Stereoselective oxidation of aryl-substituted vicinal diols into chiral α-hydroxy aldehydes by re-engineered propanediol oxidoreductase

Article abstract of DOI:10.1021/cs400824h

α-Hydroxy aldehydes are chiral building blocks used in synthesis of natural products and synthetic drugs. One route to their production is by regioselective oxidation of vicinal diols and, in this work, we aimed to perform the oxidation of 3-phenyl-1,2-propanediol into the corresponding α-hydroxy aldehyde applying enzyme catalysis. Propanediol oxidoreductase from Escherichia coli efficiently catalyzes the stereoselective oxidation of S-1,2-propanediol into S-lactaldehyde. The enzyme, however, shows no detectable activity with aryl-substituted or other bulky alcohols. We conducted ISM-driven directed evolution on FucO and were able to isolate several mutants that were active with S-3-phenyl-1,2-propanediol. The most efficient variant displayed a k_cat/K_M of 40 s^-1 M^-1 and the most enantioselective variant an E-value (S/R) of 80. Furthermore, other isolated variants showed up to 4400-fold increased activity with another bulky substrate, phenylacetaldehyde. The results with engineered propanediol oxidoreductases identified amino acids important for substrate selectivity and asymmetric synthesis of aryl-substituted α-hydroxy aldehydes. In conclusion, our study demonstrates the feasibility of tailoring the catalytic properties of propanediol oxidoreductase for biocatalytic properties.

Full text of DOI:10.1021/cs400824h

Research Article
pubs.acs.org/acscatalysis
Stereoselective Oxidation of Aryl-Substituted Vicinal Diols into Chiral
α‑Hydroxy Aldehydes by Re-Engineered Propanediol
Oxidoreductase
Cecilia Blikstad,† Kathe M. Dahlstrom,‡ Tiina A. Salminen,‡ and Mikael Widersten*,†
̈
̈
†Department of Chemistry−BMC, Uppsala University, Box 576, SE-751 23 Uppsala, Sweden
‡Structural Bioinformatics Laboratory, Åbo Akademi University, Tykistokatu 6A, FIN-20520 Turku, Finland
̈
S
* Supporting Information
ABSTRACT: α-Hydroxy aldehydes are chiral building blocks
used in synthesis of natural products and synthetic drugs. One
route to their production is by regioselective oxidation of
vicinal diols and, in this work, we aimed to perform the
oxidation of 3-phenyl-1,2-propanediol into the corresponding
α-hydroxy aldehyde applying enzyme catalysis. Propanediol
oxidoreductase from Escherichia coli efficiently catalyzes the
stereoselective oxidation of S-1,2-propanediol into S-lactalde-
hyde. The enzyme, however, shows no detectable activity with
aryl-substituted or other bulky alcohols. We conducted ISM-
driven directed evolution on FucO and were able to isolate
several mutants that were active with S-3-phenyl-1,2-propane-
diol. The most efficient variant displayed a kcat/KM of 40 s−1 M−1 and the most enantioselective variant an E-value (S/R) of 80.
Furthermore, other isolated variants showed up to 4400-fold increased activity with another bulky substrate, phenylacetaldehyde.
The results with engineered propanediol oxidoreductases identified amino acids important for substrate selectivity and
asymmetric synthesis of aryl-substituted α-hydroxy aldehydes. In conclusion, our study demonstrates the feasibility of tailoring
the catalytic properties of propanediol oxidoreductase for biocatalytic properties.
KEYWORDS: biocatalytic diol oxidation, regioselectivity, directed enzyme evolution, principal components analysis, steady-state kinetics
because of their intrinsic reactivity and thereby tendency to
participate in unwanted side reactions (e.g., ref 5). In cells,
ketones and aldehydes can be formed from oxidation of
secondary or primary alcohols, respectively. These trans-
formations are typically catalyzed by dehydrogenases utilizing
nicotinamide cofactors (NAD(P)+) as electron acceptors. As
these enzymes are generally unprecedented in their catalytic
powers and stereoselectivity, attempts to harness their
INTRODUCTION
■
The carbonyl group is arguably the most important functional
group in biomolecules. Proteins, carbohydrates, nucleic acids,
lipids, various metabolites, and a number of cofactors all
contain carbonyl groups in various forms: esters, carboxylic
acids, ketones, amides, and aldehydes. The polar character and
hydrogen bonding capacity of the carbonyl group stabilize
biomolecular structures and enable intra- and intermolecular
interactions necessary in essential life processes such as storage
and transfer of genetic information, chemical catalysis by
enzymes, and ion transport. The reactivities and physicochem-
ical properties of carbonyl containing molecules place them as
essential building blocks in metabolic reactions involving aldol
condensations, redox reactions, racemizations, sugar cycliza-
tions, amino- and acyl transfer, amidations, and isomerizations.
The same reaction types are equally important in synthetic
chemistry, in design and production of pharmaceutical drugs,
where carbonyls are utilized for interaction with target
biomolecules or as attachment points for addition of other
functional groups and site-specific derivatizations.
efficiencies for synthetic purposes have been conducted.6−9
A
limiting factor for the use of biocatalysis is the limited substrate
scope of most naturally occurring enzymes. This limitation can
now, however, be addressed in favorable cases using current
methodology for re-engineering of enzyme structures to
improve their suitability as biocatalysts.10
Propanediol oxidoreductase from Escherichia coli, FucO
(EC:1.1.1.77), participates in the catabolism of unusual
carbohydrates such as fucose and rhamnose.11 FucO catalyzes
the reversible conversion of the three-carbon α-hydroxy
aldehyde S-lactaldehyde and S-1,2-propanediol utilizing
NADH/NAD+ as cofactors (Scheme 1A). The strict
α-Hydroxy-substituted aldehydes and ketones are found
naturally in aldose and ketose sugars and are important chiral
building blocks in the synthesis of natural products and
synthetic drugs.1−4 Their synthesis is, however, not trivial
Received: July 8, 2013
Revised: November 1, 2013
Published: November 4, 2013
© 2013 American Chemical Society
3016
dx.doi.org/10.1021/cs400824h | ACS Catal. 2013, 3, 3016−3025

ACS Catalysis
Research Article
FucO is a homodimeric protein with a classical Rossman
folded coenzyme binding site15 (Figure 1A). The enzyme
Scheme 1. (A) Interconversion of S-2-Hydroxypropanal (S-
lactaldehyde) and S-1,2-Propanediol Catalyzed by FucO
Using NAD(H) as a Cofactor; (B) Serial Epoxide Hydrolysis
and Primary Alcohol Oxidation Producing a α-Hydroxyl
a
Substituted Aldehyde
Figure 1. (A) Dimeric structure of FucO. (B) The binding sites for the
cofactor and the alcohol/aldehyde. (C) The alcohol/aldehyde binding
site. The view is rotated 90 degrees around the y-axis from the view in
(B) to show the active site from the entry point of the alcohol/
aldehyde substrate. Residues mutated to create libraries of variant
FucO enzyme are indicated by stick models. Library A: T149, N151
(magenta); library C: V164 (cyan); library D: F254 and L259
(orange), and library E: V164 and C362 (cyan). NAD(H) analogue
adenosine-5-diphosphoribose (gray) and S-1,2-propanediol (gray) are
presented as stick models, active-site iron(II) is shown as a sphere
(yellow), and the polypeptide chain as surface and cartoon. The image
was created in PyMOL (ver. 1.5 www.pymol.org) from the atomic
coordinates in 1RRM.15
a
a, Hydrolysis of S- or R-(2,3-epoxypropyl)benzene by epoxide
hydrolase; b, oxidation of S- or R-4 into the corresponding α-hydroxy
aldehydes (S- or R-5) by FucO variant and NAD+.
