Obtaining optical purity for product diols in enzyme-catalyzed epoxide hydrolysis: Contributions from changes in both enantio- and regioselectivity

Article abstract of DOI:10.1021/bi3007725

Enzyme variants of the plant epoxide hydrolase StEH1 displaying improved stereoselectivities in the catalyzed hydrolysis of (2,3-epoxypropyl)benzene were generated by directed evolution. The evolution was driven by iterative saturation mutagenesis in comb

Full text of DOI:10.1021/bi3007725

Article
pubs.acs.org/biochemistry
Obtaining Optical Purity for Product Diols in Enzyme-Catalyzed
Epoxide Hydrolysis: Contributions from Changes in both Enantio-
and Regioselectivity
Åsa Janfalk Carlsson, Paul Bauer, Huan Ma, and Mikael Widersten*
Department of Chemistry-BMC, Uppsala University, Box 576, SE-751 23 Uppsala, Sweden
S
* Supporting Information
ABSTRACT: Enzyme variants of the plant epoxide hydrolase
StEH1 displaying improved stereoselectivities in the catalyzed
hydrolysis of (2,3-epoxypropyl)benzene were generated by
directed evolution. The evolution was driven by iterative
saturation mutagenesis in combination with enzyme activity
screenings where product chirality was the decisive selection
criterion. Analysis of the underlying causes of the increased
diol product ratios revealed two major contributing factors:
increased enantioselectivity for the corresponding epoxide
enantiomer(s) and, in some cases, a concomitant change in
regioselectivity in the catalyzed epoxide ring-opening half-reaction. Thus, variant enzymes that catalyzed the hydrolysis of racemic
(2,3-epoxypropyl)benzene into the R-diol product in an enantioconvergent manner were isolated.
Scheme 1. Hydrolysis of a Chiral Epoxide into Vicinal Diols
he catalytic powers of enzymes present great potentials as
replacements or complements to presently dominating
T
approaches in organic synthesis. The inherently asymmetric
nature of enzymes’ active sites facilitates specific and efficient
reactions involving chiral reactants and intermediates. Still,
adoption of enzymes in synthetic regimes is slow, which may be
attributed to different issues. Enzymes are, by evolution of their
biological roles, generally too specific in their substrate
acceptance to become useful synthetic tools in reactions
involving a wider structural range of related compounds.
Enzymes are, with exceptions, prone to denature in organic
solvents, which prevents their inclusion in presently used
synthetic protocols. The development of methodologies
affording re-engineering of both functional and physicochem-
ical properties of enzymes, however, shows promise that most,
if not all, of these issues can be addressed.1−8
We have set out to design multistep reactions catalyzed solely
by enzymes involving chiral starting compounds that are
transformed into stereochemically defined products. Naturally
occurring enzymes are identified on the basis of catalyzed
chemistry rather than substrate selectivity and are re-engineered
through the process of directed evolution9−13 into potential
biocatalysts that display desired catalytic functions and
substrate selectivities. The starting reactants are chiral epoxides
that are hydrolyzed into likewise chiral vicinal diols (Scheme 1)
that can be further derivatized in subsequent reaction steps.
Both the epoxides and the product diols are important
precursors in the synthesis of asymmetric compounds, and
the possibility of producing these molecules in optically pure
form is therefore of interest.14
ring is opened either by a base-activated water molecule, such
as in the limonene epoxide hydrolyzing enzyme, or by an
enzyme-contributed carboxylate, thereby forming a covalent
alkyl−enzyme intermediate.15−18 The latter mechanism is
present in the α/β-hydrolase-type enzymes,19−22 with the
ring opening being facilitated by enzyme-provided acids, a Tyr
pair.23,24 Our group is using the potato epoxide hydrolase
StEH125,26 as the starting scaffold for engineering new epoxide
hydrolase enzymes that can be applied as biocatalysts for chiral
epoxide ring-opening reactions.27 StEH1 differs in active-site
structure and substrate selectivity from a previously engineered
Aspergillus niger epoxide hydrolase.22,28−30 The substrate
binding site in StEH1 is larger, thus broadening the substrate
scope of the wild-type enzyme compared to that of the A. niger
isoenzyme.
The intrinsic reactivities of the oxirane carbons in a ring-
opening reaction with an electron rich nucleophile (such as a
carboxylate) depend on the additional substituents of the
epoxide. Since the transition state involves a partially positively
Received: June 11, 2012
Revised: August 23, 2012
Published: August 29, 2012
The enzyme-catalyzed reactions found for naturally occurring
epoxide hydrolases are SN2-type reactions in which the epoxide
© 2012 American Chemical Society
7627
dx.doi.org/10.1021/bi3007725 | Biochemistry 2012, 51, 7627−7637

Biochemistry
Article
charged carbon, an electron-donating group, such as phenyl,
will promote addition of the nucleophile at the substituted
carbon. Consequently, electron-withdrawing groups steer
nucleophilic attack toward the other epoxide carbon.31 In
epoxides without electron-donating or -withdrawing substitu-
ents, other factors like steric hindrance will become more
important. In enzyme-catalyzed ring opening, however, these
guiding principles of expected product outcome are overturned
and other restraints, such as active-site structure, will influence
regioselectivity. This is well illustrated by StEH1-catalyzed
hydrolysis of styrene oxide [2 (Chart 1)] and trans-2-
deciding product outcome but rather influence the overall
reaction rate.24,28,34 Therefore, it is possible to make naive
assumptions about product outcomes from docking experi-
ments or more elaborate calculations.35−38 The detailed
understanding is, however, unclear, which is unsatisfactory
from a perspective of manipulating properties such as
stereoselectivity by rational approaches guided by first
principles.
(2,3-Epoxypropyl)benzene (1) will, upon catalyzed hydrol-
ysis, form either the R-4 or the S-4 diol. The outcome depends
both on the enantiopreference of the enzyme for the epoxide
substrate and on the regioselectivity of the attack. Wild-type
StEH1 catalyzes the hydrolysis in a regiospecific manner in
which the enzyme-contributed nucleophile (D105 carboxylate)
attacks the epoxide at the (expected) least hindered nonchiral
carbon 2 (Chart 1). The catalyzed hydrolysis causes retention
of configuration, and the resulting diol products therefore
inherit the stereochemistry of the reactant epoxides. The
enantioselectivity of StEH1 with racemic 1 is low with a 2.5-
fold preference for hydrolysis of S-1 as compared to the R-
enantiomer.27
In this work, we set out to modify the stereoselectivity of the
StEH1-catalyzed reaction with 1 to generate enzyme variants
yielding an enriched enantiomer excess of the resulting chiral
product diols. The aim was set without favoring alterations in
enantio- or regioselectivities, because either, or in combination,
can potentially generate the same result of improved product
optical purity. As discussed above, lacking the necessary
understanding of the prerequisites for steering selectivity in
substrate acceptance and regioselectivity in nucleophilic attack,
we applied the semirational iterative saturation mutagenesis
(ISM) approach originally introduced by M. Reetz.39,40 This
strategy has proven to be efficient in producing enzyme variants
with improved desired properties.41−45 The approach relies on
subdividing the active site into groups of amino acid residues
deemed to affect the targeted property. In this study, the StEH1
active site was divided into five subsites, A−E (Figure 2),
maintaining one (A, F33), two (B, W106 and L109; C*, L145
and I155; D, I180 and F189; and E, L266 and V267), or three
(C, V141, L145, and I155) residues. Applying a codon-limited
mutagenesis regime (NDT) further served to generate mutant
libraries of manageable sizes. The decrease in the possible
structural diversity, from 20n to 12n variants in the NDT
mutated libraries, is a reasonable trade-off because the hit rate
has been shown to be comparable while the effort in library
constructions and screening becomes greatly reduced.46 The
approach allowed for isolation of StEH1 variants displaying
distinct stereoselectivities with epoxide 1 that also had acquired
altered selectivities with the related epoxides 2 and 3. These
substrates were included because of both their structural
similarities with epoxide 1 and their relevance in synthetic
chemistry. The aim was to see patterns and potential
similarities, concerning, e.g., enantio- and regioselectivity,
between the various substrates and enzyme variants.
