Improving the properties of bacterial r-selective hydroxynitrile lyases for industrial applications

Article abstract of DOI:10.1002/cctc.201402742

Hydroxynitrile lyases (HNLs) catalyse the reversible cleavage of cyanohydrins to carbonyl compounds and HCN. The recent discovery of bacterial HNLs with a cupin fold gave rise to a new promising class of these enzymes. They are interesting candidates for the synthesis of cyanohydrins on an industrial scale owing to their high expression levels in Escherichia coli. The activity and enantioselectivity of the manganese-dependent HNL from Granulicella tundricola (GtHNL) were significantly improved by site-saturation mutagenesis of active site amino acids. The combination of beneficial mutations resulted in a variant with 490-fold higher specific activity in comparison to the wild-type enzyme. More importantly, GtHNL-A40H/V42T/Q110H is a highly competitive alternative for the synthesis of chiral cyanohydrins, such as 2-chlorobenzaldehyde cyanohydrin, (R)-2-hydroxy-4-phenylbutyronitrile, and (R)-2-hydroxy-4-phenyl-3-butene nitrile, which serve as intermediates for the synthesis of pharmaceuticals.

Full text of DOI:10.1002/cctc.201402742

DOI: 10.1002/cctc.201402742
Full Papers
Improving the Properties of Bacterial R-Selective
Hydroxynitrile Lyases for Industrial Applications
Romana Wiedner,[a] Bettina Kothbauer,[a] Tea Pavkov-Keller,[a] Mandana Gruber-Khadjawi,[a]
Karl Gruber,[a, b] Helmut Schwab,[a, c] and Kerstin Steiner*[a]
Hydroxynitrile lyases (HNLs) catalyse the reversible cleavage of
cyanohydrins to carbonyl compounds and HCN. The recent dis-
covery of bacterial HNLs with a cupin fold gave rise to a new
promising class of these enzymes. They are interesting candi-
dates for the synthesis of cyanohydrins on an industrial scale
owing to their high expression levels in Escherichia coli. The ac-
tivity and enantioselectivity of the manganese-dependent HNL
from Granulicella tundricola (GtHNL) were significantly im-
proved by site-saturation mutagenesis of active site amino
acids. The combination of beneficial mutations resulted in
a variant with 490-fold higher specific activity in comparison to
the wild-type enzyme. More importantly, GtHNL-A40H/V42T/
Q110H is a highly competitive alternative for the synthesis of
chiral cyanohydrins, such as 2-chlorobenzaldehyde cyanohy-
drin, (R)-2-hydroxy-4-phenylbutyronitrile, and (R)-2-hydroxy-4-
phenyl-3-butene nitrile, which serve as intermediates for the
synthesis of pharmaceuticals.
Introduction
Hydroxynitrile lyases (HNLs) catalyse the stereoselective cleav-
age of cyanohydrins. More importantly, they are valuable tools
in biocatalysis owing to their ability to synthesise chiral a-cya-
nohydrins through a CÀC bond forming condensation reaction.
A chiral centre is (potentially) formed, the carbon chain is elon-
gated by one carbon atom, and an additional versatile func-
tional group—the nitrile—is introduced into the molecule.
Enantiopure cyanohydrins are versatile building blocks and in-
termediates that serve as starting materials for many enzymatic
and chemical follow-up reactions, which find application in
pharmaceutical, agrochemical, and cosmetic industries.[1–4] To
meet the requirements for industrial applications, HNLs need
to fulfil several criteria in addition to high enantioselectivity:
1) availability of sufficient quantities of proteins with consistent
quality and batch-to-batch reproducibility at low cost, 2) broad
substrate scope, 3) high stability at acidic pH and high solvent
stability, and 4) activity at low temperatures [as the unselective
chemical background reaction is significantly suppressed at
low pH (<4.5), at low temperature, and in the absence of
water].[5]
A number of R- and S-selective HNLs have been identified so
far (see tables in related reviews[4,6]); however, only a few of
them can be heterologously expressed in sufficient amounts
either in Escherichia coli or in Pichia pastoris. Bacterial HNLs
with a cupin fold, which have recently been discovered, can be
expressed in E. coli in exceptionally high yield.[7–9] The bacterial
manganese-dependent HNL from Granulicella tundricola
(GtHNL) was characterised in detail, and its structure was
solved.[7,10] GtHNL catalysed the synthesis of (R)-mandelonitrile
from benzaldehyde and HCN with 80% conversion and 90%
ee, which is not sufficient for industrial applications, but is
a promising starting point for protein engineering approaches.
