Crystal structures of cyclohexanone monooxygenase reveal complex domain movements and a sliding cofactor

Article abstract of DOI:10.1021/ja9010578

Cyclohexanone monooxygenase (CHMO) is a flavoprotein that carries out the archetypical Baeyer-Villiger oxidation of a variety of cyclic ketones into lactones. Using NADPH and O2 as cosubstrates, the enzyme inserts one atom of oxygen into the substrate in a complex catalytic mechanism that involves the formation of a flavin-peroxide and Criegee intermediate. We present here the atomic structures of CHMO from an environmental Rhodococcus strain bound with FAD and NADP⁺ in two distinct states, to resolutions of 2.3 and 2.2 A. The two conformations reveal domain shifts around multiple linkers and loop movements, involving conserved arginine 329 and tryptophan 492, which effect a translation of the nicotinamide resulting in a sliding cofactor. Consequently, the cofactor is ideally situated and subsequently repositioned during the catalytic cycle to first reduce the flavin and later stabilize formation of the Criegee intermediate. Concurrent movements of a loop adjacent to the active site demonstrate how this protein can effect large changes in the size and shape of the substrate binding pocket to accommodate a diverse range of substrates. Finally, the previously identified BVMO signature sequence is highlighted for its role in coordinating domain movements. Taken together, these structures provide mechanistic insights into CHMO-catalyzed Baeyer-Villiger oxidation.

Full text of DOI:10.1021/ja9010578

Published on Web 04/22/2009
Crystal Structures of Cyclohexanone Monooxygenase Reveal
Complex Domain Movements and a Sliding Cofactor
I. Ahmad Mirza,† Brahm J. Yachnin,† Shaozhao Wang,‡ Stephan Grosse,‡
He´le`ne Bergeron,‡ Akihiro Imura,| Hiroaki Iwaki,§ Yoshie Hasegawa,§
Peter C. K. Lau,*,‡ and Albert M. Berghuis*,†
Departments of Biochemistry and Microbiology & Immunology, McGill UniVersity, 3649 Prom
Sir William Osler, Bellini PaVilion, Room 466, Montreal, QC, Canada H3G 0B1, Biotechnology
Research Institute, National Research Council Canada, 6100 Royalmount AVenue, Montreal,
QC, Canada H4P 2R2, Department of Life Science & Biotechnology and ORDIST, Kansai
UniVersity, Suita, Osaka, 564-8680, Japan, and Process Technology Research Laboratories,
Daiichi Sankyo Co. Ltd, Kasai R&D Center, 1-16-13, Kitakasai Edogawa-ku,
Tokyo 134-8630, Japan
Received February 10, 2009; E-mail: peter.lau@cnrc-nrc.gc.ca; albert.berghuis@mcgill.ca
Abstract: Cyclohexanone monooxygenase (CHMO) is a flavoprotein that carries out the archetypical
Baeyer-Villiger oxidation of a variety of cyclic ketones into lactones. Using NADPH and O2 as cosubstrates,
the enzyme inserts one atom of oxygen into the substrate in a complex catalytic mechanism that involves
the formation of a flavin-peroxide and Criegee intermediate. We present here the atomic structures of
CHMO from an environmental Rhodococcus strain bound with FAD and NADP+ in two distinct states, to
resolutions of 2.3 and 2.2 Å. The two conformations reveal domain shifts around multiple linkers and loop
movements, involving conserved arginine 329 and tryptophan 492, which effect a translation of the
nicotinamide resulting in a sliding cofactor. Consequently, the cofactor is ideally situated and subsequently
repositioned during the catalytic cycle to first reduce the flavin and later stabilize formation of the Criegee
intermediate. Concurrent movements of a loop adjacent to the active site demonstrate how this protein
can effect large changes in the size and shape of the substrate binding pocket to accommodate a diverse
range of substrates. Finally, the previously identified BVMO signature sequence is highlighted for its role
in coordinating domain movements. Taken together, these structures provide mechanistic insights into
CHMO-catalyzed Baeyer-Villiger oxidation.
Introduction
pharmaceutical industry, as it avoids hazardous reagents like
organic peracids, chlorinated solvents, and metals that are
For reasons of exquisite chemo-, regio-, and enantioselec-
tivity, the enzyme cyclohexanone monooxygenase (CHMO),
produced by Acinetobacter sp. NCIMB 9871, has been the
subject of study for over 30 years.1-4 A member of the Baeyer-
Villiger monooxygenase (BVMO) family, and in the superfamily
of flavin-containing monooxygenases (FMO5), it is highly
valued for its ability to produce lactones and other important
chiral building blocks while producing only water as a clean
byproduct.6,7 Current increasing interests in CHMO are specif-
ically focused on possible use in asymmetric synthesis for the
otherwise used in chemical Baeyer-Villiger reactions.8-12
The mechanism of CHMO-catalyzed oxidation of cyclohex-
anone, and BVMO-catalyzed reactions in general, has been
shown by kinetic and spectroscopic data to proceed in a manner
very similar to the century-old chemical Baeyer-Villiger
reactions.12-15 In this chemical reaction, addition of an oxidant,
such as m-chloroperoxybenzoic acid, to the carbonyl group of
a ketone creates a tetrahedral intermediate that undergoes
“Criegee” rearrangement to yield the corresponding ester or
lactone. With BVMOs (Scheme 1), the tightly bound FAD
molecule is reduced by NADPH. The reduced flavin then reacts
† McGill University.
