An engineered old yellow enzyme that enables efficient synthesis of (4R,6R)-actinol in a one-pot reduction system

Article abstract of DOI:10.1002/cbic.201402555

(4R,6R)-Actinol can be stereo-selectively synthesized from ketoisophorone by a two-step conversion using a mixture of two enzymes: Candida macedoniensis old yellow enzyme (CmOYE) and Corynebacterium aquaticum (6R)-levodione reductase. However, (4S)-phorenol, an intermediate, accumulates because of the limited substrate range of CmOYE. To address this issue, we solved crystal structures of CmOYE in the presence and absence of a substrate analogue p-HBA, and introduced point mutations into the substrate-recognition loop. The most effective mutant (P295G) showed two- and 12-fold higher catalytic activities toward ketoisophorone and (4S)-phorenol, respectively, than the wild-type, and improved the yield of the two-step conversion from 67.2 to 90.1%. Our results demonstrate that the substrate range of an enzyme can be changed by introducing mutation(s) into a substrate-recognition loop. This method can be applied to the development of other favorable OYEs with different substrate preferences.

Full text of DOI:10.1002/cbic.201402555

DOI: 10.1002/cbic.201402555
Full Papers
An Engineered Old Yellow Enzyme that Enables Efficient
Synthesis of (4R,6R)-Actinol in a One-Pot Reduction
System
Shoichiro Horita,[a] Michihiko Kataoka,[b] Nahoko Kitamura,[c] Takuya Nakagawa,[c]
Takuya Miyakawa,[a] Jun Ohtsuka,[a] Koji Nagata,[a] Sakayu Shimizu,[c] and Masaru Tanokura*[a]
(4R,6R)-Actinol can be stereo-selectively synthesized from ke-
toisophorone by a two-step conversion using a mixture of two
enzymes: Candida macedoniensis old yellow enzyme (CmOYE)
and Corynebacterium aquaticum (6R)-levodione reductase.
However, (4S)-phorenol, an intermediate, accumulates because
of the limited substrate range of CmOYE. To address this issue,
we solved crystal structures of CmOYE in the presence and
absence of a substrate analogue p-HBA, and introduced point
mutations into the substrate-recognition loop. The most effec-
tive mutant (P295G) showed two- and 12-fold higher catalytic
activities toward ketoisophorone and (4S)-phorenol, respective-
ly, than the wild-type, and improved the yield of the two-step
conversion from 67.2 to 90.1%. Our results demonstrate that
the substrate range of an enzyme can be changed by introduc-
ing mutation(s) into a substrate-recognition loop. This method
can be applied to the development of other favorable OYEs
with different substrate preferences.
Introduction
Biocatalytic conversion using enzyme(s) is a promising tech-
nique for the preparation of optically active compounds be-
cause of their high enantioselectivity and substrate specificity.
However, the strict specificity sometimes results in a limited
substrate range, such that biocatalytic conversion is not always
applicable for the synthesis of target molecules. If a strategy to
create enzymes with broader substrate range can be estab-
lished, biocatalytic conversion will become a more useful tool
for the synthesis of various optically active compounds. Re-
cently, iterative saturation mutagenesis (ISM) has been report-
ed as a promising strategy to design artificial enzymes for bio-
catalytic conversion.[1] This method is an efficient approach for
the directed evolution of functional enzymes by performing
iterative cycles of high-throughput saturation mutagenesis at
rationally chosen sites.[1]
Proteins of the old yellow enzyme (OYE) family are potential
targets for protein engineering for useful biocatalysts, as they
catalyze the asymmetric reduction of C=C bonds in a,b-unsatu-
rated carbonyl compounds with high enantioselectivity (molec-
ular mechanisms below). Several trials have been performed
for changing the enantioselectivity and/or substrate range of
OYE proteins with linear or cyclic enone compounds.[2,3]
Recently, some OYE proteins were also reported to be of use
in multi-step asymmetric reduction in combination with other
biocatalysts.[4] (4R,6R)-4-Hydroxy-2,2,6-trimethylcyclohexanone
((4R,6R)-actinol) is an industrially important doubly chiral build-
ing block useful for the synthesis of xanthoxin, zeaxathin, and
related compounds.[4] A separate two-step biocatalytic conver-
sion method has been established to synthesize (4R,6R)-actinol
from the commercially available compound 2,6,6-trimethylcy-
clohex-2-ene-1,4-dione (ketoisophorone).[4,5] In this method, ke-
toisophorone is first reduced to (6R)-2,2,6-trimethyl-1,4-cyclo-
hexanedione ((6R)-levodione) by Saccharomyces cerevisiae OYE