regioselectivity displayed by the FucO enzyme in the oxidation
of vicinal diols is attractive from the prospect of possible
biocatalytic production of chiral α-hydroxy aldehydes by the
use of this enzyme. FucO, however, because of its biological
role, is not active with larger phenyl-substituted alcohols,12
which restricts its biocatalytic potential. Our intent for FucO
was to catalyze regio- and enantioselective oxidation of aryl-
substituted vicinal diols such as phenylpropanediols, R- and S-4
(Chart 1). These diols are products of epoxide hydrolase
catalyzed hydrolysis of (2,3-epoxypropyl)benzene,13,14 and one
aim was to produce, by protein engineering, FucO variants
capable of catalyzing the oxidation of diol 4 and thereby enable
a two-step transformation of (2,3-epoxypropyl)benzene into
the α-hydroxy aldehyde products 5 (Scheme 1B).
belongs to the class III alcohol dehydrogenases, which are
considered to be Fe(II) dependent.16 The entrance for the
substrate into the active site is narrow and limits the size of
allowed substrates (Figure 1B). Hence, the protein structure
requires remodeling to oxidize bulky substrates such as diol 4.
FucO was therefore subjected to directed evolution to produce
variant enzymes. Mutagenesis was guided by inspection of
available crystal structures,15 and the targeted amino acid
residues were considered to restrict entry to the active site
(Figure 1C). Mutations were introduced according to the
iterative saturation mutagenesis (ISM) approach as described
by Reetz and co-workers.17,18 By coupling the ISM-driven
mutagenesis with direct assays of enzyme activities in the
generated mutants, we have isolated variant FucO enzymes
capable of catalyzing redox transformations on aryl-substituted
substrates.
Chart 1. Compounds Applied as Substrates for FucO
Variants: 1-Propanol (1), S-1,2-Propanediol (S-2), R-1,2-
Propanediol (R-2), Phenylacetaldehyde (3), S-3-Phenyl-1,2-
propanediol (S-4), and R-3-Phenyl-1,2-propanediol (R-4)
EXPERIMENTAL SECTION
■
Chemical and Reagents. Commercially available chem-
icals were purchased from Sigma-Aldrich at the highest purity
available. Chiral purity of S- or R-1,2-propanediol were ≥96%.
S- or R-(2,3-epoxypropyl)benzene at >98% chiral purity were
purchased from TCI Europe N.V. Oligonucleotides used in
mutagenesis and library constructions were custom-ordered
from Thermo Scientific. Enzymes and reagents for molecular
biology work were purchased from Fermentas. Chromato-
graphic resins were purchased from GE Healthcare. The S- and
R-enantiomers of 4 were synthesized by Solanum tuberosum
epoxide hydrolase (StEH1) catalyzed ring-opening of the
corresponding epoxide enantiomers; 50 or 100 mM of S- or R-
(2,3-epoxypropyl)benzene, 1.5−3 μM of StEH1, and 3.3 mM
acetonitrile was added to 100 or 10 mM sodium-phosphate, pH
7.0, respectively.13 Reactions were carried out overnight at
3017
dx.doi.org/10.1021/cs400824h | ACS Catal. 2013, 3, 3016−3025

ACS Catalysis
Research Article
room temperature under mild shaking (∼20 rpm) in 15 mL
reaction volumes contained in 50 mL Erlenmeyer glass flasks.
The StEH1-catalyzed epoxide hydrolysis retains the config-
urations of the corresponding epoxides.14
Library Design. On the basis of previous functional
analyses12 and investigation of crystal structures of FucO,15
six amino acid residues were targeted for mutagenesis with the
anticipated effect of altering substrate selectivity. These residues
line the entrance of the substrate binding site. The six residues
were divided into four different focused libraries: Library A,
Thr149 and Asn151; Library C, Val164; Library D, Phe254 and
Leu259; and Library E, Val164 and Cys362. For construction of
the second-generation library AD FucO variant A5 (N151G)
was used as template gene and for library DA and DE variant
D93 (L259V) were used. For third-generation library DAE the
DE1472 (N151G, L259V) was used as template. A
combination of saturation mutagenesis and limited codon sets
were used (Table 1).19
Library Constructions. In the construction of first-
generation libraries, plasmid pGTacfucO-5H encoding wild-
type FucO12 was used as template. Sequences of mutagenic
oligonucleotides and the codon replacements are shown in
Supporting Information, Table SI1. Mutated PCR fragments
were subcloned between the XhoI and SpeI restriction sites of
pGTacStEH1-5H.20 The resulting expression plasmids were
transformed into E. coli XL1-Blue by electroporation. A pool of
500−1000 clones from each library were sequenced to confirm
expected mutations and to assess library quality based on
introduced codon/mutation distribution.21
Small-Scale Protein Expression and Lysate Prepara-
tion for Activity Screening. Expression of library enzymes
was carried out in 96-well plates. The procedure was essentially
according to a previously described protocol used for protein
expression in library screening of StEH1 mutants,13 with the
exception that after bacteria inoculation the cultures were
grown for 3.5 h and at the time of induction of transcription,
100 μM FeCl2 was added together with the IPTG inducer. The
microtiter plates were inoculated with library clones together
with four clones of each control; wild-type FucO and StEH1
(as internal negative controls) and four wells were left without
inoculate. For the second- and third-generation libraries, the
parent variants (A5, D93, and DE1472) were also included as
controls. The initial protocol as described by Gurell and
Widersten13 was followed up to the point of bacteria harvest.
After removal of the supernatant bacterial pellets were prepared
for screening. Twenty-five microliters of B-PER (Pierce)
fortified with 0.2 mg/mL DNase I, 0.2 mg/mL lysozyme and
protease inhibitor (Complete Protease Inhibitor Cocktail
Tablets EDTA-free from Roche) were added to each well.