‡ National Research Council Canada.
§ Kansai University.
(8) Alphand, V.; Carrea, G.; Wohlgemuth, R.; Furstoss, R. Trends
Biotechnol. 2003, 21, 318–323.
| Daiichi Sankyo Co. Ltd.
(1) Donoghue, N. A.; Trudgill, P. W. Eur. J. Biochem. 1975, 60, 1–7.
(2) Stewart, J. D. Curr. Org. Chem. 1998, 192–216.
(3) Schwab, J. M.; Li, W.; Thomas, L. P. J. Am. Chem. Soc. 1983, 105,
4800–4808.
(9) Constable, D. J. C.; Dunn, P. J.; Hayler, J. D.; Humphrey, G. R. Green
Chem. 2007, 9, 411–420.
(10) Kamerbeek, N. M.; Janssen, D. B.; van Berkel, W. J. H.; Fraaije, M. W.
AdV. Synth. Catal. 2003, 345, 667–678.
(4) Gibson, M.; Nur-e-alam, M.; Lipata, F.; Oliveira, M. A.; Rohr, J. J. Am.
Chem. Soc. 2005, 127, 17594–17595.
(11) Mihovilovic, M. D.; Mueller, B.; Stanetty, P. Eur. J. Org. Chem. 2002,
3711–3730.
(5) van Berkel, W. J. H.; Kamerbeek, N. M.; Fraaije, M. W. J. Biotechnol.
2006, 124, 670–689.
(12) Kayser, M. M. Tetrahedron: Asymmetry 2009, 947–974.
(13) Renz, M.; Meunier, B. Eur. J. Org. Chem. 1999, 737–750.
(14) Ryerson, C. C.; Ballou, D. P.; Walsh, C. T. Biochemistry 1982, 21,
2644–2655.
(6) Stewart, J. D.; Reed, K. W.; Martinez, C. A.; Zhu, J.; Chen, G.; Kayser,
M. M. J. Am. Chem. Soc. 1998, 120, 3541–3548.
(7) ten Brink, G. J.; Arends, I. W.; Sheldon, R. A. Chem. ReV. 2004,
104, 4105–4124.
(15) Sheng, D.; Ballou, D. P.; Massey, V. Biochemistry 2001, 40, 11156–
11167.
9
8848 J. AM. CHEM. SOC. 2009, 131, 8848–8854
10.1021/ja9010578 CCC: $40.75  2009 American Chemical Society

Crystal Structures of Cyclohexanone Monooxygenase
A R T I C L E S
Scheme 1. Reaction Mechanism of CHMOa
a The atoms of the NADPH hydride and molecular oxygen are tracked throughout the reaction mechanism using red and blue, respectively.
with molecular oxygen to form a C4A-peroxyflavin intermediate
(E·FADHOO- ·NADP+). This peroxyflavin intermediate plays
the same role as the peracid in the conventional Baeyer-Villiger
oxidation reaction,15 acting as the nucleophile in the nucleophilic
attack of the carbonyl carbon of the ketone substrate. This
produces the tetrahedral Criegee intermediate that subsequently
rearranges to give the C4a-hydroxyflavin and ε-caprolactone, a
ring-expanded product.16 A molecule of water is spontaneously
eliminated from the hydroxyflavin to regenerate the oxidized
FAD. It is noteworthy that in this reaction, the NADPH, which
is the first species to bind to the enzyme and react with the
flavin, is held in the active site in the NADP+ form until the
last step of the reaction cycle.
identified BVMO signature sequence which has thus far
remained enigmatic.17
Materials and Methods
For a complete description of the cloning of the chnB1 gene
from Rhodococcus sp. strain HI-31, and overexpression, purification,
mutagenesis of the CHMO enzyme, refer to the Supporting
Information (SI).
Substrate Profiling. Standards of racemic lactones were preparaed
by oxidation of the cyclic ketones with m-CPBA according to
literature reports of Baeyer-Villiger oxidation.6 All ketone sub-
strates used in this study were purchased from Aldrich Chemical
Co. and used without further purification.
