Biocatalytic and Structural Properties of a Highly Engineered Halohydrin Dehalogenase

Article abstract of DOI:10.1002/cbic.201300005

Two highly engineered halohydrin dehalogenase variants were characterized in terms of their performance in dehalogenation and epoxide cyanolysis reactions. Both enzyme variants outperformed the wild-type enzyme in the cyanolysis of ethyl (S)-3,4-epoxybutyrate, a conversion yielding ethyl (R)-4-cyano-3-hydroxybutyrate, an important chiral building block for statin synthesis. One of the enzyme variants, HheC2360, displayed catalytic rates for this cyanolysis reaction enhanced up to tenfold. Furthermore, the enantioselectivity of this variant was the opposite of that of the wild-type enzyme, both for dehalogenation and for cyanolysis reactions. The 37-fold mutant HheC2360 showed an increase in thermal stability of 8°C relative to the wild-type enzyme. Crystal structures of this enzyme were elucidated with chloride and ethyl (S)-3,4-epoxybutyrate or with ethyl (R)-4-cyano-3-hydroxybutyrate bound in the active site. The observed increase in temperature stability was explained in terms of a substantial increase in buried surface area relative to the wild-type HheC, together with enhanced interfacial interactions between the subunits that form the tetramer. The structures also revealed that the substrate binding pocket was modified both by substitutions and by backbone movements in loops surrounding the active site. The observed changes in the mutant structures are partly governed by coupled mutations, some of which are necessary to remove steric clashes or to allow backbone movements to occur. The importance of interactions between substitutions suggests that efficient directed evolution strategies should allow for compensating and synergistic mutations during library design.

Full text of DOI:10.1002/cbic.201300005

CHEMBIOCHEM
FULL PAPERS
DOI: 10.1002/cbic.201300005
Biocatalytic and Structural Properties of a Highly
Engineered Halohydrin Dehalogenase
[a]
[a]
[b]
[b]
[a]
Marcus Schallmey, Robert J. Floor, Bernhard Hauer, Michael Breuer, Peter A. Jekel,
[a]
[c]
[a]
Hein J. Wijma, Bauke W. Dijkstra, and Dick B. Janssen*
Two highly engineered halohydrin dehalogenase variants were
characterized in terms of their performance in dehalogenation
and epoxide cyanolysis reactions. Both enzyme variants out-
performed the wild-type enzyme in the cyanolysis of ethyl (S)-
or with ethyl (R)-4-cyano-3-hydroxybutyrate bound in the
active site. The observed increase in temperature stability was
explained in terms of a substantial increase in buried surface
area relative to the wild-type HheC, together with enhanced
interfacial interactions between the subunits that form the
tetramer. The structures also revealed that the substrate bind-
ing pocket was modified both by substitutions and by back-
bone movements in loops surrounding the active site. The ob-
served changes in the mutant structures are partly governed
by coupled mutations, some of which are necessary to remove
steric clashes or to allow backbone movements to occur. The
importance of interactions between substitutions suggests
that efficient directed evolution strategies should allow for
compensating and synergistic mutations during library design.
3,4-epoxybutyrate, a conversion yielding ethyl (R)-4-cyano-3-
hydroxybutyrate, an important chiral building block for statin
synthesis. One of the enzyme variants, HheC2360, displayed
catalytic rates for this cyanolysis reaction enhanced up to ten-
fold. Furthermore, the enantioselectivity of this variant was the
opposite of that of the wild-type enzyme, both for dehaloge-
nation and for cyanolysis reactions. The 37-fold mutant
HheC2360 showed an increase in thermal stability of 88C rela-
tive to the wild-type enzyme. Crystal structures of this enzyme
were elucidated with chloride and ethyl (S)-3,4-epoxybutyrate
Introduction
Halohydrin dehalogenases (EC 4.5.1.–) catalyze the elimination
of halides from vicinal haloalcohols, resulting in epoxide ring
halohydrin dehalogenases, to the short-chain dehydrogenase/
[8,9]
reductase (SDR) enzyme superfamily.
[
1]
formation. Although the natural role of these enzymes is in
the dehalogenation of halogenated xenobiotics such as epi-
Halohydrin dehalogenases each possess a proton-abstract-
ing catalytic triad (Ser-Tyr-Arg), which has a function similar to
that of the commonly observed catalytic triad in SDR enzymes
(Ser-Tyr-Lys). Halohydrin dehalogenases also exhibit the Ross-
mann-fold structural unit present in dehydrogenases, but lack
the typical Gly-rich motif required for nucleotide cofactor bind-
[
2]
chlorohydrin, they can also be used for the preparation of
[
3]
enantiopure haloalcohols and epoxides. The halohydrin deha-
logenase HheC from Agrobacterium radiobacter AD1 can be
recombinantly produced in Escherichia coli in high yields, and
this has allowed detailed biochemical and mechanistic investi-
[7]
ing. Instead, halohydrin dehalogenases each contain a spa-
cious anion binding pocket formed by a loop that—in SDR
proteins—participates in substrate and cofactor binding. Mech-
[
4]
[4–6]
gation, as well as engineering studies.
The enzyme is
[
7]
active as a homotetramer and belongs, together with other
[10]
[7,11]
[12,13]
anistic,
structural,
and computational
studies have
explained the observed high regioselectivity and enantiopre-
ference of the HheC-type halohydrin dehalogenase for R-con-
figured substrates.
[
a] M. Schallmey, R. J. Floor, P. A. Jekel, Dr. H. J. Wijma, Prof. Dr. D. B. Janssen
Department of Biochemistry
Groningen Biomolecular Sciences and Biotechnology Institute
University of Groningen
Nijenborgh 4, 9747 AG Groningen (The Netherlands)
HheC catalyzes haloalcohol dehalogenation in equilibrium
with the reverse reaction of halide-mediated epoxide ring
opening. As a promiscuous activity, HheC also catalyzes nucle-
ophilic epoxide ring opening with various small anions such as
E-mail: d.b.janssen@rug.nl
+
[b] Prof. Dr. B. Hauer, Dr. M. Breuer
BASF SE, Fine Chemicals and Biocatalysis Research
GVF, 67056 Ludwigshafen (Germany)
[14,15]
cyanide, azide, nitrite, and cyanate.
Some examples of
these biotechnologically interesting reactions are the prepara-
[
c] Prof. Dr. B. W. Dijkstra
[16]
Protein Crystallography, Groningen Biomolecular Sciences and
Biotechnology Institute, University of Groningen
Nijenborgh 7, 9747 AG Groningen (The Netherlands)
tion of optically active oxazolidinones
and (R)-azidoalco-
[17]
hols as valuable optically active intermediates.
The formation of carbon-carbon bonds through enzyme-cat-
alyzed epoxide cyanolysis was one of the first promiscuous ac-
tivities observed for halohydrin dehalogenases and has been
exploited for the production of optically pure b-hydroxynitriles
+
[
] Present address:
Institute of Technical Biochemistry, University of Stuttgart
Allmandring 31, 70569 Stuttgart (Germany)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201300005.
[18]
such as (R)-g-chloro-b-hydroxybutyronitrile.
Other aliphatic
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 0000, 00, 1 – 13
&
1
&
ÞÞ
These are not the final page numbers!

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
and aromatic epoxides have also been shown to be converted
to the corresponding b-hydroxynitriles with good yields and
[
14,19]
selectivities.
In particular, chiral C3 and C4 compounds,
such as 4-chloro-3-hydroxybutyric acid (1; Scheme 1), are very
[23,26]
Scheme 2. Biocatalytic route to optically pure hydroxynitrile (R)-6.
quence activity relationships (ProSAR), the best variant con-
[23,25]
tained at least 35 mutations.
Detailed biochemical proper-
ties of individual process variants have not been published,
except for patent disclosures of sequences encoding variants
with 25 and 37 mutations and at least 1.5- to twofold activity
Scheme 1. Precursors for the statin side chain [compounds (S)-1 and (R)-2]
and other halohydrin dehalogenase substrates or products (7 to 18).
improvement over the wild-type enzyme for the conversion of
[26]
(
S)-4 into (R)-6. Because the ProSAR algorithm applied by
Fox and co-workers discards neutral or deleterious muta-
[25]
tions, the substitutions present in the obtained best multi-
site mutants are highly likely to contribute to the observed
phenotype.
important building blocks for high-value pharma- and nutra-
[
20]
ceuticals. Compound (R)-1 can be used as a precursor for
the production of l-carnitine, whereas (S)-1 is the direct pre-
cursor for the preparation of the hydroxynitrile (R)-4-cyano-3-
hydroxybutyric acid [(R)-2], which is used in the production of
Knowledge of the biochemical and structural properties of
a highly evolved enzyme variant would provide useful informa-
tion on functional and structural changes that occur during ex-
tensive directed evolution. However, structures of laboratory-
evolved enzymes containing large numbers of mutations are
rare. Most structural data for enzymes improved by directed
evolution studies are for mutants with no more than four
amino acid substitutions per protein chain. Some exceptions
are represented by the Aspergillus terreus acyltransferase var-
[
21]
statins.
Statins are a class of drug compounds that inhibit 3-hy-
droxy-3-methylglutaryl-coenzyme A reductase, the key enzyme
[
22]
that regulates the cholesterol biosynthesis pathway. One of
the most widely prescribed drugs for the treatment of elevated
blood cholesterol levels is Lipitor (atorvastatin calcium), with
[
23]
[28]
[29]
global annual sales exceeding US$ 10 billion. Most routes for
the synthesis of the (3R,5S)-3,5-dihydroxyhexanoate (also as
lactone) statin side chain employ (R)-2 as a chiral building
iant LovD G5, the Bacillus subtilis lipase variant LipA 4D3,
the Bacillus megaterium P450 monooxygenase BM-3 variant
[30]
139-3, the Aspergillus sp. carboxylesterase variant NylB’ Hyb-
S4M94, the p-nitrobenzyl esterase variant 8g8 from B. subti-
[
21,24]
[31]
block;
this results in a demand for more than 100000 kg
[
23]
[30]
[33]
(
R)-2 each year. A variety of chemoenzymatic routes for the
synthesis of optically pure (R)-2 have been explored.
lis, and an E. coli aspartate aminotransferase variant ATB17,
[21,22,24]
which contain between six and 17 amino acid exchanges per
mutant. Structural information on such heavily engineered en-
zymes can also provide explanations for their modified proper-
ties, such as higher catalytic rates or increased thermal stabili-
ties, and might contribute to the design of directed evolution
libraries through, for example, adoption of strategies that en-
hance the possible occurrence of cooperative effects of muta-
tions.
These routes involve whole cells or enzymes such as dehydro-
genases, lipases, and nitrilases. One possible process employs
variants of a ketoreductase and a HheC-derived halohydrin de-
halogenase in a three-step/two-enzyme process (Scheme 2) to
produce optically pure ethyl (R)-4-cyano-3-hydroxybutyrate
[
23,25,26]
[(R)-6)] from prochiral ethyl 4-chloroacetoacetate (3).
In
this process, 3 is reduced by a ketoreductase variant from Can-
dida magnoliae to ethyl (S)-4-chloro-3-hydroxybutyrate [(S)-4].
In a second step, (S)-4 is converted into the corresponding
ethyl (S)-3,4-epoxybutyrate [(S)-5] by use of an engineered
HheC variant. Subsequently, (R)-6 is produced from (S)-5 by cy-
anolysis with use of the same HheC variant, yielding the prod-
uct with 99.5% purity and in an enantiomeric excess exceed-
In this study, we have biochemically characterized two
highly engineered HheC variants, each containing 37 muta-
[25]
tions.
One variant, HheC2360, shows improved catalytic
rates, enhanced temperature stability, and opposite enantiose-
lectivity relative to its parent HheC. Furthermore, its X-ray
structure was elucidated to assess the influence of structural
changes on its kinetic properties and temperature stability.
[
25]
ing 99.9%.
HheC variants have been tailored to the needs of this pro-
[
25]
cess by multiple rounds of directed evolution.
After 18
rounds of directed evolution, driven by establishing protein se-
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 0000, 00, 1 – 13
&
2
&
ÞÞ
These are not the final page numbers!