The pellets were resuspended by mild vortexing, and the
mixtures were incubated at mild shaking at room temperature
for 1 h. Subsequently, 100 μL of 10 mM sodium-phosphate
buffer, pH 7.0, was added to each well, and the plates were
centrifuged at 3,500 g for 60 min, at 4 °C.
Library Screening. After cell lysis and centrifugation 20 μL
of lysate from each well were transferred to a polystyrene 96-
well microtiter plate with flat bottom (Nunc). Buffer, cofactor
and substrate were mixed together and preincubated at 30 °C.
The reaction mixtures were thereafter added to the lysate-
containing microtiter plate to a final volume of 150 μL and
enzyme activity was measured for 3 to 5 min. Reaction
velocities were monitored spectrophotometrically at 30 °C
through the change in NADH concentration at 340 nm (Δε =
3018
dx.doi.org/10.1021/cs400824h | ACS Catal. 2013, 3, 3016−3025

ACS Catalysis
Research Article
Figure 2. Principal component analysis on normalized library screening activities. The loading plots (left panel) describe the relationships between
the screening substrates. Substrates are indicated by black triangles. The score plots (the three panels to the right) displays the observations
projected from K-space into two dimensions. The analysis was performed on the total data set per generation of library (1st, 2nd, and 3rd). To aid in
the interpretation of the score plots only one library is shown per plot. All library clones and a mean value of the controls are shown as a triangle.
Symbols key is shown in the lower right corner. Library clones are in gray triangles and hits are in black triangles. Clones containing template genes
were added as controls to 2nd and 3rd selection rounds, and their scores are shown by the respective symbols.
6.22 mM−1 cm−1) in real time, using a Molecular Device
SpectraMAX 190 plate reader. Oxidation of alcohol substrates
were performed in 0.1 M glycine-NaOH, pH 10.0, and in the
presence of 0.2 mM NAD+. Reduction of aldehyde 3 was
performed in 0.1 M sodium-phosphate, pH 7.0, and in the
presence of 0.2 mM NADH. First-generation library clones
were screened for enzymatic activity with, 6 mM 1-propanol
(1), 2.5 mM S-propanediol (S-2), 40 mM R-propanediol (R-2),
and 34 mM phenylacetaldehyde (3). In screening of second-
and third-generation clones the propanediols were substituted
by S- and R-3-phenylpropane-1,2-diol (S- and R-4), concen-
trations used were 5 mM 1, 20 mM 3 and 10 mM of S- or R-4.
The genes encoding selected hits were sequenced in full to
identify introduced mutations and to exclude other alterations.
Larger Scale Expression and Purification of FucO
Variants. C-terminally 5-His-tagged FucO mutants were
expressed in 0.5 L cultures according to the protocol previously
described.12 After centrifugation the cells were frozen at −80
°C until purification. Frozen bacteria pellets were thawed and
resuspended in binding buffer (20 mM imidazole, 0.5 M NaCl,
20 mM sodium-phosphate, pH 7.5 and 0.02% (w/v) NaN3), up
to a final volume of approximately 25 mL. Cells were lysed
using a cell disruptor (Constant systems Ltd.), and the lysate
was cleared by centrifugation at 38,000 g for 1 h at 4 °C.
Cleared lysate was mixed with 1 mL of Chelating Sepharose
Fast Flow resin charged with Ni2+ ions and incubated for 30
min at 4 °C. The slurry was subsequently centrifuged for 5 min
at 800 g, and the supernatant was carefully discarded.
Unspecifically bound proteins were eluted by resuspension of
resin in 20 mL wash buffer (60 mM imidazole, 0.5 M NaCl, 20
mM sodium-phosphate, pH 7.5 and 0.02% (w/v) NaN3),
incubation for 10 min, centrifugation for 5 min at 800 g and
discarding the supernatant. This wash step was repeated twice.
Bound enzyme was eluted by addition of 2.5 mL elution buffer
(300 mM imidazole, 0.5 M NaCl, 20 mM sodium-phosphate,
pH 7.5, and 0.02% (w/v) NaN3), incubation for 5 min followed
by 7 min centrifugation at 800 g. The elution was repeated
twice. The two elution fractions were desalted using a PD-10
column equilibrated with 0.1 M sodium-phosphate, pH 7.4,
fortified with 0.02% (w/v) NaN3. All purification steps were
carried out at 4 °C and all incubations were performed on a
rocking-table. Purity of protein variants were determined by
SDS-PAGE stained with Coomassie Brilliant Blue R-250.
Concentrations of purified proteins were determined from
their absorbances at 280 nm using an extinction coefficient of
41 000 M−1 cm−1.12 All kinetic measurements were conducted
within two weeks after purification
Kinetic Characterization of FucO Variant Enzymes.
Catalyzed initial rates were measured, and the steady-state
parameters kcat, KM, and kcat/KM were extracted essentially as
described in ref 12, with the exception that each substrate
concentration was measured in duplicates. Measurements with
3 were performed in a Molecular Device SpectraMAX 190 plate
reader. Activities with S- and R-4 were performed in a
Shimadzu UV-1700 spectrophotometer.
3019
dx.doi.org/10.1021/cs400824h | ACS Catal. 2013, 3, 3016−3025

ACS Catalysis
Research Article
Figure 3. (A) Bar chart of catalytic efficiencies of FucO variants obtained from the directed evolution. Bars to the left show kcat/KM (s−1 M−1) with
the respective enantiomers of diol 4 and bars to the right with aldehyde 3. (B and C) Evolutionary pathways for improvements in catalytic activity
with 3 (B) or the different enantiomers of 4.
Multivariat Data Analysis of Screening Data. Principal
component analysis was performed on initial reaction velocities
recorded during library screening using SIMCA-P+ 12.0.1
(Umetrics). Rows in the data matrix corresponded to the
different screening substrates and columns corresponded to the
reaction velocities displayed by the different clones. Clones
displaying negative values were excluded from the data table.
To not skew the analysis with data from control clones, mean
values of these reaction velocities were used. Before analysis,
the data was normalized to unit length and mean-centered.
Three PCAs were performed, one for each library generation.
For clarity, the data is graphically presented separately for each
library. Figure 2 shows PC1 versus PC2 and display library
distributions with controls and improved variants indicated.