The ability of CHMO to transform cyclic ketones besides
cyclohexanone was investigated using whole cells of Escherichia
coli BL21(DE3)pSDRmchnB1. Conditions of whole-cell biotrans-
formation were carried out as previously described.18,19 Reaction
samples were extracted with ethyl acetate, dried with MgSO4, and
analyzed by GC/MS, using a Hewlett-Packard HP6890 series gas
chromatograph (GC system) connected to an HP5971 mass-selective
detector (Hewlett-Packard) and a DB-1 capillary column (J&W
Scientific, Folsom, CA). Chiral capillary-column GC was performed
on a Perkin-Elmer Sigma 2000 gas chromatograph, employing a
0.25 mm × 30 m (0.25-µm film) ꢀ-Dex 225 column (Supelco Inc.).
Peaks were identified by comparison of their retention times and
Even though the CHMO of Acinetobacter sp. NCIMB 9871
has been the most extensively studied among numerous BV-
MOs, to date there is no structural model available for this
enzyme. The only crystal structure that is currently available
for a BVMO enzyme is that of phenylacetone monooxygenase
(PAMO), which comes from a moderate thermophilic micro-
organism, Thermobifida fusca.17 The structure of PAMO was
solved with bound FAD in the absence of NADP cofactor and/
or substrates (analogues) and suggested a multidomain archi-
tecture resembling that of the disulfide oxidoreductases. It
allowed for the identification of the active site and a number of
residues critical for catalysis but also highlighted the functional
complexity of BVMOs. Specifically, the authors speculated on
the nature of the conformational changes that must occur to
enable the sequential binding of the different substrates to the
enzyme and the formation of several distinct catalytic intermedi-
ates that are consistent with spectroscopic data.15,17
1
their mass spectra with those of authentic standards. H and 13C
nuclear magnetic resonance (NMR) analyses were performed on a
Bruker Avance-500 spectrometer, and the spectra were recorded
in a CDCl3 solution at room temperature.
Crystallization and Data Collection. Two different crystal
forms of the enzyme were obtained by either hanging drop or sitting
drop vapor diffusion methods. Crystals were grown in 4-µL drops
containing equal volumes of concentrated (10 mg/mL) protein
solution mixed in a 5 M excess of NADP+ (Sigma) with well
solution. For crystal form I, the well solution contained 0.2 M
sodium acetate trihydrate, 0.1 M sodium cacodylate pH 6.5, and
30% PEG 8000. For crystal form II, the well solution consisted of
0.2 M magnesium acetate tetrahydrate, 0.1 M sodium cacodylate
pH 6.5, and 20% PEG 8000 (Nextal). Diffraction data for
CHMOclosed were collected on a Rigaku RUH-3R generator
equipped with an R-axis IV++ detector, while CHMOopen was
collected at beamline X6A of the National Synchrotron Light
Here, we present multiple crystal structures of CHMO from
Rhodococcus sp. strain HI-31 in complex with FAD and
NADP+. The two structures reveal the nature of conformational
changes required for catalysis and clarify how the NADP
cofactor can perform different roles in the reaction mechanism.
Our results also provide insight into the structural basis for
CHMO’s ability to accommodate a diverse range of substrates.
Finally, the structural data rationalize the role of the previously
(18) Iwaki, H.; Hasegawa, Y.; Wang, S.; Kayser, M. M.; Lau, P. C. K.
Appl. EnViron. Microbiol. 2002, 68, 5671–5684.
(19) Iwaki, H.; Wang, S.; Grosse, S.; Bergeron, H.; Nagahashi, A.;
Lertvorachon, J.; Yang, J.; Konishi, Y.; Hasegawa, Y.; Lau, P. C. K.
Appl. EnViron. Microbiol. 2006, 72, 2707–2720.
(16) Alphand, V.; Furstoss, R. Tetrahedron: Asymmetry 1992, 3, 379–382.
(17) Malito, E.; Alfieri, A.; Fraaije, M. W.; Mattevi, A. Proc. Natl. Acad.
Sci. U.S.A. 2004, 101, 13157–13162.
9
J. AM. CHEM. SOC. VOL. 131, NO. 25, 2009 8849

A R T I C L E S
Mirza et al.
Table 1. Summary of Crystallographic Dataa
genomic BVMO sequences (SI-Figure 5), CHMO is positioned
in the same branch as the classical cyclohexanone monooxy-
genase from Acinetobacter sp. NCIMB9871 (CHMOAC), sharing
an overall 55% sequence identity (SI-Figure 6).
crystal form I (CHMOopen
)
crystal form II (CHMOclosed)
space group
a, Å
P212121
64.3
P212121
62.5
b, Å
c, Å
67.0
136.0
91.0
100.4
Substrate Profiling. Mirroring the sequence similarity, the
regio- and enantiospecificity of this CHMO are also highly
similar to those of CHMOAC (SI-Table 2); however, it appears
resolution range, Å
completeness, %
redundancy
Rmerge
Rfactor/Rfree
no. of reflections
no. of atoms
no. of protein atoms
no. of cofactor atoms
no. of solvent atoms
rms bond length, Å
rms bond angle, °
20.0-2.2
99.4 (97.0)
5.5 (3.4)
0.08 (0.36)
18.3/24.5
27 243 (3788)
4364
4014
101
249
0.017
50.0-2.3
99.9 (99.9)
6.8 (6.7)
0.10 (0.54)
17.9/24.8
23 460 (3390)
4446
4166
127
153
0.021
that CHMO has a broader substrate spectrum than CHMOAC
.