Chart 1. Epoxide Substrates
methylstyrene oxide (3). The enzyme-afforded ring opening
of S-2 does involve nucleophilic attack at the (expected)
benzylic carbon but preferentially at the least hindered carbon
with the R-enantiomer.32 A similar behavior has been observed
in the catalyzed hydrolysis of 3 where the S,S-enantiomer is
exclusively attacked at the benzylic carbon, whereas the
regiopreference in the hydrolysis of R,R-3 varies with changes
in temperature and pH.33 These findings suggest that
regioselectivity in StEH1 catalysis may be amenable to
manipulation by directed protein evolution.
The apparent structural requirements of the catalyzed
reaction are (i) complementarity of the epoxide and the
active-site compartment and (ii) correct spatial orientation of
the epoxide with respect to both the catalytic acid(s) (two Tyr
residues) and the nucleophilic Asp (see Figure 1 for the
working model of the catalytic mechanism of StEH1). The
hydrolytic step of the catalytic cycle, which generally is the rate-
limiting step, is expected to contribute to a lesser degree to
MATERIALS AND METHODS
Chemicals and Reagents. All chemicals used were of the
highest purity commercially available.
■
Figure 1. Current working model of the catalytic mechanism of
StEH1. The nucleophilic carboxylate of Asp105 reacts with one of the
electrophilic epoxide carbons. The subsequent ring opening is
catalyzed by hydrogen bonding of the leaving group oxyanion to the
phenols of Tyr154 and Tyr235. The so-formed covalent alkyl−enzyme
intermediate is hydrolyzed in the general base (His300)-catalyzed half-
reaction to yield the diol product.
Construction of Enzyme Libraries. The sites used for
mutations were A (F33), B (W106 and L109), C (V141, L145,
and I155), and D (I180 and F189), with the addition of C*
(L145 and I155) and E (L266 and V267). For the R-specific
variants, the previously generated quadruple mutant R-C1B127
7628
dx.doi.org/10.1021/bi3007725 | Biochemistry 2012, 51, 7627−7637

Biochemistry
Article
Figure 2. Stereo image of the active site of StEH1. Side chains of residues included in the iterative mutagenesis are shown as colored sticks: library A
(F33) in magenta, library B (W106 and L109) in yellow, library C (V141, L145, and I155) in green, library D (I180 and F189) in blue, and library E
(L266 and V267) in cyan. Catalytic residues are shown as white sticks. Library C* does not include V141 as an altered site. Image created with
PyMol (http://www.pymol.org) applying the atomic coordinates of Protein Data Bank entry 2CJP.28
was used as a template for further mutations at the A and D
sites. In accordance with the idea of the ISM approach,6,39 the
top candidate from the D library (R-C1B1D33) was
subsequently used as a template for the E library. Similarly,
the S-specific variants were constructed using the wild-type
StEH1 cDNA together with primers for the A and B libraries
where the top candidate from the B library (S-B17) then was
used as the template for both the C* and D libraries. All
primers used are listed in Table SI1 of the Supporting
Information, in which a degenerate NDT codon set, leading to
12 possible amino acids, was used for each mutation site. The
mutants were constructed using nested polymerase chain
reaction (PCR) except for the A library, using QuikChange,
and the D library, using normal PCR. Each library was
transformed by electroporation into Escherichia coli XL-1 Blue
cells carrying the pREP4 GroEL/ES vector,47 allowing for
overexpression of chaperonins. Colonies were selected on LB-
agar plates, supplemented with 100 μg/mL ampicillin and 30
μg/mL kanamycin, and the qualities of the constructed libraries
were verified by sequencing at least 20 randomly picked clones.
Cultivation and Expression of Libraries. Cells were
grown in 96-well microtiter plates. Single colonies were
inoculated into wells containing 350 μL of 2TY (16 g/L
tryptone, 10 g/L yeast extract, and 5 g/L NaCl) with 100 μg/
mL ampicillin and 30 μg/mL kanamycin added. Culture wells
were covered with gas permeable film and incubated for 18 h at
30 °C and 175 rpm. The overnight culture (25 μL) was
reinoculated into 325 μL of 2TY with 50 μg/mL ampicillin and
15 μg/mL kanamycin. After incubation for 5 h, expression of
StEH1 protein variants together with GroEL/ES chaperonins
was induced by addition of isopropyl β-D-thiogalactopyranoside
(IPTG) (final concentration of 1 mM). Expression was allowed
for 18 h, and cells were harvested by centrifugation. Cells were
lysed by addition of 25 μL of Bacterial Protein Extraction
Reagent (B-PER), with protease inhibitor added, to each well.
The cell pellets were resuspended by being vortexed and then
incubated for 40 min at room temperature. To increase the
volume, 75 μL of 0.1 M sodium phosphate buffer (pH 7.0) was
added before the cell debris was centrifuged for 1 h at 4 °C and
3500 rpm. The lysate was stored cold until it was used.
Screening of Libraries. The screening reactions were
performed in duplicate for R- and S-1, where 25 μL of lysate
from each well was transferred to each of four different flat-
bottom plates. To each well was added 25 μL of 10 mM 1
[dissolved in 0.1 M sodium phosphate buffer (pH 7.0) with
10% (v/v) DMSO]. The reaction mixtures were incubated for
10−20 min, depending on the activity of the enzyme variants,
at 30 °C and 200 rpm. To assay the remaining epoxide, a
colorimetric method as described by Agarwal48 was applied.
Twenty-five microliters of 100 mM 4-(4-nitrobenzyl)pyridine
[dissolved in an 80:20 (v/v) glycol/ethanol mixture] was
added. The reaction between nitrobenzyl pyridine and
remaining epoxide was allowed for 20 min at 80 °C and 200
rpm before being stopped by addition of 50 μL of ethanol.
After addition of 25 μL of 1 M K2CO3, color, indicative of
nonhydrolyzed epoxide, appeared. To estimate epoxide hydro-
lase activities in each well, the absorbances at 570 nm were
measured using a microtiter plate reader. The apparent
hydrolase activities were used to identify potential hits. The
corresponding StEH1 cDNAs were sequenced to identify the
inserted mutations. Unique variants were subsequently verified
as hits by a repeated round of activity measurement.
Protein Expression and Purification of Hits. The
expression plasmid encoding a selected variant was transformed
into E. coli XL-1 Blue cells, carrying the pREP4 GroEL/ES
plasmid. A single colony was inoculated into 5 mL of 2TY with
100 μg/mL ampicillin and 30 μg/mL kanamycin and grown
overnight at 30 °C and 200 rpm. The overnight culture was
diluted in 0.5 L of 2TY medium containing 50 μg/mL
ampicillin and 15 μg/mL kanamycin and grown at 30 °C and
200 rpm. When the OD600 reached 0.4, protein expression of
both the StEH1 variants and the chaperonins was induced by
the addition of IPTG (final concentration of 1 mM).
Expression was allowed for 18 h at 30 °C and 200 rpm, and
7629
dx.doi.org/10.1021/bi3007725 | Biochemistry 2012, 51, 7627−7637

Biochemistry
Article
cells were harvested by centrifugation. The cell pellet was
stored at −80 °C until it was further treated.
concentrations ranging from 0.12 to 4.5 mM) was added to
0.1 M sodium phosphate buffer (pH 7.0) and acetonitrile [3%
(v/v)]. The reaction mixtures were incubated at 30 °C, before
the enzyme (final concentrations of 50−200 nM) was added.
