophobic
interactions and π-stacking. Altering these residues (L175W,
2299
dx.doi.org/10.1021/ja110146z |J. Am. Chem. Soc. 2011, 133, 2292–2302

Journal of the American Chemical Society
ARTICLE
L179W, L290A, F316A) decreased the overall activity but had no
influence on regioselectivity or THF-ring formation. As it is
known from other P450s, for instance mammalian CYPs31 or
PikC,32 hydrophobic interactions are mostly nonspecific. Yet, for
correct substrate positioning and selectivity, polar residues that
specifically recognize corresponding substrate functionalities are
more important, which is well supported by the drastically
impaired activity of Q91 and T239 mutants.
cytochrome P450 monooxygenase. Through elucidation of
four AurH variant crystal structures, which represent the
first structures within the CYP151A family, inhibition studies,
modeling, and site-directed mutagenesis, we have identified
key residues that are required for the regio- and stereoselective
hydroxylation and subsequent THF ring formation. On
the basis of our results, we developed a model in which the
hydroxylated intermediate changes its conformation in
the binding site due to a sterical switch of the enzyme, thus
allowing the second oxygenation-heterocyclization step. Further-
more, we engineered mutants with diverging regiospecificity and
function. The isolation and full structure elucidation of three new
aureothin analogues produced by the AurH variants revealed that
these mutants favor C-9a oxygenation and produce hydroxyl,
aldehyde, and carboxylate moieties. Overall, AurH represents a
fascinating example of a fine-tuned, multifunctional biocatalyst of
high synthetic value. Thus, our work not only provides new
insights into finesse of enzyme-mediated oxygenation reac-
tions, but also sets the ground for future enzyme engineering
in the field of synthetically useful biotransformations. Since related
five- and six-membered O-heterocycles are integral components of
many bioactive polyketides,43 we believe that the dual function of
AurH-type oxygenases is more widespread in nature.
In addition to substantiating our structure and modeling
studies, the mutagenesis experiments yielded AurH variants with
an altered function. While some mutants solely lost the ability to
produce an O-heterocycle, two mutants (F89W and T239F)
exhibited a change in regioselectivity. While the wild type invariably
hydroxylates C-7 in the first step, the mutants were capable of C-9a
hydroxylation. Furthermore, the AurH variants catalyzed multiple
rounds of oxygenation, which resulted in the formation of the C-9a
aldehyde and carboxylate derivatives. Recent examples for altered
substrate specificity and regioselectivity of mutated P450s include
human CYP2A6, mutants of which were able to oxidize sub-
stituted indoles33 and B. megaterium P450BM3 with altered
substrate specificity34,35 or increased activity,36 and B. subtilis
CYP102A3, which was evolved to a higher selectivity for hydro-
xylation of terminal carbon atoms of octane.37 In a similar fashion,
the regioselectivity of PikC was changed upon mutagenesis.17
Oxidation of YC-17 at different positions yielded methymycin
and neomethymycin in equivalent amounts, whereas mutant
E85Q preferentially formed methymycin, which again high-
lights the importance of hydrophilic residues for substrate
positioning. In comparison, AurH catalyzes the regioselective
oxidation at two distinct positions. However, in stark contrast
to PikC, the reaction follows a sequential manner. In the
F89W and T239F mutants, the reaction sequence was dis-
turbed. Due to steric pressure, anchoring of the pyrone entity
requires an alternative binding mode, which does not allow the
binding of a C-7 hydroxylated product and oxidation at
position C-9a and vice versa. This is plausible as the flexibility
of loop β2 would allow antagonizing the steric hindrance of
the bulky side chains of the mutants.