The former readily accepted 4-tert-butylcyclohexanone (com-
pound 7) as a substrate with excellent enantioselectivity (>99%
ee), compared to the poor activity obtained by CHMOAC. A
similar trend was observed for compounds 8-13. Finally,
CHMO also showed very good activity with the decalone-type
bicyclic ketones (14-16). It is noteworthy that 2-tetralone was
selectively oxidized to only one isomer.
1.784
1.963
Overall Structure. Crystal structures of CHMO in complex
with FAD and NADP+ were determined in two different crystal
forms to a resolution of 2.3 and 2.2 Å (Table 1). Nearly all of
the polypeptide chain could be modeled in both crystal forms,
except for the first and last 5-7 residues. In crystal form I
(CHMOopen), residues 487-504 were also not modeled due to
the absence of appropriate density. This segment presumably
forms an unstructured loop. In both crystal forms, well-defined
density was observed for the FAD and NADP+ cofactors.
Furthermore, in crystal form II (CHMOclosed), electron density
corresponding to the nicotinamide half of a second NADP+
molecule involved in stacking interactions with W50 was
observed. The presence of this second, partially ordered NADP+
does not have any known biological function.
a Data in parentheses represent the outermost shell.
Source. Both data sets were collected under standard cryogenic
conditions and processed using the HKL2000 suite of programs20
(Table 1).
Structure Determination and Refinement. The structure of
CHMOopen was determined by molecular replacement using the
program PHASER21 and the coordinates of PAMO (PDB ID
1W4X) as the search model.17 Upon identification of the appropriate
rotation and translation matrix, the model was mutated using the
program Modeller22 to reflect the sequence of CHMO. This model
was then subjected to multiple rounds of high-temperature simulated
annealing refinement using the program CNS.23 Upon convergence
of the crystallographic R-factors, refinement was switched over to
REFMAC5 and included positional and B-factor refinement, with
subsequent TLS refinement being applied toward the end stage.24
Manual model rebuilding was carried out iteratively during the
course of refinement using the program PyMol.25 Phasing of
CHMOclosed was achieved using the nearly completed structure of
CHMOopen and employed an analogous refinement strategy as the
latter. Figures were prepared using the program PyMol.25
The atomic coordinates and structure factor amplitudes have been
deposited in the Protein Data Bank, www.pdb.org (PDB ID codes
3GWD and 3GWF).
The three-dimensional architecture of CHMO, which mirrors
that of PAMO, is highly modular in nature and can be separated
into three domains and numerous loops (Figure 1). The FAD
binding domain is composed of the first 140 N-terminal residues,
supplemented with residues 387-540 from the C-terminus. The
NADP binding domain is composed of discontinuous segments
152-208 and 335-380. Inserted within the sequence of the
NADP domain and connected by residues 209-223 and
322-334 is an entirely helical domain comprised of residues
224-322. This domain serves as a modulating region between
the two dinucleotide binding domains and also provides crucial
residues that make up the substrate binding pocket.
Results
Rhodococcus sp. strain HI-31 is a new environmental isolate,
selected by growth on cyclohexanone as a sole source of carbon,
and found to encode at least two CHMO-related sequences
(chnB1 and chnB2) within a 17.1-kilobase pair sequenced region
(DDBJ accession no. AB471785 and SI). The chnB1 gene was
cloned in E. coli, and the protein, henceforth designated simply
as CHMO, was overproduced and purified. This enzyme is a
stable, monomeric, 540 residue protein, displaying characteristic
flavoprotein absorption maxima at 376 and 477 nm and a
minimum at 405 nm, consistent with the observation that it
contains 0.94 mol of FAD per mol of enzyme. In an unrooted
phylogenetic analysis of known BVMOs and a selected list of
Connecting the FAD and NADP domains are two unstruc-
tured loop segments made up from residues 144-151 and
residues 381-386. It is noteworthy that the linkers are very
well defined in the CHMOclosed crystal form, whereas these
residues generally have higher B-factors and have less defined
density in CHMOopen, suggesting some degree of conformational
flexibility. The FAD cofactor is buried deep within the FAD
domain, consistent with the observation that this cofactor does
not dissociate from the enzyme, while the NADP+ is sandwiched
between the Rossmann fold of the NADP domain and a loop
region of the FAD domain. In both crystal forms, a substrate
binding site can be observed at the juxtaposition of the NADPH
and FAD binding sites.
