Modification of substrate specificity resulting in an epoxide hydrolase with shifted enantiopreference for (2,3-Epoxypropyl)benzene

Article abstract of DOI:10.1002/cbic.201000185

Random mutagenesis targeted at hotspots of noncatalytic active-site residues of potato epoxide hydrolase StEH1 combined with an enzyme-activity screen allowed the isolation of enzyme variants displaying altered enantiopreference in the catalyzed hydrolysi

Full text of DOI:10.1002/cbic.201000185

DOI: 10.1002/cbic.201000185
Modification of Substrate Specificity Resulting in an
Epoxide Hydrolase with Shifted Enantiopreference for
(
2,3-Epoxypropyl)benzene
[a]
Ann Gurell and Mikael Widersten*
Random mutagenesis targeted at hotspots of noncatalytic
active-site residues of potato epoxide hydrolase StEH1 com-
bined with an enzyme-activity screen allowed the isolation of
enzyme variants displaying altered enantiopreference in the
catalyzed hydrolysis of (2,3-epoxypropyl)benzene. The wild-
type enzyme favored the S enantiomer with a ratio of 2.5:1,
whereas the variant displaying the most radical functional
changes showed a 15:1 preference for the R enantiomer. This
mutant had accumulated four substitutions distributed over
two out of four mutated hotspots: W106L, L109Y, V141K, and
I151V. The underlying causes of the enantioselectivity were a
decreased catalytic efficiency in the catalyzed hydrolysis of the
S enantiomer combined with retained activity with the R enan-
tiomer. The results demonstrate the feasibility of molding the
stereoselectivity of this biocatalytically relevant enzyme.
Introduction
Environmentally friendly alternatives in the production of fine
chemicals and the synthesis of pharmaceuticals are in high
demand, which is likely to continue to grow further. As the
public’s ecological awareness increases, the possibility to de-
velop biological catalysts for the production of useful chemi-
cals has become a task of utmost importance for the scientific
community. The environmental requirements made on these
catalysts have on several occasions proved to be fulfilled by
into four “hotspots” for directed mutagenesis. Each hotspot
consisted of one to three amino acid residues (Figure 1).
[
1,2]
the use of enzymes.
Enzymes are already used as biocatalysts in industry, but
demand in relation both to substrate specificity and to stereo-
specificity is high, therefore further development of applicable
[
1,3–5]
enzymes is needed.
Soluble epoxide hydrolases
(
EC 3.3.2.10) are cofactor-independent enzymes and members
[
6]
of the a/b-hydrolase fold superfamily. In some cases they
show high catalytic activity as well as high enantioselectivity,
thus making these enzymes attractive for further development
into biocatalysts possessing desired substrate specificity and
[
7–11]
catalytic efficiency.
Epoxide hydrolases catalyze the conversion of epoxides into
the corresponding vicinal diols through the formation of a
covalent alkylenzyme intermediate between an active-site Asp
carboxylate and the substrate, which is subsequently hydro-
Figure 1. Mutated “hotspots” in the active site of potato epoxide hydrolase
StEH1. A) Phe33 (magenta), B) Trp106 and Leu109 (yellow), C) Val141,
Leu145, and Ile155 (green) and D) Ile180 and Phe189 (blue). The catalytic
residues Asp105 (nucleophile), His-300 (general base), and Tyr154 and
Tyr235 (Lewis acids) are shown in white stick representation. Image created
[
11]
lyzed in a general base-assisted reaction. The work present-
ed here concerns the epoxide hydrolase from potato: Solanum
tuberosum soluble epoxide hydrolase 1 (StEH1). The intention
of the study was to assess the feasibility of molding epoxide
substrate selectivity, including enantiopreference. The ap-
proach relied on mutagenesis of active-site residues of StEH1
that line the active site pocket but do not have any direct cata-
lytic function. Targeting amino acid residues close to the active
site has proven more effective than targeting distant ones
[21]
with PyMol (http://www.pymol.org) from the atomic coordinates in 2cjp.
Restricting the mutagenesis to these hotspots reduces the
protein sequence space that needs to be searched, which
leads to library sizes approachable by screening. Furthermore,
[
a] A. Gurell, Prof. M. Widersten
Department of Biochemistry and Organic Chemistry, Uppsala University
Box 576, 75123 Uppsala (Sweden)
Fax: (+46)18-55-8431
[12]
when aiming for improved substrate- and enantioselectivity.
A set of such residues were identified, categorized, and divided
E-mail: mikael.widersten@biorg.uu.se
1
422
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1422 – 1429

Epoxide Hydrolase with Shifted Enantiopreference
by applying a limited codon degeneracy (NDT at positions 33,
pocket of StEH1 that participate in determining substrate se-
lectivity. Residues confirmed from other studies to be impor-
tant for retention of catalytic function were excluded because
a main priority was to isolate catalytically active enzyme var-
106, 109, 155, 180, and 189 and NHA at positions 141 and 145)
the genetic amino acid alphabet was reduced to 12 possible
codons at each mutated position, further reducing the re-
quired screening efforts. In this limited assortment of side-
chain functionalities, amino acid residues with aromatic, ali-
phatic, uncharged polar, positively charged, and negatively
charged properties were still represented. The cDNAs encoding
enzyme variants judged as displaying the highest hydrolytic
activity towards the substrate epoxide in the first round of
libraries, were subsequently used as starting points for new
libraries revisiting the same hotspots. The assumption made
was that this combinatorial approach would lead both to co-
operative and to additive positive effects in enzyme variants
maintained in the second-generation libraries by applying the
iterative saturation mutagenesis (ISM) approach as described
[22–24]
iants.
Each of the four hotspots consisted of one to three residues.
