Computational Design of Enantiocomplementary Epoxide Hydrolases for Asymmetric Synthesis of Aliphatic and Aromatic Diols

Article abstract of DOI:10.1002/cbic.201900726

The use of enzymes in preparative biocatalysis often requires tailoring enzyme selectivity by protein engineering. Herein we explore the use of computational library design and molecular dynamics simulations to create variants of limonene epoxide hydrolas

Full text of DOI:10.1002/cbic.201900726

Accepted Article
Title: Computational design of enantiocomplementary epoxide
hydrolases for asymmetric synthesis of aliphatic and aromatic
diols
Authors: Hesam Arabnejad, Elvira Bombino, Dana I. Colpa, Peter A.
Jekel, Milos Trajkovic, Hein J. Wijma, and Dick Janssen
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemBioChem 10.1002/cbic.201900726
Link to VoR: http://dx.doi.org/10.1002/cbic.201900726
A Journal of
www.chembiochem.org

ChemBioChem
10.1002/cbic.201900726
Computational design of enantiocomplementary epoxide
hydrolases for asymmetric synthesis of aliphatic and aromatic diols
Hesam Arabnejad, Elvira Bombino, Dana I. Colpa, Peter A. Jekel, Milos Trajkovic, Hein J.
Wijma, Dick B. Janssen*
Biotransformation and Biocatalysis, Groningen Biomolecular Sciences and Biotechnology
Institute, University of Groningen, Nijenborgh 4, 9747AG, Groningen, The Netherlands
*
corresponding author: d.b.janssen@rug.nl
1
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
Abstract
The use of enzymes in preparative biocatalysis often requires tailoring enzyme
selectivity by protein engineering. Here, we explore the use of computational library design
and MD simulations to create variants of limonene epoxide hydrolase that produce
enantiomeric diols from meso‐epoxides. Three substrates of different sizes were targeted:
2,3‐butene oxide, cyclopentene oxide and cis‐stilbene oxide. Most of the 28 designs tested
were active and showed the predicted enantioselectivity. Excellent enantioselectivities were
obtained for the bulky substrate cis‐stilbene oxide, and enantiocomplementary mutants
produced (S,S)‐ and (R,R)‐stilbene diol with >97% enantiomeric excess. An (R,R)‐selective
mutant was used to prepare (R,R)‐stilbene diol with high enantiopurity (98% conversion to
diol, >99 % e.e.). Some variants displayed higher catalytic rates (kcat) than the original
enzyme, but in most cases K values increased as well. The results demonstrate the
M
feasibility of computational design and screening to engineer enantioselective epoxide
hydrolase variants with very limited laboratory screening.
Keywords
Computational design, enantioselectivity, epoxide hydrolase, molecular dynamics, stilbene
oxide
Introduction
The use of biocatalysis in chemistry is an attractive option for many synthetic
processes, especially for preparing fine chemicals and bioactive compounds. [1] ‐[7] Although
nature provides an enormous diversity of industrially useful enzymes, they often must be
engineered to meet industrial process requirements. [8] , [9] In case of pharmaceutical
synthesis, of special importance are chemoselectivity, compatibility with harsh reaction
conditions and product enantiopurity.[1] ,[10] Consequently, extensive studies have been
carried out on controlling and improving enzyme selectivity by protein engineering, often
through directed evolution, [11] which led to enzymes with improved selectivity in kinetic
resolution of enantiomers and better performance in asymmetric transformation of
prochiral compounds .[8] [10] [12]
Whereas directed evolution is very successful, it requires high‐throughput screening
methods. In case of enzyme enantioselectivity, screening is possible by chiral
chromatography or by the use of quasi enantiomers in NMR or MS[13] but this may be time‐
consuming and expensive. Directed evolution becomes complicated when no high‐
throughput expression is available, such as in case of enzymes that must be produced in
fungi. Several methods have been proposed to overcome these bottlenecks, such as
optimizing strategies for library construction, e.g. by focusing mutations in different
combinations around the active site, [14] [15] [16] and by incorporating structural or
[9]
2
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
phylogenetic information .[17] [18] Another option is the use of computational tools to design
improved enzymes.[19] ‐[23] Statistical methods have also been used. [24] [25] Biophysics‐based
computational protocols have emerged as powerful platforms for the engineering of
thermostable and organic‐solvent compatible enzyme variants. [26] ‐[31] Multiple mutations
can be explored simultaneously, allowing larger jumps in sequence space compared to
directed evolution and structure‐based rational mutagenesis. Furthermore, in silico
screening of enzyme variants by docking and high‐throughput molecular dynamics
simulations makes it possible to predict enzyme properties and reduce library size for
experimental evaluation from thousands to dozens.[32] [33]
In this study, we explore a computational framework (CASCO, catalytic selectivity by
computational design) [32] for obtaining enantiocomplementary epoxide hydrolases. This
framework uses the Rosetta scoring function and search algorithm [19] ,[34] to generate
libraries of primary designs. Next, high‐throughput molecular dynamics simulations (MD)
with scoring the frequency of occurrence of reactive (or near‐attack) conformations (NACs)
[35] ‐[37]
are used for ranking and to select a small set of variants that qualify for laboratory
testing. The frequency of reactive poses during MD simulations can explain the selectivity of
computationally designed enzyme variants. [38] ‐[40] For that purpose, generating multiple
MD trajectories with independent assignment of initial atom velocities gives much better
sampling of accessible conformational space than the use of a single trajectory running over
a long simulation time. [41] ‐[45] Accordingly, to enable screening of thousands of Rosetta
designs by MD, CASCO uses 20‐80 of such short MD simulations for scoring conformational
stability of designed reactive enzyme‐substrate complexes. The MD step thus examines if
the conformation of the enzyme substrate complex, which is partially constrained during
the Rosetta design step (NAC), will be maintained in short MD runs, or whether that reactive
conformation is immediately lost e.g. by movement of the substrate to a non‐reactive pose.
We used this approach earlier to predict enantioselectivity in kinetic resolutions catalyzed
by haloalkane dehalogenases.[46]
The potential of this computational approach was for limonene epoxide hydrolase
(
LEH) redesign was illustrated in a previous study [32] where we observed that the
performance of the best enzymes in a 37‐variant library obtained by the CASCO framework
was similar to that of the best variants obtained by screening approximately 4700 variants
generated by the CASTing strategy for directed evolution. [47] Nevertheless, computational
library design for enzyme engineering has serious challenges, of which reliability of
predictions is an important example. To further explore the possibilities and limitations of
computational redesign, we designed and examined a novel set of enantioselective
limonene epoxide hydrolase (LEH) variants.
LEH catalyzes the hydrolysis of epoxides by activating a bound water molecule for
nucleophilic attack directly on one of the substrate's oxirane carbon. [48] The water is
positioned by H‐bonds to Asn55 and Tyr53 while Asp132 abstracts a proton from the water,
3
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
which enables nucleophilic attack as a hydroxyl ion (Scheme 1). At the same time, Asp101
protonates the epoxide oxygen, which makes it a better leaving group. The reaction is
concerted. [49] In case of meso‐epoxides, the stereochemical outcome is determined by
regioselectivity of the attack and the products are enantiomers. LEH has been extensively
used as a model system for exploring the use of directed evolution strategies to engineer
enantioselectivity, and many variants have been described. [47] [50] ‐ [55] We previously
examined the use of computational design and screening to improve stability and control
enantioselectivity.[26] ,[29] [32]
In this work, we investigated three small sets of enantiocomplementary epoxide
hydrolase variants for converting meso‐epoxides (cis‐2,3‐butene oxide, cyclopentene oxide,
and cis‐stilbene oxide) to their corresponding (R,R)‐ or (S,S)‐diols. For each target
enantiomer, five top‐ranked new variants were experimentally characterized to explore how
challenging it is to position each of the substrates uniquely in the active site. We also
measured kcat and K
was higher than of the thermostable LEH‐P variant, from which all mutants were derived.
However, the Michaelis constants (K ) were almost always higher as well, which decreases
M
values for selected variants. This showed that the kcat of some variants
M
catalytic efficiency. The results confirmed the working hypothesis that obtaining unique
binding orientations, reflected in high enantioselectivity, was easier with the bulkier
substrates. Mutants converted cis‐stilbene oxide to diols with an enantiomeric excess (e.e.)
of > 99%. Small‐scale preparative reactions were carried out.
A
B
Scheme 1. Conversion of epoxides by limonene epoxide hydrolase. A) regioselective hydrolysis
of meso epoxides examined in this study. B) Catalytic mechanism of LEH, illustrated with proRR
hydrolysis of cyclopentene oxide.