Apart from the altered regioselectivity, the AurH mutants stand
out because of their ability to catalyze the stepwise oxygenation of
the C-9a methyl group to a carboxyl moiety. Despite extensive
research on P450s, it is surprising that to date only a handful of
P450s are known that catalyze similarly the full oxidation of methyl
groups to carboxylic acids in biosynthetic pathways. These rare
P450s include ent-kaurene-oxidase and P450-1 in gibberellic
acid (GA) biosynthesis in Arabidopsis thaliana38 and Gibber-
ella fujikuroi,39 respectively, abietadienol/abietadienal oxidase
PtAO in resin acid biosynthesis in conifers,40 and CYP71AV1
from Artemisia annua that transforms amorpha-4,11-diene to
artemisinic acid.41 The multifunctional P450 BorI from bor-
relidin biosynthesis in Streptomyces parvulus T€u4055 is also
capable of oxidizing a methyl group to an aldehyde, which may
be further oxidized to the carboxylic acid in a shunt pathway.42
However, to the best of our knowledge the change of function
from sequential hydroxylation/heterocyclization to a full six-
electron oxidation of a nonactivated methyl group is fully
unprecedented.
’ EXPERIMENTAL SECTION
Structure Determination, Refinement, and Substrate Mod-
eling. Data collection was carried out with five protein crystals (Table 1).
One crystal of nterm-AurH-II was used for phase determination by
performing a MAD experiment. Freshly prepared crystals were used for
collecting the data sets of nterm-AurH-I, nterm-AurH-II, Ancymidol:AurH,
and wt-AurH. Data reduction was performed using XDS and XSCALE.44
The iron position was located using SHELXD45 and used for phasing and
density modification as implemented in SHARP/autoSHARP.46 The final
experimental electron density map was of decent quality (figure of merit
0.408) and allowed automatic chain tracing performed by ARP/wARP,47
which resulted in a partial model of 367 residues. This model was transferred
to the high resolution data set using rigid body refinement of REFMAC5.48
Several cycles of manual building with COOT49 and refinement
with REFMAC5 completed the model. Water molecules were placed
using ARP/wARP and checked manually with COOT. The final refine-
ment steps involved TLS group refinement. Therefore, the structure
was divided into eight TLS groups yielding a model of excellent quality
(Table 1). The chloride ions located in nterm-AurH-II were inserted
after examining the chemical environment and the anomalous electron
density [Fþ(lrem) - F-(lrem)] of the low remote data set calculated
by SFTOOLS48 and FFT.48 Phasing for Ancy:AurH was done by mole-
cular replacement procedure using rigid body refinement of REFMAC5.48
The phases for wt-AurH resulted from molecular replacement using
PHASER.48 Model bias of Ancy:AurH and wt-AurH were avoided by
simulated annealing using PHENIX.50 The following refinement procedure
was similar to that described for nterm-AurH. Due to the lower amount
of reflections, however, only up to two TLS groups per protomer were
used for the final refinement steps. All structures were validated using
WHATCHECK48 and SFCHECK48 (Table 2). Figures were generated
using POVscriptþ51 and POVRAY (www.povray.org).
Molecular Docking. The structures of wt-AurH, Ancy:AurH, and
nterm-AurH-I served as templates for our docking calculations of the
substrates 3 and 2 using AUTODOCK4.29 When performing the docking
calculations, all water molecules beside the distal water of the heme group
were removed. Q91 was assigned as a flexible residue because it was
observed that this residue adopts different conformations in Ancy:AurH
and the unbound structures. All other amino acids remained as rigid.
’ CONCLUSION
In conclusion, we have unveiled the structural and biochemical
basis of a unique enzymatic THF formation catalyzed by a
2300
dx.doi.org/10.1021/ja110146z |J. Am. Chem. Soc. 2011, 133, 2292–2302

Journal of the American Chemical Society
ARTICLE
For the substrates 2 and 3 only the bonds between C9-C8, C8-C7,
C7-C6, and the ether group of the pyrone ring system were allowed
to rotate during docking calculations while the conjugated system
remained rigid. The grid box covered the complete active center, and
docking was performed with 25 cycles using Lamarckian genetic
algorithm (LGA). The docked models were visualized and analyzed
for structural and chemical sense.
Accession Codes. The structure of nterm-AurH-I, nterm-AurH-
II, and wt-AurH and the structure of AurH in complex with ancymidol
have been deposited in the PDB database under accession codes 3P3L,
3P3O, 3P3X, and 3P3Z, respectively.