Phe33 (library A) is positioned deep inside the active site
pocket in the wild-type enzyme. The phenyl ring is involved in
an edge-on interaction with the aromatic portion of one of the
catalytic Tyr residues (Tyr235). This interaction might be impor-
tant for maintaining correct positioning of this catalytically cru-
cial residue. In addition, the side chain of Phe33 participates in
forming the substrate binding cavity.
Trp106 (library B) is also involved in an aromatic edge-on
interaction with the same Tyr residue through its indole ring,
thereby sandwiching this catalytic Tyr between Trp106 and
Phe33 (Figure 1). The backbone amides of Phe33 and Trp106
also form the oxyanion hole that stabilizes the tetrahedral re-
action intermediate formed during the hydrolytic half-reaction
[
10,13–17]
by Reetz and co-workers.
The targeted reaction was the enzymatic hydrolysis of (2,3-
epoxypropyl)benzene (1; Scheme 1). This chiral epoxide has
been little studied as a potential epoxide hydrolase substrate,
[11,21,23,24]
of catalysis.
Therefore, in addition to contributing to
the boundaries of the active site cavity, these residues might
influence chemical catalysis. Leu109 (library B) is located one
turn of an a-helix away from Trp106, placing its nonpolar side
chain in van der Waals contact with the indole. The residues in
hotspot B line the entry pathway for the substrate.
The three residues in library C—Val141, Leu145, and Ile155—
are located roughly perpendicular to those mentioned above.
These residues also line the entry and exit pathways for the
epoxide and diol.
Scheme 1. Hydrolysis of (2,3-epoxypropyl)benzene (1) into either the R (2)
or the S (3) enantiomer of the resulting 3-phenylpropane-1,2 diol.
The residues in the final library D—Ile180 and Phe189—are
situated at the back of the active-site pocket (Figure 1).
although reports of whole-cell conversions have demonstrated
Screening strategy
that epoxide hydrolases from Rhodotorula species catalyze its
[
18,19]
hydrolysis with some efficiency.
The kinetics of the StEH1-
For the first round of library screening a racemic mixture of
epoxide 1 was used to probe functional StEH1 variants. The
rationale was to achieve primary identification of variant en-
zymes possessing sufficiently high catalytic activity regardless
of stereopreference.
catalyzed hydrolysis of epoxide 1 have not been established,
but the same enzyme has been shown to catalyze the hydroly-
sis of the trans isomer of 2-methylstyrene oxide with high effi-
[
20]
ciency and enantioselectivity. Initial studies in our group sug-
gested that 1 was also hydrolyzed, but with a low degree of
enantioselectivity. Hence, the aims of this study were both to
characterize the catalytic efficiency and enantioselectivity of
the native enzyme with 1 and to attempt to mold the modest
enantioselectivity towards higher levels of discrimination.
Suitable variants identified from the first screening round
were then used as templates for the subsequent round of
mutagenesis. In this second cycle, the same hotspots were re-
visited according to the scheme in Figure 2. The generated
second-generation enzyme variants were challenged for activi-
ty with the pure enantiomers of 1 to allow identification of
enantioselective enzyme forms with retained catalytic activity.
Results and Discussion
Choice of targeted residues
First-generation variants
The tertiary structure of unliganded wild-type StEH1 has been
529 clones from library A, 704 from library B, 3872 from library
C, and 616 from library D were screened. The enzyme variants
were scored as hits if high activity with rac-1 was established.
The hit frequencies of functional enzyme variants varied be-
tween 3 and 9%, depending on the library. The cDNAs from 70
picked clones were sequenced in full to deduce the distribu-
tion of mutations at each position of the amino acid sequen-
ces (Table 1).
[
21]
determined at 1.95 ꢁ resolution. The lack of crystal structure
information relating to enzyme–substrate interactions some-
what hampers the identification of active-site residues that
determine substrate selectivity. However, combining previous
knowledge of the catalytic mechanism with the available struc-
ture information makes it possible, with a relatively high
degree of certainty, to identify residues in the active-site
ChemBioChem 2010, 11, 1422 – 1429
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1423

M. Widersten et al.
tion 145 was also found to be selected for. Ile155, however,
was in all cases conservatively replaced by Val, retaining the
hydrophobic character but offering more space.
Residues 141 and 145 are located closest to the protein sur-
face and are therefore also more exposed to solvent through
the active site entrance. Neither of these residues interacts di-
rectly with any of the catalytic residues. These positions might
therefore be more permissive to nonconservative replacements
without loss of catalytic function. Ile155, on the other hand, is
sandwiched between the catalytic Tyr pair (Tyr154 and Tyr235)
and might stabilize these catalytic groups.
Figure 2. Outline of combinations of hotspots in constructed cDNA libraries.
Sites A) Phe33, B) Trp106, Leu109, C) Val141, Leu145, Ile151, and D) Ile180,
Phe189.
At position 180, residue replacements from library D variants
had conserved the nonpolar nature of the wild-type Ile by con-
taining variously Ile, Leu, or Val. In one tenth of the cases, a
Cys residue was incorporated at that position. At position 189,
the wild-type Phe residue was exchanged for other nonpolar
residues of less bulk (Table 1).
Hits from library A that were not wild-type “background”
clones had retained an aromatic residue at position 33 (50%
Phe, 25% Tyr) in 75% of all cases, whereas in the remaining
2
5% of the clones, the aromatic residue was replaced by Val.
First-generation variants’ enantioselectivities
In library B, the aromatic nature of the wild-type Trp106 was
retained by replacement with Phe in all analyzed clones. The
strong conservation of an aromatic side chain might be linked
to the previously mentioned aromatic edge-on interactions
made by Phe33, Trp106, and Tyr235 in the wild-type enzyme.
Leu109 was exchanged either for an Ile or for a Val, conserving
the hydrophobic characteristics in this position of the struc-
ture.