Reactions were performed in triplicate with aliquots taken out
after 0.5, 1.5, 2.5, and 3.5 min. The reactions were stopped by
addition of methanol (with added phenylethyl alcohol as the
internal standard) to a final concentration of 50% (v/v).
Samples were kept on ice from this point or stored at −20 °C
until they were analyzed. Samples were analyzed using reverse
phase HPLC with a mixture of 0.1 M sodium phosphate buffer
(pH 3.0) and methanol [63:37 (v/v)] as the mobile phase on a
Supelco Ascentis C18 column (250 mm × 4.6 mm, 5 μm). The
flow rate was set to 0.8 mL/min with the rest of the system
being built up as previously described. Peaks were detected at
220 nm with approximate retention times of 15 and 50 min for
diol 4 and epoxide 1, respectively.
Analysis of Regioselectivity. To identify the peaks for R-4
and S-4 in the HPLC chromatogram, 13C NMR was used to
analyze the product of a reaction catalyzed by StEH1, proven to
have a strict regioselectivity with 1. Enzyme (1 μM) was mixed
with 25 mM S-1 in 0.1 M sodium phosphate buffer [1:1 (v/v)
H216O/H218O] (pH 7.0). For the solubility of the epoxide,
acetonitrile was added to a final concentration of 3% (v/v).
Substrate was added in 2 μL aliquots over time to further
increase solubility. The reaction mixture was incubated for 16 h
at 30 °C and 400 rpm, whereafter product formation was
confirmed by reverse phase HPLC. Prior to NMR analysis,
D2O (final concentration of 10%) was added to the sample.
NMR spectra were recorded at 500 MHz (1H) and 125.7 MHz
(13C) on a Varian Inova spectrometer. For the 18O-labeled
compounds, 13C NMR spectra were recorded over a spectral
range of 30 kHz, using an acquisition time of 1.3 s, a relaxation
delay of 3 s, and a pulse flip angle of 44°. A total of 14200
transients (total experiment duration of 17 h) were acquired,
for a final signal:noise ratio of 40−70. Data were processed with
zero filling to 524288 points and Gaussian apodization. The
result allowed assignment of the respective diol peaks obtained
after chiral HPLC, as described below (Figure SI1 of the
Supporting Information).
To determine the regioselectivity of the purified enzymes,
reactions with 10−15 mM enantiopure (>98%) substrates 1−3
were performed overnight at 30 °C with slow agitation in a 500
μL total reaction volume using 0.5 μM enzyme and 3% (v/v)
acetonitrile in 0.1 M sodium phosphate buffer (pH 7.0). After
incubation, 500 μL of methanol was added to stop the
reactions, whereafter the solvents were evaporated as previously
described. The precipitate was resuspended in an 80:20 (v/v)
hexane/2-propanol mixture and kept at −20 °C until further
analysis. Samples were vortexed and centrifuged for 5 min at
13000 rpm to remove insoluble debris before being analyzed by
chiral HPLC. The diol products formed from epoxides 2 and 3
were detected by measuring the absorbance at 225 nm, whereas
the hydrolysis products of 1 were detected at 220 nm.
Protein Sequence Alignments. The StEH1 sequence was
used as query in a protein BLAST49 search of the plant subset
of the UniProtKB database using the online tool at http://
www.uniprot.org. Hit sequences of varying degrees of sequence
identity (scores of 300−550) were collected and submitted to a
ClustalW50 multiple-sequence alignment.
Protein purification was performed in a batch manner in
which all the purification steps were conducted under slow
agitation at 4 °C. First, the pellet from a 0.5 L culture was
resuspended in 20 mL of binding buffer [0.1 M sodium
phosphate, 0.5 M NaCl, and 20 mM imidazole (pH 7.4)] with
added DNaseI and protease inhibitor. Cells were lysed using a
Constant Systems Cell disruptor at 1.7 kPa, and the lysate was
centrifuged for 1 h at 30000g. The supernatant was transferred
to a 50 mL conical tube containing 1−2 mL of chelating resin,
preloaded with Ni2+. The mixture was incubated for 30 min
before being centrifuged for 5 min at 700g. The supernatant
was discarded, and 20 mL of washing buffer [0.1 M sodium
phosphate, 0.5 M NaCl, and 0.1 M imidazole (pH 7.4)] was
added. After incubation for 10 min, the resin was centrifuged at
700g and the supernatant was discarded. This washing step was
then repeated. The protein was eluted by resuspending the
resin in 2.5 mL of elution buffer [0.1 M sodium phosphate, 0.5
M NaCl, and 0.3 M imidazole (pH 7.4)]. After incubation for 5
min, the resin was centrifuged at 800g for 5 min. As a last step,
the supernatant was transferred to a PD-10 desalting column,
equilibrated with 0.1 M sodium phosphate buffer (pH 7.4), and
eluted with 3.5 mL of equilibration buffer. The elution step was
repeated, whereafter purity was confirmed by sodium dodecyl
sulfate−polyacrylamide gel electrophoresis. Purified enzyme
variants were stable for several weeks when stored in the cold.
Protein concentrations were determined by measuring UV
absorption at 280 nm, using an extinction coefficient (ε) based
on the value for wild-type StEH1,26 corrected by the following
formula:
ε = εWT + ΔTrp × 5500 + ΔTyr × 1490 + ΔCys × 125
Enantiomeric Excess of Diol Product. The substrate rac-
1 was added, together with acetonitrile [3% (v/v)], to 0.1 M
sodium phosphate buffer (pH 7.0) to a concentration of 30.4
mM. The reactions were assayed in triplicate at 30 °C. Enzyme
(0.5 μM) was added to start the reactions, which were followed
by the removal of aliquots at regular time points between 0 and
60 min. The reactions were stopped by addition of methanol to
a final concentration of 50% (v/v). Solvent was evaporated for
5 h in a vacuum centrifuge at 800 rpm and the pellet
resuspended in a hexane/2-propanol mixture [80:20 (v/v)].
The relative amounts of each R- and S-4 were analyzed by chiral
high-performance liquid chromatography (HPLC) [Daicel
Chiralpak AS-H (250 mm × 4.6 mm inside diameter)] using
a hexane/2-propanol mixture [90:10 (v/v)] as a mobile phase.
The system used a LC 20AD solvent delivery module
(Shimadzu, Tokyo, Japan), a Shimadzu SIL-10AF autosampler,
and a Shimadzu SPDM20A photometric unit with the flow rate
set at 0.5 mL/min. The retention times were 23 and 28 min for
S-4 and R-4, respectively.
Kinetic Characterization. Steady-state kinetic parameters
for substrates 2 and 3 were determined by measuring initial
rates of epoxide consumption as a decrease in absorbance at
225 nm (ε values of −2.75 and −4.3 mM−1 cm−1 for 2 and 3,
respectively). Reaction mixtures, containing pure enantiomers
of each substrate (final concentrations of 0.016−0.7 mM)
dissolved in 0.1 M sodium phosphate buffer (pH 7.5) with
acetonitrile [1% (v/v)] added, were incubated at 30 °C.
Reactions were started by addition of 0.5−1 μM enzyme.
For the reactions using epoxide 1 as the substrate, a protocol
adapted from ref 27 was used. The substrate (final
7630
dx.doi.org/10.1021/bi3007725 | Biochemistry 2012, 51, 7627−7637

Biochemistry
Article
the highest enantiomer ratio of R-4 were R-C1B1D33E3 and
-E6, L266F and L266G, respectively (Figure 3).
RESULTS
■
Enzyme Library Constructions and Screening. Increase
in the Enantiomer Excess of R-4 from Racemic 1. A previous
initial study generated an enzyme variant,27 hereafter called R-
C1B1 (Table 1), which was chosen to template further
Increase in the Enantiomer Excess of S-4 from Racemic 1.