Results
Computational design. The enantioselectivity of limonene epoxide hydrolase is
dependent on the regioselectivity of water attack within the active site. To design epoxide
hydrolase variants for enantioselective conversion of the three substrates (1a, 2a, 3a,
4
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
Scheme 1), we performed a series of design calculations for these substrates. Substrates
were docked in the active site and placed in a reactive configuration using restraints. This
included a hydrogen bond between the leaving oxygen and D101, a close distance between
the nucleophilic water and the attacked oxirane carbon, and close to linear orientation of
the nucleophilic water, the oxirane carbon, and the epoxide oxygen. Next, the Monte Carlo
search algorithm of Rosetta was used to optimize the identity and side chain geometries of
amino acids surrounding the active site for either proRR or proSS attack of the nucleophilic
water on the epoxide carbon (Scheme 1). The reactive geometries were essentially defined
A
B
Arg99
N H
H
O
Asp101
N
H
N
H
O
H
H

d2
O

1A
O

O
d3

H
d1
1B
H
O
Asp132
H

5
d4
d5

O
O
d6
H
Asn55
N
H
Tyr53
Figure 1. Design of limonene epoxide hydrolases for asymmetric conversion of meso epoxides. A) NAC
criteria used to predict the enantioselectivity of diol formation from 2,3‐butene oxide by MD‐simulations. The
same criteria were used for all three epoxides. Angles and distances are defined as followed: for the
nucleophilic attack angle 1A= 128‐163, 1B= 128‐163 and d
1
=0‐3.22 Å. For the H‐bonds 2‐6= 120‐180 and d2‐
6
=0‐3.50 Å. B) selection of target positions (cyan) in PDB structure PDB 4R9K. Catalytic residues are shown in
yellow, and the substrate in magenta. The targeted amino acid positions are either located in the peripheral
structural elements H1 (M32, L35), H3 (L74), and 3 (M78, I80, V83) which border the proRR side of the
substrate binding pocket; or in the central region, which forms the proSS side of the binding pocket and
consists of H4 (F139), 4 (L103), 5 (L114, I116) and 6 (F134). The secondary structure elements are defined
as follows: N‐loop (residues 1 to 23), H1 (24 to 35), H2 (40 to 46), H3 (64 to 75), H4 (135 to 143), 1 (52 to
5
6), 2 (60 to 62), 3 (79 to 91), 4 (94 to 105), 5 (111 to 123) and 6 (126 to 133).
as a near attack conformation (Fig. 1A). The explored sequence space was created by
targeting 11 selected positions around the active site (Fig. 1B) with randomization to any of
the nine hydrophobic residues (AFGILMPVW). This way, Rosetta generated variants forming
a low energy complex with the target substrate either in the proRR or proSS orientation,
resulting in thousands of possible proRR and proSS designs per substrate (Table 1).
5
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
These sets of primary Rosetta designs represent mutant libraries enriched in the desired
phenotype.
To computationally screen these libraries in an orthogonal manner by the likelihood
of showing the correct regioselectivity of water attack, we performed molecular dynamics
simulations. From MD trajectories, enantioselectivities were predicted by scoring the
fraction of time that the enzyme‐substrate complex is in the same reactive proRR or proSS
conformation used during the Rosetta design step (near‐attack conformations, NACs, Fig.
1
A). Instead of a single or a few long MD simulations, we performed a large number of
parallel MD simulations with independently assigned initial atom velocities (HTMI‐MD),
since this gives more extensive conformational sampling and better agreement with
experimental results than single long simulations. [41] ‐[46] For each of the 15,000 Rosetta
designs, at least 5 independently initialized MD runs of 10 ps were performed (Table 1).
Designs that passed initial selection rounds were subjected to more MD simulations, up to
80 10 ps and 5 100 ps. NACs were counted on the fly and their frequency was averaged
for each design. The ratio between averaged NAC frequencies for proRR and proSS attack
conformations was used as a predictor for enantioselectivity using Equation 1. This MD
screening of the primary libraries reduced the number of designs by a factor of 15 to 70
(Table 1).
Table 1. Molecular dynamics screening of designed enzyme variants.
No. of designs remaining for substrate
CASCO step
Criteria
1
a
2a
3a
Total Rosetta designs
Rosetta designs with a unique sequence
Designs remaining after MD screening
28795
4125
33252
4732
20714
6232
5
1
2
4
8
5
 10 ps
e.e. pred >97%
e.e. pred >97%
e.e. pred >97%
e.e. pred >98%
711
524
378
170
142
63
990
723
587
314
283
181
3.8%
2504
2205
1631
1284
0  10 ps
0  10 ps
0  10 ps
0  10 ps
 100 ps
e.e.pred >98% + [NAC] > 5%
pref
978
442
[a]
pred
[b]
e.e.
> 98%
Rosetta designs passing all criteria
1.5%
7.1%
pref
[
a]
pref
[
NAC] indicates the NAC frequency for the preferred product enantiomer. Additional criterion [NAC]
>
1
0%. [ Additional criterion [NAC] > 5%.
b]
pref
It was noticed during in silico screening by MD that a larger fraction of the stilbene
oxide designs displayed high NAC frequencies and high predicted enantioselectivities than
what was found with cyclopentene and butene oxide designs. As a result, more stilbene
oxide designs survived the final selection (7.1 %) than designs for cyclopentene oxide 1a (1.5
) and butene oxide 2a (3.8 %), even though the selection criteria were set significantly
%
more strict for stilbene oxide than for the other substrates (Table 1). This suggests a better
occupancy of reactive orientations for stilbene oxide, and more restricted conformations of
6
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
the stilbene oxide designs than of the designs with the two smaller substrates, at least
during MD simulations.
Visual inspection of the top‐ranked designs was carried out to verify that there were
no noticeable structural problems that would decrease catalytic activity or enantio‐
selectivity. For every target enantiomer, the variants predicted to have a high e.e.pred for
that product were ranked, those with the highest [NAC]pref first, and variants were inspected
until five variants were identified for each target enantiomer. It was noticed that for
cyclopentene oxide 1a and butene oxide 2a there were few designs with obvious errors. In
25 inspected designs for those two substrates there were only five with recognizable
structural errors (one was unusually flexible, three had such a spacious active site that
reorientation of the substrate seemed likely, and in one mutant there was a new H‐bond to
the epoxide oxygen, Table S1). In contrast, 10 of the 20 inspected variants for stilbene oxide
displayed structural problems. Four designs appeared too spacious (Supporting Information)
and 6 designs had a water positioned such that attack on the non‐intended carbon atom of
the epoxide seemed likely, even though the NAC analysis did not suggest this.
Experimental characterization. The top‐5 designs for each product enantiomer of
the three substrates were investigated experimentally. Two designs (26A, 45A) were
selected twice, both for proRR hydrolysis of cyclopentene oxide 1a and butene oxide 2a
(Table 2). We also included 3 designs predicted to have proRR selectivity by molecular
dynamics simulation while they were originally designed using Rosetta to have proSS
selectivity. The 28 new LEH variants were constructed in the thermostable LEH‐P template
described earlier [56] by sequential rounds of QuikChange mutagenesis. The use of a
thermostable parent enzyme increases the chance that protein function is maintained upon
introduction of mutations that may be too destabilizing in a mesostable template and is also
reported for directed evolution protocols.[57] [59] The LEH template used here (LEH‐P, Tm,app
0⁰C, pdb 4R9K) is not the most stable enzyme from our previous work, since it lacks the
disulfide bonds of the most stable variant (Tm,app= 85⁰C, pdb 4R9L). [56]
=
7
After sequence verification genes were expressed in E. coli Top10 or 10β for enzyme
production. All mutants were well expressed and could be isolated by His‐tag metal affinity
chromatography. Typical yields were 50‐150 mg per L of culture. The redesigned LEHs were
very stable and could be stored for over 2 years at ‐80⁰C without loss of activity. Most
variants showed a somewhat lower apparent melting temperature but more stable variants
were also found (Table 2). Overall thermostability was well maintained with an average
T
5
m,app of the redesigned enzymes of 62.7°C as compared to 70°C for the template LEH‐P and
0°C for the wild‐type LEH. Activity assays were done by mixing purified enzyme with
epoxide and measuring diol formation. Of the obtained 28 designs, 27 showed catalytic
activity on the substrate they were designed for. Of these active variants, 66% have an
activity that is between 10‐164% of the wild‐type activity for their respective substrate.
7
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
Table 2. predicted and observed activities of computationally redesigned limonene epoxide hydrolase variants.
Mutations[
a]
Computational
Experimental
Enzyme
NAC% NAC% eepred eeexp
Act.