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
(
E95G, K121R, A152T, Y177F, H179D, W238T) or with the C-ter-
Results and Discussion
minal extension (Y177F, G251S, M252V) of neighboring sub-
units in the tetrameric assembly.
Sequence analysis of improved HheC variants
HheC was tailored by Fox and co-workers for the improved
conversion of (S)-4 to (R)-6 under industrial process condi-
Biochemical characterization of HheC2360 and HheC2656
[
25]
tions. As a result, a set of 1151 unique HheC sequences was
described, with a maximum of 42 substitutions per se-
Synthetic genes for HheC2360 and HheC2656 were expressed
in a recombinant fashion in E. coli MC1061 from a pBAD-based
expression vector at similar levels as HheC. After purification of
the enzymes, yielding samples that gave single bands on SDS-
PAGE, catalytic activities for dehalogenation and cyanolysis
were measured. Most data were compared with those for the
variant HheC-C153S, which is wild-type HheC containing
a single mutation that makes the enzyme less susceptible to
[
26]
quence.
To obtain insight into the sequence variation responsible for
the improved catalytic properties, the substitutions occurring
in the reported sequence population were analyzed. This anal-
ysis revealed that 153 of the 254 HheC residues are mutated in
less than 1% of the cases (i.e., in 11 or fewer of the sequences)
and that a further 49 residues are mutated in less than 5% of
the reported HheC sequences. Of the remaining HheC residues,
two subsets of 25 and 27 positions were classified either as fre-
quently mutated (in >66% of the sequences) or moderately
mutated (altered in 5–37% of the sequences), respectively. An
overview of the observed substitutions is given in a multiple
sequence alignment in Figure S1 in the Supporting Informa-
tion. It was expected that the set of most frequent mutations
would include the substitutions that are of major importance
for the relevant changes in properties of HheC, such as the
reported improvement of at least 1.5-fold in the conversion of
[4,10]
oxidative damage, but otherwise has similar properties.
The dehalogenation of the chloroalcohol ester (S)-4, which
yields the corresponding (S)-5 as an intermediate, is the first
HheC-catalyzed reaction in the production of (R)-6. The catalyt-
ic constants for this dehalogenation reaction with (R)- and (S)-
4, as well as those with several other haloalcohols, were
obtained by measuring initial halide release rates. Data were
measured with both mutants and were compared with data
for HheC-C153S (Table 1). The HheC-C153S enzyme exhibited
steady-state parameters similar to previously determined con-
[
26]
(
S)-4 to (R)-6.
The 25 most frequent mutations (Q37H, K38Q, K52I, Y70L,
stants for the wild-type enzyme. The k values were in the
cat
À1
range of 2.7 to 35.1 s per active site for all tested substrates.
Q72H, F82A, A83P, P84V, F86W, Q87R, G99D, A100T, R107K,
T134A, T146S, C153S, T154A, G174A, E181G, F186Y, T189S,
N195S, K203R, A222T, M245V) map to different parts of the
enzyme structure. Only two mutations target buried residues
For achiral 1,3-dichloropropan-2-ol (7) and 1,3-dibromopropan-
2-ol (8), as well as for (R)-4, the K values were lower than the
M
lowest tested haloalcohol concentration of 0.05 mm. For the
other haloalcohols, the KM values were in the 0.16 to 4 mm
range. Because of the very low KM values with the two achiral
dihalopropanols 7 and 8 and with (R)-4, the highest catalytic
efficiencies were found with these substrates, whereas the
lowest catalytic efficiency was observed for rac-3-chloropro-
pane-1,2-diol (rac-11), which has the highest KM value. Surpris-
(
A100T, G174A), whereas the other 23 affected residues each
2
have a solvent-exposed area of ꢀ5 ꢁ . Furthermore, of these
frequent substitutions, only F186Y is among the six first-shell
residues within a 5 ꢁ distance from the catalytic triad Y145-OH.
Another 34 residues lie within 5–10 ꢁ of the catalytic triad
Y145-OH (second shell), and eight of these are frequent substi-
tutions (F82A, A83P, P84V, F86W, A100T, T134A, T146S, G174A).
The most frequently occurring mutations were analyzed fur-
ther to obtain insight into the structural basis of the properties
of the enzyme variants under study.
ingly, both enantiomers of 4 showed low K values; this means
M
that the wild type already recognizes both enantiomers of this
substrate quite well. Previously, the HheC wild type was ap-
plied in the conversion of the methyl ester of 1 with stoichio-
[34]
metric yields but no catalytic constants were reported.
HheC2360 and HheC2656 were selected out of the reported
sequences for detailed investigation because they each contain
at least 24 of the mutations strongly enriched in the ProSAR
procedure, but originate from different branches of the diversi-
In conclusion, both HheC mutants exhibited elevated KM
values relative to HheC-C153S for all tested haloalcohols
(Table 1). These elevated KM values are not surprising because
the mutants were obtained after screening for conversion of
substrate concentrations as high as 20 mm (first and second
tier) and >600 mm (third tier) (S)-4. The screening condi-
tions thus lacked selective pressure for the evolution of HheC
mutants with high affinities.
[
25,26]
ty that was generated.
Each of these variants contains 37
[25]
mutations in total. Variant HheC2360 contains all of the fre-
quent mutations mentioned above with the exception of two
different amino acid substitutions (A100M, T146A), It also con-
tains 12 less frequent mutations (A60V, V75I, V112A, K121R,
P135S, Y166H, Y177G, L178V, H179D, V201W, V205Y, I246V).
Variant HheC2656 differs from HheC2360 in its sequence at 18
positions (K10L, E95G, A100, V112, P135, T146S, A152T, Y166,
S180T, A152T, Y177F, L178, S180T, H201, V205, W238T, G251S,
M252V). As in HheC2360, the majority of the less frequently oc-
curring mutations in HheC2656 are positioned in or flank struc-
tural regions that interact either with neighboring monomers
Mutant HheC2360 exhibited the highest increase in kcat (3.1-
fold) for the target haloalcohol substrate (S)-4, whereas for the
opposite enantiomer (R)-4 it showed only a 1.6-fold increase in
kcat. This increase results in a kcat value for (S)-4 similar to those
found with other tested haloalcohols [7, 8, and rac-3-bromo-
propane-1,2-diol (rac-12)], the kcat values of which were little
affected (0.8- to 1.2-fold changes). For rac-1-chloropropan-2-ol
(rac-9), 1-chloro-2-methylpropan-2-ol (10), and rac-2-chloro-1-
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 0000, 00, 1 – 13
&
3
&
ÞÞ
These are not the final page numbers!