Eighteen of the first-round screening hits were expressed in
E. coli and purified by Ni-IMAC chromatography. Purified en-
zymes were challenged for hydrolysis activity in the presence
of rac-1, and the stereochemistry of the product diols was ana-
lyzed by chiral HPLC. The regiospecificity in the catalyzed reac-
tion has not been firmly established but is assumed to involve
nucleophilic attack and ring opening at the less hindered, achi-
ral, carbon 1. A diol product of S configuration (3 in Scheme 1)
should hence derive from the hydrolysis of (S)-1, and vice
versa for the R diol. It has been established, however, that all
enzymes described here only catalyze epoxide ring opening at
one carbon: that is, there is no detectable promiscuity in the
point of attack, with only one diol enantiomer being formed
from hydrolysis of a pure enantiomer of the epoxide 1 (data
not shown).
In library C, Val141 was in most cases changed into a polar,
positively charged Lys. The observation that the major propor-
tion of clones selected from the C-library contained this highly
polar residue at position 141 instead of the nonpolar alterna-
tives (Val, Leu, Ile) encoded in the NHA set suggests a favora-
ble influence on catalysis from Lys141. Other polar residue re-
placements were also found in this position (Table 1). In 50%
of cases the introduction of a polar residue (Thr) at posi-
Table 1. Sequence analysis of enzyme hits selected from screening for activity with 1.
Mutated position
library)
Identity and frequency of residue replacements in sequenced variants
(
A
C
D
E
F
G
H
I
K
L
N
P
Q
R
S
T
V
Y
1
st generation
Phe33 (A)
Trp106 (B)
Leu109 (B)
Val141 (C)
Leu145 (C)
Ile155 (C)
Ile180 (D)
Phe189 (D)
0.5
1.0
0.3
0.3
0.8
0.5
0.3
0.1
0.4
0.1
0.1
0.3
0.5
1.0
0.1
0.4
0.1
0.2
0.5
0.2
0.4
0.2
2
nd generation
Phe33 (A)
Trp106 (B)
Leu109 (B)
Val141 (C)
Leu145 (C)
Ile155 (C)
Ile180 (D)
Phe189 (D)
0.1
0.5
0.6
0.1
0.1
0.3
0.1
0.6
0.1
0.4
0.2
0.4
0.1
0.1
0.6
0.3
0.3
0.3
0.3
0.2
0.5
1
424
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1422 – 1429

Epoxide Hydrolase with Shifted Enantiopreference
The product ratio for the wild-type StEH1 was 2:1 in favor of
individually and in parallel. The library B enzyme variant
[W106F, L109I] contributed the starting cDNA for a library com-
bining these mutations with NDT at codon 33 (library BA).
From this library, 528 variants were screened. Variant [W106F,
L109I] was also combined with new mutations at posi-
tions 141, 145, and 155 to form a new library (library BC, 704
variants screened). The variant [V141K, I155V] was combined
with new mutations at position 33 (library CA, 352 variants
screened) or with additional mutations at positions 106 and
3
3
, resulting in an enantiomeric excess of the S diol product of
4%. The large majority of all tested first-generation variants
displayed similar enantiomeric preference for (S)-1 (Figure 3).
109 (library CB, 880 screened variants). The first-round variant
[
F189V] was combined with mutations at all other hotspots,
generating libraries DA, DB, and DC (528, 704 and 704 variants
were screened, respectively).
The hit frequencies from these libraries with (S)-1 ranged
from 2–10%. The corresponding hit frequencies when
screened with the R epoxide varied from 1–5%. The clones
were chosen so as to represent an assortment of enzyme var-
iants putatively active with either enantiomer. cDNAs derived
from 80 catalytically active clones were sequenced. The results
revealed that the overall quality of the second-generation li-
braries was poorer, with the majority of active clones being
wild-type enzyme. For instance, the D-derived libraries, origi-
nating from the first-generation variant [F189V] and combined
with hotspots A–C only resulted in three new variants (out of
43 sequenced clones). From the second recombination cycle a
total of 22 new variants were identified. A summary of the resi-
due distribution is shown in Table 1
At position 33, 50% of the isolated mutants had retained
the wild-type Phe. In the remaining variants a polar residue
had been introduced. At position 106 (Trp in wild-type StEH1)
either a Phe or a Leu was found. This is a change from the iso-
lated first-generation variants, which in all cases contained a
Phe at this position. This Trp106!Leu replacement removes
the aromatic indole side chain and would be expected to in-
troduce space at this point of the active site. This substitution
would also make an interaction with Tyr235 involving aromatic
rings impossible. Leu109 was exchanged in the second-genera-
tion variants either for Tyr or for Asn. Both these replacements
are noteworthy: introduction of a Tyr inserts an aromatic side
chain with hydrogen-bonding capacity one helical turn away
from the residue at position 106. Examination of the sequence
of individual variants reveals that in the two (identical) variants
containing Leu106, position 109 was occupied by a Tyr. This
could reflect compensatory mutations in which the aromatic
Tyr109 substitutes for the loss of Trp106, although direct inter-
action with Tyr235 would be impossible without major rear-
rangements of the structure. The anticipated effect of these
two mutations on substrate-binding properties would be in-
creased available space in the vicinity of the Asp nucleophile
Figure 3. Product outcome after catalyzed hydrolysis of rac-1 by StEH1 var-
iant enzymes. Bars represent enantiomeric excess of S diol 3 (%) in the prod-
uct mixtures: ee
dicate additional mutations introduced during PCR.
p
=([3ꢀ2])/([3]+[2])ꢂ100. Italicized residue replacements in-
The enzyme variant [W106F, L109I] showed an increased pref-
erence for the S epoxide relative to the wild-type StEH1, result-
ing in a product ratio of 4:1 in favor of the S diol. The variants
[
F189V] and [V141K, I155V] (Figure 3) both exhibited increased
tendencies to yield the R diol (2). These variants—[F189V],
V141K, I155V], and [W106F, L109I]—with distinctly altered
[
characteristics with regard to stereoselectivity were chosen to
parent the next round of mutagenesis.
