A smart library of epoxide hydrolase variants and the top hits for synthesis of (S)-β-blocker precursors

Article abstract of DOI:10.1002/anie.201402653

Microtuning of the enzyme active pocket has led to a smart library of epoxide hydrolase variants with an expanded substrate spectrum covering a series of typical β-blocker precursors. Improved activities of 6- to 430-fold were achieved by redesigning the active site at two predicted hot spots. This study represents a breakthrough in protein engineering of epoxide hydrolases and resulted in enhanced activity toward bulky substrates. Hot pockets: Microtuning of the enzyme active pocket gives a smart library of epoxide hydrolase variants with an expanded substrate spectrum covering a series of typical β-blocker precursors. Improved activities of 6- to 430-fold were achieved by redesigning the active site at two predicted hot spots, and enhanced activity toward bulky substrates was found.

Full text of DOI:10.1002/anie.201402653

Angewandte
Chemie
DOI: 10.1002/anie.201402653
Enzyme Catalysis
A Smart Library of Epoxide Hydrolase Variants and the Top Hits for
Synthesis of (S)-b-Blocker Precursors**
Xu-Dong Kong, Qian Ma, Jiahai Zhou,* Bu-Bing Zeng,* and Jian-He Xu*
Abstract: Microtuning of the enzyme active pocket has led to
a smart library of epoxide hydrolase variants with an expanded
substrate spectrum covering a series of typical b-blocker
precursors. Improved activities of 6- to 430-fold were achieved
by redesigning the active site at two predicted hot spots. This
study represents a breakthrough in protein engineering of
epoxide hydrolases and resulted in enhanced activity toward
bulky substrates.
S
ince the substrate specificity of enzymes limits their broad
application as biocatalysts in industry, various efforts have
been made toward expanding the substrate spectrum or even
diversifying the reaction type of a single biocatalyst.[1] Among
them, a minimalist active site redesign,[1a] based on the
structural information of an enzyme, would greatly change its
substrate specificity. In this study, starting from a robust
epoxide hydrolase (EH) with high activity toward a simple
model substrate, phenyl glycidyl ether (1a), we developed
a small but smart library of EH variants with improved
activity (by 6–430-fold) for nine typical b-blocker precursors
(Scheme 1), by redesigning the active site of the enzyme for
two predicted hot spots, namely Met145 and Phe128.
Scheme 1. Typical epoxide substrates chosen to assay variants of the
epoxide hydrolase BmEH in hydrolysis.
majority of b-blockers depend on their S enantiomer for
binding with the b-adrenergic receptor, and in general the
S enantiomers are 10–500-fold more potent than the R enan-
tiomers.[4] Biocatalytic synthesis of single-enantiomer b-
blockers or their precursors has been reported, and mainly
involves using lipases or esterases as biocatalysts.[5] The
enantioselective hydrolysis of racemic epoxides using cofac-
tor-independent EHs is a promising approach for obtaining
enantiopure epoxides or diols as key chiral precursors of b-
blockers. However, EHs rarely possess sufficient activity for
epoxides with bulky substituents,[6] so the industrial produc-
tion of these compounds in enantiomerically pure form with
EHs remains a big challenge.
b-Adrenergic receptor blocking agents (b-blockers) are
a group of popular drugs used widely for cardiovascular
therapies.[2] b-Blockers are commercially available on the
market as active pharmaceutical ingredients, primarily as
racemates.[3] However, the each enantiomer of a b-blocker
should be used individually because of their different
pharmacokinetic and pharmacodynamic properties. The
In the last decades, interest in microbial EHs has arisen
primarily because of their application in the synthesis of
enantiopure epoxides/diols. Nevertheless, the use of EHs in
the kinetic resolution of aryl glycidyl ethers for the synthesis
of (S)-b-blockers, or other chiral drugs like Ranexa and Flivas,
is still hindered by either the low activity or insufficient
enantioselectivity of existing EHs. We have cloned a novel
EH (BmEH) from Bacillus megaterium ECU1001 with
unusual R enantioselectivity and an activity of 83 UmgÀ1
protein toward the substrate phenyl glycidyl ether (1a).[7]
However, for various substrates with a bulky substituent on
the phenyl ring (Scheme 1), the activity of BmEH was found
to decrease significantly. For example, when BmEH was
employed for the resolution of Alprenolol and Propranolol
precursors (4a and 9a), the activity decreased to less than 5%
when compared with that for 1a. Therefore, to meet the need
of chiral-b-blocker synthesis, it is imperative to alter the
substrate specificity of BmEH toward bulky epoxides by
protein engineering.