Mutagenesis and Activity Assays. For mutational analyses, an
in vivo-approach was chosen. The aurH gene was cloned into an
E. coli-Streptomyces shuttle vector downstream of the constitutively active
ermE promoter from the erythromycin biosynthesis gene cluster. Muta-
tions were introduced into the aurH sequence with the QuikChange XL
II Kit (Stratagene) according to the manufacturer’s instructions (see
Supporting Information Table S3 for primer sequences). Correct clones
were identified by sequencing and were transferred into wild type
S. albus. Aliquots of spore suspension (10 μL) were used to inoculate
seed cultures of J medium (10 mL) supplemented with thiostrepton
(5 mg L-1). After incubation at 30 °C with orbital shaking at 160 rpm
for 24 h, 3 mL of the seed culture was used to inoculate M10 medium
(30 mL) containing thiostrepton (5 mg L-1). Cultivation was continued
for another day before deoxyaureothin (3) (0.5 mg, 43.5 μmol L-1 final
concentration) was added to each culture. After distinct time intervals of
ongoing cultivation, aliquots (5 mL) were taken and extracted once
with ethyl acetate (5 mL each). The organic phases were collected
and evaporated under reduced pressure. The dried extracts were
redissolved in methanol (1 mL), and aliquots (20 μL) were subjected
to HPLC-UV-MS measurements on an ET 250/4 Nucleosil 120-
5 C8 column from Macherey-Nagel. Separation was achieved by
elution at 1 mL min-1 in a gradient (20 min) from 20% to 98%
acetonitrile in water (0.1% formic acid). MS detection was carried out
in positive mode.
Concentration under reduced pressure afforded 3.6 mg of 9a-hydroxydeox-
yaureothin (4).
The last fraction of the silica gel chromatography was concentrated,
and the residue was dissolved in ethyl acetate (250 mL) and extracted
three times with aq Tris (100 mL, 100 mM, pH 8.0). The combined
aqueous phases were washed three times with ethyl acetate (50 mL
each), followed by acidification with aq NaH2PO4 (100 mL, 1 M, pH
2.0). The turbid solution was extracted four times with ethyl acetate
(100 mL each). The combined organic phases were dried over sodium
sulfate and concentrated under reduced pressure to yield deoxyaur-
eothin-9-carboxylic acid (6) as a mixture of E/Z-isomers. The
isomers were dissolved in a 1:3 mixture of DMSO and acetonitrile
and separated by preparative HPLC (Nucleosil C18) with isocratic
elution (42% acetonitrile in water containing 0.1% TFA). Re-extrac-
tion with ethyl acetate afforded 17 mg of (9Z,11E)-deoxyaureothin-9-
carboxylic acid (6, referred to as the genuine deoxyaureothin-9-carboxylic
acid), and 20 mg of (9Z,11Z)-deoxyaureothin-9-carboxylic acid, resulting
from light-induced isomerization.
9a-Hydroxydeoxyaureothin (4). Yellow, amorphous solid (3.6 mg).
1H NMR (500 MHz, CDCl3): 8.16 (d, J = 8.7 Hz, 2H, H-15/17), 7.33
(d, J = 8.6 Hz, 2H, H-14/18), 6.34 (s, 1H, H-12), 6.14 (s, 1H, H-10),
4.22 (s, 2H, H-9a), 3.77 (s, 3H, H-2a), 2.78 (t, J = 7.5 Hz, 2H, H-7), 2.68
(t, J = 7.5 Hz, 2H, H-8), 1.92 (s, 3H, H-11a), 1.88 (s, 3H, H-5a), 1.78 (s,
3H, H-3a). 13C NMR (125 MHz, CDCl3): 179.0 (C-4), 160.3 (C-2),
155.5 (C-6), 144.3 (C-16), 142.2 (C-13), 137.6 (C-9), 136.9 (C-11),
129.6 (C-10), 127.7 (C-14/18), 125.4 (C-12), 121.8 (C-15/17), 116.8
(C-5), 97.7 (C-3), 64.9 (C-9a), 53.3 (C-2a), 28.1 (C-7), 24.5 (C-8), 17.2
(C-11a), 8.2 (C-5a), 5.0 (C-3a). FT-IR (solid film): ν 3422 (OH), 2925
(CH3), 2857 (O-CH3), 1662 (CdO), 1581 (CdC), 1512 (NO2),
1337 (NO2), 855 (vicinal ar H) cm-1. UV(MeOH) λmax (lg ε): 247.5
(4.12), 341 (4.03) nm. (þ)-ESI-MS m/z 422 [M þ Na]þ, m/z 400
[M þ H]þ, MSn m/z 168, m/z 152. HRESI-MS m/z [M þ H]þ calcd
for C22H26NO6 400.1755; found 400.1743.