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
Table 1. Comparison of kinetic constants for halide release.
Haloalcohol
HheC-C153S
HheC2360
HheC2656
k
cat
À1
K
M
k
cat/K
M
k
[s
cat
À1
K
M
k
cat/K
M
k
[s
cat
À1
K
M
k
cat/K
À1
M
À1
À1
À1
À1
À1
[s
]
[mm]
[s mm
]
]
[mm]
[s mm
]
]
[mm]
[s mm
]
[
a]
[a]
ethyl (R)-4-chloro-3-hydroxybutyrate [(R)-4] 4.0Æ0.3
ethyl (S)-4-chloro-3-hydroxybutyrate [(S)-4] 2.68Æ0.06
<0.05
>80
6.3Æ0.2
8.3Æ0.3
4.8Æ0.4
7.8Æ0.2
0.30Æ0.01
6.5Æ0.1
2.8Æ0.2
2.3
4.4
0.5
14.5Æ0.3
16.2Æ0.5
13.5Æ0.3
22.9Æ1.5
2.5Æ0.1
3.5Æ0.3
2.8Æ0.2
5.8
4.6
4.8
0.16Æ0.01
16.8
[a]
1.9Æ0.2
[
[
a]
a]
1
,3-dichloropropan-2-ol (7)
,3-dibromopropan-2-ol (8)
6.0Æ0.3
6.6Æ0.3
<0.05
<0.05
>120
>132
10.4Æ1.9
[a]
1
0.59Æ0.05 13.2
1.5Æ0.2 15.3
rac-1-chloropropan-2-ol (rac-9)
-chloro-2-methylpropan-2-ol (10)
rac-3-chloropropane-1,2-diol (rac-11)
rac-3-bromopropane-1,2-diol (rac-12)
rac-2-chloro-1-phenylethanol (rac-13)
18.2Æ0.6
35.1Æ0.5
3.17Æ0.05
6.1Æ0.1
0.7Æ0.1
26.0
66.2
0.8
38.1
9.7
19.6Æ1.3
0.02
0.4
0.002
0.3
0.7
1.23Æ0.05 39.8Æ3.4
0.03
1.3
1
0.53Æ0.02
4.0Æ0.2
16.5Æ0.8
10.5Æ0.3
0.4Æ0.1
10.8Æ0.3
8.0Æ0.6
[
b]
[b]
[b]
>0.2
@85.5
58.0Æ21.7 0.01
0.16Æ0.02
1.3Æ0.1
5.7Æ0.2
3.8Æ0.4
22.6Æ1.8
5.3Æ1.4
57.4Æ2.8
0.2
1.8
12.6Æ0.3
6.39Æ0.08 3.6Æ0.1
M
[a] The K value was lower than the lowest measured haloalcohol concentration. [b] No substrate saturation according to Michaelis–Menten kinetics was observed
at the highest measured haloalcohol concentration.
phenylethanol (rac-13) 3.3- to 60-fold drops in kcat relative to
HheC-C153S were observed.
ingly, the highest rate was obtained with target epoxide (S)-5.
The apparent KM values for cyanide were in the 0.3 to 1.8 mm
range for all epoxides, so the catalytic efficiencies did not
The KM values of mutant HheC2360 corresponded to affini-
ties for the tested haloalcohols that were reduced four- to
À1
À1
exceed 3.7 s mm . Both 37-fold mutant enzymes displayed
>
200-fold relative to HheC-C153S. For chloroalcohol rac-11,
higher cyanolysis rates than HheC-C153S for all epoxides
the increase in KM with the HheC2360 mutant enzyme was so
high that no saturation according to Michaelis–Menten kinetics
was observed, even at the highest substrate concentration
tested (85.5 mm). As a consequence, the catalytic efficiencies
tested. The apparent K values for cyanide were also increased
M
for all tested epoxides, except for mutant HheC2656 convert-
ing epoxide (S)-5.
The apparent kcat values of epoxide cyanolysis observed with
mutant HheC2360 were ten to 120 times greater than those
observed with HheC-C153S. Whereas the highest absolute rate
(
k /K ) of mutant HheC2360 for all haloalcohols tested were
cat M
lower than those with HheC-C153S. The reduction in catalytic
À1
efficiency was lowest, however, for the target chloroalcohol (S)-
was 110 s —for chiral (S)-1,2-epoxybutane [(S)-14]—the great-
4
.
est relative increase was observed with rac-1,2-epoxyhexane
(rac-16), with the bulkiest aliphatic side chain. The apparent KM
values of mutant HheC2360 were 2.2 to 17.3 times larger than
the HheC-C153S values. Accordingly, catalytic efficiencies of
HheC2360 were 1.9 to 55 times larger than those of HheC-
C153S for all tested epoxides.
Mutant HheC2656 also showed increases in kcat values with
most haloalcohols. Absolute values were 1.6 to 4.1 times
higher than for mutant HheC2360. The KM values for mutant
HheC2656 were also 2.8 to more than 350 times higher than
those found with HheC-C153S. Here, too, catalytic efficiencies
for all haloalcohols were lower than with HheC-C153S and, as
with HheC2360, the target chloroalcohol (S)-4 exhibited the
lowest relative drop in k /K .
Mutant HheC2656 showed apparent kcat values 3.6 to 58
times higher than those obtained for HheC-C153S. The appar-
ent K values with the mutant HheC2656 and rac-14 to rac-16
cat
m
M
The enantiopreference of HheC-C153S with 4 is in favor of
the R enantiomer, with an apparent enantiomeric ratio of E >
were 1.2 to 7.0 times higher than those with HheC-C153S. For
target epoxide (S)-5, a slightly reduced apparent KM (1.3-fold
lower) was even observed. Variant HheC2656 displays higher
affinities than mutant HheC2360 towards all tested epoxides;
this seems to come at the expense of catalytic rates, which are
higher for HheC2360.
R
4
.8. The mutant HheC2656 hardly discriminated between the
R
enantiomers of 4 (E =1.3). In contrast, mutant HheC2360
S
exhibited an enantiomeric ratio of E =1.9, corresponding to
a modest preference for (S)-4 over (R)-4. Even though the
observed values are low, the enantiopreference of mutant
HheC2360 is the opposite of those of HheC-C153S or mutant
HheC2656. Enantiopreference towards S-configured haloalco-
hols and epoxides has not been observed in previous bio-
To explore the enantiopreferences of all three enzymes, cata-
lytic efficiencies of cyanolysis reactions were determined with
S
(R)- and (S)-14. Mutant HheC2360 shows an apparent E of 1.2,
indicating a very small preference for S-configured 14. In con-
[
10,11]
R
chemical studies on HheC and its variants,
but mutants
trast, the determined apparent enantiomeric ratios of E =1.4
R
with a preference for S-configured substrates were found in
and E =1.8 indicated slight preferences for (R)-14 for HheC-
[
27]
mutants of Arthrobacter sp. strain AD2.
C153S and HheC2656, respectively. The directed evolution of
HheC-C153S towards HheC2360 thus also inverted the enantio-
preference in cyanolysis, whereas both HheC-C153S and
HheC2656 exhibit the R enantiopreference for dehalogenation
shown earlier for the cyanolysis of the methyl ester of 1 by
The cyanolysis of epoxide ester (S)-5 by a HheC mutant is
the last enzymatic step in the production of (R)-6. Apparent ki-
netic constants for cyanide with fixed initial concentrations of
(
S)-5 and several aliphatic epoxides were obtained for all three
[34]
enzymes (Table 2). Enzyme HheC-C153S displayed modest cya-
HheC.
À1
nolysis apparent kcat values in the range 0.5 to 4.8 s . Interest-
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 0000, 00, 1 – 13
&
4
&
ÞÞ
These are not the final page numbers!

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
[
a]
Table 2. Apparent kinetic constants for cyanide obtained with various epoxides in ring-opening reactions.
Epoxide
HheC-C153S
HheC2360
HheC2656
k
cat
À1
K
M
k
cat/K
À1
M
k
[s
cat
À1
K
M
k
cat/K
M
k
[s
cat
À1
K
M
k
cat/K
M
À1
À1
À1
À1
À1
[s
]
[mm]
[s mm
]
]
[mm]
[s mm
]
]
[mm]
[s mm
]
ethyl (S)-3,4-epoxybutyrate [(S)-5]
rac-1,2-epoxybutane (rac-14)
4.8Æ0.4
1.1Æ0.1
1.1Æ0.1
1.3Æ0.1
2.5Æ0.2
1.8Æ0.4 2.7
0.6Æ0.1 1.8
0.3Æ0.1 3.7
0.5Æ0.1 2.6
0.9Æ0.2 2.8
46.6Æ6.0
91.7Æ6.6
88.6Æ9.1
110.3Æ5.4
4.1Æ1.1
11.4
17.5Æ0.5 1.4Æ0.1 12.5
63.9Æ8.5 3.1Æ0.9 20.6
52.6Æ2.1 2.1Æ0.2 25.0
33.4Æ2.6 2.4Æ0.4 13.9
31.3Æ2.5 1.7Æ0.4 18.4
4.1Æ0.7 22.4
5.2Æ1.1 17.0
5.4Æ0.6 20.4
(
(
R)-1,2-epoxybutane [(R)-14]
S)-1,2-epoxybutane [(S)-14]
rac-1,2-epoxy-2-methylbutane (rac-15)
rac-1,2-epoxyhexane (rac-16)
57.5Æ5.8 11.0Æ1.8
5.3
0.51Æ0.03 1.4Æ0.2 0.4
62.0Æ7.1
3.1Æ0.8 20.0
19.0Æ0.8 1.7Æ0.2
11.2
[a] All kinetic constants for cyanolysis were determined with use of a fixed starting concentration (20 mm) of epoxide.
In summary, mutant HheC2656 appears to be a rather gener-
Both wild-type HheC and mutant HheC2360 are tetramers, and
the tetrameric assembly can be regarded as a dimer of A+B
and C+D dimers. In the tetramer, subunit A is located oppo-
site C, and B opposite D. Backbone differences were only ob-
served in the loops between A83 and W86 and between T134
to W139 and in the C-terminal loop, including the last C-termi-
nal residues visible in the electron density. The three mutant
protein structures were highly similar, with RMSD values for Ca
atoms <0.3 ꢁ. A similarly high degree of similarity between
structures with substrate or product bound was also observed
when comparing wild-type enzyme with (R)-styrene oxide [(R)-
17] and chloride bound (PDB ID: 1PWZ) and with (R)-p-nitro-
phenyl-2-azidoethanol [(R)-18] bound in the active site (PDB
alist enzyme, giving higher dehalogenation rates than
HheC2360. Although the cyanolysis rates of mutant HheC2656
were not as high as those observed for HheC2360, HheC2656
is still a good catalyst for various cyanolysis reactions, as illus-
trated by the low apparent KM values for cyanide. In contrast,
mutant HheC2360 displayed significantly higher cyanolysis ac-
tivities and exhibited an enantiopreference for S-configured
molecules both for the dehalogenation and for the cyanolysis
reactions. In the statin side chain production process described
above, the chirality of (S)-5 is introduced by reduction of 3 to
[
23,25]
(
S)-4 with an S-selective ketoreductase.
This reaction is
followed by dehalogenation of (S)-4 to yield (S)-5 with the
HheC2360 mutant halohydrin dehalogenase, which displays
catalytic efficiency for (S)-4 about twice that for (R)-4. Because
the kcat and k /K values of mutant HheC2360 in cyanolysis of
[7]
ID: 1PX0).
The HheC2360 structure with hydroxynitrile (R)-6 shows for
the first time how a cyano moiety is bound in the active site of
a halohydrin dehalogenase. The cyanide binding mode in the
mutant is very similar to that of the azido moiety of (R)-18 in
the wild-type enzyme (PDB ID: 1PX0) and also to the chloride
binding mode in HheC (PDB ID: 1PWZ). The nitrogen of the
(R)-6 cyano group is thus at the same location as a chloride
ion in the wild-type anion binding site. Furthermore, a water
molecule that interacts with the halide in the wild-type is at
the same position in the HheC2360 structure, where it is in-
volved in a hydrogen-bonding interaction (2.9 ꢁ) with the ni-
trogen atom of the cyano group. Such equivalence was sug-
cat
M
(
S)-5 to (R)-6 are more than 5.6 and 2.6 times higher than the
corresponding values for the conversion of (S)-4 to epoxide
S)-5, the epoxide intermediate should be efficiently removed
(
from a chloroalcohol/epoxide reaction mixture by rapid con-
version to the desired product (R)-6. The better S activity and
a higher rate of cyanolysis that come from the combination of
an S-selective ketoreductase variant and the HheC variant
HheC2360 thus allow three enzymatic reaction steps that lead
to formation of (R)-6 from prochiral 3 (Scheme 2). In view of its
better catalytic properties, HheC2360 was selected for structur-
al analysis.
[7,12]
gested in previous studies,
but the HheC2360 mutant struc-
ture 4IXT confirms these assumptions and provides structural
information on cyanide binding by halohydrin dehalogenases.
Crystal structure of HheC2360
To explain the greatly enhanced epoxide cyanolysis rates and
the observed inversion of enantiopreference, structures of
mutant HheC2360 were solved by X-ray crystallography. Three
structures of the enzyme were obtained: HheC2360 with chlo-
ride bound (PDB ID: 4IY1), HheC2360 with both epoxide (S)-5
and chloride bound (PDB ID: 4IXW), and HheC2360 with hy-
droxynitrile (R)-6 bound in the active site (PDB ID: 4IXT). The
structures were compared with wild-type HheC crystal struc-
tures.
Structural changes near the active site
Comparison of the catalytically important residues S132, Y145,
and R149 in the mutant and wild-type enzyme structures
showed no differences in their side-chain orientations (RMSD<
0.2 ꢁ). The active-site cavity of HheC can be subdivided into
three sites: 1) the site that interacts with the substrate hydroxy
group (or epoxide oxygen), 2) the binding site for the anionic
leaving group (or incoming nucleophile), and 3) the binding
site for the epoxide ring substituent (e.g., phenyl in Figure 1A).
Whereas only minor structural changes occur in the epoxide
oxygen and anion binding sites (1 and 2), the epoxide ring-
substituent binding region (3) has undergone significant
changes in the HheC2360 mutant structure.
The structures of the three mutant tetrameric proteins were
superimposed on that of wild-type HheC (PDB ID: 1PWX),
giving RMSD values of less than 0.7 ꢁ for the equivalent Ca
atom positions. These low RMSD values indicate that the gen-
eral structure of the tetrameric enzyme assembly is unaltered.
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 0000, 00, 1 – 13
&
5
&
ÞÞ
These are not the final page numbers!