(ScOYE2)[6] or Candida macedoniensis AKU4588 (CmOYE),[7] and
(6R)-levodione is then reduced to (4R,6R)-actinol by (6R)-levo-
dione reductase from Corynebacterium aquaticum M-13
(LVR)[8,9] (upper pathway in Scheme 1).[4]
[a] Dr. S. Horita, Dr. T. Miyakawa, Dr. J. Ohtsuka, Dr. K. Nagata,
Prof. Dr. M. Tanokura
Department of Applied Biological Chemistry
Graduate School of Agricultural and Life Sciences
University of Tokyo
1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-8657 (Japan)
E-mail: amtanok@mail.ecc.u-tokyo.ac.jp
To establish a more cost-effective and environmentally
friendly, synthetic method for (4R,6R)-actinol, we developed a
one-pot system for this two-step biocatalytic conversion. How-
ever, when the two-step biocatalytic conversion was per-
formed in one pot with a mixture of ScOYE2 and LVR, the yield
of (4R,6R)-actinol was only 16%, with an accumulation of an
intermediate, 4-hydroxy-2,6,6-trimethyl-2-cyclohexanone ((4S)-
phorenol)[4] in 48% yield, a result of the inefficient reduction of
(4S)-phorenol to (4R,6R)-actinol by ScOYE2. When CmOYE was
used instead of ScOYE2, the yield of (4R,6R)-actinol increased
[b] Prof. Dr. M. Kataoka
Division of Applied Life Sciences
Graduate School of Life and Environmental Sciences
Osaka Prefecture University
1-1 Gakuencho, Naka-ku, Sakai, Osaka 599-8531 (Japan)
[c] N. Kitamura, T. Nakagawa, Prof. Dr. S. Shimizu
Division of Applied Life Sciences
Graduate School of Agriculture, Kyoto University
Oiwake, Kitashirakawa, Sakyo-ku, Kyoto 606-8502 (Japan)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201402555.
ChemBioChem 0000, 00, 0 – 0
1
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
and b-strands forming the a8b8 barrel are named as shown in
Figure S2 (each loop is numbered according to the b-strand it
follows). No electron density of NADP+ in the catalytic site of
CmOYE or CmOYE–p-HBA was observed, even though a tenfold
molar excess of NADP+ to CmOYE was present in the crystalli-
zation drops.
In the absence of p-HBA, the electron density of loop 6 was
observed only in chain A, possibly because of the formation of
intermolecular hydrogen bonds between chains A in adjacent
asymmetric units (Table S2). In the presence of bound p-HBA,
hydrophobic interactions between loop 6 and p-HBA contrib-
ute to the stable conformation of loop 6, such that electron
density for loop 6 was observed for all the chains.
Scheme 1. Two-step biocatalytic conversion of ketoisophorone to (4R,6R)-ac-
tinol. Biocatalytic synthesis of (4R,6R)-actinol from ketoisophorone is per-
formed by CmOYE (or ScOYE2) and LVR. CmOYE and ScOYE2 show less cata-
lytic activity in the reduction of (4S)-phorenol than in the other reactions.
Both CmOYE and the CmOYE–p-HBA complex form homo-
dimers through interactions of the a5 and a6 helices in the re-
spective crystals; this is consistent with the gel filtration data,[5]
which showed that CmOYE behaves as a homodimer in solu-
tion irrespective of the presence or absence of p-HBA. The
rmsd between any protomer pair of CmOYE and the CmOYE–
p-HBA complex range from 0.2 to 1.0 ꢁ (Table S3).
to 67.2%, but 28.1% still remained as (4S)-phorenol (this
study).
The molecular mechanism of asymmetric reduction by OYEs
has been elucidated in detail.[10–13] In brief, a strictly conserved
histidine/asparagine or histidine/histidine pair[11] and a tyrosine
residue[12] are involved in the asymmetric reduction; a hydride
derived from FMNH2 is stereo-selectively transferred to Cb of
the bound a,b-unsaturated carbonyl compound, and the tyro-
sine residue donates a proton to Ca of the a,b-unsaturated car-
bonyl compound from the opposite side. However, the struc-
tural basis of the different substrate preferences was unknown.
To gain insight into the structural basis of the substrate rec-
ognition, we solved structures of CmOYE, in which its flexible
lid-forming loop near the catalytic site was present in open
and closed states. We then performed structure-guided muta-
tions of the lid-forming loop, and successfully obtained mu-
tants with higher catalytic activities against a wider range of
substrates. Further, we examined the yield of (4R,6R)-actinol in
the one-pot, two-step biocatalytic conversion with the most
active CmOYE mutant and LVR, and we observed a significantly
higher yield of (4R,6R)-actinol relative to that obtained with
wild-type CmOYE. We also discuss the structural difference of
the lid-forming loops in other OYE proteins. The structural and
biochemical results in this study provide insights into the
molecular basis of the lid-forming loop important for catalytic
activity of other OYEs.