[*] X.-D. Kong, J.-H. Xu
State Key Laboratory of Bioreactor Engineering, East China
University of Science and Technology, Shanghai 200237 (China)
E-mail: jianhexu@ecust.edu.cn
Q. Ma, B.-B. Zeng
Shanghai Key Laboratory of New Drug Design, East China
University of Science and Technology, Shanghai 200237 (China)
E-mail: zengbb@ecust.edu.cn
J. Zhou
State Key Laboratory of Bio-organic and Natural Products
Chemistry, Shanghai Institute of Organic Chemistry
Chinese Academy of Sciences, Shanghai 200032 (China)
E-mail: jiahai@sioc.ac.cn
[**] Financial support was received from the National Natural Science
Foundation of China (No. 21276082), Ministry of Science and
Technology, P.R. China (Nos. 2011AA02A210 and 2011CB710800),
and Shanghai Commission of Science and Technology (No.
11431921600). We are grateful to Prof. Kui-Ling Ding at SIOC/CAS
and Prof. Uwe T. Bornscheuer at Greifswald University for their
critical reading and valuable suggestions.
We have recently determined the crystal structures of
both BmEH (PDB ID: 4NZZ) and its complex with
a substrate analogue (PDB ID: 4O08), as shown in Figure 1.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201402653.
Angew. Chem. Int. Ed. 2014, 53, 6641 –6644
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
6641

.
Angewandte
Communications
separately for each position using seven preselected amino
acids (Ala, Cys, Ile, Leu, Ser, Thr, and Val),[9] to modify the
active pocket with different levels of hindrance and hydro-
phobicity.
Aiming to screen for highly active BmEH variants for
each substrate, we first tested the activities of the lyophilized
cell-free extracts (CFEs) of 15 EH variants from the library
on 10 chosen epoxides using reverse-phase HPLC. As shown
in Figure 2 (for more details, see also Table S1 in the
Supporting Information), each of the 10 indicative substrates
could find at least one variant (best hit) with much higher
activity than the wild-type enzyme (BmEHWT). The variants
with mutations at Met145 showed the highest activity for the
ortho-substituted 3a and 4a, whereas the Phe128 mutants
were more active for the meta-substituted 7a and 9a. This
outcome is consistent with the locations of these two residues
relative to the phenyl ring of phenoxyacetamide (POA;
Figure 1). The highest activity (47.9 UmgÀ1 CFE) was
observed for 3a with the variant M145S, thus representing
a greater than tenfold increase when compared with that of
BmEHWT for 1a (4.32 UmgÀ1 CFE).
Figure 1. Locations of Met145 and Phe128 chosen for mutation
around the active site of BmEH. Molecule represented as space-filling
model represents phenoxyacetamide (POA), an analogue of epoxide
1a. PDB ID: 4O08.
To characterize the real improvement in catalytic effi-
ciency, BmEHWT and the respective best variant of each
substrate were purified and assayed for their specific activ-
ities. In comparison with BmEHWT, activities of the best hits
toward 1a–10a increased by 1.75- to greater than 400-fold
(see Table S2), which is coincident with the results of crude
enzymes (CFEs). Among all the variants, M145S for 3a and
M145A for 4a showed the highest activities, of up to 1062 and
It is worth noting that this EH, having an a/b-hydrolase fold,
has a particularly deep and hydrophobic active cavity. By
analyzing the active site of BmEH, two residues (Met145 and
Phe128) were identified as potential hot spots for enhancing
the BmEH activity toward the bulky substrates mentioned
above. By referring to several successful examples for
creating a smart library,[8] Met145 and Phe128 were mutated
Figure 2. Overview of the BmEH variants performance in hydrolytic reactions of various epoxides. Data are shown as logarithmic in the radar
map. Activities equal or lower than 0.01 UmgÀ1 cell-free extract are shown as 0.01 UmgÀ1. For detailed accounts of the numbers, see the
Supporting Information.
6642 www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 6641 –6644

Angewandte
Chemie
531 UmgÀ1 protein, respectively. Such high activity has rarely
been reported in EHs, thus revealing the amazing power of
biocatalyst redesign. However, the combination of F128S with
M145S or M145I did not have a synergistic effect on
improving the enzymatic activities, but instead caused
severe protein-folding problems (see Figure S1).