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
The shape of the HheC2360 active site is altered by a combi-
nation of several, sometimes coupled, mutations (Figure 1B).
This change is largely caused by mutation P84V (see below)
and concomitant backbone changes that sterically hinder bind-
ing of the epoxide ring substituent. Furthermore, mutation
F86W affects the rotational freedom of W139, which creates
room for a longer aliphatic substrate in the site that binds the
epoxide ring substituent (site 3). The highest relative increase
in catalytic efficiency was found for the cyanolysis of the lon-
gest tested linear aliphatic epoxide 16 (50-fold); this suggests
that the mutations create an active site geometry optimized
for longer aliphatic substrates.
The only mutation that affects the epoxide ring oxygen
binding region (site 1) is T134A, which is present in both mu-
tants. In wild-type HheC, residue T134 forms a hydrogen bond
with the catalytic triad residue S132. The role of S132 is to pro-
vide a hydrogen bonding interaction with the oxygen of the
substrate hydroxy group, which must be deprotonated by
Y145 during the epoxide formation reaction. In more than
[26]
90% of the sequences of HheC mutants this hydrogen bond
cannot be formed, due to a T134A substitution. Because this is
the most frequently observed substitution, it is presumably of
high importance for the evolved target activity.
Lastly, in mutant HheC2360 the anion binding site (2) is also
slightly modified. In the wild-type structure, anions are stabi-
lized by interactions with the backbone amide groups of Y177
and L178, the side chains of P175 and L178, the slightly posi-
tive edges of the aromatic rings of F12, F186, and Y187, and
The mutations F82A and A83P, which border the epoxide
ring substituent binding region (3), change the shape of the
active site in HheC2360 in such a way (Figure 1B) that the
[7]
through a bound water molecule. The loop between residues
P175 to Y185 forms part of the spacious halide-binding loop.
The L178V mutation increases the available space in the anion-
binding site. This allows the position of the chloride to be
shifted by 0.8 to 1.1 ꢁ relative to HheC structures with bromide
(
PDB ID: 1PWX) and chloride (PDB ID: 1PWZ), respectively. As
a consequence, the distance between chloride and the primary
epoxide ring carbon atom (which is the site of nucleophilic
attack) increases to 4 ꢁ in structure 4IXW. In wild-type struc-
tures (PDB ID: 1PWZ, 1ZMT) this distance is 3.2 ꢁ. Apparently,
this increase does not have a negative effect on the catalytic
activity in ring-opening reactions with cyanide. Similar distan-
ces, in the 3.5 to 4.5 ꢁ range, were observed in quantum
chemical models for cyanolysis and azidolysis reactions cata-
Figure 1. Comparison of active site cavities of HheC and HheC2360.
A) Active site of HheC with bound (R)-17 (yellow) and chloride (green; PDB
ID: 1PWZ). B) Active site of HheC2360 with (S)-5 (orange) and chloride (PDB
ID: 4IXW). The most important changes in the active sites are a rotation of
W139 and the intrusion of residue 84 into the active site. For a stereorepre-
sentation see Figure S2.
[12]
lyzed by wild-type HheC. Interestingly, mutations L178A and
L178F have been described before but produced inactive, di-
[6]
meric mutant enzyme assemblies in a wild-type context.
The mutation F186Y stabilizes a rare non-proline cis-peptide
bond, which is present at the anion-binding site (2) both in the
wild-type and in the mutant structures. When the cis-peptide
bond was discovered upon elucidation of the wild-type
enzyme structure, the authors proposed that it has a stabilizing
role on the anion binding architecture. In the wild type, the
cis-peptide bond is only stabilized by a single hydrogen bond
cavity can now bind (S)-5 in a catalytically productive mode,
with the epoxide ring oxygen oriented towards the catalytic
triad residues S132 and Y145. This productive binding mode of
(
S)-5 by the mutant is very different from the catalytically un-
productive binding of (S)-p-nitrostyrene oxide [(S)-19] by the
[
11]
wild-type enzyme.
In the latter case, the epoxide ring
[7]
oxygen of (S)-19 was oriented away from the catalytic triad
residues, due to inverted positions of the epoxide oxygen and
the terminal carbon of the substrate relative to the (R)-19 bind-
ing mode. This explained why the wild-type enzyme exhibits
with the backbone amide of F12. The substitution F186Y fur-
ther stabilizes this cis-peptide bond by introducing an extra hy-
drogen bond between the hydroxy group of Y186 and the
backbone amide group of F82A. The mutation F186Y was the
second most frequently observed substitution in the mutant
[11]
enantiopreference towards R-configured substrates.
For
[26]
mutant HheC2360, the mutations F82A and A83P lead to
changes in the active site shape that allow productive binding
of an S-configured epoxide.
sequence population (89%),
which highlights the impor-
tance of this cis-peptide bond stabilization for the evolved
target activity.
Neither HheC2360 nor HheC-C153S displays high enantio-
selectivity in the cyanolysis of aliphatic epoxides such as (R)-
and (S)-14. For wild-type HheC, this low enantioselectivity with
aliphatics has been observed before and contrasts with the
enantioselective positioning of chiral aromatic substrates such
as 19. The low enantioselectivity with 14 is likely the result of
a lack of specific interactions between the enzyme and the
ethyl substituent on the oxirane ring, which would otherwise
fix the orientation of the substrate in the active site.
From inspection of the structure it is clear that substitutions
flanking the active site pocket are likely to contribute to the
observed changes in catalytic properties, such as the more
than 80- and 120-fold enhanced activities with epoxides rac-14
and rac-16, respectively, and the inversion of enantioselectivity
with chloroalcohol 4. Such replacements around active site res-
idues are often reported in directed evolution studies directed
towards improving catalytic activities or expansion of substrate
[
15]
[35,36]
scope,
and have in fact been reported in most structural
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 0000, 00, 1 – 13
&
6
&
ÞÞ
These are not the final page numbers!

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
[
28,30,32,33]
2
investigations of highly mutated proteins.
For this
for the wild type and 18340 ꢁ for the mutant (Table S2). The
reason, efficient directed evolution protocols such as iterative
saturation mutagenesis selectively target residues close to the
surface area buried upon formation of the tetramer is thus in-
2
creased by 500 ꢁ in mutant HheC2360, which is predicted to
[
37]
À1
active site.
Mutations of a few key residues close to an
cause an increase in the dissociation energy of 8 kcalmol . It
active site can drastically alter substrate specificities. A carbox-
is known that such an increase in buried surface area can stabi-
lize a protein. An increase in buried surface area was found to
lead, for example, to a 68C increase in the melting tempera-
ture of cocaine esterase mutants from Rhodococcus sp.
[
31]
[38]
ylesterase and an N-acetylneuramic acid lyase, for exam-
ple, required only mutations of a few key residues to produce
variants that hydrolyze an artificial dimeric substrate or con-
dense non-natural substrates, respectively.
[40]
strain MB1.
An increase in buried surface area was also
found to be responsible for the 408C difference in melting
temperature between a mesostable and thermostable homo-
Improved thermal stability
[41]
logue of a Rhodococcus sp. naphthalene 1,2-dioxygenase.
As well as shifts in catalytic activity, the HheC mutants were
also reported to show enhanced stabilities. Indeed, mutant
HheC2360 displayed an increased thermal stability of at least
Another contribution to the increased stability comes from
intra-subunit van der Waals interactions. A new intra-subunit
van der Waals interaction is created by the relocation of resi-
due F136 induced by the nearby P135S mutation. This muta-
tion shifts the position of the phenyalanine side chain by more
than 11 ꢁ (Figure 4A). The distance between the two F136 resi-
dues in the wild type (5.9 to 7.5 ꢁ) was too large for van der
Waals interactions, whereas in HheC2360 the residues are at fa-
vorable interaction distance (4.3 ꢁ to 4.8 ꢁ). It is therefore rea-
soned that this mutation stabilizes the tetrameric ensemble by
8
8C. In measurements of residual dehalogenase activity, the
10
mutant exhibits an apparent T50 (temperature at which 50% of
the activity is lost after ten minutes of incubation) of 638C, in
contrast with a value of 558C for HheC-C153S (Figure 2).
Figure 2. Residual dehalogenase activity after 10 min incubation of enzyme
at different temperatures. HheC-C153S (open diamonds) and mutant
HheC2360 (closed squares) exhibited initial activities at 308C (100%) of
À1
À1
0
.26 Umg and 0.11 Umg , respectively.
Thermal unfolding was monitored by determining changes
in fluorescence upon heating in the presence of the fluoro-
[
39]
phore SYPRO Orange. The maximum change in fluorescence
is a measure of the apparent unfolding or dissociation temper-
ature (Tm,app). The Tm,app was 55.58C for HheC-C153S and 66.58C
for HheC2360 (Figure 3).
The tetrameric assembly of HheC is stabilized by a large hy-
drophobic surface area that is buried upon formation of the
2
oligomer. This buried surface was calculated to be 17840 ꢁ
Figure 4. Enhanced inter-subunit interactions observed in mutant HheC2360
(
magenta) relative to wild-type HheC (cyan). A) The relocation of F136,
2
which leads to an increase of 330 ꢁ in buried surface area. B) The enlarged
aromatic network in HheC2360 contains five tryptophan residues from two
different subunits; these stabilize the C-terminal W249’ of the opposite mo-
nomer in the active site. C) Mutation V205Y introduces an additional hydro-
gen bond between two adjacent subunits.
Figure 3. First derivative of fluorescence change (dF/dT) caused by binding
of SYPRO Orange to hydrophobic parts of HheC-C153S (black) and mutant
HheC2360 (gray) due to thermal unfolding in a temperature gradient from
30 to 908C.
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 0000, 00, 1 – 13
&
7
&
ÞÞ
These are not the final page numbers!