(%)
0.12 U/mg 70
T
(⁰C)
m,app
cyclopentene oxide (1a)
(%)[
14
32
9
b]
[c]
proRR proSS (%)
LEH‐P
1
4
5
4
4
3
4
2
2
2
A
M32L_M78W_I80V_L103F_F139W
5.63 0.000 100
3.86 0.040 98
2.59 0.008 99
2.44 0.000 100
2.08 0.000 100
2
29
1
<1
24
41
55
15
16
164
63.5
67.0
67.0
‐
57.0
‐
48.5
68.5
73.5
45.5
Designs
for
3A L74I_L103V_F134Y_F139W
9A M78L_L103V_L114G_I116F_F139I
5A M32L_L74I_L103V_L114W_I116L_F134G_F139W
6C M32L_L103V_L114A_I116F_F139W
A
C
4A M32A_M78I_I80F_L103I_I116V_F139L
5A M78I_I80F_L103I_I116V_F139L
6A L35F_M78F_I80G_I116V_F139W
6
(
R,R)‐1b
d
‐
46
M32L_L35W_I80A_I116V_F139W
M32L_L35W_I80G_V83I_I116V_F139W
0.280 36.5
0.208 24.1
0.248 23.6
0.080 19.4
0.240 19.3
‐98
‐98
‐98
‐99
‐98
‐74
‐80
‐85
‐84
‐60
Designs
for
(
S,S)‐1b
cis‐butene oxide (2a)
Act. (%)c
LEH‐P
‐2
57
24
18
31
‐3
‐73
‐41
‐82
6
0.23 U/mg
4
4
4
4
5
3
3
3
3
2
7B M32L_L35M_M78I_I80L_V83L_I116M_F134Y
8A M32L_L35G_M78L_I80W_L103I_F139L
24.1 0.176 99
9.00 0.152 97
5.72 0.112 96
4.14 0.048 98
2.90 0.040 97
3
60
1
56.5
56.0
‐
59.0
73.0
70.5
56.0
57.5
78.0
45.5
Designs
for
e
5A M32L_L74I_L103V_L114W_I116L_F134G_F139W
9A L103V_L114W_I116L_F134G_F139W
0A M32L_L103V_F134Y_F139M
0A L74W_I80F_L103I_I116V_F139L
1A L35F_M78F_I80A_I116V_F139W
2A L35W_L74F_I80G_I116V_F139L
3B M32L_I80W_L103I_F139L
(
R,R)‐2b
6
15
11
100
15
1
0.496 25.2
0.320 23.7
0.080 21.8
0.168 21.0
0.080 20.4
‐96
‐97
‐99
‐98
‐99
Designs
for
(
S,S)‐2b
6A L35F_M78F_I80G_I116V_F139W
‐60
87
cis‐stilbene oxide (3a)
Act. (%)c
LEH‐P
92
91
‐97
>99
92
>99
‐
40
‐
‐95
‐28
0.35 U/mg
5
5
6
6
3
6
6
6
4
6
1Ae M32L_L35G_I80W_L103V_F139W
28.2 0.104 99
24.9 0.144 99
22.7 0.048 100
22.1 0.136 99
22.0 0.352 97
0.016 15.5 ‐100
51
22
52
3
11
<1
7
<1
21
2
59.5
61.0
65.0
66.5
75.5
71.0
67.0
57.5
62.0
66.0
Designs
for
2A M32L_L35M_M78I_L103I_L114M_I116F
e
0A M32L_L35G_I80W_L103V_F139L
(
R,R)‐3b
1B M32A_I80V_L103V_L114W_I116V_F134G_F139L
8A M32L_M78L_I80V_L103V_F134W_F139L
2A M32L_I80V_L103V_L114W_I116A_F134G_F139L
3B M78I_I80L_L103V_L114W_I116V_F134G _F139L
4C M32A_L103V_L114W_I116A_F134G_F139L
1B M32L_L35M_L103I_L114M_I116F_F139L
Designs
for
0.240 15.4
0.152 15.3
0.072 15.0
‐97
‐98
‐99
(
S,S)‐3b
5B L74I_M78F_L103V_L114A_I116V_F134W_F139M
0.008 14.7 ‐100
[
a]
Mutations at the peripheral (proRR side) of the substrate binding pocket are marked with a gray background.
Other mutations line the center (proSS side) of the substrate binding pocket.
Positive numbers: (R,R)‐diol preference; negative numbers: (S,S)‐diol preference.
Relative activities expressed in percentage of the activity with the template enzyme (indicated). Data from
duplicate measurements with the same enzyme batch.
[
[
b]
c]
[
[
d]
e]
‐, no activity.
Variants designed by Rosetta to exhibit (S,S)‐product selectivity but predicted by HTMI‐MD and found
experimentally to produce (R,R)‐diol 3b.
8
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
Chiral analysis of the formed diols revealed that 21 (77%) of the designed active variants
showed the enantioselectivity predicted by MD simulations (10% e.e. cutoff, Table 2).
The cyclopentene oxide (1a) designs were generated to examine if the design and
selection protocol gave comparable results to an earlier study in which 34 LEH variants were
tested for the same substrate. [32] Of the 10 new cyclopentene oxide designs, nine were
active and had the predicted enantioselectivity. The inactive design was 45A, which was also
selected as a design for butene oxide 2a, for which it did have a low catalytic activity. The
five proSS designs produced (S,S)‐diol 1b with e.e. values of 56‐85%, while lower e.e.'s were
obtained with proRR designs for (R,R)‐1b. The higher enantioselectivity of proSS designs is in
agreement with previous observations.[32]
The butene oxide designs were generated to test the possibility to control
regioselectivity of water attack with a very small prochiral epoxide substrate. All 10 variants
were active and 8 of them had the predicted enantioselectivity. As with cyclopentene oxide,
the proSS designs performed better than the proRR designs, with the best mutant (32A)
providing (S,S)‐diol 2b with 77% e.e. The wrong predicted designs had very low enantio‐
selectivity (e.e. of 2‐6 %). Thus, for both of the small substrates the MD simulations were an
effective tool for predicting LEH enantioselectivity.
For the largest substrate, stilbene oxide (3a), 6 of the 8 active variants showed the
predicted enantioselectivity. This included two variants (51A, 60A) that were originally
designed using Rosetta to have (S,S)‐3b selectivity but for which the HTMI‐MD screening
predicted preferential formation of (R,R)‐diol, which was in agreement with what was
observed experimentally. In these cases, MD corrected the Rosetta design prediction. On
the other hand, the combination of Rosetta and MD for design and prediction of
enantioselectivity still gave 2 mismatches between prediction and experiment in case of 3a
designs. Of these, variant 52A was exceptional since it was predicted to give (R,R)‐diol 3b
whereas experimentally it produced (S,S)‐diol with high e.e. (97%).
Despite the 2 prediction errors, the results show that highly enantioselective variants
could also be designed computationally for stilbene oxide. The thermostable template
enzyme had 92% (R,R)‐diol selectivity, and 4 of the 8 active new variants also displayed very
high (R,R)‐preference (>91% e.e.) whereas 2 other designed variants displayed high
(
S,S)‐diol preference (>91% e.e.). The more extensive protein‐substrate interactions with a
bulky substrate likely result in more restricted reactive conformations of enzyme‐substrate
complexes, accompanied by high product enantioselectivity. Of the experimentally
3
characterized variants for all three substrates, 3 have a relatively large (> 100 Å ) increase in
volume of active site, as calculated from reduced side chain volume of introduced amino
acids, and indeed these three variants have low or no catalytic activity (see Supporting
Information). When variants with a predicted increase of the active site volume of >100 Å3
would have been removed from the libraries, only the primary Rosetta libraries for cis‐
stilbene oxide would have shrunken (elimination of 178 of the 442 variants listed in Table 1).
9
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
Catalytic properties of the best variants. To examine how the use of Rosetta for
redesign of LEH towards production of a specific diol enantiomer influences catalytic
activity, we examined the kinetic properties (kcat and K ) of the best mutants (Table 3). The
M
LEH variants were produced using 1 L cultures, giving again 50‐150 mg purified protein/L
broth, which was comparable to the yield of the parent thermostable enzyme. Variants RR8
and SS16 were selected earlier as the best variants from a set of 37 designs for cyclopentene
oxide 1a [32] and were included for comparison (Table 3). The results showed that 46C and
RR8 had catalytic rate constants (kcat) for cyclopentene oxide that were comparable to that
of the thermostable template enzyme LEH‐P. Variants 43A and 24A had lower catalytic
constants (2.5‐fold and 7‐fold respectively). Variant SS16 variant displayed an almost 2‐fold
higher kcat than the template LEH‐P. Thus, catalytic rates were quite well maintained.
However, in most cases the K values were higher (8‐ to 76‐fold) for the redesigned LEHs.
M
This might be due to an increase in the volume of the substrate
Table 3. Kinetic properties of computationally redesigned epoxide hydrolases.