Second-generation variants
(
Asp105). The Leu109!Asn replacement introduces a highly
The cDNAs encoding the first-round variants [W106F, L109I],
polar amide into this otherwise totally nonpolar micro-environ-
ment. Polar residues at positions 141 and 145, in contrast to
the nonpolar wild-type Val141 and Leu145, also appear to be
preferred in the second-round hits. The hydrophobic character
of Ile155, however, was retained, similarly to the first-genera-
tion variants.
[V141K, I155V], and [F189V] were used to template the second
round of mutagenesis. In order to minimize the screening
efforts, not all permutations of hotspots were visited in this
second construction of cDNA libraries.
The second-generation enzyme variants (Figure 2) were
screened for hydrolytic activity with the pure enantiomers of 1,
ChemBioChem 2010, 11, 1422 – 1429
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1425

M. Widersten et al.
Second-generation variants’ enantioselectivities
Enantioselectivity is the ratio of the catalytic efficiencies for
either substrate enantiomer. In the analysis here these values
clearly illustrate a radical shift in stereospecificity from the
wild-type enzyme, via the double mutant, to the quadruple
mutant. The E value for (R)-1 increases from 0.4!1.1!15, an
almost 40-fold difference between the wild-type StEH1 and the
[W106L, L109Y, V141K, I155V] variant.
Eight selected mutants were purified and analyzed for enantio-
selectivity with rac-1 as substrate. Variant [V141Q, I155L], from
the DC library, showed a modest preference for the (S)-1 enan-
tiomer (1.2-fold). The variant [V141K, I155V], from the CA li-
brary, did not distinguish between enantiomers significantly.
Interestingly, at the protein level, this mutant is identical to the
first-generation mutant [V141K, I155V] that had been used as
starting point in the generation of the CA library. The variant
selected from the CA library, however, was a distinct clone as
could be seen from the codon usage (TTT rather than the wild-
type TTC).
Conclusions
The main conclusion we can draw from these results is that
StEH1 is a promising candidate biocatalyst amenable to re-
design of stereoselectivity by mutagenesis in the active-site
cavity.
The library CB variant [W106L, L109Y, V141K, I155V], which
differs from [V141K, I155L] only in residues 106 and 109, dis-
played new stereoselective properties. This mutant displayed a
The intial aim of this work was to improve on the poor
enantioselectivity displayed by the wild-type StEH1 enzyme.
The first round of activity screening, however, was performed
in the presence of both epoxide enantiomers and was mainly
intended to target high enzyme activity rather than selectivity.
This approach might have limited the success rate in finding
variant enzymes with drastically improved selectivity for (S)-1
but did, on the other hand, allow for selection and isolation of
other variants that had retained their activity with the R enan-
tiomer, thereby widening the scope of potentially interesting
new catalysts. The epoxide hydrolase from Aspergillus niger has
previously been subjected to redesign of its enantiospecificity
5.7-fold preference for the R enantiomer of 1, which suggested
a radical shift in enantiopreference from (S)-1 in the wild-type
enzyme to (R)-1 in this mutant. The remaining purified variants
displayed poor activities and/or purification yields and were
not analyzed further.
Steady-state kinetics
The kinetic properties of the wild-type StEH1, together with
those of the variants [V141K, I155V] and [W106L, L109Y, V141K,
I155V], were determined in catalyzed hydrolysis reactions with
either (R)-1 or (S)-1. The steady-state parameters are presented
in Table 2.
[25]
by an approach similar to that described here for StEH1. In
that work, the modest enantioselectivity displayed by the wild-
type enzyme in favor of the S enantiomer of phenylglycidol
ether was enhanced 64-fold, from an initial E value of 3 to 193
in the most selective mutant. The variant exhibiting this re-
markable enantioselectivity had accumulated nine residue re-
placements relative to the parent enzyme, as a result of five
cycles of hotspot recombinations. In the work presented here
only two mutagenesis cycles were performed but a drastic
effect (38-fold) on enantioselectivity could still be achieved.
The kinetic data reveal that the basis for the evolved prefer-
The initial studies of enantiopreference were determined at
high concentrations of epoxide substrate, so these compari-
sons would be expected primarily to be comparisons of the
kcat efficiencies of the enzyme variants. A comparison of the
determined values of kcat exhibited by these three different
enzyme forms with the product ratios obtained from the race-
mic substrate illustrates that the wild-type enzyme displays a
S
R
k /k ratio of 3.3. This value is consistent with the obtained S
ence for (R)-1 is not increased catalytic efficiency; the k /KM
cat cat
cat
diol/R diol product ratio (Figure 3). The corresponding turnover
ratio of the [V141K, I155V] mutant is 1.4, to be compared to an
approximately unit product ratio. The same comparison
cannot be performed on the (R)-1-specific variant [W106L,
L109Y, V141K, I155V] because the KM values for both enantio-
mers are substantially elevated in this enzyme form, prevent-
ing accurate determinations of kcat.
values displayed by the wild-type enzyme and by the [W106L,
L109Y, V141K, I155V] mutant are quite similar (Table 2). Rather,
the effect can be traced to a crippled enzyme activity with the
S enantiomer, which is decreased 55-fold as a result of these
residue changes. Because the double mutant [V141K, I155V]
has retained 20% of the activity of the wild-type enzyme, the
most drastic negative effect on the activity with (S)-1 appears
Table 2. Steady-state kinetic parameters for the catalyzed hydrolysis of (R)-1 and (S)-1.