Deoxyaureothin-9a-aldehyde (5). Yellow, amorphous solid (20
1
mg). H NMR (600 MHz, CDCl3): 9.48 (s, 1H, H-9a), 8.24 (d, J =
8.6 Hz, 2H, H-15/17), 7.42 (d, J = 8.6 Hz, 2H, H-14/18), 6.95 (s, 1H,
H-10), 6.75 (s, 1H, H-12), 3.86 (s, 3H, H-2a), 2.86 (t, J = 7.4 Hz, 2H,
H-8), 2.76 (t, J = 7.4 Hz, 2H, H-7), 2.13 (s, 3H, H-11a), 1.88 (s, 3H,
H-5a), 1.79 (s, 3H, H-3a). 13C NMR (125 MHz, CDCl3): 194.9 (C-
9a), 180.7 (C-4), 162.1 (C-2), 156.7 (C-6), 154.5 (C-10), 146.8 (C-
16), 142.4 (C-13), 140.3 (C-9), 136.8 (C-11), 134.3 (C-12), 129.9
(C-14/18), 123.6 (C-15/17), 118.9 (C-5), 99.4 (C-3), 55.2 (C-2a),
29.8 (C-7), 22.6 (C-8), 17.7 (C-11a), 10.0 (C-5a), 6.8 (C-3a). FT-IR
(solid film): ν 2927 (CH3), 2866 (O-CH3), 1663 (CdO), 1592
(CdC), 1518 (NO2), 1341 (NO2), 855 (vicinal ar H) cm-1. UV
(MeOH) λmax (lg ε): 259 (4.30), 333.5 (4.28) nm. (þ)-ESI-MS m/z
420 [M þ Na]þ, m/z 398 [M þ H]þ. MSn m/z 168, m/z 153.
HRESI-MS m/z [M þ H]þ calcd for C22H24NO6 398.1598; found
398.1596.
Production, Purification, and Structure Elucidation of
Novel Aureothin Derivatives. For production of suitable amounts
of the new deoxyaureothin derivatives, S. albus strains were generated
that harbor an integrative plasmid with genes coding for deoxyaureothin
(3) biosynthesis (i.e., aureothin biosynthesis gene cluster lacking the
aurH gene)13 and a plasmid coding for AurH mutant F89W. A seed
culture of J medium (10 mL), supplemented with thiostrepton (5 mg L-1
)
and apramycin (30 mg L-1), was inoculated with a spore suspension
(10 μL) of the S. albus mutant strain and incubated at 30 °C with orbital
shaking at 200 rpm. After 24 h, a larger seed culture of identical medium
(1 L) was inoculated with the first seed culture, and incubation was
continued at 28 °C/120 rpm for another two days. M10 medium
[14 L, supplemented with thiostrepton (5 mg L-1) and apramycin
(30 mg L-1)] was inoculated using the 1 L seed culture in a stainless
steel fermenter. Cultivation was carried out for six days at 28 °C and
continuous stirring at 400 rpm without pH adjustment.