Libraries S-A and S-B were constructed and analyzed, following
the same procedure described above, using wild-type StEH1
cDNA as the template. From the S-A library, 13 hits were
isolated and sequenced (Table SI2 of the Supporting
Information). The F33C and F33Y mutants were analyzed
for their abilities to increase the enantiomer excess of S-4 with
only F33C displaying a slight improvement (Figure 3). From
the S-B library, where 17 hits were sequenced, a variant (S-B17)
showing a more pronounced increase was found (Table 2 and
Figure 3). S-B17 was therefore subsequently used to template
mutagenesis at the C* or D site. From these respective libraries,
variants S-B17C*10 and S-B17D14 were identified as catalyzing
the formation of S-4 with a slightly increased enantiomer excess
compared to that of the parent S-B17 variant (Figure 3).
Primary Structures of Selected Variants. In general, the
hydrophobic nature of the active-site residues targeted for ISM
was retained also in the isolated variants, and the observed
replacements, with few exceptions, can be considered
conservative (Table SI2 of the Supporting Information).
Among the variants displaying increased product ratios of the
respective diol enantiomers, all substitutions except V141K and
L145R retained the nonpolar character of the initial residues
(Table 2). The L145R replacement in S-B17C*10, however,
appears to represent a phenotypically silent mutation because it
was functionally indistinguishable from the related variant S-
B17C*9 (L145H) in the reactions with epoxides 2 and 3
(Table SI3 of the Supporting Information).
Table 1. Library Data
mutated
residues
library
size
no. of screened no. of
template
library
clones
hits
R-C1B1
R-C1B1
R-C1B1D33
StEH1 wt
StEH1 wt
S-B17
A
F33
12
144
144
12
168
504
504
168
504
504
504
2
42
8
D
E
I180, F189
L266, V267
F33
A
13
17
19
18
B
W106, L109
L145, I155
I180, F189
144
144
144
C*
D
S-B17
mutagenesis at subsites A and D (Figure 2). The constructed
protein expression libraries R-C1B1A and R-C1B1D main-
tained 12 and 144 members, respectively. To ensure a full
coverage of potential protein variants, well above a theoretical
coverage of >95%, all libraries were oversampled at least 3-fold
when screening for epoxide hydrolase activity with 1 (Table 1).
Two hits were isolated from the R-C1B1A library, which
both had retained wild-type F33 through silent mutations
(TTT in the NDT codon set) (Table SI2 of the Supporting
Information). From the R-C1B1D library, 42 hits were
collected and sequenced. A subset of these clones, chosen on
the basis of apparent enzyme activity in the screening assay,
were purified and analyzed via chiral HPLC to map the abilities
to afford an increased R-4:S-4 ratio. The variant displaying the
highest ratio compared to the parent was R-C1B1D33 (Figure
3 and Table 2). The selection process was repeated with R-
C1B1D33 as the template for mutagenesis at the E site. From
this R-C1B1D33E library, eight clones were isolated (Table SI2
of the Supporting Information) where the variants displaying
In the S-B library selection, the aromatic character of the
original Trp residue was conserved by Tyr or Phe substitutions
in 13 of 17 selected hits (Table SI2 of the Supporting
Information).
Catalytic Activities. The steady state kinetic data are
presented in Table SI3 of the Supporting Information and are
described here.
R-4 Selected Variants. A desired property of these mutants
was an acquired ability to catalyze the production of R-4 with
refined optical purity, as compared to that of the parental
StEH1 enzyme. Therefore, with retained regioselectivity, an
increased ratio of the catalytic efficiencies between R-1 and S-1
was the expected outcome for the variants displaying an
increased enantiomer excess of R-4. This was also the general
trend; the kcat/KM for S-1 decreased, from 42 to <1 s−1 mM−1,
during the iterations of mutagenesis and selection, while the
activity with the R-enantiomer decreased less, from 17 to 12 s−1
mM−1 in the furthest evolved mutant R-C1B1D33E3 (Table
SI3 of the Supporting Information). The resulting E values of 3
for variant R-C1B1D33 (an 8-fold increase compared to that of
the wild-type enzyme) and 14 for mutant R-C1B1D33E3,
however, did not solely explain the improvements that were
observed in the R-4:S-4 ratios. A general observation was that
the individual parameters kcat and KM were affected in the
direction of maintained turnover of R-1 and increased values of
KM for both substrate enantiomers with the exception of R-
C1B1D33E3, which maintained a comparably low KMR‑1 (Table
SI3 of the Supporting Information).
Figure 3. Ratios of diol products R-4 and S-4 over time after 0.5 μM
enzyme-catalyzed hydrolysis of 30.4 mM racemic 1: wild-type StEH1
The selected variants were also analyzed for their acquired
catalytic functions toward related epoxides 2 and 3. With
styrene oxide (2), the most dramatic effect was the increase in
enantioselectivity for S-2, from 69 to 5800 for R-C1B1D33.
The effect can be attributed to an 80-fold decrease in the
○
△
▲
( ), R-C1 ( ), R-C1B1-K141V (◮), R-C1B1 ( ), R-C1B1D33
▽
◆
▼
(
(
), R-C1B1D33E3 (⧩), R-C1B1D33E6 ( ) S-A8 (◑), S-B17
■
), S-B17C*10 ( ), and S-B17D14 (◨). Data points are mean
values of three independent measurements.
7631
dx.doi.org/10.1021/bi3007725 | Biochemistry 2012, 51, 7627−7637

Biochemistry
Article
Table 2. Primary Structures of Selected Mutants
library B
library C
library D
library E
library A
F
106
W
109
L
141
145
L
155
180
I
189
F
266
L
267
V
wild-type
V
K
K
K
K
K
I
a
R-C1
°
V
R-C1B1
L
L
L
L
Y
Y
Y
Y
°
°
°
°
V
R-C1B1D33
R-C1B1D33E3
R-C1B1D33E6
V
°
°
°
L
L
L
V
F
°
°
V
G
library A
library B
library C*
library D
33
C
106
109
145
R
155
Y
180
F
189
S-A8
S-B17
Y
Y
Y
I
I
I
S-B17C*10
S-B17D14
°
a
A circle indicates a silent mutation.
catalytic efficiency with R-2 (Table SI3 of the Supporting
Information). A similar behavior was also observed in the
hydrolysis of epoxide 3, albeit to a lesser degree. The E value
for S,S-3 increased from 84 to >1300 for variant R-
C1B1D33E6, whereas the activity with R,R-3 was below the
detection limit. Clearly, the mutations introduced at L266 in
the E3 and E6 variants had a negative impact on the catalytic
efficiencies with both phenyl-substituted epoxides 2 and 3.
S-4 Selected Variants. The wild-type enzyme displays an E
value for S-1 of 2.5, which was increased to approximately 6 for
the furthest improved variant, S-B17D14. As for the R-4
selected mutants, the KM for both substrate enantiomers was
increased (2−4-fold), which was accompanied by decreases in
kcat that were larger with the nonselected R-epoxide (Table SI3
of the Supporting Information). The observed increase in the
enantiomer excess of S-4 could be linked to the steady state
kinetic parameters, here leading toward kinetic resolution of the
S-4 enantiomer.
The hydrolysis of epoxide 2 by the S-4 selected variants was
not substantially different from that of the wild-type enzyme,
albeit with 5−10-fold lower activities for both enantiomers. A 3-
fold increase in E for S-2 was displayed by variant S-B17D14,
mainly caused by a more pronounced decrease in activity with
R-2.
In the reaction with epoxide 3, the S-4 selected variants
displayed 15−30-fold increases in E for S,S-3. Variants S-
B17C*10 and S-B17D14 both exhibited turnover numbers for
S,S-3 that exceeded that of the wild-type enzyme, while the
catalytic efficiencies with the R,R-enantiomer were 70−140-fold
decreased. The resulting E values were between 1300 and 2700
(Table SI3 of the Supporting Information).