Substrates
(
R)-1
(S)-1
ꢀ
1
ꢀ1
ꢀ1
[a]
ꢀ1
ꢀ1
ꢀ1
[a]
enzyme
k
cat [s
]
K
M
[mm]
k
cat/K
M
[s mm
]
E (2/3)
k
cat [s
]
K
M
[mm]
kcat/K
M
[s mm
]
E (3/2)
wild-type
V141K, I155V
W106L, L109Y, V141K, I155V
13ꢁ0.9
9.1ꢁ0.2
0.78ꢁ0.1
1.0ꢁ0.04
17ꢁ2
0.4
1.1
15
43ꢁ3
13ꢁ4
1.0ꢁ0.1
1.6ꢁ0.4
42ꢁ3
2.6
0.89
0.07
9.3ꢁ0.2
12ꢁ0.6
8.3ꢁ0.9
[
c]
[c]
[d]
[d]
7ꢁ3
2ꢁ1
n.d.
n.d.
0.77ꢁ0.08
R
S
S
R
[a] (kcat/K
M
) /(kcat/K
M
) . [b] (kcat/K
M
) /(kcat/K
M
) . [c] Poorly determined due to low degree of substrate saturation. [d] Not determined due to inability to reach
substrate saturation.
1
426
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1422 – 1429

Epoxide Hydrolase with Shifted Enantiopreference
to be due to the W106L and L109Y substitutions. The shift in
enantioselectivity can be visualized as differences in activation
energies in the reaction on going from the racemic epoxide to
either diol enantiomer (Figure 4). The differences in activation
The variant [W106F, L109I], in which the introduced residue
replacements conserve the aromatic character of residue 106
and the aliphatic nature of residue 109, exhibits a pronounced
preference for (S)-1 in relation to the parent enzyme, with an
enantiomeric excess of product (ee ) of 60% for the S diol
p
product. The variant [W106L, L109Y, V141K, I155V], on the
other hand, exhibits the opposite enantioselectivity, with an
ee for the R diol of 70%. Clearly further experimental evidence
p
is required to better assess the importance of this active-site
region as a determinant of enantiospecificity, but the results
demonstrate that, with this epoxide, the positioning of an aro-
matic residue in the vicinity of the nucleophile is of major
importance in deciding enantiopreference. This work supple-
ments other recent reports that point to iterative saturation
mutagenesis (ISM) as an emerging approach to directed pro-
[26–31]
tein evolution.
Experimental Section
Chemicals, reagents, and bacterial strains: Chemicals were pur-
chased from Sigma–Aldrich at the highest purity available unless
stated otherwise. The enantiomers of (2,3-epoxypropyl)benzene (1)
were purchased from TCI Europe N.V. and their enantiomeric puri-
ties were determined by chiral HPLC to be >98%. Oligonucleo-
tides were custom-made by Thermo Scientific. Restriction and
DNA-modifying enzymes, as well as reagents for molecular biology
work, were supplied by Fermentas. Escherichia coli XL1-Blue was
supplied by Stratagene. Chromatographic resins were purchased
from GE Healthcare.
Figure 4. Differences in activation energies in the formation of either the R
diol (2) or the S diol (3) from rac-1. Relative energy levels were calculated
°
2
3
from DDG =ꢀRTln(kcat/K
M M
) /(kcat/K ) .
energies between reactions leading to the S diol, relative to
ꢀ
1
those leading to the R diol, increase by 9.2 kJmol on going
from the wild-type enzyme to the quadruple mutant. The
underlying reason for this change is at present not clear but
could be due either to suboptimal alignment of (S)-1 in the
active site for efficient nucleophilic attack by the Asp105 car-
boxylate, or to acid catalysis contributed by the Tyr154/Tyr235
pair.
Library construction: Four different sites composed of amino acid
residues lining the active site pocket were identified and defined
as hotspots. They were divided into A (F33), B (W106, L109), C
(V141, L145, I155), and D (I180, F189). The mutant variants were
constructed by inclusion of mutagenic primers in the PCR reac-
tions. The primers incorporate codon degeneracy at defined posi-
tions in the cDNA (Table 3). In the first round of library construc-
tions the plasmid pGTacStEH1–5H coding for wild-type StEH1 was
used as template. In the second round of library constructions the
plasmids encoding selected first-generation library hit variants
were used as templates. The plasmid constructs coding for protein
identified as hits were sequenced in full to identify the inserted
mutations and to confirm that no further alterations in the se-
quence had occurred. In the first round of library construction all
four hotspots were visited. In the second round of libraries three
enzyme variants from three different first-generation libraries were
chosen as starting materials for the second round of library con-
struction. The enzyme variant [W106F, L109I] from library B was
used as the template in the second-generation library for introduc-
In the A. niger enzyme variant mentioned above, developed
for enhanced enantiospecificity, this mutant had lost catalytic
efficiency with the R enantiomer of phenylglycidol ether by a
factor of 700, whereas the activity with the S enantiomer had
[
25]
only decreased tenfold. Steric hindrance of the R epoxide
was the suggested basis for the increased selectivity for the S
enantiomer. It was proposed that this steric hindrance caused
an increased distance between the nucleophile carboxylate
and the attacked epoxide carbon, leading to less efficient catal-
ysis.
Table 3. Mutagenic primers used in this study. The corresponding mutant codons are indicated in bold type.
Name
Mutation
Primer sequence (5’!3’)
ATCCTCTTCATCCACGGCNDTCCTGAACTCTGGTAC
GTCGTTGCGCACGACNDTGGCGCGNDTATTGCTTGGCATTTA
CCCATATATAGCCTTTDNCCCCTCAACTDNATTCATCTTTGGGTTC
AAGGCTATATATGGGGAAGATCATTATNDTTCGAGATTT
[
a]
PEH-53
PEH-54
PEH-55
PEH-56
PEH-57
F33NDT
W106NDT, L109NDT
[
b]
V141NHA, L145NHA
I155NDT
I180NDT, F189NDT
TTGCTCCAATTGGTGCTAAGTCTGTTCTAAGAAANDTTTGACATACCGCGATCCTGCACCANDTTATTTTCCTA
[a] N=G/A/T/C, D=G/A/T. [b] H=A/T/C.