Table S4 lists the hydrogen bonds between CmOYE and
FMN. These hydrogen bonds contribute to the orientation of
bound FMN, which is oriented in a similar manner to FMNs
bound to other OYEs, with its re face buried in the protein and
its si face directed toward the solvent (Figure S1). The amino
acid residues involved in the recognition of bound FMN are
highly conserved (Figure S3). The bound FMN contains a
planar isoalloxazine ring, consistent with the proposal that a
methionine residue at the position corresponding to Leu36 in
CmOYE would result in a “butterfly” conformation flavin,
whereas amino acid residues other than methionine would
not.[14]
Figure 1A shows the active site of the CmOYE–p-HBA com-
plex. The FobsꢀFcalcd electron density map of p-HBA is shown in
blue mesh contoured at 2.5s. p-HBA binds to CmOYE with
three hydrogen bonds: O1 in the aldehyde group of p-HBA
forms a hydrogen bond to the side-chain hydroxyl group of
Tyr375, and O4 in the hydroxyl group of p-HBA forms hydro-
gen bonds to NE2 of His191 and ND2 of Asn194. In contrast,
no electron density was observed for any substrates used in
the co-crystallization. This is probably because these substrates
bind less tightly to CmOYE than p-HBA, without forming a
hydrogen bond with Tyr375.
Results and Discussion
A comparison of the crystal structures of CmOYEs in the
presence and absence of p-HBA shows the high flexibility of
loop 6. In the presence of bound p-HBA, loop 6 adopts a
closed conformation in all CmOYE chains, whereas in the
absence of bound p-HBA, loop 6 is observed only in chain A,
where it adopts an unprecedented open conformation
(Figure 2). All reported OYE structures show loop 6 adopting
the closed conformation. Flexibility of loop 6 was also suggest-
ed for Saccharomyces pastorianus OYE (SpOYE),[15] because
NAD(P)H could not access either the catalytic site or the FMN
if the loop adopted only the closed conformation.
Crystal structures of CmOYE and CmOYE–p-HBA
The crystal structures of both CmOYE (403 aa) and a CmOYE–
p-HBA (p-hydroxybenzaldehyde) complex were determined at
1.8 ꢁ resolution (Figure S1 in the Supporting Information; PDB
IDs: 4TMB and 4TMC; X-ray diffraction data and structural
refinement statistics in Table S1). The crystals of both CmOYE
and the CmOYE–p-HBA complex contained two CmOYE dimers
in an asymmetric unit. Each protomer of CmOYE or CmOYE–p-
HBA consists of eight a-helices and eight b-strands, and folds
into an a8b8 barrel (the conserved motif among OYEs) with
a bound FMN molecule at the top of the b barrel. The secon-
dary structure elements are shown in Figure S2. The a-helices
To gain insights into the substrate preferences of CmOYE,
the binding model of (4S)-phorenol to CmOYE was built based
on the crystal structures of the CmOYE–p-HBA complex (Fig-
&
ChemBioChem 0000, 00, 0 – 0
www.chembiochem.org
2
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
bulky dimethyl group at C6 of (4S)-phorenol would not be
favored by CmOYE, because the dimethyl group at C6 can col-
lide with Pro295 in loop 6 and Phe250 in loop 5 (Figures 1B
and S4).
To examine the effects of Phe250, Pro295, and Phe296 on
the substrate preference of CmOYE, mutants F250G, P295G,
and F296G were constructed to avoid the possible collision
between the dimethyl group in (4S)-phorenol and these three
residues in CmOYE. The results demonstrated that, compared
to wild-type CmOYE, the P295G mutant showed higher catalyt-
ic activities toward ketoisophorone and (4S)-phorenol. This
could be explained by Pro295 colliding with the dimethyl
group of ketoisophorone and (4S)-phorenol. Thus, Pro295 in
loop 6 of CmOYE acts as a substrate filter. Because of the lack
of side-chain atoms in Gly and the much greater flexibility of
torsion angles (f and y) in Gly compared with those in Pro,
the P295G mutant has more space in the active site and in-
creased conformational flexibility of loop 6, the lid of the cata-
lytic site. Then we examined the importance of Phe296 in
loop 6 by saturation mutations, and demonstrated that this
residue has a less significant effect on the catalytic activity
than Pro295 (Figure S6).