Activities, which were two orders of magnitude higher,
were observed with the variant M145S for the epoxides 3a–5a
and 10a, and the variant F128S for the substrate 9a (see
Table S2). To elucidate the reason for the increases in activity
from a biochemical perspective, the apparent steady-state
kinetic parameters of these substrate–enzyme pairs was
measured (Table 1). The Km constants of the EH variants
Figure 3. Relationship between the substrate polarity and the highest
activity observed among BmEH variants. Substrates are ranked in the
order of their logP values (obtained from the SciFinder database; see
Table S5).
Table 1: Apparent kinetic parameters of BmEHWT and its variants.[a]
Substrate–Enzyme
Km
[mm]
kcat
[sÀ1
kcat/Km
Fold[e]
]
[sÀ1 mÀ1
]
highest activity observed among all the BmEH variants, we
found a trend: for substrates with lower hydrophobicity,
usually lower enzyme activities were also observed (Figure 3).
However, the epoxides 8a and 9a did not exhibit additional
increases in activity, as expected. This observation is probably
because they have substitutions on both the ortho and
meta positions of the phenyl ring, thus resulting in severe
steric hindrance.
Some clues could also be found from the kinetic
parameters of the variants (Table 1). When BmEH was
successfully engineered toward a certain substrate (e.g., 3a–
5a and 9a), its kcat value achieved a level of 102–103 sÀ1, but
the Km value for the substrates had more remarkable differ-
ences. For 5a and 10a, having hydrophilic substitutions,
a Km value which was nearly one order of magnitude higher
was observed, thus the catalytic efficiency tended to be
limited by its weaker substrate binding affinity. A possible
explanation might be based on the deeply buried hydrophobic
active site of BmEH, and makes the hydrophobic interactions
of the protein with the epoxides the main driving force for
substrate binding.
3a–WT
3a–M145S
4a–WT
4a–M145S
5a–WT
5a–M145S
9a–WT
9a–F128S
10a–WT
10a–M145S
3.96
1.67
1.61
10.8
1245
1.95
2.73ꢀ103
7.46ꢀ105
1.21ꢀ103
8.54ꢀ105
n.a.[d]
273
705
362
896
169
0.82
701
>20[b]
>20[b]
1.32
>0.22[c]
>135[c]
0.60
n.a.[d]
4.50ꢀ102
4.03ꢀ105
n.a.[d]
0.49
199
>20[b]
6.53
>0.10[c]
16.7
2.55ꢀ103
[a] The kinetic parameters were determined with a substrate concen-
tration of 0.6–10 mm for 3a, 4a, 9a, and 0.6–20 mm for 5a and 10a. See
the Supporting Information for experimental conditions. [b] The
K
m values were beyond the concentration range of the substrate.
[c] Calculated based on the highest rate detected. [d] Not available in this
case. [e] The folds of kcat/Km improved for particular variants over the
wild-type. In the case of 5a and 10a, the folds of specific activity
improvement are listed instead.
for substrates 3a, 4a, 9a, and 10a decreased significantly
when compared with that of BmEHWT, thus indicating a two-
to threefold stronger binding affinity. However, the variants
exhibited kcat values which were as much as 100–300-fold
higher than that of BmEHWT. Therefore, the activity increase
of the BmEH variants is mainly a result of the dramatic
increase in their turnover frequencies.
Subsequently, several best-variants for each substrate
were examined for their enantioselectivity (see Table S3). For
the majority of the substrates, the variants have a moderate to
good enantioselectivity (E = 30 ~ 200). When compared with
BmEHWT, at least for epoxides 7a and 9a (E = 19 and 25),[7b]
the variants had a significantly improved enantioselectivity
(E = 70 and 53, respectively), thus indicating greater potential
for bioresolving these epoxides. As shown in Table S4, the
improved enantioselectivities of BmEHM145I for 7a and
BmEHF128S for 9a are mainly derived from the more dominant
increases in kcat toward R epoxides than S epoxides.
As mentioned above, some BmEH variants could effec-
tively convert epoxides into the corresponding diols with
moderate to good enantioselectivity. For epoxides 2a–4a and
7a–9a, the most active variants were chosen for enzymatic
resolution at 10 gLÀ1 (ca. 50–60 mm). To facilitate the epoxide
dissolution and to prevent spontaneous hydrolysis, a biphasic
system of isopropyl ether/water (1/4, v/v) was composed for
the preparative bioresolution (Table 2). Under these reaction
Table 2: Preparative bioresolution of epoxides (10 gLÀ1) using BmEH
variants in a biphasic system containing 20% diisopropyl ether.[a]
Sub.