ChemBioChem 2010, 11, 1422 – 1429
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1427

M. Widersten et al.
ing new mutations at position A (F33) and C (V141, L145, I155).
The enzyme variant [V141K, I155V] from library C was used as the
template for introducing mutations at position 33 and at posi-
tions 106 and 109. The enzyme variant [F189V] was used as the
template for introducing mutations at positions A (F33), B (W106,
L109), and C (V141, L145, I155). For a schematic view of the paths
taken during library construction see Figure 2.
A well giving rise to a low absorbance value combined with a col-
orless well in both duplicate plates was scored as a hit.
Protein expression and purification of hits: Expression plasmids
encoding mutant variants of StEH1–5H were transformed into
E. coli XL1-Blue bacteria by electroporation. An overnight culture
(
5 mL) of 2TY medium [tryptone (1%, w/v), yeast extract (1.6%, w/
ꢀ
1
v), NaCl (0.5%, w/v)] containing ampicillin (100 mgmL ), inoculated
with a single bacteria colony grown at 308C (200 rpm), was used
to start a culture of 2TY media (0.5 L) fortified with ampicillin
Protein expression for screening: All procedures were performed
under sterile conditions in 96-well microtiter plates (Nunc, poly-
propylene, round-bottomed, accommodating at least 400 mL).
Expression plasmids encoding mutant variants of StEH1–5H were
transformed into E. coli XL1-Blue bacteria by electroporation with
the aid of a Bio-Rad Gene Pulser. The bacteria were plated on agar
ꢀ1
(50 mgmL ). The culture was grown at 308C (200 rpm) until the
optical density at 600 nm reached 0.3. Isopropyl b-d-thiogalacto-
pyranoside (final concentration 1 mm) was then added to induce
protein expression. The culture was grown for 18 h and the cells
were collected by centrifugation at 5000 g for 12 min and then
resuspended in buffer [sodium phosphate (pH 7.0, 10 mm) and
sodium azide (0.02%, w/v)] with added protease inhibitor (Com-
plete Protease Inhibitor Cocktail Tablets EDTA-free, Roche). The
bacterial resuspension was then ultrasonicated with a Vibra Cell
Sonifier (power setting 7.5) for five pulses of 30 s with more than
one minute cooling time in between. The suspension was centri-
fuged for 45 min at 30000g and the supernatant, containing the
enzyme of interest, was filtered through a 0.2 mm filter with a glass
pre-filter (Pall Life Science). The lysate was then directly applied
ꢀ
1
plates (ampicillin, 100 mgmL ). From these plates single colonies
were picked and inoculated into wells containing 2TY medium
[
3
tryptone (1%, w/v), yeast extract (1.6%, w/v), NaCl (0.5%, w/v),
50 mL] containing ampicillin (100 mgmL ). Four wells were left
ꢀ
1
uninoculated and four were inoculated with wild-type StEH1 for
reference reasons. The plates were covered with gas-permeable
adhesive seal film and incubated overnight at 308C (200 rpm).
After 18 h incubation, overnight culture (25 mL) from each well was
used to inoculate a new well on a new plate with fresh 2TY media
ꢀ
1
fortified with ampicillin (50 mgmL ). The bacteria were grown for
h (308C, 200 rpm), after which isopropyl b-d-thiogalactopyrano-
II
3
onto a HisTrap HP 1 mL column preloaded with Ni ions and equili-
side was added (final concentration 1 mm) to induce protein
expression (glycerol was added to the wells of the plate with the
overnight cultures to a final concentration of 15%; the plate was
stored in ꢀ808C for later recovery of screening hits). After 18 h of
incubation, the plate was centrifuged (48C, 3500 rpm) for 20 min.
The supernatant was removed and the bacteria pellet was resus-
pended in sodium phosphate, (0.1m, pH 7.0) with added protease
inhibitor (Complete Protease Inhibitor Cocktail Tablets EDTA-free
from Roche) and stored at ꢀ808C until screened.
brated with imidazole-containing buffer [imidazole (20 mm), NaCl
(0.5m), sodium phosphate (10 mm), pH 7.0]. Loosely bound pro-
teins were washed off by use of an increase in imidazole concen-
tration to 100 mm. Tightly bound proteins were eluted with the
same buffer fortified with imidazole (300 mm). The enzyme-con-
taining fractions from the last elution step were pooled and the
buffer was changed to sodium phosphate (0.1m, pH 7.4) with the
aid of a PD-10 column. The homogeneities of purified protein sam-
ples were determined by SDS-PAGE stained with Coomassie Bril-
liant Blue R-250 and the enzyme was established to be at least
Activity screening: Buffer was added to the frozen bacteria pellets
9
5% pure. Protein concentrations of collected fractions were deter-
[
1
(
final concentrations in the wells: sodium phosphate (pH 7.0,
mined from the absorbance values at 280 nm. The molar absorb-
ance coefficient used, calculated from the amino acid composition,
ꢀ
1
0 mm), lysozyme (0.2 mgmL ),
MgCl2
(5 mm), DNaseI
ꢀ
1
0.2 mgmL )]. The pellets were thawed at room temperature and
ꢀ1
ꢀ1
was 59030m cm both for wild-type StEH1 and for mutant var-
were then incubated at 48C for 1 h, after which four freeze/thaw-
ing cycles were performed (16 min at ꢀ808C, followed by 20 min
incubation at 308C). The plate was then centrifuged at 3500 rpm
for 60 min, after which an aliquot (25 mL) of lysate from each well
was transferred to a corresponding microtiter plate (polystyrene,
iants. The coefficient was recalculated and used according to the
formula (nWꢂ5500)+(nYꢂ1490)+(nCꢂ125), where
n is the
number of the specific amino acid residue, if the enzyme had lost
or gained any tryptophan, tyrosine, or cysteine residues. The calcu-
lated molecular weight used was 37.1 kDa. Around 5–20 mg of
purified enzyme per liter of culture was obtained. All preparations
were made at 48C starting from the centrifugation of the bacteria
and the enzyme was stored in the fridge with retention of full
activity during the timespan of the experimental data collection.