To investigate further, a series of mutants with polyglycine in
loop 6 of CmOYE were constructed, and their catalytic activi-
ties were examined. The mutants were named CmOYE-
(loop 6!Gn) to indicates a CmOYE mutant in which the loop 6
of CmOYE (289EPRVTDPFLPEFEKWFKEGT308) is replaced with
a polyglycine linker Gn (n=2–5). All four mutants exhibited
higher catalytic activity toward (4S)-phorenol (Figure 3). These
results support the hypothesis that the loop 6 plays a signifi-
cant role in substrate recognition. Although the kinetic param-
eters of the series of mutants were examined, not all data
were obtained because of the low affinity of each mutant
toward ketoisophorone and/or (4S)-phorenol (Table S5).
The binding model of (4S)-phorenol to CmOYE suggests that
the bulky dimethyl group at C6 of (4S)-phorenol would not be
Figure 1. A) The catalytic site of CmOYE in complex with p-HBA. The elec-
tron density map OMIT jmFoꢀDFc j of p-HBA (shown both in a stick model
and blue mesh contoured to 2.5s). The amino acid residues shown in
orange, blue, and magenta are residues on loop 5, loop 6, and others, re-
spectively. B) Putative binding model of (4S)-phorenol to CmOYE. The di-
methyl groups at C6 are shown as yellow spheres. The oxygen and carbon
atoms in p-HBA are red and white, respectively. Dotted lines represent hy-
drogen bonds.
Figure 2. Superposition of CmOYE structures in the absence (green) and
presence (magenta) of p-HBA in the catalytic pockets. The structures shown
in green and magenta represent open and closed forms of CmOYE (loop 6),
respectively. Amino acid residues in the catalytic sites, FMN (yellow), and
p-HBA (gray), are shown as stick models.
ures 1B, S4, and S5). The O, C, Ca, and Cb atoms of (4S)-phore-
nol were superimposed onto the position of the O, C, Ca, and
Cb atoms of p-HBA, respectively, as it has been proposed that
the position of the carbonyl group of the substrate is trapped
by the histidine/asparagine or histidine/histidine pair,[11] and
that the targeted C=C bond is placed on top of N5 of FMN.
The binding model of (4S)-phorenol to CmOYE implies that the
Figure 3. Specific activity of CmOYE and its mutants toward cyclohex-2-en-
1-one (black), ketoisophorone (gray), and (4S)-phorenol (white). It should be
noted that specific activity does not reflect binding affinity.
ChemBioChem 0000, 00, 0 – 0
www.chembiochem.org
3
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
favored by CmOYE, because the dimethyl group in (4S)-phore-
nol would collide with Phe250 in loop 5 and Pro295 in loop 6
(Figure 1B, Figure S5). This is supported by the fact that
CmOYE(F250G) and CmOYE(F250A), both of which should have
larger substrate binding pockets than CmOYE(WT), showed
increased catalytic activities toward (4S)-phorenol (Figure 3). In
contrast, CmOYE(F250A/P295G) and CmOYE(F250G/P295G/
F296G), which should have even larger substrate binding pock-
ets than CYE(F250G) and CmOYE(F250A), showed less catalytic
activities towards cyclohex-2-en-1-one, ketoisophorone, and
(4S)-phorenol (Figure 3), thus suggesting that these substrate-
binding pockets are too big. We chose CmOYE(P295G), which
showed two- and 12-fold higher catalytic activities toward
ketoisophorone and (4S)-phorenol, respectively, than the wild-
type enzyme, as the most suitable enzyme for the efficient bio-
catalytic conversions of ketoisophorone to (6R)-levodione and
(4S)-phorenol to (4R,6R)-actinol.
pounds B and C in Figure S7), both of which have a 3-methyl
group, whereas some catalytic activity was detected toward
ketoisophorone and cyclohex-2-en-1-one (compounds D and A
in Figure S7), both of which lack a 3-methyl group, thus sug-
gesting that ketoisophorone and cyclohex-2-en-1-one enter
the catalytic pocket of CmOYE in a similar orientation. In other
words, the carbonyl group at position 1 of ketoisophorone is
equivalent to the carbonyl group of cyclohex-2-en-1-one when
recognized by CmOYE. The molecular model constructed
based on the crystal structure of CmOYE indicates that the 3-
methyl group of 3-methylcyclohex-2-en-1-one and isophorone
would collide with Thr37 and Trp116 in CmOYE (Figure S8),
thus clearly explaining why these compounds are not suitable
substrates for CmOYE.