Variant Designed to
produce
Assay
substrate
e.e.[a]
%)
Prefe‐
rence
k
cat
K
M
k
cat/K
M
‐
1 [b]
(mM)[b]
4.2±0.2
189±20
344±17
35±4[c]
‐1 ‐1
(
(s )
(M s )
7.9
LEH‐P
‐
1a
1a
1a
1a
1a
1a
2a
2a
3a
3a
3a
3a
13
85
(R,R)
(R,R)
(R,R)
(R,R)
(S,S)
(S,S)
(R,R)
(S,S)
(R,R)
(R,R)
(S,S)
(S,S)
0.035±0.004
0.039±0.009
0.022±0.002
0.017±0.005[c]
0.062±0.001
0.005±0.001
0.056±0.010
0.063±0.002
0.147±0.012
0.052±0.006
0.003±0.001
0.002±0.001
RR8
(R,R)‐1b
(R,R)‐1b
(S,S)‐1b
(S,S)‐1b
(S,S)‐1b
‐
0.20
0.06
0.48
1.14
1.9
4
4
6C
3A
34
8
SS16
4A
LEH‐P
2A
LEH‐P
‐90
‐85
24
54±4
2
3±0.2
18±1
3.1
3
(S,S)‐2b
‐
‐82
92
225±9
0.28
406
890
19
0.37±0.02
0.06±0.01
0.16±0.03
0.19±0.02
6
5
4
0A
2A
1B
(R,R)‐3b
(R,R)‐3b
(S,S)‐3b
>99
‐89
‐94
10.5
[
[
[
a]
Calculated from multiple data points i.e. product enantiomers in different reactions.
Averages of duplicate measurements with standard deviation.
b]
c]
Single measurement; margins from average coefficient of variation for 1a data.
3
binding cavity, which grew by 37 Å in mutant 46C that had a 76‐fold increase in K
M
for 1a.
The exception was 24A, which has a 1.4‐fold lower K
constant (kcat/K ) was reduced (Table 3). The kinetic measurements show that the main
cause of the lower catalytic activities of most designs found during initial tests (Table 2) is
due to a higher K , not a lower kcat
M
. In all designs for 1a, the specificity
M
M
.
1
0
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
The results for the variants designed to convert the other small substrate, cis‐butene
oxide 2a, are similar. For this substrate, we examined 32A, which has the highest
enantioselectivity for producing (S,S)‐diol and inverted enantioselectivity in comparison to
the template. It showed an insignificant increase in kcat, but the K
M
was much higher,
resulting in a drop in kcat/K in comparison to the template LEH‐P.
M
The results were different for the variants designed to convert the bulky epoxide
stilbene oxide 3a to (R,R)‐ or (S,S)‐diol. Here, kcat values were lower than with wild‐type,
whereas K values were better. The observation that kcat values are lowered in the designs
M
for 3a whereas NAC percentages during MD simulations (Table 2) were fine indicates that
prediction of reactivities across different substrates using such short MD simulations is
troublesome. This is not unexpected as MD does not account for energy barriers along
reaction coordinates.
The high enantioselectivities obtained for stilbene oxide variants, in comparison to
the designs for the two other substrates, are likely due a more restricted conformational
freedom in case of the bulky stilbene oxide. The better K values relative to those with small
M
substrates might be due to the design procedure making the active site too spacious for
small substrates, preventing a snug fit with good hydrophobic binding interactions. Stilbene
oxide is also more bulky than the natural substrate limonene epoxide and the mutations
3
created enough additional space (ca. 67 Å for mutant 60A, calculated from reduced side
chain volumes) for tighter binding and a low K
mutant 60A with 3a was accompanied by a 6‐fold better K
catalytic efficiency in 60A. Furthermore, the tight substrate binding caused the specificity
constants (kcat/K ) to be higher for stilbene oxide 3a than for cyclopentene oxide 1a and cis‐
,3‐butene oxide 2a (Table 3). The improved K values for 3a in comparison to wild type was
seen with all 3 tested stilbene oxide designs.
M
. Consequently, the 3‐fold lower kcat of
M
, leading to an improved
M
2
m
Preparative scale conversions. The stilbene oxide enantioselectivities of designs 41B
and 52A are higher than reported for other LEH variants tested on this substrate. [50] [52] To
examine if these redesigned LEHs could be used in preparative scale conversions, the
conversion of cis‐stilbene oxide 3a to the (R,R)‐ and (S,S)‐diols by variants 60A and 41B was
examined under different reaction conditions, including varying temperatures and
cosolvents (Table 4). Cosolvents were tested because the solubility of the substrates and
products in water is low. Even in the presence of 10% dioxane in 50 mM HEPES, pH 8.0, both
cis‐stilbene oxide and the diols were only partially soluble when added at 50 mM. Under
these conditions, the cis‐stilbene oxide remained visible as globular crystals while the (R,R)‐
diol and the (S,S)‐diol formed needles. The LEH variants 41B and 60A were active in this
suspension. Addition of cosolvents 1,4‐dioxane and THF and the presence of a biphasic
system with a layer of ethyl acetate were studied. No conversion was observed with variant
1
1
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
6
0A in the biphasic system with ethyl acetate. Addition of 10% dioxane gave the best
conversion, yielding up to 78% diol in 44 h at 30 ºC. Remarkably, adding THF as cosolvent
similar properties as 1,4‐dioxane) gave lower conversion when compared to reaction
(
conditions without cosolvent or with 10% dioxane. The conversion of cis‐stilbene oxide by
variant 60A was further improved (63% to 80%) by increasing the reaction temperature
from 30 to 40 ºC. The best conditions (10% dioxane and 40 ºC) were combined and gave a
conversion of 86% and 63% for variants 60A and 41B, respectively. Increasing the dioxane
concentration to 15% was beneficial for the proRR variant 60A, yielding a conversion of 98%,
but drastically decreased conversion of 3a by the proSS variant 41B. The enantiomeric
excess of the stilbene diol products was analyzed by chiral HPLC (Fig. S2‐S4). For the best
conversions the following results were obtained: >99% e.e. for (R,R)‐stilbene diol (LEH 60A)
and 88% e.e. for the (S,S)‐stilbene diol (LEH 41B).
Table 4. Asymmetric synthesis of stilbene diols by computationally engineered enantiocomple‐
mentary epoxide hydrolases.[a]
Temp (C)
Time (h)
Conversion (%)
e.e. (%)
Production of (R,R)‐diol by LEH 60A
HEPES 50 mM
30
30
30
40
40
40
40
44
44
44
44
24
48
48
63
78
52
80
76
86
98
‐[b]
‐
‐
‐
‐
‐
1
1
0% 1,4‐dioxane
0% THF
At 40 ºC
1
1
1
0% 1,4‐dioxane
0% 1,4‐dioxane
5% 1,4‐dioxane
>99% (R,R)
Production of (S,S)‐diol by LEH 41B
HEPES 50 mM
30
40
40
40
44
24
48
48
38
50
63
18
‐
‐
1
1
1
0% 1,4‐dioxane
0% 1,4‐dioxane
5% 1,4‐dioxane
88% (S,S)
‐
[
a]
Reaction mixtures (total volume 1 ml) contained 10 mg cis‐stilbene oxide (final conc. 50 mM,
suspension) and 6.37 mg enzyme (final conc. 320 μM) in 50 mM HEPES, pH 8.0.
[
b]
‐, not determined.
Structural origin of enantioselectivity. In order to provide a structural explanation
for the observed mutant enantioselectivities we inspected the Rosetta designed structures
and the average HTMI‐MD structures of selected mutants. According to these models, the
nucleophilic water molecule stays in virtually the same position (Fig. 1, 2). This agrees with
the X‐ray structure, in which the catalytic water has an unusually low B‐factor [48] indicating
a precise orientation due to H‐bonds from Tyr53, Asn55, and Asp132. The enantioselectivity
of the enzyme is therefore determined by the positioning of substrate relative to this water.
1
2
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
In all of the modelled structures, the positional differences that influence enantioselectivity
can globally be described as a sliding motion of the epoxide carbon atoms in front of the
nucleophilic water molecule (Fig. 2). Mutations causing the substrate to reside more
towards the center of the dimeric enzyme (i.e. nearβ strands 4, 5, 6, and helix H4, see
legend of Fig. 1 for residue numbers) will lead to attack on the (R)‐configured carbon of the
epoxide ring, resulting in an (S,S)‐diol. Vice versa, proRR attack will dominate if the substrate
is positioned more towards the peripheral side (i.e. at near β strand 3 and helices H1 and
H3).
In the models of the proSS selective variants that convert substrates 1a and 2a with
high enantioselectivity (e.e. >75%), the dominant substrate orientations are achieved by
steric hindrance introduced by mutations at the proRR side (e.g. L35W/F, L74W/F, M78F,
and I80W/F). These mutations will promote positioning of the substrate more towards the
central (proSS) side. At the same time, space‐creating mutations on the proSS side (I116V
and F139L) will further increase the preference for (S,S)‐diol formation. Indeed, designs for
substrate 1a and 2a carrying both L35W and I116V gave an (S,S)‐diol preference of > 73%
e.e. (Table 2). Furthermore, also in mutant 32A the steric hindrance mutations L35F and
L74F on the peripheral (proRR) side are accompanied by I116V, leading to (S,S)‐butanediol
formation with the best e.e. of 77%.