Regioselectivity in Epoxide Ring Opening. StEH1
displays a strong preference for nucleophilic attack at C-2 in
the catalyzed epoxide ring opening of both enantiomers of 1
(Figure 4 and Table SI3 of the Supporting Information). For
phenyl-substituted epoxides 2 and 3, the S- and S,S-
enantiomers are specifically reacting at the benzylic carbon
(C-1). The R-2 epoxide, however, reacts primarily at the least
hindered C-2, and R,R-3 is hydrolyzed via ring opening at either
carbon, with a slight preference for the benzylic carbon.
Analysis of the isolated mutants underlined the plasticity of
regioselectivity. The results are summarized in Figure 4 and
described below.
Figure 4. Regiopreferences in nucleophilic attack of the epoxide
carbons of StEH1 and isolated variants. See Chart 1 for number
designation.
R-4 Selected Variants. These mutants mimicked the
regiopreference of the wild-type enzyme with epoxide R-1, S-
2, and S,S-3 with one exception, mutant R-C1 (V141K, I155V),
7632
dx.doi.org/10.1021/bi3007725 | Biochemistry 2012, 51, 7627−7637

Biochemistry
Article
displaying distinct regioselectivity with both S-2 and S,S-3. The
original strong preference for attack at the benzylic carbon had
been shifted to a 70:30 preference for ring opening at C-2. With
epoxide S-1, the regiopreference was slightly shifted toward
attack at C-1. This shift was accentuated in later generation
variants R-C1B1 and R-C1B1D33. Mutations at position 266,
however, reverted the regioselectivity back to attack at C-2;
both E library variants R-C1B1D33E3 and R-C1B1D33E6
displayed a stronger preference for reaction at the least
hindered epoxide carbon, with R-C1B1D33E6 being highly
selective for attack at C-2. The shift in regioselectivity displayed
by mutants R-C1B1 and R-C1B1D33, and the increased E
values for R-1 in the E library mutants, could be linked to the
observed increases in R-4:S-4 product ratios.
The V141K mutation found in R-C1B1 was originally seen as
an accepted, but phenotypically silent, mutation because of the
polar character of the side chain ammonium group that stood
out in an otherwise hydrophobic active site. The regioselectivity
of the R-C1B1-K141V revertant, however, displayed a lower
degree of attack at C-1, resulting in a lower enantiomer excess
of the R-diol product, demonstrating a role for K141 in steering
regiopreference.
For the R-enantiomer of styrene oxide, the wild-type
preference for attack at the least hindered C-2 was shifted
toward higher degrees of reaction at the benzylic carbon, even
though all variants, except R-C1B1D33E6, displayed promis-
cuity in regioselectivity. With R,R-3, the initial (slight)
preference for attack at the benzylic carbon was altered to a
selective reaction at C-2 in several mutants. The involvement of
position 266 for steering regioselectivity in this series of
variants, for both R-2 and R,R-3, is clear from a comparison of
the E library variants; R-C1B1D33E3, which contains an L266F
replacement, preferentially reacts at the benzylic carbon
(70:30), whereas the corresponding Gly substitution directs
epoxide ring opening to occur exclusively at C-2. It should,
however, be kept in mind that the overall enzymatic activity
with R,R-3 is comparably very small versus that with the S,S-
enantiomer.
S-4 Selected Variants. These enzymes displayed unaltered
regiopreferences with both enantiomers of 1, S-2, and S,S-3.
With R-2, all displayed a relaxed regioselectivity with
approximately equal degrees of reaction at either epoxide
carbon. The behavior with R,R-3 was similar to that of the R
variants (Figure 4), with a shift in regiopreference from C-1 to
C-2 (75% C-2). Again, the actual enzyme activities with this
epoxide were humbled by the corresponding activities with the
S,S-enantiomer.
The main goals of this study were (i) to test the influence of
changes in enantioselectivity versus regioselectivity on the
outcome of the chirality of the product diols, (ii) to analyze
how a structure−function trajectory generated through a
pathway of stepwise mutagenesis in combination with defined
selection criteria, in this case product configuration after
hydrolysis of epoxide 1, would be expressed also in reactions
with other related epoxides, and (iii) to map the enzyme active
site for determinants of stereoselectivity in hydrolysis of
epoxides 1−3.
Structure−Activity Relationships. The mutated residues
were selected on the basis of their structural contributions to
the active-site pocket and anticipated interactions with substrate
and reaction intermediates. Although the same general
approach of targeting mutation sites was applied with the A.
niger enzyme, the altered positions in the respective active sites
are in all cases different, as judged by structure alignments. For
a comparison with other naturally occurring plant enzymes,
sequence alignments were conducted to evaluate the variability
in positions corresponding to those chosen for ISM of StEH1
(Figure SI3 of the Supporting Information). F33, which was
mutated to generate A library variants, and W106, which
contributes to the B libraries, are both fully conserved in these
19 plant epoxide hydrolase sequences. The reason for this strict
conservation may lie in their roles in contributing their peptide
backbone amides to the oxyanion hole stabilizing the negatively
charged tetrahedral intermediate occurring in the hydrolytic
half-reaction of catalysis. Although these interactions do not
involve side chain functionalities, the conservation may reflect
strict structural constraints for this stabilizing function. F33 was
also strongly conserved from insertion of silent mutations in the
isolated variants in this study. Although several mutants that
were catalytically active were isolated, these did not display
substantially improved enantiomeric excesses of the different
diol 4 enantiomers and were therefore discarded from further
evolution efforts.
The NDT codon set does not include a Trp codon leading to
W106 being by default mutated in the library variants. From the
mutants isolated and characterized, however, no evidence of
special importance of the indole side chain at this position
could be established; insertion of Gly, Leu, Phe, or Tyr, instead
of Trp, generated functional StEH1 variants. An additional
observation of the nonessential role of W106 in catalysis was
made with mutant S-B16 that had retained W106, because of an
artifact in the library construction. This mutant can be viewed
as an F106W revertant of mutant S-B9. The ability of this single
mutant to produce S-4 from racemic 1 was poorer than that of
the double mutant, illustrating that the absence of Trp codons
in the NDT codon set did, in this case, not present limitations
to the available structure−function space. This residue adjacent
to the nucleophile, however, appears to be important in
steering stereoselectivity because both enantio- and regiopre-
ferences are affected by mutations at this site. This is further
supported by a study in which the corresponding position in
the isoenzyme from Agrobacterium radiobacter was mutated,
resulting in variants with altered enantioselectivities.52
DISCUSSION
■
Manipulating Stereoselectivity in Epoxide Hydrolase.
Productive enzyme engineering by rational approaches is
dependent on a detailed understanding of structure−function
relationships, including substrate−enzyme complementarity,
dynamics, and electrostatics during the catalytic cycle. When
considering fine-tuning of substrate selectivity, such as enantio-
and/or regioselectivity in transformation of chiral substrates,
the issue becomes even more challenging because of the
relatively small differences in energy barriers. Such systems are
nontrivial to describe with confidence using current theoretical
methods38 but may be approached by extensive sampling
methods.51 Hence, experimental data are dearly needed to
improve our present theoretical models of stereoselectivity in
these enzymes.
Naturally occurring substitutions found in the other
corresponding mutated sites are primarily conservative and
consist of nonpolar residues. One exception is V141 that in the
majority of these related proteins is substituted with Pro.
However, V141 is located in a loop preceding an α-helix in the
lid region of the protein structure,28 and a Pro replacement may
7633
dx.doi.org/10.1021/bi3007725 | Biochemistry 2012, 51, 7627−7637

Biochemistry
Article
therefore also, by virtue of the nonpolar nature of this residue,
be considered as conservative.
enzymes, may shift the equilibria toward conformations
allowing functionalities different from that of the original wild
type. Hence, the described StEH1 variants may accordingly
display distinct conformer distributions compared to the wild
type, reflected in their acquired functional properties. The R-C1
variant has accumulated two mutations, V141K and I155V.