Domain Movements. Identification of two different NADP+
cocrystallization conditions has allowed for the determination
of two distinct structures that reveal significant differences. The
two main variations between the CHMOopen and CHMOclosed
crystal forms are in the relative domain orientations and an
order/disorder transition of loop 487-504. These differences
perpetuate alternate cofactor binding modes and effect significant
changes in the substrate binding pocket.
(20) Otwinowski, Z.; Minor, W. Methods in Enzymology; Academic Press:
New York, 1997; Vol. 276, pp 307-326.
(21) McCoy, A. J.; Grosse-Kunstleve, R. W.; Storoni, L. C.; Read, R. J.
Acta Crystallogr. D: Biol. Crystallogr. 2005, 61, 458–464.
(22) Eswar, N.; Eramian, D.; Webb, B.; Shen, M. Y.; Sali, A. Methods
Mol. Biol. 2008, 426, 145–159.
(23) Brunger, A. T.; Adams, P. D.; Clore, G. M.; DeLano, W. L.; Gros, P.
Acta Crystallogr. D: Biol. Crystallogr. 1998, 54, 905–921.
(24) Murshudov, G. N.; Vagin, A. A.; Dodson, E. J. Acta Crystallogr. D:
Biol. Crystallogr. 1997, 53, 240–255.
(25) DeLano, W. L. The PyMOL molecular graphics system; DeLano
Scientific: Palo Alto, CA, 2002.
9
8850 J. AM. CHEM. SOC. VOL. 131, NO. 25, 2009

Crystal Structures of Cyclohexanone Monooxygenase
A R T I C L E S
Figure 1. Color-coded cartoon representation of overall domain structure of CHMO and examination of conformational changes: (A) wheat, FAD domain;
dark blue, NADP domain; light green, linker regions; brown, helical domain; orange, mobile loop region; and yellow, BVMO signature motif. (B) FAD
domain-anchored structural superposition of CHMOopen (blue) and CHMOclosed (green). (C) Equivalent atoms vector diagram illustrating changes in relative
backbone atoms positions between CHMOopen and CHMOclosed (dark blue and purple regions from panel a). Notable is the large rotation of the NADP
domain with a concurrent similar albeit less pronounced movement of the helical domain.
Using the core FAD binding residues as an anchor (backbone
rms 0.3 Å), comparison of the two structures reveals a 5°
rotation of the NADP domain centered at the active-site cavity,
which translates to an over ∼5 Å backbone shift for residues
distal to the axis of rotation (Figure 1). Displaying a similar
albeit less prominent movement, the helical domain returns an
average backbone rotation of ∼3°, resulting in shifts in the order
of 2 Å for distal residues. The movement of both the NADP
domain and the helical substrate domain is made possible by
the flexing of the four connector loops.
Another important difference that can be observed between
CHMOopen and CHMOclosed is in the conformation of residues
487-504. In CHMOopen, this loop is not visible and likely adopts
a flexible, solvent-exposed conformation similar to that observed
in the structure of PAMO. This is in sharp contrast to the
CHMOclosed structure, where this loop folds back onto itself,
Figure 2. Comparison of CHMOopen (blue) and CHMOclosed (green) NADP+
binding modes. The cofactor binds to the NADP domain through interactions
with T210, R209, T186, T189, and Q192, which are the same in both the
open and closed conformations. By contrast, large shifts can be noted in
internalizing the center portion of this segment (residues
490-493).
residues D59, K328, R329, and W492 between the conformations leading
to the “push” configuration observed in CHMOclosed
NADP+ Binding. In both the CHMOopen and CHMOclosed
crystal forms, the NADP+ cofactor adopts an extended confor-
mation, with its diphosphate backbone forming hydrogen bond
interactions with a GXXGXG motif (residues 182-187). Further
stabilization of the cofactor is afforded through interactions
between the 2′ phosphate of NADP+ and OG of T210 and NE
of R209. The guanidinium moiety of R209 is also involved in
stacking interactions with the adenine base. In both crystal
structures, the cofactor is oriented in such a manner that the
B-face of the nicotinamide ring is facing the isoalloxazine ring,
suggesting pro-S hydride transfer during flavin reduction (Figure
2), which differs from what has been suggested for PAMO on
the basis of kinetic studies.26 The presence, however, of similarly
oriented cofactors in two independently determined crystal forms
as well as the observation of the same orientation in multiple
FMO structures suggests that this manner of cofactor orientation
is likely the more frequently observed, biologically relevant
mode of binding.27,28
.
For instance, between the two structures, there is an average
1.7 Å backbone shift for residues 326-330. The two residues
most affected by this shift are K328 and R329. The strictly
conserved K328 has previously been implicated in NADPH vs
NADH selectivity;29 however, in CHMOopen it does not appear
to interact with this cofactor, given a distance of over 10 Å
between the closest possible H-bonding partners. Alternatively,
in CHMOclosed, a 4.2 Å shift and concurrent rotation result in
decreased distance between NZ K328 and a phosphoryl oxygen,
suggesting a possible indirect interaction between these atoms.