The origin of the opposite (R,R)‐selectivity of variants with substrates 1a and 2a can
be explained in a similar way. It is likely that increased steric hindrance due to mutations on
the central side of the substrate binding cavity (e.g. mutations L114W, I116F/M) encourage
positioning of the substrate closer to the peripheral (proRR) side of the binding pocket. This
is visible in mutant 46C for substrate 1a (e.g. mutation I116F) and possibly in 47B for
substrate 2a (e.g. mutation I116M). However, such a steric effect is not clear in all predicted
proRR mutant structures, and some designs indeed show low enantioselectivity (e.g. 45A,
5
0A). The weakly (R,R)‐diol selective variant 63B contains mutation I116V, creating space on
the central (proSS) side, but it is accompanied by L114W, reducing that space. The combined
effect of such mutations appears difficult to rationalize in view of effects of side chain
interactions and dynamics.
The mutants designed for stilbene oxide 3a have high product enantioselectivities
but the mutations that cause them appear to not only involve steric effects. The majority of
the 3a designs showed (R,R)‐preference, including variant 63B designed for (S,S)‐
enantioselectivity. Surprisingly, the two most (S,S)‐diol selective mutants for 3a (52A, 41B)
carried mutation I116F, a mutation that introduces steric hindrance at the central side and
in case of substrates 1a and 2a favors (R,R)‐selectivity (see above). The unexpected (S,S)‐diol
preference might be due to ‐ interactions between the substrate and the newly
introduced aromatic ring of Phe116. The same selectivity by attraction might hold for one of
the best (R,R) selective mutants: variant 60A (e.e. >99%) carries the I80W mutation at the
peripheral (proRR) side and no steric hindrance introducing mutation on the proSS site. For
1
3
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
the other strongly (R,R)‐selective mutant, 38A with e.e. > 99%, the introduction of steric
hindrance at the proSS side (F134W) can explain the improvement in enantioselectivity
Figure 2. Structural basis of redesigned enantioselectivity. Shown are the active‐site cavities of three proSS‐
variants (blue shades, names indicated) and three proRR variants (yellow‐orange shades). The variants were
designed for the substrates indicated at the left of each pair of panels. The reacting water molecules are
shown. Hydrogen atoms of substrates are hidden for clarity. In each panel, both the designed enzyme with its
substrate is shown, as well the position of the same substrate in the opposite design (substrates in proSS
designs in cyan, substrates in proRR designs in yellow). This shows pronounced differences in substrate
positioning and how steric hindrance steers product enantioselectivity.
along the same lines as for substrates 1a and 2a. The wild type (e.e >90% (R,R)‐diol)
also has aromatic functionality with Phe134 at the peripheral (proRR) region. Thus, it
appears possible that enantiomeric preference with stilbene oxide is partially determined by
an influence of attractive ‐ interactions on binding orientations of the substrate.
Instead of a translational shift in the position of substrate, the stereoselectivity of
LEH variants could also be influenced by rotating the substrate in the binding pocket by 180
along an axis formed by the epoxide oxygen and the spot in between the two epoxide
carbon atoms. This would switch the orientation of the epoxide carbon atoms relative to the
nucleophilic water. However, such substrate rotations were not observed in any of the
1
4
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
models or during MD simulations. Also attempts by us to dock substrates into productive
orientations featuring such a rotation were unsuccessful.
Discussion
Changing enantioselectivity of limonene epoxide hydrolase by directed evolution has
been extensively investigated by Reetz, Sun and coworkers, focusing on improving directed
evolution strategies including optimizing target positions and positional diversity in libraries.
[47] [50] [52]
Such well‐optimized directed evolution protocols still rely on substantial
experimental screening by chiral chromatography, which triggered us to examine if
computational methods could be developed to enrich libraries and replace most of the
laboratory screening by computational screening of protein libraries.[60] In silico screening
methods have been used earlier to design cocaine hydrolyzing enzymes with improved
catalytic efficiency, [61] [62] but also to design libraries of a cytochrome P450 harboring
variants with controlled selectivity, [63] to increase amidase activity in an esterase.[64]
The results show that design of small sets of mutants with Rosetta and screening by
MD simulations could indeed generate LEH variants with desired enantioselectivity. Rosetta
design targeted 11 positions at the same time, with a 9 amino acid diversity per position.
MD screening was done by multiple ultra‐short simulations with on the fly scoring of
reactive conformations and allowed to screen of thousands of Rosetta designs. This CASCO
protocol[32] reduced the size of the library required, and for each substrate only 5 variants
per desired enantiomer were used to find mutants with enhanced enantioselectivity. In case
[
47] [50] ‐ [52]
of cyclopentene oxide 1a, directed evolution
and previous computational design
gave (S,S)‐selective LEH variants producing diol with similar e.e. as found here (60%‐95%
e.e., Table 2).
A comparison of predicted enantioselectivities as calculated from NAC percentages
Table 2) shows that there is good overall agreement only in qualitative terms, i.e. (R,R)‐ or
S,S)‐diol stereopreference was correctly predicted for 77% of the variants using multiple
[32]
(
(
short MD simulations with independent initialization. On the other hand, within a set of 5
designs, experimental activities for individual variants did not correlate with their computed
NAC%, showing that the MD simulations as performed here do not provide quantitative
information on catalytic rates. Note, however, that rates shown in Table 2 are strongly
influenced by both K values and do not reflect kcat. Furthermore, in a broader sense, for
m
each of the three substrates examined, the set of designs that gave the highest NAC% in MD
also showed the highest average activity. Thus, on average the proRR designs for 3a were on
average more active than its proSS designs and also give the highest NAC percentages for
3
a. For the other two substrates proSS designs were more active and gave higher NAC%.
In this study, the best enantioselectivities were clearly obtained with stilbene oxide
a. Whereas most designs produced (R,R)‐diol 3b (with e.e. >99% for 2 variants),
3
enantiocomplementary mutants yielding (S,S)‐3b were also found, with a highest e.e. of
1
5
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
9
7%. With butene oxide 1a; [51] [52] the obtained variants showed high and modest
enantioselectivity in the production of (R,R)‐ and (S,S)‐diols, respectively. The LEH mutants
found for stilbene oxide could be used to produce enantiopure (R,R)‐diol and (S,S)‐diol at
preparative scale, indicating the potential of this approach to generate a practically useful
biocatalyst. Ring opening of stilbene oxide was tested earlier using mutants optimized on
cyclopentene oxide and cyclohexene oxide, which resulted in highly (R,R) selective variants
(
with e.e. 99%) but only modest (S,S)‐selective variants (e.e. 44%). [65] The results suggest
that the likelihood of obtaining high enantioselectivity is much better with the bulkier
stilbene oxide than with the smaller substrates, with the two phenyl rings of stilbene oxide
offering more opportunities for steric hindrance as well as Van der Waals and π‐π
interactions.
Similar to what we found with redesigned aspartases catalyzing asymmetric
hydroamination of acrylates,[33] we observed that the lower activity found with most
designs for 1a and 2a as compared to the template (Table 2) was not due to a reduced kcat
but due to an increase in K (Table 3). From a practical point of view, this is less disturbing
M
than the opposite, because for reasons of process economy preparative‐scale applications
must be carried out at high substrate concentrations anyway. Recently, Sun et al. [65]
attributed low activity of LEH variants obtained by directed evolution to misalignment of the
nucleophilic water, epoxide carbon and epoxide oxygen for an S 2 reaction, caused by
N
increased flexibility in the active site, but without reporting kinetic data. This explanation
likely does not hold for the computationally redesigned enzyme studied by us because such
a misalignment would reduce kcat and probably also K
except for some stilbene oxide designs (Table 3). Proper alignment of the nucleophilic water
and reacting substrate atoms for S 2 is one of the constraints in the Rosetta design process
M
, which is not what we observe,
N
and a NAC criterion during MD screening.
Surprisingly, four of the observed enantioselectivities with stilbene oxide 3a were
opposite to the Rosetta design target (Table 2). Two of these were corrected in MD
simulations. To understand how these incorrect designs could emerge, we examined
sequences and structures of Rosetta‐optimized enzyme‐substrate complexes of all
substrates (see above). For substrates 1a and 2a, the observed enantioselectivities could be
explained by the combination of steric hindrance and space‐creating mutations from the
peripheral and central side of the substrate binding pocket, acting together to steer the
substrate in a proRR or proSS binding mode. This did not hold for several of the mutants
designed for stilbene oxide 3a. For both variants that are strongly proSS selective for 3a the
only bulk introducing mutation is I116F, which would be expected to introduce steric
hindrance at the central side of the cavity and thus stimulate proRR selectivity. Also, for one
of the two most proRR selective variants (60A, e.e. > 99%) the I80W mutation would be
expected to decrease space at the peripheral side and thereby stimulate proSS selectivity.