This mutant displays special regioselectivities with the S- and
S,S-enantiomers of styrene oxide and methylstyrene oxide,
respectively (Figure 4). This feature was a nonselected for
property, silent during the selection process that targeted
improved formation of R-4, but illustrates that also minor
structural changes can result in substantial effects. In the
second- and third-generation mutants originating from R-C1,
this unexpected behavior reverted back to the initial selectivity,
concomitant with a shift in regiopreference from C-2 to C-1 in
the ring-opening reaction with S-1. Therefore, it appears that
intrinsic preexisting potentials of different regiopreferences and
catalytic efficiencies are disclosed by mutagenesis and/or the
presence of a particular substrate, e.g., styrene oxide, in this
case.
Regio- and Enantioselectivities Affecting Product
Ratios. Regioselectivity in the StEH1-catalyzed hydrolysis of
phenyl-substituted epoxides is apparently plastic32,33 and is
expected to depend on structural constraints for the formation
of a productive ES complex promoting nucleophilic attack at a
reactive carbon. In cases where the enzyme displays
promiscuity in regiopreference, two diol products will be
formed at ratios determined by the flux through the different
reaction pathways (Scheme 2).
Scheme 2. Model of StEH1-Catalyzed Hydrolysis of
Epoxides
Because of the lack of detailed structural data for the isolated
mutants, attempts to link the observed regiopreferences of
these variants with the structural alterations would become
speculative. The difference in energy barriers between attack at
C-1 or C-2 can be calculated from the relative ratios of reacting
epoxide carbon [ΔΔG = −RT ln(C-1/C-2)]. In the case of
wild-type StEH1, displaying 93% attack at C-2 of S-1 (Table
SI3 of the Supporting Information), the energy difference is
+6.5 kJ/mol. In mutant R-C1B1, the same difference is only −1
kJ/mol, so the introduced mutations account for a total change
of <8 kJ/mol in changing the regiopreference from C-2 to C-1.
The structural changes that contribute to the major shift are
W106L and L109Y. W106 fulfills dual roles in the catalysis by
StEH1: it contributes via the main chain amide to the oxyanion
hole stabilizing the tetrahedral intermediate that is formed
during the hydrolytic half-reaction of catalysis, and through its
indole side chain to substrate interactions. L109, situated one
α-helix turn from W106 (Figure 1), likewise contributes its side
chain for substrate interactions. Via comparison of the
regiopreferences of mutants R-C1B1 and S-B17, it is note-
worthy that retaining an aromatic residue at position 106 also
preserves the regioselectivity of the wild-type enzyme in the
reaction with S-1.
The shift in regiopreferences from C-2 to C-1 for mutant R-
C1B1 (and R-C1B1D33) reverts in the next-generation E
library variants, R-C1B1D33E3 and -E6 (Figure 4 and Table
SI3 of the Supporting Information). Both these variants are
single mutants, L266F and L266G, respectively, of the parent.
In the wild-type enzyme, the side chain C-δ atom of L266 is 3.8
Å from both the phenolic hydroxyl of Y154 and the ε-2 carbon
of the F189 phenyl ring. In variant R-C1B1D33 and its
offspring, F189 is replaced with Leu, introducing more space in
this region of the active site. With this in mind, the observed
effects on the regiopreferences in mutants R-C1B1D33E3 and
-E6 are puzzling; in the L266F variant, the introduction of the
benzyl substituent is expected to create a clash with the side
chain of Y154 and cause local rearrangements of these residues.
Such rearrangements may affect the hydrogen bonding with the
substrate epoxide oxygen and/or the stabilization of the alkyl−
enzyme intermediate causing a shift in the regiopreference. In
the E6 mutant, however, L266 is replaced with Gly, thus
removing all side chain bulk. Possibly, also an increased
accessible volume in the active-site cavity may result in local
The validity of the model in Scheme 2 has been
demonstrated previously.33 The model accounts for all cases
of varying degrees of regioselectivity and also explains the
hysteretic behavior that is frequently observed with this
enzyme, such as the sigmoid substrate concentration depend-
ence of initial velocities. Several variants isolated in this study
displayed tendencies of cooperative substrate dependency,
although not to the extent that addition of extra parameters to
the Michaelis−Menten equation was significantly justified. This
observation, however, emphasizes that in a complex kinetic
mechanism, such as this, mutations introduced when moving
through evolutionary trajectories should be expected to shift
the flux through the different branches of the pathway and
thereby also affect the product ratios.
For several enzyme variants in this study, showing improved
preferences for formation of either of the products S- or R-4,
only modest increases in E values were obtained. This was
puzzling, but the explanation could be traced to alterations in
regioselectivity. Consequently, if there is a change in
regioselectivity, or a tendency of promiscuity, considering
strictly the E values of an enzyme variant would not reveal the
full story. This was exemplified by the R-4 selected variants R-
C1B1, R-C1B1-K141V, and R-C1B1D33 with a significant
change, from attack at C-2, for wild-type StEH1, to a preferred
attack at C-1, with epoxide S-1 (Figure 4). This, in combination
with an unaltered (strict) regioselectivity toward C-2 for R-1,
led to a higher percentage of R-4 than otherwise expected from
the E values alone. The resulting enantioconvergence is a
desired property because the limiting yield of enantiopure
product of 50% from a kinetic resolution of a racemic substrate
can be exceeded. This phenomenon of enantioconvergence has
previously been successfully demonstrated for an epoxide
hydrolase from A. niger, using a related substrate.53
In the formulation of a possible mechanism for the observed
effects on enantio- and regioselectivities of the evolved StEH1
mutants, the concept of structural dynamism54 is fitting. This
model assumes the existence of an ensemble of protein
conformations with a given equilibrium distribution where
different conformers possess distinct functional properties, as is
also suggested by Scheme 2. Mutations in the protein and/or
subjection to nonphysiological substrates, in the case of
7634
dx.doi.org/10.1021/bi3007725 | Biochemistry 2012, 51, 7627−7637

Biochemistry
Article
rearrangements of the catalytic Y154 with similar end results in
the observed regiopreferences. In a recent theoretical study of
stereoselectivity of the murine soluble epoxide hydrolase, the
importance of hydrogen bonding between the active-site
tyrosines and the epoxide oxygen was emphasized.38
Improved enantioselectivity of an evolved enzyme may arrive
at the cost of a decrease in catalytic activity, and a high
enantiomeric excess might not be practically significant if it
sacrifices high catalytic efficiency. It is therefore important to
also include information describing, e.g., product formation,
which has been done in Figure 5. Upon comparison of the
properties with the structural alterations, the functional data
presented here suggest that the amino acid residue positioned
C-terminally and adjacent to the nucleophile influences
stereoselectivity in these reactions.
ASSOCIATED CONTENT
* Supporting Information
■
S
Tables and figures with details of results from NMR analysis,
sequence alignments, library constructions and characterization,
and enzyme kinetics. This material is available free of charge via
the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*E-mail: mikael.widersten@kemi.uu.se. Phone: +46 (0)18 471
4992.
■
Funding
The work was supported by the Swedish Research Council
(Grants VR 621-2008-3579 and 621-2011-6055).
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Prof. Adolf Gogoll for assistance with the NMR
analysis, Dr. M. Stieger (F. Hoffmann-La Roche, Basel,
Switzerland) for providing the pREP4 GroEL/ES plasmid,
and Dr. Lynn Kamerlin for manuscript review.
Figure 5. Evolution of conversion of rac-1 into either S-4 (left) or R-4
(right) as a function of R-4:S-4 product ratios during ISM. rac-1 (30.4
mM) was incubated with 0.5 μM StEH1 variant enzymes for 1 h.