Mutation in loop 6 can alter the substrate specificity of
OYEs
The time-courses of the production of (6R)-levodione, (4S)-
phorenol, and (4R,6R)-actinol from ketoisophorone in the one-
pot two-step biocatalytic conversion by CmOYE(WT) or
CmOYE(P295G) with LVR are shown in Figure 4. The conversion
Based on crystallographic data, the structures of the catalytic
sites of OYEs can be divided into two groups: the CmOYE
family and BsOYE family. In the case of the CmOYE family,
loop 6 acts as a mobile lid of the catalytic site, whereas in the
BsOYE family, loop 6 is a relatively rigid loop (Figure 5). Based
on this structural classification, C. macedoniensis OYE (CmOYE;
PDB ID: 4TMB; this study), S. pastorianus OYE (SpOYE; PDB ID:
1OYB),[15] and Shewanella oneidensis OYE (SoOYE; PDB ID:
2GQ9)[14] belong to the CmOYE family; Bacillus subtilis OYE
(BsOYE, also known as YqjM; PDB ID: 1Z42),[16] Thermus scoto-
ductus SA-01 OYE (TsOYE, also known as CrS; PDB ID: 3HGJ)[17]
and Pseudomonas putida xenobiotic reductase A (XenA; PDB
ID: 3L66)[18] belong to the BsOYE family. OYE can be used for
highly stereo-selective reduction, but an (R)-enantiomer of 6-
membered cyclic enone has not been yet obtained with OYE.
~
Figure 4. Time-course of the production for (6R)-levodione ( ), (4S)-phorenol
~
*
*
(
), and (4R,6R)-actinol ( ) from ketoisophorone ( ) in a simultaneous two-
step biocatalytic conversion by A) CmOYE(WT) and B) CmOYE(P295G). The
reactions were stopped when the yields of (4R,6R)-actinol approached the
plateau phase (2.5 h).
by CmOYE(P295G) and LVR showed a higher yield of (4R,6R)-
actinol than by CmOYE(WT) and LVR. After 2.5 h, CmOYE(WT)
and LVR produced (4R,6R)-actinol in 67.2% yield with 28.1%
accumulation of the (4S)-phorenol intermediate, whereas
CmOYE(P295G) and LVR produced (4R,6R)-actinol in 90.1%
yield with only 9.1% accumulation of (4S)-phorenol. These re-
sults demonstrate that introducing mutations into loop 6 (the
lid of the catalytic site) enhances the catalytic activities of
CmOYE with (4S)-phorenol, and contributes to the significant
improvement in the yield of (4R,6R)-actinol in the one-pot,
two-step biocatalytic conversion.
Ketoisophorone (compound D in Figure S7) is a good sub-
strate for CmOYE(WT); it has two carbonyl groups that would
be recognized by the His191/Asn194 pair of CmOYE. To exam-
ine the modes of recognition of ketoisophorone by CmOYE,
the catalytic activities toward cyclohex-2-en-1-one, 3-methylcy-
clohex-2-en-1-one, isophorone, and ketoisophorone were ex-
amined (Figure S7). No catalytic activity of CmOYE was detect-
ed toward 3-methylcyclohex-2-en-1-one or isophorone (com-
Figure 5. Catalytic site and loop 6 of A) the CmOYE family and B) the BsOYE
family. CmOYE, SpOYE, SoOYE, BsOYE, TsOYE, and XenA are shown in green,
cyan, yellow, magenta, brown and blue, respectively.
&
ChemBioChem 0000, 00, 0 – 0
www.chembiochem.org
4
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
plate.[7] The PCR primers used were 5’-CGCGC GCGCA TATGA
AAAAC AATAA AGAAC GACAA GGAAA-3’ (including an NdeI site,
underlined) and 5’-GGGGC CCCGG ATCCT TATTA AGAGA GGGGA
AGGTG CACTT CA-3’ (including a BamHI site, underlined). The PCR
product was digested with NdeI and BamHI, and ligated into the
expression vector pET-28a(+) (Novagen). The point and deletion
mutants were introduced by site-directed mutagenesis using the
pET-28a plasmid. To obtain CmOYE with a thrombin-cleavable N-
terminal His6-tag, Escherichia coli BL21(DE3) harboring the expres-
sion plasmid was grown at 378C in LB medium, and protein ex-
pression was induced by adding IPTG (1 mm) and further incubat-
ing at 378C for 3 h. Cells were harvested by centrifugation, resus-
pended in lysis buffer (Tris·HCl (50 mm, pH 8.0), NaCl (300 mm),
imidazole (10 mm)), and disrupted by sonication on ice. After cen-
trifugation, CmOYE with an N-terminal His6-tag was purified with
Ni Sepharose 6 Fast Flow (GE Healthcare). The His6 tag was re-
moved by thrombin protease (Novagen) digestion. CmOYE was fur-
ther purified by anion exchange chromatography with a Resource
Q 6 mL column, followed by size exclusion chromatography with
a Superdex 200 HR 10/30 column (GE Healthcare). The purified
sample was then concentrated (~10 mgmLꢀ1) in a 20 mL Vivaspin
concentrator (10 kDa cutoff; Sartorius, Gçttingen, Germany). The
In a previous study, an iterative saturation mutagenesis (ISM)
method[1] was applied to expand the substrate range of BsOYE
by using 3-methylcyclohex-2-en-1-one as a model substrate.[19]
However, the results in our study indicated that not only
BsOYE-family but also CmOYE-family proteins would be candi-
dates for ISM to increase the catalytic activities toward 3-meth-
ylcyclohex-2-en-1-one and to control the R and S selectivities.