9
6 wells, flat-bottomed, Nunc). This was done in duplicate for each
plate. The following screening protocol is adapted from ref. [32].
The substrate 1 was dissolved in sodium phosphate [0.1m, pH 7.0,
containing dimethyl sulfoxide (10%, v/v)]. For the screening of the
first-generation libraries the racemic mixture of 1 was used. In the
screening of the second-generation libraries the pure enantiomers
were used. Equal volumes of lysate and substrate were mixed, re-
sulting in a final concentration of 1 of 5 mm. The enzyme was incu-
bated with the substrate with agitation for 10 min at 308C, after
which the epoxide hydrolase reaction was stopped by addition of
Enantioselectivities of purified enzyme variants: The diol product
distribution after epoxide-hydrolase-catalyzed hydrolysis of rac-1
was established with purified enzyme variants and separation of
diols by chiral HPLC. Purified enzyme (0.5 mm) was mixed with rac-
1 (30 mm) in sodium phosphate (0.1 mm, pH 7.0), containing a
final concentration of 3%, (v/v) acetonitrile. The reaction mixture
was incubated at 308C (200 rpm) for five minutes, after which
methanol (50% final concentration) was added to stop the reac-
tion. The solvents were then evaporated and the residual analytes
were dissolved in hexane/isopropanol (80:20, v/v). The separation
of the formed diols was performed by injection of the analytes
with the aid of a Shimadzu SIL-10AF autosampler over a Daicel
Chiralpak AS-H (250ꢂ4.6 mm ID) column with a Shimadzu Promi-
2
-(4-nitrobenzyl)-pyridine (25 mL of 100 mm dissolved in ethylene
glycol/ethanol 80:20, v/v). The plate was incubated for 20 min at
08C, after which ethanol (50 mL, 95%) and potassium carbonate
1m, 25 mL) were added. A blue colored conjugate forms in the
8
(
wells if epoxide is still present, whereas a colorless well indicates
total conversion of epoxide into diol. The absorbance from the de-
veloped color was measured directly at 570 nm with a microtiter-
plate reader spectrophotometer (SpectraMax190, Molecular Devi-
ces). Manual visual identification of hits was also performed, due
to precipitation in some of the wells giving rise to false negatives.
ꢀ1
nence LC-20AD pump and a flow rate of 0.5 mLmin . The mobile
phase consisted of hexane/isopropanol (95:5, v/v). Peaks were de-
1
428
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1422 – 1429

Epoxide Hydrolase with Shifted Enantiopreference
tected at 220 nm with a Shimadzu SPDM20A diode array detector.
The retention times for the different diols were 47 and 59 min. The
measurements were made at least in triplicate.
[3] A. Schmid, J. S. Dordick, B. Hauer, A. Kiener, M. Wubbolts, B. Witholt,
Nature 2001, 409, 258–268.
[
4] E. G. Hibbert, F. Baganz, H. C. Hailes, J. M. Ward, G. J. Lye, J. M. Woodley,
P. A. Dalby, Biomol. Eng. 2005, 22, 11–19.
Steady-state kinetics: The steady-state kinetic parameters for the
reaction with the pure enantiomers of 1 was established for the
wild-type StEH1 as well as for two second-generation enzyme var-
iants—[V141K, I155V] and [W106L, L109Y, V141K, I155V]—by re-
versed-phase HPLC. Both substrate and internal standard (2-phe-
nylethanol) were dissolved in acetonitrile and added to the reac-
tion mixture, resulting in a final concentration of 1.5%, (v/v) aceto-
nitrile in the reaction buffer (0.1 mm sodium phosphate, pH 7.0).
The reaction was followed at 308C and after enzyme addition, ali-
quots from the reaction mixture was taken and added to methanol
[5] M. D. Toscano, K. J. Woycechowsky, D. Hilvert, Angew. Chem. 2007, 119,
3274–3300; Angew. Chem. Int. Ed. 2007, 46, 3212–3236.
[
6] D. L. Ollis, E. Cheah, M. Cygler, B. Dijkstra, F. Frolow, S. M. Franken, M.
Harel, S. J. Remington, I. Silman, J. Schrag, J. L. Sussman, K. H. G. Ver-
schueren, A. Goldman, Protein Eng. 1992, 5, 197–211.
[
[
[
7] A. Steinreiber, K. Faber, Curr. Opin. Biotechnol. 2001, 12, 552–558.
8] A. Archelas, R. Furtoss, Curr. Opin. Chem. Biol. 2001, 5, 112–119.
9] K. M. Koeller, C.-H. Wong, Nature 2001, 409, 232–240.
[
10] M. T. Reetz, C. Torre, A. Eipper, R. Lohmer, M. Hermes, B. Brunner, A. Mai-
chele, M. Bocola, M. Arnad, A. Cronin, Y. Genzel, A. Archelas, R, Fur-
stoss, Org. Lett. 2004, 6, 177–180.
(
final concentration 50%) to stop the reaction at defined time
[11] M. Widersten, A. Gurell, D. Lindberg, Biochim. Biophys. Acta Gen. Subj.