Deoxyaureothin-9-carboxylic Acid (6). Yellow, amorphous solid
1
(17 mg). H NMR (300 MHz, DMSO-d6): 12.9 (br,H-9a), 8.20 (d,
J = 8.9 Hz, 2H, H-15/17), 7.55 (d, J = 8.8 Hz, 2H, H-14/18), 6.58 (s, 1H,
H-10), 6.33 (s, 1H, H-12), 3.97 (s, 3H, H-2a), 2.81 (t, J = 7.1 Hz, 2H,
H-7), 2.64 (t, J = 7.1 Hz, 2H, H-8), 1.97 (d, J = 1.0 Hz, 3H, H-11a), 1.81
(s, 3H, H-5a), 1.68 (s, 3H, H-3a). 13C NMR (125 MHz, DMSO-d6):
179.3 (C-4), 170.3 (C-9a), 161.8 (C-2), 157.1 (C-6), 145.7 (C-16),
143.7 (C-13), 138.3 (C-11), 136.5 (C-10), 133.6 (C-9), 130.1 (C-14/
18), 129.5 (C-12), 123.5 (C-15/17), 117.8 (C-5), 97.5 (C-3), 55.7 (C-
2a), 32.4 (C-8), 29.5 (C-7), 16.8 (C-11a), 9.6 (C-5a), 6.9 (C-3a). FT-IR
(solid film): ν 2927 (CH3), 2874 (O-CH3), 2521 (OH), 1660 (CdO),
1590 (CdC), 1514 (NO2), 1338 (NO2), 856 (vicinal ar H) cm-1. UV
(MeOH) λmax: 253, 339 nm. (þ)-ESI-MS m/z 436 [M þ Na]þ, m/z
414 [M þ H]þ. MSn m/z 168, m/z 153. HRESI-MS m/z [M þ H]þ
calcd for C22H24NO7 414.1547; found 414.1542.
The cultivation broth was separated from cellular material by centrifuga-
tion and extracted twice with equal volumes of ethyl acetate. The combined
organic phases were dried over sodium sulfate and concentrated under
reduced pressure. The residue was dissolved in chloroform and subjected to
open column chromatography on silica gel using chloroform as the initial
mobile phase. Several fractions containing deoxyaureothin derivatives 4-6
were obtained using a chloroform-methanol gradient for elution. After
further purification on Sephadex LH-20 (methanol), the final separation of
5 was achieved by preparative HPLC (Nucleosil C18) using a water (0.1%
trifluoracetic acid)/acetonitrile gradient (up to 83% acetonitrile), yielding
20 mg of deoxyaureothin-9a-aldehyde (5).
Fractions containing 4 were purified via preparative HPLC (Nucleosil
C18) using a gradient from 0.1% aqueous (NH4)HCO3 to 83% acetonitrile.
2301
dx.doi.org/10.1021/ja110146z |J. Am. Chem. Soc. 2011, 133, 2292–2302

Journal of the American Chemical Society
ARTICLE
’ ASSOCIATED CONTENT
(21) Makino, M.; Sugimoto, H.; Shiro, Y.; Asamizu, S.; Onaka, H.;
Nagano, S. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 11591–6.
(22) Nagano, S.; Li, H.; Shimizu, H.; Nishida, C.; Ogura, H.; Ortiz de
Montellano, P. R.; Poulos, T. L. J. Biol. Chem. 2003, 278, 44886–93.
(23) Gotoh, O. J. Biol. Chem. 1992, 267, 83–90.
(24) Lee, D. S.; Yamada, A.; Sugimoto, H.; Matsunaga, I.; Ogura, H.;
Ichihara, K.; Adachi, S.; Park, S. Y.; Shiro, Y. J. Biol. Chem. 2003, 278,
9761–7.
S
Supporting Information. Sequence alignment of AurH
b
homologues, primer sequences for mutagenesis, additional in-
formation about the activity of mutants, spectra of new aureothin
derivatives. This material is available free of charge via the
Internet at http://pubs.acs.org.
(25) Cojocaru, V.; Winn, P. J.; Wade, R. C. Biochim. Biophys. Acta
2007, 1770, 390–401.
’ AUTHOR INFORMATION
(26) Nagano, S.; Poulos, T. L. J. Biol. Chem. 2005, 280, 31659–63.
(27) Schlichting, I.; Berendzen, J.; Chu, K.; Stock, A. M.; Maves,
S. A.; Benson, D. E.; Sweet, R. M.; Ringe, D.; Petsko, G. A.; Sligar, S. G.
Science 2000, 287, 1615–22.
Corresponding Author
georg.zocher@uni-tuebingen.de (G.Z., for stuctural biology
correspondence); Christian.Hertweck@hki-jena.de (C.H., for
biochemistry/chemistry correspondence)
(28) Cupp-Vickery, J. R.; Garcia, C.; Hofacre, A.; McGee-Estrada, K.
J. Mol. Biol. 2001, 311, 101–10.
Author Contributions
#These authors contributed equally.