In recent years, enzyme engineering by either random or di-
rected methods was applied to HNLs to increase their stability
(at elevated temperature, at acidic pH, or in organic solvents),
to broaden their substrate scope, to reverse or increase their
enantioselectivity, and to increase their expression
levels.[4,6,11–13] The stability of the R-selective AtHNL from Arabi-
dopsis thaliana at acidic pH was improved by one pH unit by
exchanging eleven surface amino acids with the corresponding
amino acids of the S-selective MeHNL from Manihot esculenta,
which belongs to the same fold and is more stable.[14] More-
over, the activity, enantioselectivity, and substrate scope of the
R-selective PaHNL5 from Prunus amygdalus were significantly
improved by enzyme engineering.[15–18]
[a] Dr. R. Wiedner, B. Kothbauer, Dr. T. Pavkov-Keller, Dr. M. Gruber-Khadjawi,
Prof. K. Gruber, Prof. H. Schwab, Dr. K. Steiner
ACIB, Austrian Centre of Industrial Biotechnology GmbH
Petersgasse 14, 8010 Graz (Austria)
Fax: (+43)3168739346
E-mail: kerstin.steiner@acib.at
[b] Prof. K. Gruber
Institute of Molecular Biosciences
University of Graz
Humboldtstraße 50, 8010 Graz (Austria)
The aim of the present work was the engineering of GtHNL
to improve its activity and stability at low pH as well as to
broaden its substrate scope. For this purpose, random muta-
genesis of the entire coding region and site-saturation muta-
genesis of active site amino acids (designed evolution) were
performed. Promising variants were tested as catalysts for the
[c] Prof. H. Schwab
Institute of Molecular Biotechnology
TU Graz
Petersgasse 14, 8010 Graz (Austria)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201402742.
ChemCatChem 2015, 7, 325 – 332
325
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
synthesis of mandelonitrile and several other industrially rele-
vant cyanohydrins.
Results and Discussion
Engineering of cupin HNLs
GtHNL catalysed the synthesis of R-selective mandelonitrile in
a biphasic reaction system [100 mm sodium acetate buffer,
pH 4.0, methyl tert-butyl ether (MTBE) (1:2) containing 0.5m
benzaldehyde and 2m HCN, 1000 rpm, 58C, 24 h] with a good
conversion of 80% and promising ee of 90%.[7] Unfortunately,
the substrate scope of the wild-type enzyme was rather limit-
ed. Thus, GtHNL had to be improved with respect to its gener-
al activity and substrate scope by protein engineering. The ac-
tivity and stability at acidic pH were targeted to suppress the
unspecific chemical background reaction, which significantly
reduces the enantiopurity of the reaction products. Different
mutant libraries were created by random mutagenesis, includ-
ing either the whole coding region or three parts of the gene
excluding the metal-binding amino acids (Figure S1). In addi-
tion, site-saturation mutagenesis of active site amino acids was
performed.
Figure 1. Crystal structure of GtHNL-A40H/V42T/Q110H superimposed with
that of GtHNL-WT (PDB entry 4bif; grey). Active site amino acids are shown
as sticks (c: WT; GtHNL-A40H/V42T/Q110H: c: metal-binding amino
acids, c: active site amino acids). The figure was prepared by using the
program PyMOL.
expressed as soluble proteins at the same level as the wild-
type GtHNL (GtHNL-WT; Figure S3). As the hits were detected
in a colony assay, their activity and pH stability had to be con-
firmed in the cyanogenesis of (R)-mandelonitrile at different
pH values by using purified GtHNL variants in a quantitative
photometric assay.
Screening of the libraries for improved HNL activity and
stability at acidic pH
Even though many of the chosen variants demonstrated in-
creased activities than GtHNL-WT, hits from site-saturation mu-
tagenesis (Table 1) were significantly more active than most
hits from random mutagenesis (Figure S4), which were mainly
in the activity range of the wild-type enzyme.
Whole gene and fragment random mutagenesis of GtHNL re-
sulted in reasonable mutation rates (corresponding to one to
three amino acid mutations per gene for the whole gene li-
brary). In total, 22000 variants obtained from different random
libraries were screened for improved activity and stability to-
wards (R)-mandelonitrile at pH 3.5 with a colony-based colori-
metric HNL assay,[19] detecting HCN released during cyanogen-
esis (Figure S2). From 122 preliminary selected hits, 30 variants
with apparently improved activity in the cleavage of (R)-man-
delonitrile were identified after rescreening. Five variants har-
boured an amino acid exchange at position Q110 and one var-
iant had an amino acid exchange at position V42—the two po-
sitions that were also targeted in site-saturation mutagenesis.
In addition, the site-saturation mutagenesis libraries (264
variants of each) of eight active site positions (A40, V42, F44,
T50, L61, H96, H106, and Q110, lining the cavity of the pro-
posed active site;[7] see Figure 1) were screened for improved
activity towards (R)-mandelonitrile. The libraries at position
F44, T50, L61, and H96 contained no improved variants, where-
as two hits were obtained from the A40 library, one improved
variant resulted from the V42 library, and two and one hit
were discovered in the H106 and Q110 libraries, respectively
(Table S1).