010, 1800, 316–326.
2
points. Each reaction was performed in duplicate. The enzyme con-
centrations used were 100 nm for the wild type and 200 nm for the
enzyme variants. Both (R)-1 and (S)-1 were varied over the same
concentration range (0.03–2 mm). The detection and quantification
of the diols formed was established by injection of the analytes
with the aid of a Shimadzu SIL-10AF autosampler over a Kromasil
C-18 (250ꢂ3.2 mm ID) column with a Shimadzu Prominence LC-
[
[
12] K. L. Morley, R. J. Kazlauskas, Trends Biotechnol. 2005, 23, 231–237.
13] M. T. Reetz, M. Bocola, J. D. Carballeira, D. Zha, A. Vogel, Angew. Chem.
2
005, 117, 4264; Angew. Chem. Int. Ed. 2005, 44, 4192–4196.
14] M. T. Reetz, L.-W. Wang, M. Bocola, Angew. Chem. 2006, 118, 1258–
263; Angew. Chem. Int. Ed. 2006, 45, 1236–1241.
[
[
1
15] M. T. Reetz, J. D. Caralleira, J. Peyralans, H. Hçbenreich, A. Maichele, A.
Vogel, Chem. Eur. J. 2006, 12, 6031–6038.
[16] M. T. Reetz, J. D. Carballeira, Nat. Protoc. 2007, 2, 891–903.
ꢀ
1
2
0AD pump and a flow rate of 0.8 mLmin . The mobile phase
consisted of methanol/0.1m sodium phosphate (37:63, v/v, pH 3.0).
Peaks were detected at 220 nm with a Shimadzu SPDM20A diode
array detector. The retention times for the peaks were 16 min for
the diols and 28 min for the internal standard.
[17] M. T. Reetz, D. Kahakeaw, R. Lohmer, ChemBioChem 2008, 9, 1797–
1
804.
18] J. Lotter, A. L. Botes, M. S. van Dyk, J. C. Breytenbach, Biotechnol. Lett.
004, 26, 1197–1200.
19] M. Labuschagne, J. Albertyn, Yeast 2007, 24, 69–78.
[
2
[
Kinetic data analysis: The steady-state kinetic parameters k , K ,
[20] D. Lindberg, A. Gogoll, M. Widersten, FEBS J. 2008, 275, 6309–6320.
cat
M
[
21] S. L. Mowbray, L. T. Elfstrçm, K. M. Ahlgren, C. Andersson, M. Widersten,
Protein Sci. 2006, 15, 1628–1637.
and k /K for wild-type StEH1, [V141K, I155V], and [W106L, L109Y,
cat
M
V141K, I155V] were extracted first by fitting a linear equation to
the experimentally determined increasing product concentration
data as a function of time with use of LINFIT from the SIMFIT pro-
gram (http://www.simfit.man.ac.uk). The Michaelis–Menten equa-
tion was then fitted to the data by use of the MMFIT or RFFIT pro-
grams in SIMFIT.
[
[
22] L. T. Elfstrçm, M. Widersten, Biochem. J. 2005, 290, 633–640.
23] A. Thomaeus, J. Carlsson, J. ꢁqvist, M. Widersten, Biochemistry 2007, 46,
2
466–2479.
24] A. Thomaeus, A. Naworyta, S. L. Mowbray, M. Widersten, Protein Sci.
008, 17, 1275–1284.
[
2
[25] M. T. Reetz, M. Bocola, L. W. Wang, J. Sanchis, A. Cronin, M. Arand, J.
Zou, A. Archelas, A. L. Bottalla, A. Naworyta, S. L. Mowbray, J. Am. Chem.
Soc. 2009, 131, 7334–7343.
[
26] A. G. Sandstrçm, K. Engstrçm, J. Nyhlen, A. Kasrayan, J.-E. Bꢃckvall, Pro-
tein Eng. Des. Sel. 2009, 22, 413–420.
Acknowledgements
[
27] K. Okrasa, C. Levy, M. Wilding, M. Goodall, N. Baudendistel, B. Hauer, D.
Leys, J. Micklefield, Angew. Chem. 2009, 121, 7827–7830; Angew. Chem.
Int. Ed. 2009, 48, 7691–7694.
Dr. Diana Lindberg is gratefully acknowledged for her aid in the
HPLC measurements. The work of Dirk Maurer, Mikaela Gçrlin,
and Norris Al-Obaidi is also gratefully acknowledged. This work
was supported by the Swedish Research Council and The Carl
Trygger Foundation.
[
28] R. Hawwa, S. D. Larsen, K. Ratia, A. D. Mesecar, J. Mol. Biol. 2009, 393,
36–57.
[
[
29] L. Liang, J. Zhang, Z. Lin, Microb. Cell Fact. 2007, 6, 36.
30] M. R. M. De Groeve, M. DeBaere, L. Hoflack, T. Desmet, E. J. Vandamme,
W. Soetaert, Protein Eng. Des. Sel. 2009, 22, 393–399.
[
31] D. J. Bougioukou, S. Kille, A. Taglieber, M. T. Reetz, Adv. Synth. Catal.
Keywords: enantioselectivity · enzyme catalysis · epoxides ·
hydrolases · mutagenesis
2
009, 351, 3287–3305.
[32] F. Cedrone, T. Bhatnagar, J. C. Baratti, Biotechnol. Lett. 2005, 27, 1921–
927.
1
[
1] H. E. Schoemaker, D. Mink, M. G. Wubbolts, Science 2003, 299, 1694–
697.
2] J. M. Woodley, Trends Biotechnol. 2008, 26, 321–327.
1
Received: March 23, 2010
Published online on June 11, 2010
[
ChemBioChem 2010, 11, 1422 – 1429
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1429