Thus, it appears that effects of mutations on bulkiness are poorly related to
1
6
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
stereopreference in case of aromatic substrate 3a, suggesting that electronic effects that
are not well modeled, such as π‐π interactions, dominate over steric factors in determining
substrate positioning or reactivity.
Standard computational design and MD simulations do not explicitly account for π‐ π
interactions to save computation time. MD simulations can give more realistic results in
cases where aromatic interactions play a role when the force field is adapted with an
additional noncovalent interaction term. [66] Whether the main effect of π‐ π interactions is
on binding, conformational dynamics, or reactivity of bound substrates is unclear at present.
Recently, Zaugg et al. [67] investigated the origin of enantioselectivity of Aspergillus niger
epoxide hydrolase in the conversion of the chiral substrate phenyl glycidyl ether (PGE).
Molecular dynamics simulations suggested that the protein does not differentiate
enantiomers based on binding mode, and free energy calculations did not show significant
differences between (R)‐ and (S)‐PGE binding either. The authors suggested that the
enantioselectivity is due to kinetic differences. For such an α/β‐hydrolase fold epoxide
hydrolase a computational analysis is more complicated due to the multiplicity of reaction
pathways and chemical steps. Earlier, Lau et al. [68] studied murine epoxide hydrolase with
(
1S,2S)‐trans‐2‐methylstyrene oxide using ab initio and density functional calculations, and
suggested the importance of interactions between the substrate's phenyl group and
aromatic residues in the binding pocket. Moreover, Lind and Himo [69] published the reaction
mechanism of a soluble epoxide hydrolase (StEH1) converting styrene oxide. They proposed
coplanarity of the oxirane C1‐C2 carbons with the substrate's phenyl substituent, and π‐π
interactions between this phenyl group and a histidine and phenylalanine to be important
for the stabilization of the transition state and for the selectivity of the enzyme. Rinaldi et
al.[70] proposed that substrate‐dependent LEH regioselectivity is related to reorganization of
the active site towards each ligand. Based on QM/MM calculations, they confirmed that
substrate‐specific LEH regioselectivity is due to both conformational and electronic
parameters.
We conclude that computational design and MD simulations are well able to predict
and screen enantioselectivity of LEH variants in case of small aliphatic substrates. Whereas
highly selective variants for production of aromatic diols can be obtained, prediction
accuracy is lower. In view of the effect of interactions involving aromatic groups on epoxide
hydrolase enantioselectivity, rapid scoring methods that more accurately include effects of
‐ interactions appear necessary to further improve computational screening of LEH
variants acting on aromatic substrates.
Experimental Section and Computational Methods
Materials. The meso‐epoxides and their corresponding diols, oligonucleotides for
mutagenesis, organic solvents and glycerol were purchased from Sigma‐Aldrich. Restriction enzymes
and PfuUltra Hotstart PCR Master Mix were obtained from New England Biolabs and Agilent,
1
7
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
respectively. Ni‐NTA resin was purchased from GE Healthcare Life Sciences. SYPRO orange was
obtained from ThermoFisher Scientific. Complete protease inhibitor cocktail tablets were bought
from Roche. Media components were obtained from Difco (BD Biosciences).
Computational design. To design LEH variants for production of highly enantioenriched diols
from meso‐epoxides the previously developed CASCO strategy was used with only minor
modifications. [32] The X‐ray structure of the wild type LEH (Protein Databank 1NWW) was used for
computational design. Eleven positions around the active site (M32, L35, L74, M78, I80, V83, L103,
L114, I116, F134 and F139) were selected to mutate simultaneously to any of the nine hydrophobic
residues (AFGILMPVW). Each of the three substrates was docked in the enzyme active site, either in
a proRR or proSS conformation using Rosetta enzyme design [34] [71] Catalytically productive binding
modes were defined using a constraint file as previously. [32] This geometric description of how the
substrate should be bound included the obligation to form H‐bonds between the epoxide oxygen
and D101 and between the nucleophilic water and D132, Y53, and N55. Furthermore, the water
oxygen had to be close (1.8 Å) to the attacked carbon atom while the angle of nucleophilic attack
(i.e. from water oxygen, attacked carbon, and epoxide oxygen) should be close to 180°. Another
constraint was that the distance between the nucleophilic water and the non‐attacked epoxide
carbon atom should be > 3.8 Å. To hinder undesired substrate‐binding orientations, a bulky residue
(
W, F or Y) was introduced at one of the eleven target positions, since this may reduce binding poses
not contributing to the desired selectivity. [32] Rosetta enzyme design was used to simultaneously
mutate the remaining ten residues to any of the nine hydrophobic residues and sequence‐
conformational space was searched for substrate‐bound structures with low energy and a
catalytically productive binding mode.
High‐throughput‐multiple independent MD simulations (HTMI‐MD) were used for in silico
screening of the generated libraries and to rank the primary designs with an orthogonal tool (Wijma
et al., 2014). Independent initialization of multiple trajectories increases the conformational space
sampled by molecular dynamics and reduces the computational cost of the screening step as
compared to a single long MD run. [46] The reactivity and selectivity of each mutant were predicted
by scoring the fraction of snapshots in which the enzyme‐substrate complex is in a proRR or proSS
near‐attack conformation (NAC). The latter are defined by geometric constraints (Fig. 1), which
should be fulfilled for a reaction to become feasible. The geometric criteria for proRR and proSS
attack conformations were as defined using published quantum mechanical modelling. [72] The ratio
of proRR and proSS NAC frequencies were considered to reflect regioselectivity of attack and thus
product enantioselectivity according to Equation 1,
ꢇꢈꢉꢊꢊ
ꢋꢂꢃꢄꢅꢆꢌꢍꢎꢏꢏꢐ
ꢁ
ꢂꢃꢄꢅꢆ
e. e. ꢀ ꢁ
(Equation 1)
ꢂꢃꢄꢅꢆꢇꢈꢉꢊꢊꢑꢂꢃꢄꢅꢆꢇꢈꢉꢏꢏ
ꢐ
in which e.e. is the predicted enantiomeric excess, [NAC]proRR and [NAC]proSS indicate the fraction of
snapshots in which the enzyme‐substrate complex is in a proRR or proSS conformation, respectively.
Positive values indicate predicted (R,R)‐diol preference, negative values (S,S) selectivity.
The only modification from the existing procedures are listed in this paragraph. More design
calculations were done than previously, and also more seeds per MD simulation. [32] For the current
study, approximate 25 thousand design calculations were run per target substrate (Table 1), which is
two times more than previously. Like earlier, [32] a stepwise scheme to rank the variants was adopted
in which variants were eliminated as soon as they failed a criterion (Table 1). The selection criteria
1
8
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
were based on e.e.pred and [NAC] values for the preferred enantiomer. The criteria differed per target
substrate and are listed in Table 1. For each designed new variant 20‐80 independently started MD
simulations of 10 ps were used (previously maximally 20 MD simulations). For the final variants also
5
MD simulations of 100 ps were performed.
Finally, the best ranked mutants were visually inspected. For each of the targeted product
enantiomers, only variants predicted to have a high enantioselectivity were visually inspected,
starting with those variants that were predicted to have the highest fractions of NACs. The main
reasons for elimination of designs were a too spacious active site cavity or an orientation of the
substrate relative to the water that seemed in disagreement with the predicted enantioselectivity
(
Table S1). Only 15 of the 45 inspected designs were eliminated at the stage of visual inspection.
Furthermore, no mutations were added at this stage even though this is common in the field. [73] As
a result, the visual inspection only took a few hours.
Mutagenesis, expression and purification. For the expression of LEH and variants thereof in
E. coli a pBAD based expression vector was used. This vector contained the gene of the thermostable
variant LEH‐P with an N‐terminal hexa‐histidine tag. [56] The computationally designed variants of
LEH‐P were constructed by QuikChange site‐directed mutagenesis using Pfu Ultra Hotstart PCR
Mastermix (Agilent), combining multiple mutations in a single primer when possible, and omitting
sequence verification between individual mutation steps. PCR reactions, transformations, plating
and final sequencing were done in microtiter plate format.[74] The obtained plasmids were used to
transform chemically competent E. coli Top10 or E. coli NEB10β cells (Thermo Fischer Scientific). For
expression, cells were grown overnight in 5 mL Luria‐Bertani broth at 37°C. All cultures were
‐
1
supplemented with 50 µg mL ampicillin. The resulting culture was used to inoculate 500 mL Terrific
Broth medium and incubated at 37°C and 135 rpm. When an OD600 of 0.6 was reached, expression
was induced by adding 0.04% (w/v) arabinose and growth was continued at 30°C and 135 rpm. After
2
4 h the cells were harvested by centrifugation at 6,700 g and 4°C for 15 min.