Conclusion
We first aimed to establish a biocatalytic method for the syn-
thesis of (4R,6R)-actinol (an important carotenoid precursor)
from a commercially available compound, and found that a
two-step conversion from ketoisophorone to (4R,6R)-actinol is
possible by using the two enzymes CmOYE and LVR, both of
which showed high enantioselectivity. However, the one-pot
two-step conversion with a mixture of these enzymes gave
only 67.2% yield of (4R,6R)-actinol. As this low yield was a
result of the narrow substrate preference of CmOYE, we then
aimed to create an artificial CmOYE with higher activity toward
a broader range of substrates, by characterizing a flexible
region of CmOYE near the catalytic site. We first solved the
crystal structures of CmOYE in the absence and presence of p-
HBA (a substrate analogue) to visualize the substrate recogni-
tion mechanism of CmOYE. We observed two different states
of loop 6 (open and closed states, in the absence and presence
of p-HBA, respectively) and found that this loop acts as a lid at
the catalytic site. Based on the ligand-free and ligand-bound
structures, we propose that Pro295 and Phe296 in loop 6 as
well as Phe250 in loop 5 are key residues in substrate recogni-
tion, and are amenable to mutational analysis. We found that
CmOYE(P295G) shows 2- and 12-fold higher catalytic activity
toward ketoisophorone and (4S)-phorenol, respectively, than
the wild-type enzyme, and significantly improved the yield of
(4R,6R)-actinol (from 67.2 to 90.1%) in a one-pot two-step
transformation.
protein concentration was determined based on absorbance (e280
=
8940mꢀ1 cmꢀ1) obtained with a NanoDrop ND-1000 spectropho-
tometer (Thermo Scientific).
Crystallization and data collection: Crystallization experiments
were performed with the sitting drop vapor diffusion method at
293 K. Crystals of CmOYE and CmOYE–p-HBA were obtained by
mixing protein (1.0 mL, 10 mgmLꢀ1) with NADP+ (5 mm) with/with-
out p-HBA (5 mm; Sigma-Aldrich) in a reservoir solution (1.0 mL)
containing PEG 3350 (25%, v/v), Tris·HCl (100 mm, pH 8.0), and am-
monium sulfate (200 mm). Diffraction data were collected at NW-
12 A at PF-AR (Ibaraki, Japan). Data processing was carried out
with the programs HKL2000[20] and XDS.[21] Crystals of CmOYE
belong to space group C2 with unit cell dimensions a=287.5 ꢁ,
b=59.6 ꢁ, c=100.3 ꢁ, and b=109.98; crystals of CmOYE–p-HBA
belong to space group P212121 with unit cell dimensions a=52.3 ꢁ,
b=150.9 ꢁ, and c=199.7 ꢁ. The asymmetric units for both CmOYE
and CmOYE–p-HBA contained four molecules.
Structure determination of CYE and the CYE–p-HBA complex:
The structures of CmOYE and CmOYE–p-HBA were determined by
molecular replacement with the program MolRep[22] with the
atomic coordinates of S. pastorianus OYE (SpOYE; PDB ID: 1OYA;
69% sequence identity to CmOYE) as the initial model. Further
model building and refinements were performed with the programs
ARP/wARP,[23] Coot,[24] and Refmac.[25] FMN, p-HBA, and water mole-
cules were modeled in the final stages of refinement based on the
FobsꢀFcalcd electron density map. The refined structure of CmOYE
was used as the template for the molecular replacements of the
CmOYE–p-HBA. The refined structures were visualized with PyMol
(http://pymol.sourceforge.net/). To evaluate the structural similarity,
the DaliLite server[26] was used to calculate the rms deviations of
protomers among OYEs. The STRAP program was used for struc-
ture-based sequence alignment (http://www.bioinformatics.org/
strap/).
Notably, the CmOYE variants lacking the lid-forming loop
showed higher and lower catalytic activity toward (4S)-phore-
nol and 3-methylcyclohex-2-en-1-one, respectively, thus sug-
gesting that the lid-forming loop of CmOYE acts as a substrate
filter. Comparisons of catalytic sites of CmOYE-family and
BsOYE-family proteins revealed notable structural differences:
CmOYE-family proteins have flexible lid-forming loops, whereas
BsOYE-family proteins have no lid-forming loops (Figure 5).