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
creating an additional intramolecular van der Waals interaction
between neighboring monomers.
bridges does not appear prominent from the HheC2360 crystal
structures, however, even though six surface mutations influ-
ence electrostatics (Q37H, K38Q, Q78R, G99D, K121R, K203R).
The mutations H201W and F86W further enhance the oligo-
meric stability by extending an aromatic interaction network
that is not present as such in the wild-type enzyme (Fig-
ure 4B). Tryptophan residues from opposing monomers inter-
act in this cluster located at the inter-subunit interface. Earlier
research showed that residue W192 is essential for the forma-
tion of the tetramer and that the W192F mutation leads to in-
Proline mutations
The modifications of the HheC2360 active site are not caused
merely by side chain substitutions, but also by protein back-
bone changes resulting from the introduction or replacement
of proline residues. Proline units stabilize protein structures by
decreasing the difference in entropy between the folded and
unfolded state. The presence of proline residues also restricts
the local backbone f dihedrals to a relatively narrow area of
the Ramachandran plot. This reduces rotational freedom and
might restrict backbone conformations, including conforma-
tions that occur as early intermediates in an unfolding path-
way. Structures of evolved enzymes containing mutations in-
volving prolines are rare, but have been described in cases of
[
5]
active, monomeric or dimeric enzyme. In HheC2360, the tryp-
tophan residues W192, W86, W139, W201, and W249’ (protrud-
ing into the active site from the opposite monomer) partici-
pate in multiple tryptophan–tryptophan interactions. Whereas
in the wild-type structure the distance between H201 and
W139 was too large for aromatic interactions (5.3 ꢁ), the larger
aromatic side chains introduced by the F86W and H201W mu-
tations allow for closer contacts, especially with residue W249’.
Thus, in the HheC2360 enzyme a partially new network of aro-
matic interactions affects the shape of the active site and sta-
bilizes the positioning of the C-terminal extension into the
active site of an opposite subunit, thereby stabilizing a region
in the protein that is involved in tetramer formation.
[32]
a p-nitrobenzyl esterase variant and a glycine oxidase var-
[47]
iant. Because changes in backbone conformation could influ-
ence active site geometry, we inspected the structural effects
of the HheC2360 mutations that involve prolines.
Inter-subunit hydrogen bonds also contribute to the oligo-
meric stabilization of HheC2360 by, for example, mutation
V205Y. This interfacial mutation creates a new inter-subunit hy-
drogen bond with residue V205Y’ from the opposite monomer
The HheC2360 37-fold mutant contains one substitution
through which a proline residue is introduced (A83P) and two
mutations through which proline is removed (P84V and
P135S). All three mutations produce significant local backbone
changes relative to the wild type, whereas no large differences
in backbone structure are present in regions that do not con-
tain proline mutations. The A83P and P84V mutations lead to
Ca positional changes (relative to the wild-type enzyme) of up
to 2.5 ꢁ in the loop between A83 and W86. The P135S muta-
tion is accompanied by a drastic change in the f dihedral of
residue 135 from À608 to À308, which in turn results in a 3.4 ꢁ
shift of the position of Ca of F136 and an average shift of
2.8 ꢁ in the position of the Ca atoms between T134A and
W139. The altered backbone conformations in the 134 to 139
(2.8 ꢁ) (Figure 4C). Furthermore, the aromatic ring of Y205 par-
ticipates in van der Waals interactions with the hydrophobic
carbon atoms of the K204 side chain (3.6 ꢁ), as well as in an
edge-to-face interaction with W201 (3.8 ꢁ). Other possibly sta-
bilizing substitutions include the surface-exposed mutation
Q87R, which leads to an additional hydrogen bond with the
peptide bond nitrogen of K91 (Figure S3), and Y166H, which
extends the existing hydrogen bonding network between
E165 and K140’ from a neighboring monomer to residue K123
(Figure S4).
These results suggest that the stabilization of HheC is to a
region are accompanied by a shift of the F136 C atom by 11 ꢁ
z
large extent governed by improved inter-subunit interactions.
Consequently, protein stabilization experiments might profita-
bly focus, in cases of oligomeric proteins, on improving hydro-
phobic and hydrogen-bonding interactions between subunits.
Improved interactions of this type have also been found in
other thermostable mutants, such as l-isoaspartyl-O-methyl-
(Figure 4A and S5). This relocation of the side chain of F136
enhances the inter-subunit interactions and leads to an in-
crease in buried surface area, which contributes to enhanced
stability as discussed above.
Similar effects are caused by the A83P and P84V mutations,
which are located in a loop between residues A83 and W86,
surrounding the active site. The y and f dihedrals at the posi-
tion of the introduced proline residue are changed in relation
to the wild-type enzyme, which results in different backbone
conformations between residues 83 and 86, but wild-type
backbone conformations are re-established from position W86.
Apparently, the introduced proline locally modifies the back-
bone, but nearby strands and helices fix the backbone back to
its wild-type conformation outside the affected loop. This is
similar to what has been described for an evolved p-nitroben-
[
42]
transferase from Sulfolobus tokodaii, and in comparisons of
[43]
thermostable and mesostable alcohol dehydrogenases.
Other mutations in the HheC2360 mutants suggest further
stabilizing effects. Both mutation M245V and the previously
[
4]
found substitution C153S remove oxidation-prone sulfur-con-
taining residues from the protein. The majority of mutant
sequences generated by Fox et al. contain the substitutions
[
26]
C153S (87%) or M245V (82%). Removal of sulfoxidation-sus-
[44,45]
ceptible sites is indeed important for protein stability.
In-
[32]
troduction of charged groups on the protein surface, causing
formation of new surface salt bridges or extension of existing
zyl esterase variant. Although not located in the active site,
the A83P mutation significantly alters the shape of the active
site by inducing backbone changes, which in turn cause a dif-
ferent orientation of the side chain of V84. This reorientation
makes atoms Cg1 and Cg2 protrude into the active site
[
46]
salt-bridge networks, can also provide stabilization. This was
observed in, for example, the crystal structure of a thermosta-
[
29]
ble variant of B. subtilis lipase LipA. Formation of new salt
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 0000, 00, 1 – 13
&
8
&
ÞÞ
These are not the final page numbers!

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
(
Figure 1), which results in a narrowing of the binding region
Another example of paired mutations is given by the F82A+
A100M couple. Mutation F82A removes an aromatic ring from
the hydrophobic interior of the protein, creating an unfavora-
ble cavity. The substitution A100M compensates for this detri-
mental effect by positioning the new large methionine side
chain into this cavity (Figure 5B).
for the R group connected to the epoxide ring. The modified
binding pocket is more suitable for binding an aliphatic group
than an aromatic ring. These results suggest that experiments
directed towards changing enzyme activity can be benefit
from the introduction of proline residues in loops that flank
residues shaping active site. These resulting backbone changes
of first- or second-shell residues can induce side chain reorien-
tations, which ultimately modify substrate binding and cataly-
sis.
Whereas many directed evolution studies have targeted resi-
dues individually, such as by single-site saturation mutagenesis,
[32,48–50]
these observations and other reports
suggest that it is
more efficient to introduce mutations in a combined fashion.
Cooperative effects of mutations have been found to be re-
sponsible for large changes in activity in a p-nitrobenzyl ester-
Synergistic mutations
[32]
[48]
[49]
ase, in glucose dehydrogenase, and in lipase mutants. If
synergistic mutations are allowed by design of pairwise or
three-site libraries, with use of structural information, a larger
set of substitutions at each site will be allowed, resulting in
a much larger number of possibly functional variants than is
accessed by the sum of all functional single-site mutants.
To confirm that certain mutation pairs act in a cooperative
The structures of the HheC2360 enzyme reveal several coupled
mutations that mainly alleviate steric clashes that would result
from some mutations when present individually. For example,
the indole group introduced by substitution H201W would
sterically clash with the side chain of Y177 (0.6 ꢁ). Mutation
Y177G resolves this clash (Figure 5A). Another steric clash
Fold
manner, the changes in folding energy (DDG ) were compu-
tationally predicted for two pairs of mutations, both for the
separate mutations and their combination. Both for the
Fold
F82A+A100M and the Y177G+H201W pairs, the DDG
values calculated for the individual mutations in a wild-type
background were very large and unfavorable (Table 3). When
Table 3. Computationally predicted cooperative effect of mutations.
Fold
À1
Mutation pairs
Predicted DDG [kJmol
FoldX calculation Rosetta calculation
Sum of single Combined Sum of single Combined
]
mutations
mutations mutations
mutations
F82A+A100M
125Æ27
11Æ27
11Æ27
172Æ40
240Æ40
53Æ40
4Æ40
74Æ40
85Æ40
52Æ40
8Æ40
Y117G+H201W 118Æ27
P135S+I246V
T146A+T154A
72Æ27
72Æ27
À15Æ27
À15Æ27
Fold
Figure 5. Synergistic mutations in the evolved HheC2360 enzyme. A) The
synergistic mutations H201W and Y177G prevent a steric clash between resi-
dues W201 and Y177. B) The substitution F82A creates a destabilizing hydro-
phobic cavity in the protein, which is filled by the methionine side chain in-
troduced by the A100M mutation. The structures of HheC2360 are in ma-
genta and those of wild-type HheC in cyan.
the DDG
values were calculated for the pairwise combina-
tions, however, the destabilizing effects were much smaller.
The calculations with FoldX predicted that the effects of the
Fold
combined mutations on DDG are negligible when the error
of the calculation method is taken into consideration. The
calculations thus confirm that the F82A and A100M mutations
act in a cooperative manner, as do the Y177G and H201W sub-
stitutions.
would occur between the Cg atom of wild-type I246 and the
2
side chain of P138 (0.9 ꢁ), resulting from the 3 ꢁ backbone
shift of the loop between residues 134 and 139, which is in
turn caused by the P135S substitution discussed above. This
clash is resolved by the I246V mutation and associated back-
bone changes in the C-terminal loop, which increase the dis-
For the P135S+I246V and the T146A+T154A mutations, the
calculations predicted no compensating effects. This might be
the result of the large backbone changes that accompany the
differences in side chain structures and conformations (see
Fold
above), which make predictions of the effect on DDG
in-
tance between the side chains of P138 and the Cg atom of
accurate with computational methodology that was used
(Table 3). These observations underscore the necessity of im-
2
V246 to 3.4 ꢁ. Furthermore, the relocation of the loop harbor-
ing F136, caused by the P135S substitution discussed above,
would cause a steric clash between the side chains of T146
and T154, but the substitutions T146A and T154A prevent this.
[51]
proving backbone remodeling protocols if the aim is compu-
tational design of stabilizing mutations based on shifts in back-
bone conformations.
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 0000, 00, 1 – 13
&
9
&
ÞÞ
These are not the final page numbers!