For protein isolation, cells were resuspended in 50 mM HEPES buffer, pH 8, containing 500
mM NaCl (3 mL per gram of cells) and half of a protease inhibitor cocktail tablet to prevent
proteolysis (Roche Applied Science). After sonication (60 ꢒ 10 s with 20 s intervals, Labsonic M), the
extract was centrifuged at 35,200 g and 4°C for 1 h. The supernatant was collected and the enzyme
of interest was purified by gravity‐flow affinity chromatography under native conditions using Ni‐
NTA agarose resin (Thermo Fischer Scientific). The protein concentration of the collected fractions
was determined by a Bradford assay, and selected fractions were desalted by Econo‐Pac 10DG
desalting columns (Bio‐Rad). The purity of the prepared enzymes (yield 50‐150 mg per L TB medium)
was analyzed by SDS‐PAGE. Enzymes were stored at ‐80°C until further use.
Catalytic properties. Chiral chromatography was used to determine the enantioselective
hydrolysis of three meso‐epoxides (cis‐2,3‐butene oxide, cyclopentene oxide and cis‐stilbene oxide)
to chiral diols. In case of cis‐2,3‐butene oxide and cis‐stilbene oxide, 5 mg of purified enzyme was
added to 50 mM substrate (virtual concentration of the suspension) in 50 mM HEPES pH 8 (800 µL
total reaction volume). After incubation of the reaction mixture at 30°C for 1 h, 500 μL of 5 M NaCl
was added and the samples were extracted three times with 600 μL ethyl acetate. The combined
extracts containing diols were dried by adding anhydrous sodium sulfate, concentrated under
vacuum, and dissolved in 100 µL ethyl acetate. For cyclopentene oxide, reactions were done in a
similar way after which 250 mg K
2
CO was added to the reaction mixture followed by extraction for
3
1
9
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
two times by 600 µL of n‐butanol. The combined extract was dried, concentrated under vacuum and
resuspended in 100 µL n‐butanol.
In case of cis‐2,3‐butene oxide and cyclopentene oxide, chiral analysis was carried out by
injecting 2 µL of the extracts into an Agilent gas chromatograph equipped with a flame ionization
detector and a Hydrodex β‐TBDAc column (Aurora Borealis, initial temp. 40C, 10C/min to 150C,
hold 20 min). For cis‐stilbene oxide and its diols, samples were analyzed by HPLC on a Luxcellulose‐3
column (Phenomenex, Utrecht, the Netherlands) with heptane/2‐propanol (90/10) as the mobile
phase (flow rate 1 ml/min, detection at 254 nm). Samples from preparative scale reactions with cis‐
stilbene oxide and the diols were also analyzed by HPLC on a Chiralpak AS‐H column (Daicel Corp,
Illkirch, France) with n‐hexane/2‐propanol (90:10 (v/v) as the mobile phase (1 ml/min, detection at
2
54 nm). Enantiomeric excess (e.e.) values were calculated from concentrations of the (R,R) and (S,S)
product enantiomers.
To obtain steady‐state kinetic parameters, initial velocities at different substrate
concentrations were determined and fitted with the Michaelis‐Menten equation.
Determination of the apparent melting temperature. The ThermoFluor assay was used to
ꢔꢌꢌ
determine the apparent melting temperatures (Tꢓ ) of the purified enzyme variants.[75] This
method is based on monitoring the change in fluorescence of Sypro Orange dye during the thermal
unfolding of a protein. The dye binds to the unfolded and exposed hydrophobic protein core,
increasing its fluorescence signal. The assays were done as described before.[26]
Synthesis of stilbene diols. Reaction mixtures (total volume 1 mL) contained 10 mg of cis‐
stilbene oxide (final conc. 50 mM, suspension) and 6.37 mg of enzyme (final conc. 320 μM) in 50 mM
HEPES, pH 8.0, and were incubated (at 30 or 40 ºC, 135 rpm) for 48 h. Substrate and products were
extracted three times by 4 mL ethyl acetate, dried over MgSO and filtered. The solvent was
4
1
removed by a rotary evaporator. The residue was analyzed by H‐NMR for conversion and by chiral
HPLC to determine the enantiomeric excess. Chiral HPLC was used as described above.
Author's contribution
HJW designed the mutants; HA constructed, isolated and characterized the mutants,
HA, PJ and DIC measured catalytic activities; HJW, HA, and EB interpreted the mutants; DIC
and MT performed preparative experiments; HA, HJW, EB, MT and DBJ wrote the paper;
HJW and DBJ supervised the work.
Acknowledgment
This research was supported by the Dutch Ministry of Economic Affairs through BE‐
Basic [FS07.001] and the European Union’s Horizon 2020 Programme (Marie Curie Action ‐
ITN Es‐Cat) under GA No. 722610.
References
[
[
1] D. Muñoz Solano, P. Hoyos, M. J. Hernáiz, A. R. Alcántara, J. M. Sánchez‐Montero, Biores.
Technol. 2012, 115, 196‐207.
2] B. M. Nestl, S. C. Hammer, B. A. Nebel, B. Hauer, Angew. Chem. Int. Ed. 2014, 53, 3070‐3095.
2
0
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
[
[
[
[
[
[
[
3] R. N. Patel, Bioorg Med Chem. 2018, 26, 1252‐1274.
4] M. T. Reetz, J. Am. Chem. Soc. 2013, 135, 12480‐12496.
5] J. M. Choi, S. S. Han, H. S. Kim, Biotechnol. Adv. 2015, 33, 1443‐54.
6] R. A. Sheldon, J. M. Woodley, Chem. Rev. 2018, 118, 801‐838.
[7]
7] H. Sun, H. Zhang, E. L. Ang, H. Zhao, Bioorg. Med . Chem. 2018, 26, 1275‐1284
8] N. J. Turner, Nat. Chem. Biol. 2009, 5, 567‐573.
9] U. T. Bornscheuer, G. W. Huisman, R.J. Kazlauskas, S. Lutz, J.C. Moore, K. Robins, Nature 2012,
85, 185‐194.
4
[
[
[
[
[
[
10] J. Lalonde, Curr. Opin. Biotechnol. 2016, 42, 152‐158.
11] F. H. Arnold, Angew. Chem. Int. Ed. 2019, 58, 14420‐14426.
12] M. T. Reetz, Angew. Chem. Int. Ed. 2011, 50, 138‐174.
13] M. T. Reetz, Methods Mol. Biol. 2003, 230, 259‐282.
14] Z. Sun, Y. Wikmark, J. E. Bäckvall, M. T. Reetz, Chemistry 2016, 22, 5046‐54.
15] G. J. Williams, T. Woodhall, L. M. Farnsworth, A. Nelson, A. Berry, J. Am. Chem. Soc. 2006, 128,
1
6238‐16247.
16] J. G. van Leeuwen, H. J. Wijma, R. J. Floor, J. M. van der Laan, D. B. Janssen, ChemBiochem
012, 13, 137‐48.
17] K. Steiner, H. Schwab, Comput. Struct. Biotechnol. J. 2012, 2, e201209010.
18] J. F. Chaparro‐Riggers, K.M. Polizzi, A. S. Bommarius, Biotechnol. J. 2007, 2, 180‐191.
19] R. Das, D. Baker, Annu. Rev. Biochem. 2008, 77, 363‐82.
[
2
[
[
[
[
20] G. Kiss, N. Çelebi‐Ölçüm, R. Moretti, D. Baker, K.N. Houk, Angew. Chem. Int. Ed. 2013, 52,
5
700‐5725.
[
[
21] J. Damborsky, J. Brezovsky, Curr. Opin. Chem. Biol. 2014, 19, 8‐16
22] H. K. Privett, G. Kiss, T. M. Lee, R. Blomberg, R. A. Chica, L.M. Thomas, D. Hilvert, K. N. Houk, S.
L. Mayo, Proc. Natl. Acad. Sci. USA 2012, 109, 3790–3795.
[
[
23] N. S. Scrutton, IUCrJ. 2016, 4, 5‐6.
24] R. J. Fox, S. C. Davis, E. C. Mundorff, L. M. Newman, V. Gavrilovic, S. K. Ma, L. M. Chung, C.
Ching, S. Tam, S. Muley, J. Grate, J. Gruber, J. C. Whitman, R. A. Sheldon, G. W. Huisman, Nat.
Biotechnol. 2007, 25, 338‐344.
[
[
[
[
[
[
[
[
[
[
25] Z. Wu, S. B. J. Kan, R. D. Lewis, B. J. Wittmann, F. H. Arnold, Proc. Natl. Acad. Sci. 2019, 116,
8
852‐8858.
26] H. J. Wijma, R. J. Floor, P. A. Jekel, D. Baker, S. J. Marrink, D. B. Janssen, Protein Eng. Des. Sel.
014, 27, 49‐58.
2
27] H. Arabnejad, M. Dal Lago, P.A. Jekel, R.J. Floor, A.W.H. Thunnissen, A.C. Terwisscha van
Scheltinga, H.J. Wijma, D.B. Janssen, Protein Eng. Des. Sel. 2017, 30, 173‐187.