REFERENCES
■
resulting diol production from racemic epoxide mixtures with
product outcomes calculated from kinetic data of reactions with
pure enantiomers (see Table SI3 of the Supporting
Information), certain differences are observed. The kinetics of
racemate hydrolysis is more complex, and simulations taking
into account only the kinetic parameters for each of the pure
enantiomers therefore result in misleading results. This is
probably due to trivial issues of one enantiomer acting as an
inhibitor of the other and/or product inhibition. For the
reactions shown in Figure 5, the maximal conversion is 50% if
strict regioselectivity is assumed. This means that in the
reactions with wild-type StEH1 and the S-B17 variant, all
substrates have been consumed. It should also be pointed out
that it is possible to regain enzyme activity while moving along
the evolutionary trajectory, as indicated by the R variants where
R-C1B1D33E6 displays 25% conversion per hour.
(1) Bornschauer, U. T., Huisman, G. W., Kazlauskas, R. J., Lutz, S.,
Moore, J. C., and Robins, K. (2012) Engineering the third wave of
biocatalysis. Nature 485, 185−194.
(2) Chen, K., and Arnold, F. H. (1991) Enzyme engineering for
nonaqueous solvents: Random mutagenesis to enhance activity of
subtilisin E in polar organic media. Nat. Biotechnol. 9, 1073−1077.
(3) Chen, K., Robinson, A. C., Van Dam, M. E., Martinez, P.,
Economou, C., and Arnold, F. H. (1991) Enzyme engineering for
nonaqueous solvents. II. Additive effects of mutations on the stability
and activity of subtilisin E in polar organic media. Biotechnol. Prog. 7,
125−129.
(4) Chartrain, M., Salmon, P. M., Robinson, D. K., and Buckland, B.
C. (2000) Metabolic engineering and directed evolution for the
production of pharmaceuticals. Curr. Opin. Biotechnol. 11, 209−214.
(5) Hult, K., and Berglund, P. (2003) Engineered enzymes for
improved organic synthesis. Curr. Opin. Biotechnol. 14, 395−400.
(6) Reetz, M. T., Bocola, M., Carballeira, J. D., Zha, D., and Vogel, A.
(2005) Expanding the range of substrate acceptance of enzymes:
Combinatorial active-site saturation test. Angew. Chem., Int. Ed. 44,
4192−4196.
(7) Turner, N. J. (2009) Directed evolution drives the next
generation of biocatalysts. Nat. Chem. Biol. 5, 567−573.
(8) Clouthier, C. M., and Pelletier, J. N. (2012) Expanding the
organic toolbox: A guide to integrating biocatalysis in synthesis. Chem.
Soc. Rev. 41, 1585−1605.
(9) Harris, J. L., and Craik, C. S. (1998) Engineering enzyme
specificity. Curr. Opin. Chem. Biol. 2, 127−l32.
CONCLUSIONS
■
The iterative saturation mutagenesis approach as applied to
StEH1 allowed isolation of new epoxide hydrolase variants
producing enriched fractions of diol R-4 or, albeit to a lesser
extent, diol S-4 from a racemic epoxide starting material. The
increased product purities were due to increased E values for
the corresponding substrate enantiomer(s) or in combination
with a change in the regiopreference of the initial enzyme
afforded epoxide ring opening. A number of variants also
displayed altered catalytic properties with epoxide 2 or 3 as
exemplified by R-C1B1D33, showing an E value for S-2 that
increased from 69 to 5800. Thus, the libraries that we
constructed showed potential of being applicable for a broader
substrate range than was initially screened for. Although
detailed structural data of the isolated mutants, supported by
theoretical studies, are needed to link the observed functional
(10) Bloom, J. D., Meyer, M. M., Meinhold, P., Otey, C. R.,
MacMillan, D., and Arnold, F. H. (2005) Evolving strategies for
enzyme engineering. Curr. Opin. Struct. Biol. 15, 447−452.
(11) Chica, R. A., Doucet, N., and Pelletier, J. N. (2005) Semi-
rational approaches to engineering enzyme activity: Combining the
benefits of directed evolution and rational design. Curr. Opin.
Biotechnol. 16, 378−384.
(12) Kaur, J., and Sharma, R. (2006) Directed evolution: An
approach to engineer enzymes. Crit. Rev. Biotechnol. 26, 165−199.
7635
dx.doi.org/10.1021/bi3007725 | Biochemistry 2012, 51, 7627−7637

Biochemistry
Article
(13) Kirschner, A., and Bornscheuer, U. T. (2008) Directed
evolution of a Baeyer-Villiger monooxygenase to enhance enantiose-
lectivity. Appl. Microbiol. Biotechnol. 81, 465−472.
by the Solanum tuberosum epoxide hydrolase. Tetrahedron: Asymmetry
15, 2801−2805.
(33) Lindberg, D., de la Fuente Revenga, M., and Widersten, M.
(2010) Temperature and pH dependence of enzyme catalyzed
hydrolysis of trans-methylstyrene oxide: A unifying kinetic model for
observed hysteresis, cooperativity and regioselectivity. Biochemistry 49,
2297−2304.
(14) Larrow, J. F., and Jacobsen, E. N. (2004) Assymetric processes
catalyzed by chiral (salen) metal complexes. Top. Organomet. Chem. 6,
123−152.
(15) Lacourciere, G. M., and Armstrong, R. N. (1993) The catalytic
mechanism of microsomal epoxide hydrolase involves an ester
intermediate. J. Am. Chem. Soc. 115, 10466−10467.
(16) Borhan, B., Jones, A. D., Pinot, F., Grant, D. F., Kurth, M. J., and
Hammock, B. D. (1995) Mechanism of soluble epoxide hydrolase.
Formation of an α-hydroxy ester-enzyme intermediate through Asp-
333. J. Biol. Chem. 270, 26923−26930.
(17) Pinot, F., Grant, D. F., Beetham, J. K., Parker, A. G., Borhan, B.,
Landt, S., Jones, A. D., and Hammock, B. D. (1995) Molecular and
biochemical evidence for the involvement of the Asp-333-His-523 pair
in the catalytic mechanism of soluble epoxide hydrolase. J. Biol. Chem.
270, 7968−7974.
(18) Tzeng, H.-F., Laughlin, T. L., Lin, S., and Armstrong, R. N.
(1996) The catalytic mechanism of microsomal epoxide hydrolase
involves reversible formation and rate-limiting hydrolysis of the alkyl−
enzyme intermediate. J. Am. Chem. Soc. 118, 9436−9437.
(19) Arand, M., Cronin, A., Oesch, F., Mowbray, S. L., and Jones, T.
A. (2003) The telltale structures of epoxide hydrolases. Drug Metab.
Rev. 35, 365−383.
(20) Arand, M., Cronin, A., Adamska, M., and Oesch, F. (2005)
Epoxide hydrolases: Structure, function, mechanism, and assay.
Methods Enzymol. 400, 569−588.
(21) Morisseau, C., and Hammock, B. D. (2005) Epoxide hydrolases:
Mechanisms, inhibitor designs, and biological roles. Annu. Rev.
Pharmacol. Toxicol. 45, 311−333.
(22) Widersten, M., Gurell, A., and Lindberg, D. (2010) Structure-
function relationships of epoxide hydrolases and their potential use in
biocatalysis. Biochim. Biophys. Acta 1800, 316−326.
(34) Thomaeus, A., Carlsson, J., Åqvist, J., and Widersten, M. (2007)
Active site of epoxide hydrolases revisited: A noncanonical residue in
potato StEH1 promotes both formation and breakdown of the
alkylenzyme intermediate. Biochemistry 46, 2466−2479.