Structural Modeling and Docking. The three-dimen-
sional structures of FucO mutants DA1472 (N151G, L259V)
and DE461 (V164C, L259V, C362G) were modeled based on
the crystal structure of FucO in complex with NAD+ (PDB
code 2BL415). A set of ten models with NAD+ and Fe(II) in
the active site was created with MODELLER,22 and the model
with the lowest value of the MODELLER objective function
was used for docking studies. 3 was directly taken from the
crystal structure of E. coli amine oxidase (PDB code 1D6U23)
for ligand docking, while (2R, 3S)-3-amino-3-phenylpropane-
1,2-diol was derived from the crystal structure of Scytalidium
lignicola Scytalidopepsin B (PDB code 2IFR24) and edited to S-
4 with Maestro Molecular Modeling Interface (Version 9.3.,
RESULTS
■
Enzyme Libraries Designed to Facilitate the Isolation
of Functional FucO Variants. To streamline screening of
variants, the gene libraries were in all cases constructed from
limited codon subsets25 (Table 1). To increase the chances of
generating mutants capable of binding bulky substrates,
mutagenesis in the first-generation libraries introduced mainly
replacements of smaller amino acid residues in the lining of the
entrance to the active site. In the subsequent second and third-
generation libraries, codon sets allowing for insertion of also
other residues were included to increase the repertoire of
functional groups. The codon and mutation distributions of the
different libraries were assayed by sequencing randomly picked
sets of unselected library clones. Although minor skewing of the
theoretical codon distribution was observed at some sites, the
library qualities were considered to be adequate.
Isolation of Dehydrogenase Variants Exhibiting
Activity with Aryl-Substituted Substrates. Although the
aim was to isolate FucO variants that had achieved activity with
the 3-phenyl-1,2-propanediols (4; Chart 1), phenylacetalde-
hyde 3 was used in the initial rounds as surrogate substrate. The
motives for this choice were as follows. (i) The reduction
reaction is thermodynamically favorable and proceeds with at
least 1 order of magnitude faster rates as compared to the
oxidation reaction;12 hence, an increased sensitivity during
screening, facilitating detection of also low-activity variants, was
anticipated. (ii) 3 was judged to be a reasonable structure
analogue of the aldehyde derivatives of 4. It should be noted,
however, that 3 does not contain a sec-alcohol at the α-position
and lacks one methylene group. Therefore, any (critical)
influence that this structural difference might cause was
expected to impact the selection outcome.
Schrodinger, Inc.). All protein structures and the substrates
̈
were prepared for docking in Discovery Studio (Version 3.5,
Accelrys Inc.), which was also used to detect receptor cavities
and the active site cavities. In the active sites of the proteins,
O7N was constrained as acceptor of hydrogen bonds, and
H23/H25 of NAD+ were defined as hydrogen bond donor and
acceptor, respectively. In S-4, O1 and O5 were constrained as
hydrogen bond acceptors, and H22 and H23 as donors, while
O9 in 3 was a possible acceptor. GOLD via Discovery Studio
(Version 3.5, Accelrys Inc.) was used to perform the docking of
3 to the DA1472 mutant, S-4 to the DE461 mutant, and both
substrates to the wild-type FucO. The docking poses were
analyzed and scored with the Score Ligands function in
Discovery Studio (Version 3.5, Accelrys Inc.), and the pose
with the highest PLP2 score was chosen as the best pose.
In the first screening round, S- and R-1,2-propanediol (2)
were included as screening substrates to probe the
enantioselectivity of new dehydrogenase variants. During
screening of second and third-generation libraries, the
respective enantiomers of 4 replaced the propanediol
substrates. 1-Propanol (1) was included in all screening rounds
as a control substrate to assay the presence of (any) alcohol
dehydrogenase activity.
Screening enzyme libraries for activity with several substrates
rapidly generates a very large data set. To get an overview of the
screening outcome, principal component analysis (PCA) was
performed on normalized enzyme activities. The PCAs for the
PyMOL (Version 1.5, Schrodinger, LLC) was used to prepare
̈
pictures of the complexes.
3020
dx.doi.org/10.1021/cs400824h | ACS Catal. 2013, 3, 3016−3025

ACS Catalysis
Research Article
different libraries are presented in Figure 2. The analysis
provided a comprehensive view of the clone distributions in the
respective libraries and was also consulted in scoring of possible
hits. The 10−20 best variants from each library were
rescreened, and their corresponding genes were sequenced to
confirm their initial scoring as hits and to deduce their protein
sequences.
FucO variant hits were expressed and purified by Ni(II)-
IMAC and their activity profiles were determined using
substrates 3 and S- and R-4. The selected clones are marked
in the PCA (Figure 2). Enzymatic efficiencies for clones active
with these substrates are graphically presented in Figure 3 and
listed in Table 2, together with determined values of kinetic
parameters. Activities for purified and characterized clones,
which showed no elevated activity with the desired substrates,
are not presented.
First-Generation Clones. The PCA of screening activities
displayed that clones in the first generation were grouped into
three distinct clusters (Figure 2). One cluster showed activities
similar to the wild-type enzyme (circle in Figure 2). Another
group displayed increased activity with 3, but had lost activity
with both enantiomers of 2. Clones populating the third cluster
displayed reasonable activity with 3 and had retained some
activity with S- or R-2. Approximately 70% of the analyzed
clones did not display activity above background rates with any
of the screening substrates and were considered to represent
inactive clones.
Library A. Eight of the FucO variants were selected for
larger-scale expression, purification, and characterization. Of the
characterized mutants, variant A5 (N151G) displayed the most
interesting properties with a marked increase in the catalytic
activity with aldehyde 3; 90- and 150-fold increases in kcat and
kcat/KM, respectively. Although A5 displayed considerable
activity with 3, this variant did not catalyze the oxidation of
the S- or R-4 diols.
Table 2. Steady-State Kinetic Parameters of FucO Variants
enzyme
variant
k
cat/KM
mutation
kcat (s−1
Phenylacetaldehyde (3)
)
KM (mM)
(s−1 M−1
)
wild type
A5
0.023 0.01
2.1 0.1
23 10
1.1 0.2
N151G
L259V
12 0.7
48 12
4.4 0.7
n.s.
170
5
D93
0.62 0.1
13 0.8
DA1472
DE199
N151G, L259V
21
1
4800 500
4.9 0.3
a
V164I, L259V,
C362A
n.s.
DE452
DE461
DE1028
V153I, V164I,
4.0 0.3
2.5 0.4
1.4 0.2
0.90 0.1
6.2
1
640 60
L259V
V164C, L259V,
C362G
100 20
25
48
24
1
4
1
V164M, L259V,
C362Y
29
8
6
DE1130
DAE739
V164I, L259V
38
N151G, V164I,
L259V, C362N
21
2
5.7
1
3700 400
S-3-Phenylpropane-1,2-diol (S-4)
b
wild type
A5
n.d.
N151G
L259V
n.d.
D93
0.15 0.004
0.20 0.01
6.6 0.4
16
20
1
DA1472
DE199
N151G, L259V
n.d.
V164I, L259V,
C362A
2
12 0.6
DE452
DE461
DE1028
V153I, V164I,
n.d.
L259V
V164C, L259V,
C362G
0.32 0.01
0.25 0.005
0.29 0.01
7.3 0.4
6.3 0.3
9.3 0.5
43
40
2
1
1
V164M, L259V,
C362Y
DE1130
DAE739
V164I, L259V
31
N151G, V164I,
L259V, C362N
n.d.