Similarly, R329, which makes H-bonding contact with the amide
nitrogen of NADP+ in both crystal forms, undergoes an overall
2.7 Å shift. This results in this residue moving closer toward
the interior of the enzyme, effecting concurrent movements in
the cofactor NADP+.
In CHMOopen, the nicotinamide base is held in place through
an H-bond with D59 and stacking interactions with the xylene
portion of the FAD isoalloxazine ring system. Contrasting this,
in CHMOclosed, flexing of the phosphate backbone brought about
by an interaction between the 2′-hydroxyl of the nicotinamide
ribose and W492 as well as R329 and nicotinamide amide
nitrogen results in a sliding of the nicotinamide head, allowing
The similar yet distinct movements of the NADP and helical
domains serve to modulate large shifts while at the same time
permitting more subtle changes in the active-site binding pocket.
(26) Torres Pazmin˜o, D. E.; Baas, B. J.; Janssen, D. B.; Fraaije, M. W.
Biochemistry 2008, 47, 4082–4093.
(27) Alfieri, A.; Malito, E.; Orru, R.; Fraaije, M. W.; Mattevi, A. Proc.
Natl. Acad. Sci. U.S.A. 2008, 105, 6572–6577.
(28) Eswaramoorthy, S.; Bonanno, J. B.; Burley, S. K.; Swaminathan, S.
Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 9832–9837.
(29) Kamerbeek, N. M.; Fraaije, M. W.; Janssen, D. B. Eur. J. Biochem.
2004, 271, 2107–2116.
9
J. AM. CHEM. SOC. VOL. 131, NO. 25, 2009 8851

A R T I C L E S
Mirza et al.
Figure 3. Comparison of active-site pockets observed in CHMOclosed and CHMOopen. (A) Stereodiagram showing an enclosed cavity of ∼150 Å3 in CHMOclosed
that is formed by residues with a predominately hydrophobic character. (B) In CHMOopen, the same pocket is enlarged and surface accessible due to the
movement of the mobile loop formed by residues 487-504.
for cofactor-protein interactions at Q192 and D59. Intriguingly,
W492 is a strictly conserved residue among all known CHMOs
and is located on a mobile loop formed from residues 487-504.
To examine the relevance of this interaction, we mutated W492
to Ala, which resulted in a protein with only 14% of the original
activity using cyclohexanone as substrate (0.85 U/mg compared
to 6.0 U/mg of specific activity for CHMO).
Focusing on the NADP+ and using the FAD domain as an
anchor, the rmsd of the NADP+ from CHMOopen to CHMOclosed
is ∼2 Å. The largest movement is a result of rotations in the
diphosphate backbone that effect a lateral translation of the
nicotinamide head (Figure 2). Consequently, the nucleotide
slides parallel to the FAD xylene moiety. In both the CHMOopen
and CHMOclosed conformations, the distance between NADP+-
NC4 and FAD-NC6 is ∼3.3 Å, albeit shifted by 45°. It is noted
that neither conformation appears to be conducive to flavin
reduction through hydride transfer between the nicotinamide
and isolloaxazine rings.
mobile loop 487-504, allowing for some degree of plasticity
and accommodation with respect to substrate binding.
Discussion
We present here the first structures of a CHMO and the first
structures of a BVMO in the presence of both FAD and NADP+.
Although this protein originated from a Rhodococcus species,
its sequence in a phylogenetic analysis is found to cluster with
those of Acinetobacter, Brachymonas, and Xanthobacter CHMOs,
making it representative of this family of BVMOs (SI-Figure 5).31
The dual cofactor bound structures clarify how this protein
adopts multiple conformations during catalysis, highlighting the
diverse roles played by the nicotinamide cofactor during the
enzymatic process and the effect this has on changing the size
and accessibility of the substrate binding pocket. Because of
this, we can now precisely identify mobile portions of the
enzyme that can be used to predict other catalytically relevant
conformations.
NADP+ Binding. While the backbone contacts with the
NADP+ are essentially identical in both crystal structures, the
relative orientation of the NADP domain results in a cofactor
that undergoes a significant lateral translation between the two
crystal forms. Examination of the CHMOopen structure reveals
that the nicotinamide and isoalloxazine ring systems are parallel
to each other, with a closest distance of 3.4 Å between C4 of
the nicotinamide and C6 of the xylene moiety. Despite such a
close interaction, it is unlikely that this conformation is the mode
of binding responsible for the NADPH-dependent reduction of
the flavin, as hydride transfer would normally be expected to
occur between C4 of the nicotinamide and N5 of the flavin.
Contrasting this, the CHMOclosed structure is in a somewhat
unique and surprising conformation. Rotation of the NADP
domain results in a lateral translation of the cofactor, pushing
Substrate Binding Pocket. A substrate binding site can be
identified in close proximity to the flavin ring in CHMOclosed
.