(35) Lau, E. Y., Newby, Z. E., and Bruice, T. C. (2001) A theoretical
examination of the acid-catalyzed and noncatalyzed ring-opening
reaction of an oxirane by nucleophilic addition of acetate. Implications
to epoxide hydrolases. J. Am. Chem. Soc. 123, 3350−3357.
(36) Schiøtt, B., and Bruice, T. C. (2002) Reaction mechanism of
soluble epoxide hydrolase: Insights from molecular dynamics
simulations. J. Am. Chem. Soc. 124, 14558−14570.
(37) Hopmann, K. H., and Himo, F. (2006) Theoretical study of the
full reaction mechanism of human soluble epoxide hydrolase.
Chemistry 12, 6898−6909.
(38) Lonsdale, R., Hoyle, S., Grey, D. T., Ridder, L., and Mulholland,
A. D. (2012) Determinants of reactivity and selectivity in soluble
epoxide hydrolases from quantum mechanics/molecular mechanics
modeling. Biochemistry 51, 1774−1786.
(39) Reetz, M. T., and Carballeira, J. D. (2007) Iterative saturation
mutagenesis (ISM) for rapid directed evolution of functional enzymes.
Nat. Protoc. 2, 891−903.
(40) Reetz, M. T., Prasad, S., Carballeira., J. D., Gumulya, Y., and
Bocola, M. (2010) Iterative saturation mutagenesis accelerates
laboratory evolution of enzyme stereoselectivity: Rigorous comparison
with traditional methods. J. Am. Chem. Soc. 132, 9144−9152.
(41) Reetz, M. T., Carballeira, J. D., and Vogel, A. (2006) Iterative
saturation mutagenesis on the basis of B factors as a strategy for
increasing protein thermostability. Angew. Chem., Int. Ed. 45, 7745−
7751.
(42) Bougioukou, D. J., Kille, S., Taglieber, A., and Reetz, M. T.
(2009) Directed evolution of an enantioselective enoate-reductase:
Testing the utility of iterative saturation mutagenesis. Adv. Synth. Catal.
351, 3287−3305.
(23) Rink, R., Kingma, J., Lutje Spelberg, J. H., and Janssen, D. B.
(2000) Tyrosine residues serve as proton donor in the catalytic
mechanism of epoxide hydrolase from Agrobacterium radiobacter.
Biochemistry 39, 5600−5613.
(24) Elfstrom, L. T., and Widersten, M. (2006) Implications for an
̈
ionized alkyl-enzyme intermediate during StEH1-catalyzed trans-
stilbene oxide hydrolysis. Biochemistry 45, 205−212.
(43) Zheng, H., and Reetz, M. T. (2010) Manipulating the
stereoselectivity of limonene epoxide hydrolase by directed evolution
based on iterative saturation mutagenesis. J. Am. Chem. Soc. 132,
15744−15751.
(25) Stapleton, A., Beetham, J. K., Pinot, F., Garbarino, J. E.,
Rockhold, D. R., Friedman, M., Hammock, B. D., and Belknap, W. R.
(1994) Cloning and expression of soluble epoxide hydrolase from
potato. Plant J. 6, 251−258.
́
(44) Engstrom, K., Nyhlen, J., Sandstrom, A. G., and Backvall, J.-E.
̈ ̈
̈
(2010) Directed evolution of an enantioselective lipase with broad
substrate scope for hydrolysis of α-substituted esters. J. Am. Chem. Soc.
132, 7038−7042.
(26) Elfstrom, L. T., and Widersten, M. (2005) Catalysis of potato
̈
epoxide hydrolase, StEH1. Biochem. J. 390, 633−640.
(27) Gurell, A., and Widersten, M. (2010) Modification of substrate
specificity resulting in an epoxide hydrolase with shifted enantioprefer-
ence for (2,3-epoxypropyl)benzene. ChemBioChem 11, 1422−1429.
(45) Prasad, S., Bocola, M., and Reetz, M. T. (2011) Revisiting the
lipase from Pseudomonas aeruginosa: Directed evolution of substrate
acceptance and enantioselectivity using iterative saturation muta-
genesis. ChemPhysChem 12, 1550−1557.
(28) Mowbray, S. L., Elfstrom, L. T., Ahlgren, K. M., Andersson, C.
̈
E., and Widersten, M. (2006) X-ray structure of potato epoxide
hydrolase sheds light on substrate specificity in plant enzymes. Protein
Sci. 15, 1628−1637.
(46) Reetz, M. T., Kahakeaw, D., and Lohmer, R. (2008) Addressing
the numbers problem in directed evolution. ChemBioChem 9, 1797−
1804.
(29) Reetz, M. T., Wang, L.-W., and Bocola, M. (2006) Directed
evolution of enantioselective enzymes: Iterative cycles of CASTing for
probing protein-sequence space. Angew. Chem., Int. Ed. 45, 1236−
1241.
(30) Reetz, M. T., Bocola, M., Wang, L.-W., Sanchis, J., Cronin, A.,
Arand, M., Zou, J., Archelas, A., Bottalla, A.-L., Naworyta, A., and
Mowbray, S. L. (2009) Directed evolution of an enantioselective
epoxide hydrolase: Uncovering the source of enantioselectivity at each
evolutionary stage. J. Am. Chem. Soc. 131, 7334−7343.
(31) Parker, R. E., and Isaacs, N. S. (1959) Mechanisms of epoxide
reactions. Chem. Rev. 59, 737−799.
(32) Monterde, M. I., Lombard, M., Archelas, A., Cronin, A., Arand,
M., and Furstoss, R. (2004) Enzymatic transformations. Part 58:
Enantioconvergent biohydrolysis of styrene oxide derivatives catalysed
(47) Dale, G. E., Schonfeld, H. J., Langen, H., and Stieger, M. (1994)
̈
Increased solubility of trimethoprim-resistant type S1 DHFR from
Staphylococcus aureus in Escherichia coli cells overproducing the
chaperonins GroEL and GroES. Protein Eng. 7, 925−931.
(48) Agarwal, S. C., Van Duuren, B. L., and Kneip, T. J. (1979)
Detection of epoxides with 4-(p-nitrobenzyl) pyridine. Bull. Environ.
Contam. Toxicol. 23, 825−829.
(49) Altschul, S. F., Gish, W., Miller, W., Myers, E. W., and Lipman,
D. J. (1990) Basic local alignment search tool. J. Mol. Biol. 215, 403−
410.
(50) Thompson, J. D., Higgins, D. G., and Gibson, T. J. (1994)
CLUSTAL W: Improving the sensitivity of progressive multiple
sequence alignment through sequence weighting, position specific gap
penalties and weight matrix choice. Nucleic Acids Res. 22, 4673−4680.
7636
dx.doi.org/10.1021/bi3007725 | Biochemistry 2012, 51, 7627−7637

Biochemistry
Article
(51) Frushicheva, M. P., and Warshel, A. (2012) Towards
quantitative computer-aided studies of enzymatic enantioselectivity:
The case of Candida antarctica lipase A. ChemBioChem 13, 215−223.
(52) Van Loo, B., Kingma, J., Heyman, G., Wittenaar, A., Lutje
Spelberg, J. H., Sonke, T., and Janssen, D. B. (2009) Improved
enantioselective conversion of styrene epoxides and meso-epoxides
through epoxide hydrolases with a mutated nucleophile-flanking
residue. Enzyme Microb. Technol. 44, 145−153.
(53) Zheng, H., Kahakeaw, D., Acevedo, J. P., and Reetz, M. T.
(2010) Directed evolution of enantioconvergency: The case of an
epoxide hydrolase-catalyzed reaction of a racemic epoxide. Chem-
CatChem 2, 958−961.
(54) Tokuriki, N., and Tawfik, D. S. (2009) Protein dynamism and
evolvability. Science 324, 203−207.
7637
dx.doi.org/10.1021/bi3007725 | Biochemistry 2012, 51, 7627−7637