R-3-Phenylpropane-1,2-diol (R-4)
wild type
A5
n.d.
n.d.
N151G
L259V
D93
n.s.
n.s.
n.s.
n.s.
0.50 0.1
n.d.
Library C. One clone was scored as a positive hit with
aldehyde 3. However, the purified enzyme did not show any
noteworthy increase in activity with any of the aryl-substituted
substrates.
DA1472
DE199
N151G, L259V
V164I, L259V,
C362A
0.19 0.03
DE452
DE461
DE1028
V153I, V164I,
n.d.
L259V
Library D. Two clones were chosen for characterization;
one displaying activity with 3 and one clone displaying a
remarkably higher activity than the wild-type with R-2. The
clone with 3-activity turned out to have no increased activity
with the aryl-substituted substrates. Nevertheless, variant D93
(L259V), selected because of its increased R-2 activity, turned
out to display reasonable activity also with the targeted diol 4.
D93 showed activity with S-4 with a kcat of 0.15 s−1 and a KM of
6.6 mM. The kcat/KM value for the two different enantiomers
were determined to be 20 s−1 M−1 for S-4 and 0.5 s−1 M−1 for
R-4, resulting in an E-value (S/R) of 40. This mutant also
displayed a 10-fold increase in kcat/KM with 3 as compared to
the wild-type, mainly due to a 30-fold increase in kcat.
Second Generation Clones. Variant A5 (plus sign in
Figure 2) was used as template in the construction of library
AD and D93 (cross in Figure 2) for libraries DA and DE. PCA
of clone activities displayed two clusters (Figure 2). As
compared to the wild-type, one cluster had elevated activity
with S- and/or R-4, similar to D93. The other cluster had
elevated activity with aldehyde 3, similar to the A5. All three
libraries displayed similar patterns. Approximately 90% of the
screened clones did not show activities with any of the
screening substrates.
V164C, L259V,
C362G
0.050 0.005
0.010 0.001
0.014 0.003
26
4
1.9 0.1
1.0 0.1
V164M, L259V,
C362Y
9.8
2
DE1130
DAE739
V164I, L259V
35 11
0.40 0.05
n.d.
N151G, V164I,
L259V, C362N
a
n.s., Enzyme saturation was not reached within the practically useful
substrate concentrations. Only kcat/KM could therefore be determined.
b
n.d., No enzyme activity detected.
Library AD. From the initial screen 10 clones were selected
as potential hits. However, after rescreening none of these
variants were selected for further characterization, because of
their apparent low absolute activities.
Library DA. After rescreening and sequencing three variant
enzymes were selected for purification and further character-
ization. Although the template D93 displays activity with both 4
diols, two of the three characterized enzymes had lost most of
their diol oxidation activities but instead showed improved
activity with aldehyde 3, which lacks the α-hydroxyl group. The
most efficient catalyst was DA1472 (N151G, L259V) with kcat
and KM values for 3 of 21 s−1 and 4.4 mM, respectively. This is
so far the most active variant with this substrate, and it
3021
dx.doi.org/10.1021/cs400824h | ACS Catal. 2013, 3, 3016−3025

ACS Catalysis
Research Article
Figure 4. (A) 3 (pink) binds in a bent conformation to the DA1472 (N151G, L259V) mutant (cyan), which enables it to hydrogen bond to NAD+
and π−π stack with F254. Residues within 4 Å from 3 are shown as pink sticks. The iron-binding residues are shown as gray sticks. (B) In the wild-
type protein, 3 binds in a more extended conformation (yellow) than in the mutant (pink). This makes it impossible for 3 to π−π stack with F254 in
the wild-type protein. Residues within 4 Å from 3 are shown as sticks in pink (mutant) and yellow (wild type). The mutations are marked in cyan.
(C) S-4 (cyan) binds in a bent conformation to the DE461 (V164C, L259V, C362G) mutant (green) and hydrogen bonds to NAD+ and N151,
while π−π stacking with F254. Residues approximately 4 Å from S-4 are shown as cyan sticks. (D) In the wild-type protein, S-4 is docked in a more
extended conformation (orange) than in the mutant (cyan). This makes it impossible for S-4 to π−π stack with F254 in the wild-type protein.
Additionally, only one of the hydroxyl groups of S-4 is able to hydrogen bond to NAD+ in the wild-type complex, while S-4 in the DE461 mutant
hydrogen bonds via both hydroxyl groups. Residues approximately 4 Å from S-4 are shown as sticks in cyan (mutant) and orange (wild type). The
mutations are marked in green. Hydrogen bonds and π−π interactions between the docked substrates and the enzymes are shown in black dashed
lines for the mutant proteins and in red dashed lines for the wild-type.
displayed a substantial 4400-fold increase in kcat/KM as
compared to the wild-type enzyme and a 30-fold increase as
compared to the A5 mutant.
and DE1028 displayed 4- and 2-fold higher kcat/KM,
respectively, as compared to the D93 variant. DE1130 had
retained the relatively low activity with R-4, resulting in an
improved enantioselectivity of S-4 (E = 80). In addition, DE452
(V153I, V164I, L259V) had no detectable activity with diols 4
but had gained activity with aldehyde 3 with a kcat/KM of 640
s−1 M−1. The V153I mutation had not deliberately been
included in the library but was an artifact during library
construction.
Library DE. Eight enzyme variants were chosen for further
characterization. Out of these, five displayed activity with the
desired substrates. In this library, the most efficient clones with
S- and R-4 (Table 2) were found. DE461 (V164C, L259V,
C362G), DE1028 (V164M, L259V, C362Y), and DE1130
(V164I, L259V) all displayed kcat values for the S-enantiomer of
approximately 0.3 s−1 and kcat/KM in the range 30−40 s−1 M−1,
which corresponds to a doubling as compared to the D93
template. With respect to the R-enantiomer, variants DE461
Third Generation Clones, Library DAE. The DA1472
variant (N151G, L259V, star in Figure 2) was used to template
this library. PCA of screening activities revealed hits displaying
3022
dx.doi.org/10.1021/cs400824h | ACS Catal. 2013, 3, 3016−3025

ACS Catalysis
Research Article
activity with 3, S-4, or R-4 (Figure 2). However, after
rescreening, only one enzyme variant DAE739 (N151G,
V164I, L259V, C362N) was selected for further character-
ization. It displayed a catalytic profile similar to that of the
parental enzyme, with a kcat of 21 s−1 and a KM of 5.7 mM for 3
and without detectable activity with S- and R-4.
aldehydes, into variant enzymes capable of catalyzing the
oxidation and reduction of also bulkier substrates. From only
two rounds of (semi)saturation mutagenesis of noncatalytic
active-site residues, we succeeded to generate enzyme variants
that catalyze the oxidation of 3-phenyl-1,2-propanediol (4), a
substrate with which the wild-type FucO shows no detectable
activity. All these variants displayed retained enantiopreferences
favoring the S-enantiomer although R-4 was also accepted as
substrate to different degrees. The most stereoselective variant
DE1130, displays an E-value of 80 preferring the S-enantiomer
of 4 and can thus be classified as adequately selective for
synthetic purposes.32 Enzyme variants with increased activity
with phenylacetaldehyde (3) were also obtained, with the
second-generation mutant DA1472 exhibiting a 4400-fold
increased catalytic efficiency as compared to the wild-type.