This enclosed cavity is defined by residues 145-146, 248, 279,
329, 434-435, 437, 492, and 507, as well as the FAD and
NADP+, and has a calculated volume of ∼150 Å3.30 A similar
pocket is also present in CHMOopen, but it is less defined, is not
enclosed, and is solvent accessible (Figure 3). While the
observed CHMOclosed substrate binding pocket should be able
to hold the primary substrate, cyclohexanone (SI-Figure 7), it
does not appear large enough to hold some of the larger
substrates that CHMO can oxygenate (SI-Table 2). However,
at least one wall of the active site is formed by elements from
(30) Kleywegt, G. J.; Jones, T. A. Acta Crystallogr. D: Biol. Crystallogr.
1994, 50, 178–185.
(31) Rial, D.; Cernuchova, P.; van Bielen, J.; Mihovilovic, M. J. Mol. Catal.
B: Enzymatic 2008, 50, 61–68.
9
8852 J. AM. CHEM. SOC. VOL. 131, NO. 25, 2009

Crystal Structures of Cyclohexanone Monooxygenase
A R T I C L E S
the nicotinamide head into a shallow pocket located between
residues 55 and 59. In this conformation the amide nitrogen is
within hydrogen-bonding distance of the flavin N5 and O4. More
importantly, however, this positively charged moiety is also
available for the stabilization of the requisite flavin-C4A
peroxide moiety (Figure 2). This negative charge-stabilizing role
is analogous to that carried out by NADP+ in FMO, with the
two enzymes adopting similar conformations with respect to
the orientation of the two cofactors.27
Immediately adjacent to the active site, K328 has been
implicated in cofactor binding and NADPH vs NADH selectiv-
ity.29 However, in neither CHMOclosed nor CHMOopen does the
K328 appear to interact directly with the cofactor. This lack of
an observed interaction despite biochemical evidence to the
contrary suggests that there exists at least one more NADP(H)-
bound conformation involved in flavin reduction. It is also
interesting to note that for K328 to interact with the phosphate
group would require only a small rotation of the NADP domain.
This would also have the effect of repositioning the nicotinamide
head laterally so that it would be more accessible for the
expected hydride transfer between C4 of the NADPH and N5
of the FAD.
Substrate Binding Site. Despite having the same overall fold,
there are two important exceptions to the structural homology
observed between CHMO and PAMO. The first is the Gly-Phe
dipeptide insert in CHMO at positions 278-279, which results
in a slight bulging of the loop connecting helix 262-275 and
helix 281-285 (CHMO numbering). The second main structural
deviation is between residues 433 and 434, where PAMO has
a two-residue insertion. Both of these inserts are located in the
CHMOclosed substrate binding site and are undoubtedly respon-
sible for the difference in substrate specificity observed in
PAMO.32 Another intriguing aspect of substrate specificity is
the effect loop 487-504 plays in completing the active-site
pocket. In CHMOopen, this loop is not visible and likely forms
an unstructured loop similar to that found in PAMO. Alterna-
tively, in CHMOclosed, this loop forms part of the closed active-
site pocket, permitting interactions between W492 and the
NADP+ ribose. Mutagenesis of this conserved residue to alanine
resulted in a protein with reduced activity. The exact nature of
the interaction between this loop and the substrate, and the role
it has in substrate specificity, will require further structural
characterization. The expanded substrate spectrum of CHMO
compared to that of CHMOAC may be attributable to differences
in this loop.
Previously, it had been suggested that the strictly conserved
R329 plays an important role in catalysis, with the equivalent
R337 having been observed in multiple conformations in the
PAMO structure. These “in” and “out” conformations were
speculated to be important in stabilization of the peroxyanion
intermediate, as well as in shifting during catalysis to accom-
modate NADPH binding.17 In CHMOopen, R329 occupies a
conformation similar to the “out” structure. In this position, there
is a distance of 3.2 Å between NH of R329 and the amide
nitrogen of NADP+. This interaction may effect the relatively
tight binding of oxidized NADP+ to the enzyme, post catalysis.
Contrasting the above similarities between the PAMO “out”
and CHMOopen structures, R329 appears to occupy a new locus
in CHMOclosed. In this conformation, a slightly tighter interaction
can be noted between NH and the NADP+ at 2.9 Å; however,
given the translation of the cofactor, R329 fills a position not
previously observed. Unlike the PAMO structure, where the
variations between “in” and “out” can be attributed to differences
in rotomers, in CHMOopen and CHMOclosed R329 actually does
not significantly change rotomeric conformations. Here, the large
shift between the two structures is due to the flexing of loop
327-330 between the two structures, resulting in the pushing
of the nicotinamide head by R329. For this reason, we have
given this configuration the name “push”. This movement is
greatly facilitated by the closing of loop 487-504, which also
serves to push NADP+ deeper into the enzyme via interactions
with W492. In this manner, R329 is now ideally situated to
facilitate the reaction of molecular oxygen with reduced FAD
and subsequently stabilize either the peroxyflavin or Criegee
intermediates (Figure 2 and SI-Figure 7).