(29) Morris, G. M.; Huey, R.; Lindstrom, W.; Sanner, M. F.; Belew,
R. K.; Goodsell, D. S.; Olson, A. J. J. Comput. Chem. 2009, 30, 2785–91.
(30) Eswar, N.; Eramian, D.; Webb, B.; Shen, M. Y.; Sali, A. Methods
Mol. Biol. 2008, 426, 145.
(31) Yano, J. K.; Hsu, M. H.; Griffin, K. J.; Stout, C. D.; Johnson, E. F.
Nat. Struct. Mol. Biol. 2005, 12, 822–3.
(32) Li, S.; Ouellet, H.; Sherman, D. H.; Podust, L. M. J. Biol. Chem.
2009, 284, 5723–30.
(33) Sansen, S.; Hsu, M. H.; Stout, C. D.; Johnson, E. F. Arch.
Biochem. Biophys. 2007, 464, 197–206.
(34) Dietrich, M.; Do, T. A.; Schmid, R. D.; Pleiss, J.; Urlacher, V. B.
J. Biotechnol. 2009, 139, 115–7.
(35) Whitehouse, C. J.; Bell, S. G.; Wong, L. L. Chemistry 2008, 14,
10905–8.
(36) Whitehouse, C. J.; Bell, S. G.; Yang, W.; Yorke, J. A.; Blanford,
C. F.; Strong, A. J.; Morse, E. J.; Bartlam, M.; Rao, Z.; Wong, L. L.
ChemBioChem 2009, 10, 1654–6.
(37) Lentz, O.; Feenstra, A.; Habicher, T.; Hauer, B.; Schmid, R. D.;
Urlacher, V. B. ChemBioChem 2006, 7, 345–50.
’ ACKNOWLEDGMENT
We gratefully acknowledge the BESSY beamlines BL14.1 and
BL14.2 for offering beamtime. We thank Prof. Dr. Udo Heinemann
for the opportunity to work in his lab and Dr. Yvette Roske for
the introduction to the hydra-II crystallization robot. We also
thank A. Perner for MS and HPLC/MS measurements, and
F. Rhein for NMR measurements. Financial support by the DFG
is gratefully acknowledged.
’ REFERENCES
(1) Nelson, D. R. Arch. Biochem. Biophys. 1999, 369, 1–10.
(2) Guengerich, F. P. J. Biochem. Mol. Toxicol. 2007, 21, 163–8.
(3) Isin, E. M.; Guengerich, F. P. Biochim. Biophys. Acta 2007, 1770,
314–29.
(38) Helliwell, C. A.; Poole, A.; Peacock, W. J.; Dennis, E. S. Plant.
Physiol. 1999, 119, 507–10.
(39) Rojas, M. C.; Hedden, P.; Gaskin, P.; Tudzynski, B. Proc. Natl.
Acad. Sci. U.S.A. 2001, 98, 5838–43.
(40) Ro, D. K.; Arimura, G.; Lau, S. Y.; Piers, E.; Bohlmann, J. Proc.
Natl. Acad. Sci. U.S.A. 2005, 102, 8060–5.
(41) Teoh, K. H.; Polichuk, D. R.; Reed, D. W.; Nowak, G.; Covello,
(4) Takahashi, S.; Yeo, Y. S.; Zhao, Y.; O’Maille, P. E.; Greenhagen,
B. T.; Noel, J. P.; Coates, R. M.; Chappell, J. J. Biol. Chem. 2007, 282,
31744–54.
(5) Glieder, A.; Farinas, E. T.; Arnold, F. H. Nat. Biotechnol. 2002,
20, 1135–9.
(6) Bistolas, N.; Wollenberger, U.; Jung, C.; Scheller, F. W. Biosens.
Bioelectron. 2005, 20, 2408–23.
P. S. FEBS Lett. 2006, 580, 1411–6.
(7) M€uller, M.; He, J.; Hertweck, C. ChemBioChem 2006, 7, 37–39.
(8) He, J.; Hertweck, C. Chem. Biol. 2003, 10, 1225–1232.