The turnover number displayed by this mutant approaches the
highest kcat values so far measured for FucO with any
aldehyde.12
The choice of FucO as starting scaffold for the engineering
was based on the following criteria. (i) The enzyme catalyzes
the oxidation of the primary alcohol carbon of vicinal diols with
high (absolute) regioselectivity. This is a rare feature in alcohol
dehydrogenases and is, besides FucO, found only in certain
carbohydrate dehydrogenases such as glycerol dehydrogenase.
(ii) The enzyme is readily produced in an expression system
allowing for directed evolution. (iii) The enzyme structure,
both tertiary and oligomeric, is comparably simple which
further facilitates enzyme purification, characterization, and
manipulation. In comparison to the glycerol dehydrogenases,
which are octameric proteins,33,34 FucO is a homodimer and a
native E. coli enzyme.
The screening method used (detection of NADH
absorbance) is simple and sensitive enough to make it possible
to search for activities with different substrates in parallel. Since
library screening often represents a bottleneck in directed
evolution, this simplicity and sensitivity of the screen is of great
advantage. The use of principal component analysis also helped
to achieve an overview and to visualize the activity distributions
within the different mutant-libraries. This allowed us to explore
multiple simultaneous evolutionary pathways in parallel (Figure
3B, C).
FucO-catalyzed rates of aldehyde reductions are at least 1
order of magnitude faster than alcohol oxidations. Hence, to
increase the sensitivity of the screen and still select for variants
with low but improved activity with aryl-substituted substrates,
3 was used as a surrogate substrate in the search for enzymes
active with diols 4. It has been shown in a related study of in
vitro evolution of steroid hormone receptors, that coevolution
of highly specialized proteins can be successfully accomplished
by iteratively selecting for activity toward structural inter-
mediates of the ultimate target ligand.35 Our results show that 3
was not an optimal intermediate, since variants active with one
substrate were often not active with the other. It can be
speculated whether these differences depend on the absence of
the secondary alcohol group in 3 or the fact that the distance
between the redox-active carbon and the phenyl substituent is
shorter in 3 as compared to 4. Nevertheless, the results
demonstrate that the FucO enzyme has the inherent ability to
be modulated to catalyze reactions with different types of
substrates.
Structural Modeling and Docking Studies of Mutants.
The docking results show that the DA1472 (N151G, L259V)
and DE461 (V164C, L259V, C362G) mutants can accom-
modate the aryl-containing compounds, which makes it
possible for them to use these as substrates. In the DA1472
mutant, the aldehyde oxygen of 3 hydrogen bonds with the
amide of NAD+ (Figure 4A). Additionally, the phenyl ring of 3
interacts through π−π stacking with the phenyl ring of F254.
Docking of the same substrate to the wild-type protein
indicates that, because of a different conformation of 3 in the
wild-type active site, the π−π stacking interaction between the
phenyl ring and F254 is lost (Figure 4B). In the wild type, the
substrate adopts an extended conformation with the phenyl
ring protruding from the active site cavity, while the same
substrate binds in a bent conformation to the DA1472 mutant.
When the conformation of 3 in the mutant is compared to the
wild-type structure, it becomes clear that N151 in the wild-type
protein inflicts structural clashes, which prevents binding of this
compound in the same conformation as in mutant DA1472
(Figure 4B). Similarly to the DA1472 mutant complex, S-4 is
docked to the DE461 mutant in a bent conformation, which
enables the phenyl ring of the substrate to form π−π stacking
interactions with F254 (Figure 4C). Once again, this
interaction is lost in the wild type-S-4 complex, where the
extended conformation of S-4 causes the phenyl ring to point
away from F254 (Figure 4D). In the DE461 mutant, the L259V
mutation makes the active site cavity bigger, which enables S-4
to bind in a bent conformation. In the wild-type enzyme, the
extended conformation of S-4 allows hydrogen bonding to
N151 through both hydroxyl groups, while only the α-hydroxyl
group is able to hydrogen bond to NAD+ (Figure 4D). When S-
4 is docked to the DE461 mutant the α-hydroxyl group
hydrogen bonds to N151, and both hydroxyl groups make a
hydrogen bond to NAD+. This is due to the C362G mutation,
which creates more space for the ligand and enables S-4 to bind
in a different conformation to the DE461 mutant as compared
to the wild-type enzyme (Figure 4D).
DISCUSSION
■
In today’s world of increasing environmental pressure, there is a
growing need for “greener” solutions within chemical
manufacturing. During the past 10 years, the use of enzymes
in industrial applications such as synthesis of fine chemicals,26
pharmaceuticals,27 bioenergy,28 agriculture,29 textile industry,30
and degradation of pollutants31 has increased.10 Biocatalysts are
being established as environmentally friendly alternatives to
contemporary technologies since they are efficient also under
mild reaction conditions: moderate temperatures, aqueous
solutions, and usually without requirement for noble and other
rare metals. Furthermore, enzymes are produced from
renewable sources, are fully biodegradable and nontoxic. To
utilize the catalytic efficiency of enzymes in useful applications,
however, they must often undergo re-engineering to become
customized for the reaction at hand.
In this study, we successfully modified a propanediol
oxidoreductase, specialized by nature to catalyze the
oxidation/reduction of low-molecular 1,2-diols/α-hydroxy
Unveiling Silent Enzyme Activities. The FucO enzyme
can be viewed as evolutionary very well adapted for catalyzing
the reduction of S-lactaldehyde into S-1,2-propanediol. The
3023
dx.doi.org/10.1021/cs400824h | ACS Catal. 2013, 3, 3016−3025

ACS Catalysis
Research Article
reaction plays an important role in the anaerobic catabolism of
L-fucose since the reaction replenishes the NAD+ pool and the
product diol is readily excreted as a terminal fermentation
product.36 No other metabolic role has been assigned to this
enzyme, and its structure/activity profile, which describes high
selectivity for the S-enantiomer of small aliphatic alcohol/
aldehyde substrates, further supports its assigned role in
intermediary metabolism.12,15 The evolvability of such
specialized enzymes has been questioned and “generalist”
enzymes with broad substrate scope are considered to be more
malleable (cf. ref 37). The outcome from the present work is
partly in agreement with this view since relatively few amino
acid residue replacements were apparently allowed at the
chosen sites of mutagenesis to generate functional variants
(Figure 3B, C). On the other hand, our results in the directed
evolution of FucO certainly also support the notion that the
evolvability of enzyme function relies on underlying, undis-
closed promiscuous activities that may be observed upon
limited structural changes, that is, by few mutation events.38
The structural difference caused by the single mutation in
variant D93 (L259V) corresponds to only one methylene
bridge less. This minor modification still unveils the (hidden)
potential of this enzyme to catalyze the conversion of the bulky
S-4 diol into the S-α-hydroxy aldehyde. To give a measure of
the magnitude of improvement in catalytic activity with this
substrate is problematic since the wild-type enzyme does not
accept this substrate at all (from an estimated level of detection
of any catalytic activity one may assume a more than 1000-fold
improvement in catalytic activity from the L259V mutation).