This observation agrees with the hypothesis that in vivo sub-
strates of CmOYE-family proteins and BsOYE-family proteins
are different, although it needs to be considered whether the
lids of CmOYE-family proteins evolved primarily for optimizing
turnover rates of NAD(P)H and substrates. From an industrial
point of view, the observation that loop 6 in OYEs acts as a sub-
strate filter is of great interest for the development of novel
OYE biocatalysts.
Catalytic activities: Cyclohex-2-en-1-one and 3-methylcyclohex-2-
en-1-one were purchased from Wako Pure Chemical Industries
(Osaka, Japan); ketoisophorone and (4S)-phorenol were from
Nippon-Roche Co. (Tokyo, Japan); and menadione was from Naca-
lai Tesque Inc. (Kyoto, Japan). Catalytic activity was examined in
Tris·HCl (200 mm, pH 7.5, 0.5 mL) containing NADPH (0.32 mm, co-
factor) and an appropriate amount of the wild-type or mutant
Experimental Section
Protein expression and purification: The gene encoding CmOYE
was amplified by PCR using the cloned CmOYE gene as the tem-
ChemBioChem 0000, 00, 0 – 0
www.chembiochem.org
5
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
[1] M. T. Reetz, J. D. Carballeira, Nat. Protoc. 2007, 2, 891–903.
[2] M. Hall, C. Stueckler, H. Ehammer, E. Pointner, G. Oberdorfer, K. Gruber,
B. Hauer, R. Stuermer, W. Kroutil, P. Macheroux, K. Faber, Adv. Synth.
Catal. 2008, 350, 411–418.
[3] S. K. Padhi, D. J. Bougioukou, J. D. Stewart, J. Am. Chem. Soc. 2009, 131,
3271–3280.
[4] M. Wada, A. Yoshizumi, Y. Noda, M. Kataoka, S. Shimizu, H. Takagi, S. Na-
kamori, Appl. Environ. Microbiol. 2003, 69, 933–937.
[5] M. Kataoka, A. Kotaka, A. Hasegawa, M. Wada, A. Yoshizumi, S. Naka-
mori, S, Shimizu, Biosci. Biotechnol. Biochem. 2002, 66, 2651–2657.
[6] K. Stott, K. Saito, D. J. Thiele, V. Massey, J. Biol. Chem. 1993, 268, 6097–
6106.
[7] M. Kataoka, A. Kotaka, R. Thiwthong, M. Wada, S. Nakamori, S. Shimizu,
J. Biotechnol. 2004, 114, 1–9.
[8] M. Wada, A. Yoshizumi, S. Nakamori, S. Shimizu, Appl. Environ. Microbiol.
1999, 65, 4399–4403.
[9] S. Sogabe, A. Yoshizumi, T. A. Fukami, Y. Shiratori, S. Shimizu, H. Takagi,
S. Nakamori, M. Wada, J. Biol. Chem. 2003, 278, 19387–19395.
[10] R. E. Williams, N. C. Bruce, Microbiology 2002, 148, 1607–1614.
[11] B. J. Brown, Z. Deng, P. A. Karplus, V. Massey, J. Biol. Chem. 1998, 273,
32753–32762.
[12] R. M. Kohli, V. Massey, J. Biol. Chem. 1998, 273, 32763–32770.
[13] B. J. Brown, J.-W. Hyun, S. Duvvuri, P. A. Karplus, V. Massey, J. Biol. Chem.
2002, 277, 2138–2145.
CmOYE (enzyme). After 2 min of incubation at 303 K without sub-
strate, the reaction was started by addition of substrate (cyclohex-
2-en-1-one (4 mm), ketoisophorone (4 mm), (4S)-phorenol (4 mm),
3-methylcyclohex-2-en-1-one (4 mm), or menadione (0.2 mm)), and
then the decrease in absorbance at 340 nm was monitored at
303 K. One unit (1 U) of enzyme activity was defined as the
amount catalyzing the oxidation of 1 mmol of NADPH per minute.
The enzyme activity of the wild-type CmOYE was used as the refer-
ence. The kinetics parameters were calculated by using the same
method described above with a variety of substrate concentra-
tions.
Enzymatic production: Recombinant cells that overproduce
LVR,[27] CmOYE(WT), or CmOYE(P295G) were suspended in Tris·HCl
buffer (20 mm, pH 7.4), and then disrupted by sonication on ice.
The supernatant after centrifugation was used as a cell-free extract.