(9) Werneburg, M.; Busch, B.; He, J.; Richter, M. E.; Xiang, L.;
Moore, B. S.; Roth, M.; Dahse, H. M.; Hertweck, C. J. Am. Chem. Soc.
2010, 132, 10407–13.
(10) Richter, M. E.; Traitcheva, N.; Knupfer, U.; Hertweck, C.
Angew. Chem., Int. Ed. 2008, 47, 8872–5.
(11) Werneburg, M.; Hertweck, C. ChemBioChem 2008, 9, 2064–6.
(12) Traitcheva, N.; Jenke-Kodama, H.; He, J.; Dittmann, E.;
Hertweck, C. ChemBioChem 2007, 8, 1841–9.
(42) Olano, C.; Wilkinson, B.; Sanchez, C.; Moss, S. J.; Sheridan, R.;
Math, V.; Weston, A. J.; Brana, A. F.; Martin, C. J.; Oliynyk, M.; Mendez,
C.; Leadlay, P. F.; Salas, J. A. Chem. Biol. 2004, 11, 87–97.
(43) Hertweck, C. Angew. Chem., Int. Ed. 2009, 48, 4688–4716.
(44) Kabsch, W. J. Appl. Crystallogr. 1993, 26, 795–800.
(45) Sheldrick, G. M. Acta Crystallogr., Sect. D 2010, 66, 479–485.
(46) Bricogne, G. Methods Enzymol. 1997, 277, 14–8.
(47) Langer, G.; Cohen, S. X.; Lamzin, V. S.; Perrakis, A. Nat. Protoc.
2008, 3, 1171–9.
(48) Collaborative Computational Project, Number 4. Acta Crystallogr.,
Sect. D: Biol. Crystallogr. 1994, 50, 760-3.
(49) Emsley, P.; Cowtan, K. Acta Crystallogr., Sect. D: Biol. Crystal-
(13) He, J.; Muller, M.; Hertweck, C. J. Am. Chem. Soc. 2004, 126,
16742–3.
(14) Fischer, M.; Knoll, M.; Sirim, D.; Wagner, F.; Funke, S.; Pleiss,
logr. 2004, 60, 2126–32.
J. Bioinformatics 2007, 23, 2015–7.
(50) Adams, P. D.; Afonine, P. V.; Bunkoczi, G.; Chen, V. B.; Davis,
I. W.; Echols, N.; Headd, J. J.; Hung, L. W.; Kapral, G. J.; Grosse-
Kunstleve, R. W.; McCoy, A. J.; Moriarty, N. W.; Oeffner, R.; Read, R. J.;
Richardson, D. C.; Richardson, J. S.; Terwilliger, T. C.; Zwart, P. H. Acta
Crystallogr. Sect. D: Biol. Crystallogr. 2010, 66, 213–21.
(51) Fenn, T. D.; Ringe, D.; Petsko, G. A. J. Appl. Crystallogr. 2003,
36, 944–947.
(15) Ralston, L.; Kwon, S. T.; Schoenbeck, M.; Ralston, J.; Schenk,
D. J.; Coates, R. M.; Chappell, J. Arch. Biochem. Biophys. 2001, 393, 222–35.
(16) Holm, L.; Sander, C. J. Mol. Biol. 1993, 233, 123–38.
(17) Sherman, D. H.; Li, S.; Yermalitskaya, L. V.; Kim, Y.; Smith,
J. A.; Waterman, M. R.; Podust, L. M. J. Biol. Chem. 2006, 281, 26289–97.
(18) Sugimoto, H.; Shinkyo, R.; Hayashi, K.; Yoneda, S.; Yamada,
M.; Kamakura, M.; Ikushiro, S.; Shiro, Y.; Sakaki, T. Biochemistry 2008,
47, 4017–27.
(19) Cryle, M. J.; Schlichting, I. Proc. Natl. Acad. Sci. U.S.A. 2008,
105, 15696–701.
(20) McCoy, J. G.; Johnson, H. D.; Singh, S.; Bingman, C. A.; Lei,
I. K.; Thorson, J. S.; Phillips, G. N., Jr. Proteins 2009, 74, 50–60.
2302
dx.doi.org/10.1021/ja110146z |J. Am. Chem. Soc. 2011, 133, 2292–2302