Introducing a smaller steric constraint (more volume) at an
appropriate position was in this case enough to expose this new
catalytic activity of FucO.
Creating additional volume in the active site is not the only
solution to achieve catalytic activity with S-4. This is
demonstrated by mutagenesis at the E-site (V164 and C362
in Figure 1C). In this case, several of the mutations introduced
bulkier amino acid residues; variant DE1028 (V164M, L259V,
C362Y), DE461 (V164C, L259V, C362G), and DE1130
(V164I, L259V) all display twice as high activity as the
template L259V with this substrate. It is also noteworthy that
throughout the evolution process, F254, was retained in all
variants displaying improved catalytic activity with the aryl-
substituted substrates implying a role for its aromatic benzyl
side-chain, as also suggested from the modeling.
The mutation of asparagine to glycine at position 151 is an
even clearer example of how activity with bulkier substrates can
be disclosed by simply removing sterical hindrance in the
entrance to the active site. This mutation recurs in all variants
(except DE452) which display activity with 3. The cause, as
deduced from the docking, is that F254 stabilizes the
interactions between the substrates and the mutant proteins,
thereby facilitates catalysis. Hence, it is possible that the
hydrogen bonding capacity of 3 and S-4 are insufficient to form
adequately stable and productive enzyme−substrate complexes.
The removal of the sterical hindrance resulting from the
N151G mutation enables a stable binding of 3 which may
contribute to the observed improvement in the catalytic
efficiency with this substrate. The further improved activity
resulting from the additional L259V mutation might be because
the substrate aryl ring can come even closer to F254 and further
stabilize productive substrate−protein interactions. Since the
activity of DE739 (N151G, V164I, L259V, C362N) is similar to
that of the DAE1472 variant, the mutations at positions 164
and 362 do not seem to affect the activity to any appreciable
level.
The introduction of the N151G mutation renders the FucO
variants nearly inactive with diol 4. This may be because
hydrogen bonding interactions between the α-hydroxyl of S-4
and the N151 side-chain amide are lost (Figure 4C). The
observation that clones active with 4 are inactive with 3 and
vice versa is displayed also in the screening result (Figure 2),
where the library distribution in generation two and three are
divided into two quite distinct clusters, one displaying activity
with 4 and one with 3.
In variant DE452 an unexpected mutation, V153I, was
introduced. Mutagenesis at this site was not designed for but
became incorporated through imperfect PCR. Position 153 is
situated further into the active site tunnel, in the second sphere
from the residues included in the constructed libraries. DE452
and DE1130 only differ by this point-mutation. The resulting
structural change is minor, with one γ-methyl group introduced
in DE452. The activity profile, however, is drastically shifted:
DE1130 is active with both enantiomers of 4 but displays
comparably poor activity with aldehyde 3. DE452, in contrast,
is inactive with diols 4 but is highly active with 3. Apparently,
the additional methyl group in I153 prevents productive
binding of the phenylpropanediols but favors binding of
aldehyde 3 by 8.3 kJ/mol (as calculated from the respective
kcat/KM values of DE452 and DE1130). These results also
demonstrate that there is room for further tailoring of the
substrate selectivity and that positions other than those visited
in this study should be included.
CONCLUSION
■
The employment of enzymes in chemical synthesis requires the
availability of biocatalysts with the adequate functional and
physicochemical properties. We have in this work demonstrated
the feasibility of evolving the substrate scope of a specialist
enzyme, E. coli propanediol oxidoreductase FucO, to also
accept bulky aryl-substituted substrates. The evolution process
has revealed that few, often conservative but critically
positioned, substitutions are enough to achieve large improve-
ments in catalytic activities (Figure 3). The modeling and
docking results present a possibility that the larger active site
cavities in the mutant enzymes compared to the cavity in the
wild type enzyme causes a gain in the important stabilizing
interactions, which cannot be found in the complexes between
the wild-type protein and studied substrates. On the contrary,
the smaller active site cavity in the wild-type enzyme makes it
impossible for it to use the studied bulky aryl-substituted
molecules as substrates. The inherent enantioselectivity of the
wild-type enzyme was retained in some evolved variants also
when catalyzing the oxidation of the 3-phenyl-substituted
derivatives of the native substrate 1,2-propanediol. We now
have access to FucO variants that allows for the direct synthesis
of the chiral α-hydroxy aldehydes R- and S-5 as shown in
Scheme 1B from a racemic mixture of an elemental starting
material such as (2,3-epoxypropyl)benzene.
ASSOCIATED CONTENT
■
S
* Supporting Information
Sequences of oligonucleotide primers are presented in Table
SI1. This material is available free of charge via the Internet at
http://pubs.acs.org.
3024
dx.doi.org/10.1021/cs400824h | ACS Catal. 2013, 3, 3016−3025

ACS Catalysis
Research Article
Devine, P. N.; Huisman, G. W.; Hughes, G. J. Science 2010, 329, 305−
308.
AUTHOR INFORMATION
Corresponding Author
*E-mail: mikael.widersten@kemi.uu.se. Phone: +46 18 471
4992.
■
(28) Liu, X.; Bastian, S.; Snow, C. D.; Brustad, E. M.; Saleski, T. E.;
Xu, J.; Meinhold, P.; Arnold, F. H. J. Biotechnol. 2012, 164, 188−195.
(29) Oliveira, G. R.; Silva, M. C.; Lucena, W. A.; Nakasu, E. Y.;
Firmino, A. A.; Beneventi, M. A.; Souza, D. S.; Gomes, J. E., Jr; de
Souza, J. D., Jr; Rigden, D. J.; Ramos, H. B.; Soccol, C. R.; Grossi-de-
Sa, M. F. BMC Biotechnol. 2011, 11, 85.
Notes
The authors declare no competing financial interest.
(30) Cherry, J. R.; Lamsa, M. H.; Schneider, P.; Vind, J.; Svendsen,
A.; Jones, A.; Pedersen, A. H. Nat. Biotechnol. 1999, 17, 379−384.