28] R.J. Floor, H.J. Wijma, D.I. Colpa, A. Ramos‐Silva, P.A. Jekel, W. Szymański, B.L. Feringa, S.J.
Marrink, D.B. Janssen, ChemBiochem, 2014, 15, 1660‐1672.
29] R. J. Floor, H. J. Wijma, P. A. Jekel, A. C. Terwisscha van Scheltinga, B. W. Dijkstra, D. B. Janssen,
Proteins 2015, 83, 940‐951.
30] A. Goldenzweig, M. Goldsmith, S. E. Hill, O. Gertman, P. Laurino, Y. Ashani, R. L. Lieberman,
Molec. Cell, 2016, 63, 337‐346.
31] D. Bednar, K. Beerens, E. Sebestova, J. Bendl, S. Khare, R. Chaloupkova, J. Damborsky, PLoS
Comp. Biol. 2015, 11, e1004556.
32] H. J. Wijma, R. J. Floor, S. Bjelic, S. J. Marrink, D. Baker, D. B. Janssen, Angew. Chem. Int. Ed.
2
015, 54, 3726‐3730.
33] R. Li, H. J. Wijma, L. Song, Y. Cui, M. Otzen, Y. Tian, J. Du, T. Li, D. Niu, Y. Chen, J. Feng, J. Han,
H. Chen, Y. Tao, D. B. Janssen, W. Wu. Nature Chem. Biol. 2018, 14, 664‐670.
34] A. Leaver‐Fay, M. Tyka, S. M. Lewis, O. F. Lange, J. Thompson, R. Jacak, K. Kaufman, P. D.
Renfrew, C. A. Smith, W. Sheffler, I. W. Davis, S. Cooper, A. Treuille, D. J. Mandell, F. Richter, Y.
E. Ban, S. J. Fleishman, J. E. Corn, D. E. Kim, S. Lyskov, M. Berrondo, S. Mentzer, Z. Popović, J. J.
2
1
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
Havranek, J. Karanicolas, R. Das, J. Meiler, T. Kortemme, J. J. Gray, B. Kuhlman, D. Baker, P.
Bradley, Methods Enzymol. 2011, 487, 545‐574.
35] T. C. Bruice. Acc. Chem. Res. 2002, 3, 139‐148.
36] T. C. Bruice, S. Benkovic. Biochemistry 2000, 21, 6267‐6274.
37] A. Warshel, P.K. Sharma, M. Kato, Y. Xiang, H. Liu, M.H.M. Olsson. Chem. Rev. 2006, 8, 3210‐
[
[
[
3
235.
[
[
[
38] G. Kiss, D. Rothlisberger, D. Baker, K.N. Houk. Protein Sci. 2010, 9, 1760‐1773.
39] G. Kiss, V.S. Pande, K.N. Houk. Methods Enzymol. 2013, 523, 145‐170.
40] H.J. Wijma. In: A.S. Svendsen (ed.), Understanding enzymes; function, design, engineering and
analysis. Taylor and Francis, New York, pp 805‐833.
[
[
41] L.S.D. Caves, J.D. Evanseck, M. Karplus. Protein Sci. 1998, 3, 649‐666.
42] L. Monticelli, E.J. Sorin, D. P. Tieleman, V.S. Pande, G. Colombo. J. Comput. Chem. 2008, 11,
1
740‐1752.
[
[
[
[
[
[
43] S. Genheden, U. Ryde. J. Comput. Chem. 2010, 4, 837‐846.
44] S. Genheden, U. Ryde. J. Comput. Chem. 2011, 2, 187‐195.
45] S. Genheden, C. Diehl, M. Akke, U. Ryde. J. Chem. Theory Comput. 2010, 7, 2176‐2190.
46] H. J. Wijma, S. J. Marrink, D. B. Janssen, Chem. Inf. Model. 2014, 54, 2079‐92.
47] H. Zheng, M. T. Reetz, J. Am. Chem. Soc. 2010, 132, 15744‐15751.
48] M. Arand, B. M. Hallberg, J. Zou, T. Bergfors, F. Oesch, M. J. van der Werf, J. A. de Bont, T. A.
Jones, S. L. Mowbray, EMBO J. 2003, 22, 2583‐2592.
[
[
49] K. H. Hopmann, B. M. Hallberg, F. Himo, J. Am. Chem. Soc. 2005, 127, 14339‐14347.
50] Z. Sun, R. Lonsdale, X. D. Kong, J. H. Xu, J. Zhou, M. T. Reetz, Angew. Chem. Int. Ed. 2015, 54,
1
2410‐12415.
[
[
[
51] Z. Sun, R. Lonsdale, G. Li, M. T. Reetz, ChemBiochem 2016a, 17, 1865‐1872.
52] Z. Sun, R. Lonsdale, L. Wu, G. Li, A. Li, J. Wang, M. T. Reetz, ACS Catalysis 2016, 6, 1590‐1597.
53] A. Li, C. G. Acevedo‐Rocha, Z. Sun, T. Cox, J. L. Xu, M. T. Reetz, ChemBiochem 2018, 19, 221‐
2
28.
[
[
[
54] G. Li, H. Zhang, Z. Sun, X. Liu, M. T. Reetz, ACS Catalysis, 2016, 6, 3679‐3687.
55] G. Li, J. B. Wang, M.T. Reetz, Bioorg Med Chem. 2018, 26, 1241‐1251.
56] R.J. Floor, H.J. Wijma, P. A. Jekel, A.C. Terwisscha van Scheltinga, B.W. Dijkstra, D.B. Janssen,
Proteins, 2015, 83, 940‐951.
[
[
[
[
57] J. Bloom, S. Labthavikul, C. Otey, F.H. Arnold, Proc. Natl. Acad. Sci. USA, 2006, 103, 5869–5874.
58] W. Besenmatter, P. Kast, D. Hilvert, Protein Struct. Funct. Bioinform., 2007, 66, 500–506.
59] N. Tokuriki, D.S. Tawfik, Curr. Opin. Struct. Biol., 2009, 19, 596–604.
60] B. A. Amrein, F. Steffen‐Munsberg, I. Szeler, M. Purg, Y. Kulkarni, S. C. Kamerlin, IUCrJ. 2017,
5
0‐64.
61] D. Gao, H. Cho, W. Yang, Y. Pan, G. Yang, H. Tai, C. Zhan, Angew. Chem. Int. Ed. 2006, 45, 653–
57.
[
[
6
62] F. Zheng, W. Yang, M.C. Ko, J. Liu, H. Cho, D. Gao, M. Tong, H.H. Tai, J.H. Woods, C.G. Zhan, J.
Am. Chem. Soc. 2008, 130, 12148–12155.
[
[
63] A. Seifert, M. Antonovici, B. Hauer, J. Pleiss, ChemBiochem 2011, 12, 1346–1351.
64] P.O. Syren, P. Hendil‐Forssell, L. Aumailley, W. Besenmatter, F. Gounine, A. Svendsen, M.
Martinelle, K. Hult, ChemBiochem 2012, 13, 645‐648.
[
65] Z. Sun, L. Wu, M. Bocola, H. C. S. Chan, R. Lonsdale, X. D. Kong, S. Yuan, J. Zhou, M. T. Reetz,
J. Am. Chem. Soc. 2018, 140, 310‐318.
[
[
66] H. Yuki, Y. Tanaka, M. Hata, H. Ishikawa, S. Neya, T. J. Comput. Chem. 2007, 28, 1091‐1099.
67] J. Zaugg, Y. Gumulya, M. Bodén, A. E. Mark, A. K. Malde, J. Chem. Inf. Model. 2018, 58, 630‐
6
40.
[
[
68] Y. E. Lau, E. Z. Newby, T. C. Bruice, J. Am. Chem. Soc. 2001, 123, 3350‐3357.
69] M. E. S. Lind, F. Himo, ACS Catalysis 2016, 6, 8145‐8155.
2
2
This article is protected by copyright. All rights reserved.

ChemBioChem
10.1002/cbic.201900726
[
70] S. Rinaldi, M. W. van der Kamp, K. E. Ranaghan, A. J. Mulholland, G. Colombo, ACS Catalysis
018, 8, 5698‐5707.
2
[
[
[
[
[
71] F. Richter, A. Leaver‐Fay, S. D. Khare, S. Bjelic, D. Baker, PLoS One 2011, 6, e19230.
72] M. E. S. Lind, H. Fahmi, Angew. Chem. Int. Ed. 2013, 52, 4563‐4567.
73] L. G. Nivón, S. Bjelic, C. King, D. Baker, Proteins 2014, 82, 858‐866.
74] M. L. J. Fürst, C. Martin, N. Lončar, M.. W. Fraaije, Methods Enzymol. 2018, 608, 151‐87.
75] U. B. Ericsson, B. M. Hallberg, G. T. Detitta, N. Dekker, P. Nordlund, Anal. Biochem. 2006, 357,
2
89‐98.
2
3
This article is protected by copyright. All rights reserved.