Role of the BVMO Signature Motif in Cofactor Binding. The
presence of a FXGXXXHXXXWP motif has been shown to
be an identifier of type I BVMOs.33 In the structure of the
CHMO, this signature sequence is located at the start of the
NADP domain spanning residues 160-171. A similar motif has
been observed in the structurally similar FMOs (FXGXXX-
HXXX(Y/F).5 It has been demonstrated that this segment plays
an essential role in cofactor binding and catalysis in that
mutations of the central histidine have dramatic effects on
enzyme activity. Somewhat paradoxically, the localization of
the motif far from the active site has made it difficult to ascribe
a direct function in catalysis, and thus the role was assumed to
be structural. Here, the presence of the multiple crystal structures
permits us to gain more insight into the role of this critical
segment.
Comparison of the CHMO, PAMO, and available FMO
structures reveals that the strictly conserved aromatic ends of
the motif adopt nearly identical conformations. That is to say,
in all available structures, the terminal residues are buried deep
within the NADP binding domain and serve as “sticky ends”
that securely fix the signature motif to the NADP domain. The
role of the central histidine, however, is more complex. Based
on the PAMO structure, it appears that this residue is completely
solvent exposed in the absence of NADP+/H and does not
appear to have any direct or indirect role in substrate binding.17
In the NADP+-bound crystal structure of CHMO, however, this
critical residue is rotated inward such that the imidazole ring
of the histidine forms a hydrogen bond to the backbone of one
of the flexible linker segments (residues 381-386). In addition
to linking the FAD and NADP domains, this same linker region
is also important in positioning the NADP+ through steric
interactions. It thus appears that although the signature motif
does not play a role in catalysis directly, it is important as an
atomic switch that tightly coordinates multiple unconnected
portions of the enzyme to allow for binding and precise
positioning of the NADP+/H (Figure 4). This structural modula-
Given the two different modes of NADP+ binding, we
speculate that these structures represent two specific time points
in a very dynamic catalytic cycle, as indicated by spectroscopic
data.15 Our results suggest that the CHMOclosed structure mimics
the conformation of the enzyme in the post flavin reduction
(substrate binding) state, E ·FADH- ·NADP+, and the two
subsequent steps (Scheme 1). Given the buried positioning of
the NADP+, this conformer is ideal for stabilization of the
Criegee intermediate. Complementing this, the CHMOopen
structure likely represents the final step in the catalytic cycle,
namely NADP+ release (E ·FADH- ·NADP+ in Scheme 1).
(32) Clouthier, C. M.; Kayser, M. M.; Reetz, M. T. J. Org. Chem. 2006,
71, 8431–8437.
(33) Fraaije, M. W.; Kamerbeek, N. M.; van Berkel, W. J.; Janssen, D. B.
FEBS Lett. 2002, 518, 43–47.
9
J. AM. CHEM. SOC. VOL. 131, NO. 25, 2009 8853

A R T I C L E S
Mirza et al.
of multiple CHMO structures in complex with both FAD and
NADP+ provides snapshots of domain rotation and movements
undertaken by this biocatalyst. These structures provide insight
into how this enzyme can exploit the NADP+/H cofactor for
multiple purposes during the reaction cycle, as well as the critical
role of the BVMO signature sequence. This work provides
directions for further analysis and understanding of how the
biocatalytic potential of these enzymes might be developed in
the future.
Acknowledgment. The authors thank past and present members
of the Berghuis and Lau laboratories for their assistance and
suggestions. We also thank Marc Allaire for performing diffraction
data collection at the National Synchrotron Light Source. The work
at McGill was made possible through grants from the Natural
Sciences and Engineering Research Council of Canada and
Canadian Institute of Health Research awarded to A.M.B. A.M.B.
holds a Canada Research Chair in Structural Biology. Work at
Kansai University was supported in part by the Strategic Project
to Support the Formation of Research Bases at Private Universities
program. We thank Ping Xu for help in NMR analysis and G.
Ottolina for helpful discussions.
Figure 4. The BVMO signature sequence, shown here in yellow, is
anchored at each end by F160 and W170, which interact with hydrophobic
pockets in the NADP binding domain. In the middle of the motif, H166 is
shown interacting with the backbone of G381, which is part of one of the
linker segments shown in cyan that connect the FAD and NADP domains.
tion underscores the complexity and intricacy of the various
rotations CHMO undergoes to facilitate catalysis.
Supporting Information Available: Cloning, overexpression,
purification, mutagenesis, and substrate profiling of CHMO. This
material is available free of charge via the Internet at http://
pubs.acs.org.
Conclusion
Despite a tremendous body of literature developed for the
study of CHMOs, until now, there has been no available crystal
structure of a representative CHMO. Presently, the availability
JA9010578
9
8854 J. AM. CHEM. SOC. VOL. 131, NO. 25, 2009