(31) van Leeuwen, J. G.; Wijma, H. J.; Floor, R. J.; van der Laan, J.
M.; Janssen, D. B. ChemBioChem 2012, 13, 137−148.
(32) Bartsch, S.; Kourist, R.; Bornscheuer, U. T. Angew. Chem., Int.
Ed. 2008, 47, 1508−1511.
(33) Ruzheinikov, S. N.; Burke, J.; Sedelnikova, S.; Baker, P. J.;
Taylor, R.; Bullough, P. A.; Muir, N. M.; Gore, M. G.; Rice, D. W.
Structure 2001, 9, 789−802.
(34) Lesley, S. A.; Kuhn, P.; Godzik, A.; Deacon, A. M.; Mathews, I.;
Kreusch, A.; Spraggon, G.; Klock, H. E.; McMullan, D.; Shin, T.;
Vincent, J.; Robb, A.; Brinen, L. S.; Miller, M. D.; McPhillips, T. M.;
Miller, M. A.; Scheibe, D.; Canaves, J. M.; Guda, C.; Jaroszewski, L.;
Selby, T. L.; Elsliger, M. A.; Wooley, J.; Taylor, S. S.; Hodgson, K. O.;
Wilson, I. A.; Schultz, P. G.; Stevens, R. C. Proc. Natl. Acad. Sci. U.S.A.
2002, 99, 11664−11669.
ACKNOWLEDGMENTS
■
The work of Hilde-Marlene Bergman is acknowledged. Prof. M.
Johnson and Biocenter Finland (bioinformatics, structural
biology, and translational activities) are acknowledged for the
excellent computational infrastructure at the Structural
Bioinformatics Laboratory, Åbo Akademi University and the
National Doctoral Programme in Informational and Structural
Biology is also acknowledged. This work was funded by the
Swedish Research Council (MW, Grant #621-2011-6055),
Sigrid Juselius Foundation (T.A.S., K.M.D.), Tor, Joe and
Pentti Borg Foundation (T.A.S., K.M.D.), and Åbo Akademi
Graduate School (K.M.D.).
REFERENCES
■
(35) Chen, Z.; Zhao, H. J. Mol. Biol. 2005, 348, 1273−1282.
(36) Zhu, Y.; Lin, E. C. C. J. Bacteriol. 1989, 171, 862−867.
(37) Tracewell, C. A.; Arnold, F. H. Curr. Opin. Chem. Biol. 2009, 13,
3−9.
(38) Khersonsky, O.; Roodveldt, C.; Tawfik, D. S. Curr. Opin. Chem.
Biol. 2006, 10, 498−508.
(1) Adams, R.; Levine, I. J. Am. Chem. Soc. 1923, 45, 2373−2377.
(2) Enders, D.; Bhushan, V. Tetrahedron Lett. 1988, 29, 2437−2440.
(3) Kratzer, R.; Nidetzky, B. Chem. Commun. 2007, 1047−1049.
(4) Hoyos, P.; Sinisterra, J.-V.; Molinari, F.; Alcan
́
tara, A. R.; De
María, P. D. Acc. Chem. Res. 2009, 43, 288−299.
(5) Lynch, J. E.; Eliel, E. L. J. Am. Chem. Soc. 1984, 106, 2943−2948.
(6) Goldberg, K.; Schroer, K.; Lutz, S.; Liese, A. 2007, 76, 237−248.
̈
(7) Blank, L. M.; Ebert, B. E.; Buehler, K.; Buhler, B. Antioxid. Redox
Signaling 2010, 13, 350−393.
(8) Hall, M.; Bommarius, A. S. Chem. Rev. 2011, 111, 4088−4110.
(9) Monti, D.; Ottolina, G.; Carrea, G.; Riva, S. Chem. Rev. 2011,
111, 4111−4140.
(10) Bornscheuer, U. T.; Huisman, G. W.; Kazlauskas, R. J.; Lutz, S.;
Moore, J. C.; Robins, K. Nature 2012, 485, 187−194.
(11) Cocks, G. T.; Aguilar, T.; Lin, E. C. J. Bacteriol. 1974, 118, 83−
88.
(12) Blikstad, C.; Widersten, M. J. Mol. Catal. B: Enzym. 2010, 66,
148−155.
(13) Gurell, A.; Widersten, M. ChemBioChem 2010, 11, 1422−1429.
(14) Janfalk Carlsson, Å.; Bauer, P.; Ma, H.; Widersten, M.
Biochemistry 2012, 51, 7627−7637.
́
(15) Montella, C.; Bellsolell, L.; Perez-Luque, R.; Badía, J.; Baldoma,
L.; Coll, M.; Aguilar, J. J. Bacteriol. 2005, 187, 4957−4966.
(16) Reid, M. F.; Fewson, C. A. Crit. Rev. Microbiol. 1994, 20, 13−56.
(17) Reetz, M. T.; Carballeira, J. D.; Vogel, A. Angew. Chem., Int. Ed.
2006, 45, 7745−7751.
(18) Reetz, M. T.; Prasad, S.; Carballeira, J. D.; Gumulya, Y.; Bocola,
M. J. Am. Chem. Soc. 2010, 132, 9144−9152.
(19) Tang, L.; Gao, H.; Zhu, X.; Wang, X.; Zhou, M.; Jiang, R.
Biotechniques 2012, 52, 149−156.
(20) Elfstrom, L. T.; Widersten, M. Biochem. J. 2005, 390, 633−640.
̈
(21) Kille, S.; Acevedo-Rocha, C. G.; Parra, L. P.; Zhang, Z. G.;
Opperman, D. J.; Reetz, M. T.; Acevedo, J. P. ACS Synth. Biol. 2013, 2,
83−92.
(22) Sali, A.; Blundell, T. L. J. Mol. Biol. 1993, 234, 779−815.
(23) Wilmot, C. M.; Hajdu, J.; McPherson, M. J.; Knowles, P. F.;
Phillips, S. E. Science 1999, 286, 1724−1728.
(24) Pillai, B.; Cherney, M. M.; Hiraga, K.; Takada, K.; Oda, K.;
James, M. N. J. Mol. Biol. 2007, 365, 343−361.
(25) Reetz, M. T.; Kahakeaw, D.; Lohmer, R. ChemBioChem 2008, 9,
1797−1804.
(26) Agudo, R.; Roiban, G.; Reetz, M. T. ChemBioChem. 2012, 13,
1465−1473.
(27) Savile, C. K.; Janey, J. M.; Mundorff, E. C.; Moore, J. C.; Tam,
S.; Jarvis, W. R.; Colbeck, J. C.; Krebber, A.; Fleitz, F. J.; Brands, J.;
3025
dx.doi.org/10.1021/cs400824h | ACS Catal. 2013, 3, 3016−3025