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
with a preference for S-configured substrates were found in
Conclusions
[27]
mutants of Arthrobacter sp. strain AD2.
Directed evolution experiments can yield highly evolved
enzyme variants with dramatic improvements in stability and
catalytic efficiency, without requiring much insight into the
molecular basis of the target properties. The investigation of
the highly engineered halohydrin dehalogenase variant
HheC2360 described here provides insight into the structural
effects of the mutations that are introduced during directed
evolution. We show that the increases, of up to 100-fold, in
rates of cyanide-mediated epoxide ring opening, as well as the
modified enantiopreference occurring in HheC2360, can be
explained by inspection of crystal structures. Specifically, side
chain substitutions that allow formation (F186Y) or removal
Experimental Section
Chemicals and commercially available enzymes: The hydroxyni-
trile ethyl (R)-4-cyano-3-hydroxybutyrate [(R)-6)], the haloalcohols
ethyl (R)-4-chloro-3-hydroxybutyrate [(R)-4], ethyl (S)-4-chloro-3-hy-
droxybutyrate [(S)-4], 1,3-dichloropropan-2-ol (7), 1,3-dibromopro-
pan-2-ol (8), rac-1-chloropropan-2-ol (rac-9), 1-chloro-2-methylpro-
pan-2-ol (10), rac-3-chloropropane-1,2-diol (rac-11), rac-3-bromo-
propane-1,2-diol (rac-12), and rac-2-chloro-1-phenylethanol (rac-
13), as well as the epoxides ethyl (S)-3,4-epoxybutyrate [(S)-5], rac-
1
,2-epoxybutane (rac-14), (R)-1,2-epoxybutane [(R)-14], (S)-1,2-ep-
oxybutane [(S)-14], rac-1,2-epoxy-2-methylbutane (rac-15), and rac-
,2-epoxyhexane (rac-17), were purchased in the highest available
(T134A) of hydrogen bonds affect essential features of the
1
HheC active site, such as the cis-peptide bond in the anion
binding loop and the hydrogen bonding pattern of the catalyt-
ic triad residue S132, respectively. Furthermore, the mutations
reshape the active site cavity for improved binding of S-config-
ured target substrates, not only through altered side chains
purity from Sigma–Aldrich (St. Louis, MO, USA). Restriction enzymes
NdeI and XhoI and T4 ligase were ordered from New England Bio-
labs (Ipswich, MA, USA). SYPRO Orange was obtained from Life
Technologies (Carlsbad, CA, USA).
Cloning, expression, and enzyme preparation: The directed evo-
lution study by Fox and co-workers yielded 1422 HheC variants,
(
F86W, L178V), but also through backbone modifications that
result from the introduction and removal of proline residues
A83P, P84V). The enhanced thermal stability of the evolved de-
[26]
a multiple sequence alignment of which was prepared by using
[52]
(
ClustalX v2.1. After removal of identical protein sequences from
the alignment, the observed mutations were inspected and
mapped to the wild-type crystal structure (PDB ID: 1PWZ). Sequen-
ces HheC2360 and HheC2656 were obtained as synthetic DNA con-
structs from Geneart AG (Regensburg, Germany). Synthetic genes
for HheC2360 and HheC2656 were cloned into expression vector
halogenase is explained by several mutations that improve oli-
gomeric interface contacts (V205Y, H201W), and also by a sub-
stantial increase in buried surface area that further stabilizes
the tetrameric enzyme assembly. The latter is again due to
changes in the side chain location and orientation (F136) re-
sulting from the removal of a proline (P135S). Several of the
mutations that contribute to the change in functional proper-
ties appear to be facilitated by or are dependent on additional
coupled mutations.
[53]
pBADNk by standard protocols.
structed plasmids into E. coli MC1061, expression and enzyme
purification were carried out as described before. As a control,
After transformation of con-
[54]
HheC-C153S was expressed in recombinant fashion from plasmid
pBAD (C. Tarabiono, unpublished results) and purified in a similar
[
54]
From the structure–function analysis of HheC2360 it can be
concluded that backbone changes to loop regions caused by
mutations that introduce or remove proline residues give
access to additional functional changes in active site geometry,
relative to mutations that only influence side chains. Targeting
of proline residues can thus be highly useful when engineering
enzymes to obtain modified catalytic properties. Through co-
operative and synergistic effects, combinations of mutations
allow greater functional diversity than individual mutations.
Simultaneous introduction of mutations at two, three, or four
structurally close sites in libraries can increase the functional
diversity explored in directed evolution without strongly dilut-
ing libraries with inactive variants.
way.
(
Enzyme was stored at À208C in TEMG buffer [Tris·SO
4
10 mm), ethylenediaminetetraacetic acid (EDTA, 1 mm), b-mercap-
toethanol (1 mm), glycerol (10%), pH 7.5)]. Protein concentrations
were determined spectrophotometrically after staining with Coo-
massie Brilliant Blue.
Dehalogenase assays: Dehalogenase assays were performed in
duplicate by monitoring halide release by the colorimetric assay
[
55]
described by Bergmann and Sanik. Briefly, initial halide release
was monitored at 460 nm in 96-well plates on mixing of assay re-
agent (150 mL) with sample (150 mL). The assay reagent was pre-
pared freshly before measurements, by mixing equal volumes of
solution I [NH Fe(SO ) (0.25m) in HNO (9m)] and solution II [satu-
4
4 2
3
rated solution of Hg(SCN)2 in absolute ethanol]. Reactions were
carried out in Tris·SO (50 mm, pH 7.5 in a total volume of 2 mL) at
08C. Reactions were initiated by addition of enzyme, and halide
4
3
The enantiopreference of HheC-C153S with 4 is in favor of
R
concentrations were monitored over time. Kinetic parameters were
obtained from initial rates by use of the SimFit v6.2.3 package
the R enantiomer, with an apparent enantiomeric ratio of E >
4
.8. The mutant HheC2656 hardly discriminated between the
(http://www.simfit.man.ac.uk/) for curve fitting by Michaelis–
R
enantiomers of 4 (E =1.3). In contrast, mutant HheC2360
exhibited an enantiomeric ratio of E =1.9, corresponding to
Menten kinetics. The apparent enantiomeric ratio E was calculated
from the ratio of catalytic efficiencies (k /K ) for (R)- and (S)-4 as
described by Straathof and Jongejan.
S
cat
M
a modest preference for (S)-4 over (R)-4. Even though the
observed values are low, the enantiopreference of mutant
HheC2360 is the opposite of those of HheC-C153S or mutant
HheC2656. Enantiopreference towards S-configured haloalco-
hols and epoxides has not been observed in previous bio-
[56]
Cyanolysis assays: Cyanide consumption was monitored by utiliz-
ing the specific absorption (at 267 nm) of the [(CN) Ni] complex
2
À
4
2+
ion, formed after mixing of a cyanide-containing sample and Ni
[57]
in aqueous NH solution. In detail, samples (150 mL) were mixed
3
[
10,11]
chemical studies on HheC and its variants,
but mutants
with NiCl (2 mm, 150 mL) in an aqueous solution of NH (1m), and
2
3
the absorbance at 267 nm was read in 96-well UV-Star microplates
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 0000, 00, 1 – 13
&10
&
ÞÞ
These are not the final page numbers!