The amount of wild-type or mutant CmOYE in cell-free extract was
determined by using cyclohex-2-en-1-one as the substrate. Enzy-
matic production was examined in a 2.5 mL reaction mixture con-
taining ketoisophorone (6.6 mm 1 mgmLꢀ1
; substrate), NADH
(20 mm; cofactor), NADPH (10 mm; cofactor), cell-free extract (con-
taining 0.8 mg of CmOYE(WT) or CmOYE(P295G), enzyme amount
calculated based on the specific activity with cyclohex-2-en-1-one),
cell-free extract containing LVR (0.2 U; enzyme amount calculated
based on the specific activity for (6R)-levodione), of glucose dehy-
drogenase (40 U), and glucose (278 mm) in potassium phosphate
buffer (200 mm, pH 7.0). The mixture was incubated at 308C for
2.5 h. The concentrations of the substrate and products were mea-
sured by gas chromatography as described previously.[7]
[14] D. van den Hemel, A. Brigꢂ, S. N. Savvides, J. Van Beeumen, J. Biol.
Chem. 2006, 281, 28152–28161.
[15] K. M Fox, P. A. Karplus, Structure 1994, 2, 1089–1105.
[16] K. Kitzing, T. B. Fitzpatrick, C. Wilken, J. Sawa, G. P. Bourenkov, P. Macher-
oux, T. Clausen, J. Biol. Chem. 2005, 280, 27904–27913.
[17] D. J. Opperman, B. T. Sewell, D. Litthauer, M. N. Isupov, J. A. Littlechild, E.
van Heerden, Biochem. Biophys. Res. Commun. 2010, 393, 426–431.
[18] O. Spiegelhauer, S. Mende, F. Dickert, S. H. Knauer, G. M. Ullmann, H.
Dobbek, J. Mol. Biol. 2010, 398, 66–82.
[19] D. J. Bougioukou, S. Kille, A. Taglieber, M. T. Reetz, Adv. Synth. Catal.
2009, 351, 3287–3305.
Acknowledgements
[20] Z. Otwinowski, W. Minor, Methods Enzymol. 1997, 276, 307–326.
[21] W. Kabsch, J. Appl. Crystallogr. 1993, 26, 795–800.
[22] A. Vagin, A. Teplyakov, Acta Crystallogr. Sect. D Biol. Crystallogr. 2000, 56,
1622–1624.
[23] A. Perrakis, M. Harkiolaki, K. S. Wilson, V. S. Lamzin, Acta Crystallogr. Sect.
D Biol. Crystallogr. 2001, 57, 1445–1450.
[24] P. Emsley, K. Cowtan, Acta Crystallogr. Sect. D Biol. Crystallogr. 2004, 60,
2126–2132.
[25] A. A. Vagin, R. A. Steiner, A. A. Lebedev, L. Potterton, S. McNicholas, F.
Long, G. N. Murshudov, Acta Crystallogr. Sect. D Biol. Crystallogr. 2004,
60, 2184–2195.
We thank the beamline staff at Photon Factory for their kind
help with the data collection, Dr. Yusuke Ogura and Dr. Keisuke
Yamamoto for sincere discussion, Dr. John Salvatore Scotti for
help revising the manuscript. The synchrotron-radiation experi-
ments were done with the approval of Photon Factory, KEK
(2009G199). This work was supported in part by the National
Project on Protein Structural and Functional Analyses, the Target-
ed Proteins Research Program, the Platform for Drug Discovery,
Informatics, and Structural Life Sciences, and Grants-in-Aid for
Scientific Research from the Ministry of Education, Culture,
Sports, Science, and Technology of Japan.
[26] L. Holm, J. Park, Bioinformatics 2000, 16, 566–567.
[27] A. Yoshisumi, M. Wada, H. Takagi, S. Shimizu, S. Nakamori, Biosci. Bio-
technol. Biochem. 2001, 65, 830–836.
Keywords: biocatalysis · enzyme catalysis · enzymes · old
yellow enzyme · X-ray crystallography
Received: September 23, 2014
Published online on && &&, 0000
&
ChemBioChem 0000, 00, 0 – 0
www.chembiochem.org
6
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

FULL PAPERS
Crystal structures of old yellow en-
zyme from C. macedoniensis (CmOYE)
identify a flexible lid-forming loop,
which is revealed to be important for
substrate recognition. The yield of
(4R,6R)-actinol was improved from 67.2
to 90.1% in a one-pot, two-step biocat-
alytic reduction system, by introducing
mutations into this loop. Our results will
be of use in developing new OYE bio-
catalysts.
S. Horita, M. Kataoka, N. Kitamura,
T. Nakagawa, T. Miyakawa, J. Ohtsuka,
K. Nagata, S. Shimizu, M. Tanokura*
&& – &&
An Engineered Old Yellow Enzyme
that Enables Efficient Synthesis of
(4R,6R)-Actinol in a One-Pot Reduction
System
ChemBioChem 0000, 00, 0 – 0
www.chembiochem.org
7
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