Enzyme/
Reaction
Volume [L]/t [h]
S Epoxide
ee/yield [%]
R Diol
ee/yield [%]
Load [gLÀ1
]
2a
3a
4a
7a
8a
9a
M145I/0.5
0.1/6
98.8/37.3
97.8/38.6
97.4/43.4
99.5/38.7
96.6/44.1
99.5/44.6
67.9/32.4
71.0/51.7
92.0/46.0
74.9/39.8
96.6/39.9
99.9/37.2
M145S/0.3
M145A/0.3
M145I/0.5
M145T/1.0
F128S/0.2
0.1/15
0.1/16
1.0/11
0.1/16
1.0/16
The activity of an enzyme for a certain substrate is
determined by the structure of both protein and ligand, and
might be analyzed based on hydrophobicity and steric
hindrance. By ranking the substrates in the order of their
hydrophobicity (logP values) and comparing them with the
[a] See the Supporting Information for experimental conditions.
Angew. Chem. Int. Ed. 2014, 53, 6641 –6644
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 6643

.
Angewandte
Communications
conditions, even for the poorest epoxide, 3a, for which the
best variant gave a relatively low enantioselectivity, the
product (S)-3a could be obtained in fairly good purity (97.8%
ee) and an acceptable yield (38.6%). For the resolution of the
epoxides 7a and 9a on a greater than 10 gram scale, their
S enantiomers (> 99% ee) were recovered in 38.7 and 44.6%
yield, respectively. Furthermore, the dosage of biocatalyst
could be reduced to merely 0.2 gLÀ1 for 9a, thus resulting in
a substrate-to-enzyme (S/E) ratio of 50 (w/w), which is in
contrast to 10 gLÀ1 for BmEHWT resolving its best substrate,
ortho-methyl phenyl glycidyl ether (S/E = 3).[7b] Notbaly,
during the process of 2a and 9a resolution, the diols formed
were rarely soluble in either aqueous or organic phase, thus
making it easy to separate the epoxide from diol by simple
filtration. This filtration should greatly facilitate downstream
processing and improve the enantiopurity of the isolated
diols. For instance, the ee value of 9b recovered from
filtration was as high as greater than 99%, which is in
contrast to the 36% ee of the residual 9b in the reaction
solvent. Such an innovation is considered to be extremely
suitable for practical application on the large scale.
During the course of this work, we succeeded in expand-
ing the substrate scope of BmEH by engineering two hot
spots close to its active site, and it had a remarkable impact on
its preference for bulky epoxides. The racemic epoxide
precursors of Moprolol (2a), SR-59230 A (3a), Aprenolol
(4a), Toliprolol (7a), Xibenolol (8a), and Propranolol (9a)
were biochemically resolved, thus affording the S epoxides in
96.6–99.5% ee and 37–45% yield. To the best of our knowl-
edge, no suitable EH has been reported for the majority of
these practically useful epoxides.
[1] a) M. D. Toscano, K. J. Woycechowsky, D. Hilvert, Angew.
Chem. 2007, 119, 3274 – 3300; Angew. Chem. Int. Ed. 2007, 46,
3212 – 3236; b) H. Jochens, M. Hesseler, K. Stiba, S. K. Padhi,
R. J. Kazlauskas, U. T. Bornscheuer, ChemBioChem 2011, 12,
1508 – 1517.
[2] a) H. L. Kennedy, Am. J. Med. 2001, 110, 2S – 6S; b) R. Mehvar,
D. R. Brocks, J. Pharm. Pharm. Sci. 2001, 4, 185 – 200.
[3] J. Agustian, A. H. Kamaruddin, S. Bhatia, Process Biochem.
2010, 45, 1587 – 1604.
[4] a) M. Muthukrishnan, D. R. Garud, R. R. Joshi, R. A. Joshi,
Tetrahedron 2007, 63, 1872 – 1876; b) J. A. Nathanson, J. Phar-
macol. Exp. Ther. 1988, 245, 94 – 101; c) K. Stoschitzky, G.
Egginger, G. Zernig, W. Klein, W. Lindner, Chirality 1993, 5, 15 –
19.