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
(
Greiner Bio One, Frickenhausen, Germany). Standard assays for cy-
Computational methods: Buried surface areas, free energy gains
anide-mediated epoxide ring opening contained epoxide (20 mm)
in potassium phosphate buffer (50 mm, pH 8.0) and varying con-
centrations of sodium cyanide. The reactions were performed in
duplicate in a total volume of 2 mL at 308C after addition of
enzyme. Apparent kinetic constants and apparent catalytic efficien-
cies were obtained as described for the dehalogenase reactions.
upon formation of the tetrameric assembly (DG ), and free ener-
gies of tetrameric assembly dissociation (DGdiss) were calculated by
int
use of the PISA webserver at PDBe with structures 4IXW and
[64]
2
1ZMT. A solvent-accessible surface area of <5 A was used as
a cut-off to discriminate buried from solvent-accessible residues in
tetrameric assemblies. The predicted changes in folding energies
Fold
[65]
[66]
(
DDG ) were calculated both with FoldX and with Rosetta.
Stability assays: The thermal stabilities of variant HheC2360 and
wild-type HheC were determined by measuring residual dehaloge-
nase activity after 10 min incubation of enzyme at temperatures
between 30 and 958C with subsequent storage on ice (less than
With Rosetta an established protocol that employs a soft-repulsion
packing of all side chains within 8 ꢁ of the mutated residue was
used. FoldX has a standard error of 3.4 kJmol (excluding 5%
outliers). For the Rosetta method, no standard error has been re-
ported, but its correlation coefficient versus experimental data (r=
[67]
À1
1
0 min, to minimize potential enzyme refolding). After heat treat-
ment, halide release was measured at 308C in reaction mixtures
0
.68) was reported to be better than for FoldX (r=0.5), so assump-
À1
with 8 (10 mm) and enzyme (5 mgmL final concentration) as de-
À1
tion of an error of ꢁ5 kJmol appears prudent. Because these
scribed for the dehalogenase assays. Initial activities were deter-
mined, and relative dehalogenase activities were plotted as a func-
tion of the corresponding temperature. The reported residual activ-
errors are for single mutations in monomeric proteins, whereas the
current protein is tetrameric and double mutations are involved,
À1
we used values eight times higher (27.2 kJmol
40 kJmol
for FoldX,
10
ities thus represent T50 values (that is, the temperatures at which
0% of the activity is lost upon 10 min of incubation). In addition,
thermal unfolding of enzyme was monitored by a fluorescence-
À1
for Rosetta) to analyze whether predicted changes in
5
folding energy were significant.
[39]
based thermal unfolding assay (thermofluor).
This method is
based on fluorescence increase upon binding of SYPRO Orange to
hydrophobic protein surfaces that become exposed upon thermal
protein unfolding or multimer dissociation. The fluorescence in-
crease was monitored by excitation at 490 nm and recording of
the emission at 575 nm with a MyiQ real-time PCR machine (Bio-
Acknowledgements
HJW was funded by NWO ECHO Grant 08.B3.051, and RJF was
supported by the Metaexplore (22625) project, funded by the
European Union (7th Framework KBBE-2007-3-3-05).
Rad, Hercules, CA, USA) while increasing block temperature from
À1
3
0 to 908C by 18Cmin . In iQ 96-well real-time PCR plates (Bio-
Rad), protein solutions containing 300-fold diluted SYPRO Orange
À1
Keywords: directed evolution · enzyme catalysis · haloalcohol
(
7.5 mL) in MilliQ water and purified protein (17.5 mL, 0.4 mgmL )
dehalogenases
· protein engineering · structure–activity
in TEMG buffer were prepared. After sealing (iQ 96-well PCR plate
seals, Bio-Rad), the temperature gradient was started. The first
derivative of measured fluorescence versus temperature was calcu-
lated. It gives the apparent melting temperature at the local maxi-
mum.
relationships · X-ray diffraction
[
1] a) D. B. Janssen, M. Majeric-Elenkov, G. Hasnaoui, B. Hauer, J. H. Lutje
Spelberg, Biochem. Soc. Trans. 2006, 34, 291–295; b) M. Schallmey, R. J.
Floor, W. Szymanski, D. B. Janssen in Comprehensive Chirality, Vol. 7
Crystallization and structure determination of HheC2360: The
(
Eds.: E. M. Carreira, H. Yamamoto), Elsevier, Amsterdam, 2012, pp. 143–
HheC2360 mutant did not crystallize under wild-type condi-
155; c) Z. Y. You, Z. Q. Liu, Y. G. Zheng, Appl. Microbiol. Biotechnol. 2013,
[7,11,58]
tions.
Instead, HheC2360 crystals were obtained by sitting
9
7, 9–21.
drop vapor diffusion at room temperature. Purified protein (1 mL,
[2] A. J. van den Wijngaard, D. B. Janssen, B. Witholt, J. Gen. Microbiol.
1989, 135, 2199–2208.
À1
1
4.7 mgmL in HEPES buffer [4-(2-hydroxyethyl)-1-piperazineeth-
[
[
[
[
3] J. H. Lutje Spelberg, J. E. T. van Hylckama Vlieg, T. Bosma, R. M. Kellogg,
D. B. Janssen, Tetrahedron: Asymmetry 1999, 10, 2863–2870.
4] L. Tang, J. E. T. van Hylckama Vlieg, J. H. Lutje Spelberg, M. W. Fraaije,
D. B. Janssen, Enzyme Microb. Technol. 2002, 30, 251–258.
5] L. Tang, A. E. J. van Merode, J. H. Lutje Spelberg, M. W. Fraaije, D. B. Jans-
sen, Biochemistry 2003, 42, 14057–14065.
anesulfonic acid, 50 mm, pH 8.0] supplemented with NaCl
(
(
300 mm)) was mixed with reservoir solution [1 mL, MgCl2
200 mm), Tris·Cl (pH 8.5, 100 mm), and PEG (poly(ethylene glycol))
8
000 (12.5%, w/v)]. After two weeks very small crystals had ap-
peared, and these could be used as seeds for the growth of larger
50ꢂ150ꢂ150 mm crystals. Crystals were flash-frozen after having
1
6] L. Tang, D. E. Torres Pazmino, M. W. Fraaije, R. M. de Jong, B. W. Dijkstra,
D. B. Janssen, Biochemistry 2005, 44, 6609–6618.
been transferred to a cryo solution (similar to the mother liquor,
except that the PEG 8000 concentration was increased to 45%). To
obtain substrate- and product-bound structures, either (S)-5 or (R)-
[7] R. M. de Jong, J. J. W. Tiesinga, H. J. Rozeboom, K. H. Kalk, L. Tang, D. B.
Janssen, B. W. Dijkstra, EMBO J. 2003, 22, 4933–4944.
[
8] J. E. T. van Hylckama Vlieg, L. Tang, J. H. Lutje Spelberg, T. Smilda, G. J.
Poelarends, T. Bosma, A. E. J. van Merode, M. W. Fraaije, D. B. Janssen, J.
Bacteriol. 2001, 183, 5058–5066.
6
(7 mm) was added to the cryo solution, and crystals were soaked
for 30 min prior to flash freezing. Diffraction data were collected at
beam lines ID4-1 and ID4-4 at the ESRF in Grenoble, France. The
diffraction data were processed by use of iMOSFLM and scaled
by use of SCALA. No phasing was necessary, because a rigid
body refinement using the wild-type structure of HheC (PDB ID:
[59]
[9] K. L. Kavanagh, H. Jçrnvall, B. Persson, U. Oppermann, Cell. Mol. Life Sci.
008, 65, 3895–3906.
10] L. Tang, J. H. Lutje Spelberg, M. W. Fraaije, D. B. Janssen, Biochemistry
003, 42, 5378–5386.
2
[60]
[
2
1
PWZ) showed clear electron density. The model was further
[11] R. M. de Jong, J. J. W. Tiesinga, A. Villa, L. Tang, D. B. Janssen, B. W. Dijk-
stra, J. Am. Chem. Soc. 2005, 127, 13338–13343.
[12] K. H. Hopmann, F. Himo, Biochemistry 2008, 47, 4973–4982.
[61]
refined by use of Refmac5 and manually improved by use of
Coot. Finally, the model was validated by use of MolProbity.
[62]
[63]
[
[
13] P. E. M. Siegbahn, F. Himo, J. Biol. Inorg. Chem. 2009, 14, 643–651.
14] J. H. Lutje Spelberg, L. Tang, M. van Gelder, R. M. Kellogg, D. B. Janssen,
Tetrahedron: Asymmetry 2002, 13, 1083–1089.
15] G. Hasnaoui-Dijoux, M. Majeric Elenkov, J. H. Lutje Spelberg, B. Hauer,
D. B. Janssen, ChemBioChem 2008, 9, 1048–1051.
Structures with bound chloride, chloride and (S)-5, and (R)-6 as
ligands have been deposited at the PDB with entry codes 4IY1,
4
IXW, and 4IXT, respectively.
[
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 0000, 00, 1 – 13
&
11
&
ÞÞ
These are not the final page numbers!

CHEMBIOCHEM
FULL PAPERS
www.chembiochem.org
[
[
16] M. Majeric Elenkov, L. Tang, A. Meetsma, B. Hauer, D. B. Janssen, Org.
Lett. 2008, 10, 2417–2420.
17] a) J. H. Lutje Spelberg, J. E. T. van Hylckama Vlieg, L. Tang, D. B. Janssen,
R. M. Kellogg, Org. Lett. 2001, 3, 41–43; b) M. Majeric Elenkov, H. W.
Hoeffken, L. Tang, B. Hauer, D. B. Janssen, Adv. Synth. Catal. 2007, 349,
[41] L. Gakhar, Z. A. Malik, C. C. R. Allen, D. A. Lipscomb, M. J. Larkin, S. Ram-
aswamy, J. Bacteriol. 2005, 187, 7222–7231.
[42] Y. Tanaka, K. Tsumoto, Y. Yasutake, M. Umetsu, M. Yao, H. Fukada, I.
Tanaka, I. Kumagai, J. Biol. Chem. 2004, 279, 32957–32967.
[43] Y. Korkhin, A. J. Kalb, M. Peretz, O. Bogin, Y. Burstein, F. Frolow, Protein
Sci. 1999, 8, 1241–1249.
2279–2285; c) J. H. Schrittwieser, I. Lavandera, B. Seisser, B. Mautner, W.
Kroutil, Eur. J. Org. Chem. 2009, 2293–2298; d) L. S. Campbell-Verduyn,
W. Szymanski, C. P. Postema, R. A. Dierckx, P. H. Elsinga, D. B. Janssen,
B. L. Feringa, Chem. Commun. 2010, 46, 898–900; e) C. Molinaro, A.-A.
Guilbault, B. Kosjek, Org. Lett. 2010, 12, 3772–3775.
[44] V. G. H. Eijsink, A. Bjørk, S. Gꢄseidnes, R. Sirevꢄg, B. Synstad, B. van den
Burg, G. Vriend, J. Biotechnol. 2004, 113, 105–120.
[45] Y. H. Kim, A. H. Berry, D. S. Spencer, W. E. Stites, Protein Eng. 2001, 14,
343–347.
[
18] a) T. Nakamura, T. Nagasawa, Y. Fujio, I. Watanabe, H. Yamada, Biochem.
Biophys. Res. Commun. 1991, 180, 124–130; b) T. Nakamura, T. Nagasa-
wa, F. Yu, I. Watanabe, H. Yamada, Tetrahedron 1994, 50, 11821–11826.
19] M. Majeric Elenkov, B. Hauer, D. B. Janssen, Adv. Synth. Catal. 2006, 348,
[46] a) S. Kumar, C.-J. Tsai, R. Nussinov, Protein Eng. 2000, 13, 179–191; b) A.
Karshikoff, R. Ladenstein, Trends Biochem. Sci. 2001, 26, 550–557.
[47] M. Pedotti, E. Rosini, G. Molla, T. Moschetti, C. Savino, B. Vallone, L. Pol-
legioni, J. Biol. Chem. 2009, 284, 36415–36423.
[48] S.-H. Baik, F. Michel, N. Aghajari, R. Haser, S. Harayama, Appl. Environ. Mi-
crobiol. 2005, 71, 3285–3293.
[49] a) M. Bocola, N. Otte, K. E. Jaeger, M. T. Reetz, W. Thiel, ChemBioChem
2004, 5, 214–223; b) M. T. Reetz, M. Puls, J. D. Carballeira, A. Vogel, K.
Jaeger, T. Eggert, W. Thiel, M. Bocola, N. Otte, ChemBioChem 2007, 8,
106–112; c) T. Ema, Y. Nakano, D. Yoshida, S. Kamata, T. Sakai, Org.
Biomol. Chem. 2012, 10, 6299–6308.
[
[
579–585.
20] a) N. Kasai, T. Suzuki, Y. Furukawa, Chirality 1998, 10, 682–692; b) S. Shi-
mizu, M. Kataoka in Biotransformations (Ed.: K. Faber), Springer, Berlin,
1999, pp. 109–123; c) N. Kasai, T. Suzuki, Adv. Synth. Catal. 2003, 345,
4
8
37–455; d) S.-H. Lee, O.-J. Park, Appl. Microbiol. Biotechnol. 2009, 84,
17–828.
[21] M. Mꢃller, Angew. Chem. 2005, 117, 366–369; Angew. Chem. Int. Ed.
2
005, 44, 362–365.
[50] J. D. Bloom, F. H. Arnold, Proc. Natl. Acad. Sci. USA 2009, 106, 9995–
10000.
[
22] J. M. Patel, J. Mol. Catal. B 2009, 61, 123–128.
[
23] S. K. Ma, J. Gruber, C. Davis, L. Newman, D. Gray, A. Wang, J. Grate, G. W.
Huisman, R. A. Sheldon, Green Chem. 2010, 12, 81–86.
24] A. Liljeblad, A. Kallinen, L. T. Kanerva, Curr. Org. Synth. 2009, 6, 362–379.
25] 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–
[51] a) A. A. Canutescu, R. L. Dunbrack, Jr., Protein Sci. 2003, 12, 963–972;
b) I. Georgiev, D. Keedy, J. S. Richardson, D. C. Richardson, B. R. Donald,
Bioinformatics 2008, 24, i196–i204; c) P. M. Murphy, J. M. Bolduc, J. L.
Gallaher, B. L. Stoddard, D. Baker, Proc. Natl. Acad. Sci. USA 2009, 106,
9215–9220; d) P.-S. Huang, Y.-E. A. Ban, F. Richter, I. Andre, R. Vernon,
W. R. Schief, D. Baker, PLoS ONE 2011, 6, e24109.
[
[
[
52] M. A. Larkin, G. Blackshields, N. P. Brown, R. Chenna, P. A. McGettigan, H.
McWilliam, F. Valentin, I. M. Wallace, A. Wilm, R. Lopez, J. D. Thompson,
T. J. Gibson, D. G. Higgins, Bioinformatics 2007, 23, 2947–2948.
53] J. Jin, H. Mazon, R. H. H. van den Heuvel, A. J. Heck, D. B. Janssen, M. W.
Fraaije, FEBS J. 2008, 275, 5191–5200.
3
44.
[
26] S. C. Davis, R. J. Fox, G. W. Huisman, V. Gavrilovic, E. C. Mundorff, L. M.
Newman, Halohydrin Dehalogenases and Related Polynucleotides, U.S.
Patent 7588928, 2009.
[
[
[
[
[
[
[
[
[
[
[
[
27] L. Tang, X. Zhu, H. Zheng, R. Jiang, M. Majeric-Elenkov, Appl. Environ. Mi-
54] R. M. Haak, C. Tarabiono, D. B. Janssen, A. J. Minnaard, J. G. de Vries, B. L.
Feringa, Org. Biomol. Chem. 2007, 5, 318–323.
crobiol. 2012, 78, 2631–2637.
28] X. Gao, X. Xie, I. Pashkov, M. R. Sawaya, J. Laidman, W. Zhang, R. Cacho,
T. O. Yeates, Y. Tang, Chem. Biol. 2009, 16, 1064–1074.
[
[
55] J. G. Bergmann, J. Sanik, Anal. Chem. 1957, 29, 241–243.
56] A. J. J. Straathof, J. A. Jongejan, Enzyme Microb. Technol. 1997, 21, 559–
29] S. Ahmad, M. Z. Kamal, R. Sankaranarayanan, N. M. Rao, J. Mol. Biol.
5
71.
2
008, 381, 324–340.
[
[
57] M. W. Scoggins, Anal. Chem. 1972, 44, 1294–1296.
58] R. M. de Jong, H. J. Rozeboom, K. H. Kalk, L. Tang, D. B. Janssen, B. W.
Dijkstra, Acta Crystallogr. Sect. D 2002, 58, 176–178.
59] T. G. G. Battye, L. Kontogiannis, O. Johnson, H. R. Powell, A. G. W. Leslie,
Acta Crystallogr. Sect. D 2011, 67, 271–281.
60] P. Evans, Acta Crystallogr. Sect. D 2006, 62, 72–82.
61] A. A. Vagin, R. A. Steiner, A. A. Lebedev, L. Potterton, S. McNicholas, F.
Long, G. N. Murshudov, Acta Crystallogr. Sect. D 2004, 60, 2184–2195.
62] P. Emsley, K. Cowtan, Acta Crystallogr. Sect. D 2004, 60, 2126–2132.
63] I. W. Davis, A. Leaver-Fay, V. B. Chen, J. N. Block, G. J. Kapral, X. Wang,
L. W. Murray, W. B. Arendall, J. Snoeyink, J. S. Richardson, D. C. Richard-
son, Nucleic Acids Res. 2007, 35, W375–W383.
64] E. Krissinel, K. Henrick, J. Mol. Biol. 2007, 372, 774–797.
65] a) R. Guerois, J. E. Nielsen, L. Serrano, J. Mol. Biol. 2002, 320, 369–387;
b) J. Schymkowitz, J. Borg, F. Stricher, R. Nys, F. Rousseau, L. Serrano, Nu-
cleic Acids Res. 2005, 33, W382–W388.
30] B. Spiller, A. Gershenson, F. H. Arnold, R. C. Stevens, Proc. Natl. Acad. Sci.
USA 1999, 96, 12305–12310.
31] T. Ohki, N. Shibata, Y. Higuchi, Y. Kawashima, M. Takeo, D. Kato, S.
Negoro, Protein Sci. 2009, 18, 1662–1673.
32] B. Spiller, A. Gershenson, F. H. Arnold, R. C. Stevens, Proc. Natl. Acad. Sci.
USA 1999, 96, 12305–12310.
33] S. Oue, A. Okamoto, T. Yano, H. Kagamiyama, J. Biol. Chem. 1999, 274,
[
[
[
2
344–2349.
34] M. Majeric Elenkov, L. Tang, B. Hauer, D. B. Janssen, Org. Lett. 2006, 8,
227–4229.
35] P. E. Tomatis, R. M. Rasia, L. Segovia, A. J. Vila, Proc. Natl. Acad. Sci. USA
005, 102, 13761–13766.
[
[
4
2
[
[
36] A. Shimotohno, S. Oue, T. Yano, S. Kuramitsu, H. Kagamiyama, J. Bio-
chem. 2001, 129, 943–948.
[
[
37] M. T. Reetz, J. D. Carballeira, Nat. Protoc. 2007, 2, 891–903.
38] I. Campeotto, A. H. Bolt, T. A. Harman, C. Dennis, C. H. Trinh, S. E. V. Phil-
lips, A. Nelson, A. R. Pearson, A. Berry, J. Mol. Biol. 2010, 404, 56–69.
39] U. B. Ericsson, B. M. Hallberg, G. T. DeTitta, N. Dekker, P. Nordlund, Anal.
Biochem. 2006, 357, 289–298.
[
[
66] R. Das, D. Baker, Annu. Rev. Biochem. 2008, 77, 363–382.
67] E. H. Kellogg, A. Leaver-Fay, D. Baker, Proteins Struct. Funct. Bioinf. 2011,
[
[
7
9, 830–838.
40] D. Narasimhan, M. R. Nance, D. Gao, M.-C. Ko, J. Macdonald, P. Tamburi,
D. Yoon, D. M. Landry, J. H. Woods, C.-G. Zhan, J. J. G. Tesmer, R. K. Suna-
hara, Protein Eng. Des. Sel. 2010, 23, 537–547.
Received: January 3, 2013
Published online on && &&, 0000
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 0000, 00, 1 – 13
&12
&
ÞÞ
These are not the final page numbers!

FULL PAPERS
Synergistic mutations: A halohydrin
dehalogenase with 37 mutations and
improved catalytic properties for statin
side chain synthesis has been biochemi-
cally characterized. Crystal structures
with different ligands in the active site
give insight into the way in which in-
dividual mutations contribute to en-
hanced stability and faster cyanolysis of
epoxides and illustrate the importance
of synergistic mutations in directed evo-
lution.
M. Schallmey, R. J. Floor, B. Hauer,
M. Breuer, P. A. Jekel, H. J. Wijma,
B. W. Dijkstra, D. B. Janssen*
&& – &&
Biocatalytic and Structural Properties
of a Highly Engineered Halohydrin
Dehalogenase
ꢀ
2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 0000, 00, 1 – 13
&13
&
ÞÞ
These are not the final page numbers!