[5] a) T. N. B. Kaimal, R. B. N. Prasad, T. C. Rao, Biotechnol. Lett.
1992, 14, 21 – 26; b) N. Thakkar, A. Banerji, H. Bevinakatti,
Biotechnol. Lett. 1995, 17, 217 – 218; c) S. Pedragosa-Moreau, C.
Morisseau, J. Baratti, J. Zylber, A. Archelas, R. Furstoss,
Tetrahedron 1997, 53, 9707 – 9714; d) M. Kapoor, N. Anand, K.
Ahmad, S. Koul, S. S. Chimni, S. C. Taneja, G. N. Qazi, Tetrahe-
dron: Asymmetry 2005, 16, 717 – 725; e) J. J. Wang, C. Min, G. J.
Zheng, Ann. Microbiol. 2010, 60, 59 – 64; f) K. Wꢀnsche, U.
Schwaneberg, U. T. Bornscheuer, H. H. Meyer, Tetrahedron:
Asymmetry 1996, 7, 2017 – 2022.
[6] a) Y. Xu, J.-H. Xu, J. Pan, Y.-F. Tang, Biotechnol. Lett. 2004, 26,
1217 – 1221; b) Y. Sheng, C. Wei, Z. Zhang, S. Wang, Q. Zhu,
Appl. Biochem. Biotechnol. 2011, 164, 125 – 132; c) N. Bala, S. S.
Chimni, H. S. Saini, B. S. Chadha, J. Mol. Catal. B 2010, 63, 128 –
134; d) W. Choi, E. Huh, H. Park, E. Lee, C. Choi, Biotechnol.
Tech. 1998, 12, 225 – 228.
[7] a) Y. F. Tang, J. H. Xu, Q. Ye, B. Schulze, J. Mol. Catal. B 2001,
13, 61 – 68; b) J. Zhao, Y.-Y. Chu, A.-T. Li, X. Ju, X.-D. Kong, J.
Pan, Y. Tang, J.-H. Xu, Adv. Synth. Catal. 2011, 353, 1510 – 1518.
[8] a) S. Bartsch, R. Kourist, U. T. Bornscheuer, Angew. Chem. 2008,
120, 1531 – 1534; Angew. Chem. Int. Ed. 2008, 47, 1508 – 1511;
b) H. Jochens, U. T. Bornscheuer, ChemBioChem 2010, 11,
1861 – 1866; c) X. Feng, J. Sanchis, M. T. Reetz, H. Rabitz,
Chem. Eur. J. 2012, 18, 5646 – 5654.
The results of our study illustrate the significant potential
of semirational design for engineering biocatalysts from the
perspective of expanding the synthetic application of one
single enzyme. Although the double mutation of the two hot
spots might potentially provide even more active variants,
simple combination of the best single-site mutants did not
have a synergistic effect in our case. Therefore, the combina-
torial multipoint mutation methods such as CASTing may be
an alternative route to further improve our BmEH.[10]
[9] Preliminary experiments showed that the mutation of these two
sites to ionizable residues would significantly cause the expres-
sion of inclusion bodies.
[10] a) M. T. Reetz, L. W. Wang, M. Bocola, Angew. Chem. 2006, 118,
1258 – 1263; Angew. Chem. Int. Ed. 2006, 45, 1236 – 1241; b) Q.
Wu, P. Soni, M. T. Reetz, J. Am. Chem. Soc. 2013, 135, 1872 –
1881; c) H. Zheng, D. Kahakeaw, J. P. Acevedo, M. T. Reetz,
ChemCatChem 2010, 2, 958 – 961; d) M. T. Reetz, C. Torre, A.
Eipper, R. Lohmer, M. Hermes, B. Brunner, A. Maichele, M.
Bocola, M. Arand, A. Cronin, Y. Genzel, A. Archelas, R.
Furstoss, Org. Lett. 2004, 6, 177 – 180; e) M. T. Reetz, M. Bocola,
J. D. Carballeira, D. X. Zha, A. Vogel, Angew. Chem. 2005, 117,
4264 – 4268; Angew. Chem. Int. Ed. 2005, 44, 4192 – 4196.
Received: February 23, 2014
Revised: April 1, 2014
Published online: May 19, 2014
Keywords: enantioselectivity · enzyme catalysis ·
.
kinetic resolution · protein engineering · structure–
activity relationships
6644 www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 6641 –6644