Table 1. Specific activities of different GtHNL variants obtained from site-
saturation mutagenesis libraries in different buffers at various pH values
(100 mm citrate/phosphate and oxalate buffer at pH 5.0 and 5.5) with
18 mm (R)-mandelonitrile as the substrate.[a]
GtHNL variant
Specific activity [UmgÀ1
Oxalate buffer Citrate/phosphate buffer
]
pH 5.0
pH 5.5
pH 5.0
pH 5.5
WT
0.20Æ0.00
3.21Æ0.13
4.45Æ0.30
0.70Æ0.01
0.29Æ0.01
3.69Æ0.06
5.03Æ0.20
1.37Æ0.14
0.12Æ0.02
0.48Æ0.04
0.57Æ0.01
0.11Æ0.01
0.28Æ0.03
3.34Æ0.10
1.83Æ0.01
0.35Æ0.03
A40H
V42T
Q110H
[a] Depending on the activity of the variant, different protein concentra-
tions (from 5 mgmLÀ1 to 1 mgmLÀ1) were used. The increase in benzalde-
hyde absorption was followed at 280 nm in a plate reader.
The best variant obtained from random mutagenesis
(GtHNL-F29L/L36R/V42A) exhibited an approximately 5 times
higher specific activity than GtHNL-WT, whereas GtHNL-A40H,
GtHNL-V42T, and GtHNL-Q110H showed 12, 17, and 5 times
higher specific activities at pH 5.5 in oxalate buffer than the
wild-type enzyme. However, all variants showed substantially
reduced or even no activity at pH 4.5 and 4.0 (data not
shown). In many cases, denaturation of the enzymes was visi-
ble as precipitation. This finding was confirmed by thermal
shift measurements, which showed a significant decrease in
Confirmation of improved HNL activity and stability at
acidic pH
The most promising variants obtained from both approaches
were recloned into a pET26b expression vector and expressed
in shake flasks under standard conditions.[7] They were all well
ChemCatChem 2015, 7, 325 – 332
www.chemcatchem.org
326
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
the melting temperature (Tm) of GtHNL-WT and all variants in
different buffers, from pH 5.5 (e.g., sodium oxalate: Tm values
in the range of 60–708C, depending on the variant) to pH 4.5
(at which several variants were already unfolded at the start of
the measurement at 208C or had Tm values in the range of 35–
508C) and pH 4.0 (at which all proteins were unfolded at the
beginning, except for GtHNL-WT and variants A40H and A40R).
Most variants were slightly more stable in citrate/phosphate
buffer and sodium acetate buffer than in sodium oxalate
buffer. However, none of the variants had a higher melting
temperature than the wild-type enzyme in any of the tested
buffers. These results reveal that colony-based assays have to
be handled with care with regard to pH stability. Although the
colonies were treated with freeze–thaw cycles to break the
cells and enable equilibration at low pH, the results indicate
that either the cells were not sufficiently broken and the pH of
the unbroken cells was higher than assumed or the HNL pro-
tein was still stabilised or protected against the low pH by cell
debris.
version (38.4%) and reduced ee (73.7%). GtHNL-H106Y con-
verted 97.9% of benzaldehyde in 24 h, but completely lost its
enantioselectivity.
Screening of random libraries for activity towards racemic
2-chloromandelonitrile
(R)-2-Chloromandelonitrile is a precursor of (R)-2-chloromandel-
ic acid, a key intermediate in the synthesis of an antiplatelet
agent, clopidogrel. The synthesis of (R)-2-chloromandelonitrile
from 2-chlorobenzaldehyde has previously been improved by
rational engineering of HNL isoenzyme 5 from P. amygdalus.[15]
However, PaHNL5 can be overexpressed only in P. pastoris, and
an enzyme catalyst expressed in E. coli in high yield would be
preferable. The screening of the whole gene random library for
activity towards (R/S)-2-chloromandelonitrile originated nine
variants that showed increased activity. Seven of these variants
shared a variation at position F19, F19V, or L as a single amino
acid exchange or in combination with silent mutations or addi-
tional amino acid exchanges (data not shown). The two re-
maining hits harboured the amino acid exchange at W120R.
These candidates were not discovered during the screening of
the same library with the (R)-mandelonitrile substrate. This
finding was later confirmed by the fact that purified GtHNL-
F19V and -W120R were both less active in the cyanogenesis of
(R)-mandelonitrile in the plate reader assay as compared with
the wild-type enzyme (data not shown). Whereas GtHNL-WT
was almost inactive with 2-chloromandelonitrile (S enantiomer:
no activity; R enantiomer: 0.03 UmgÀ1 in citrate/phosphate
buffer, pH 5.5), GtHNL-F19V was particularly active towards (S)-
2-chloromandelonitrile (S enantiomer: 2.08 UmgÀ1; R enantio-
mer: 0.39 UmgÀ1) and GtHNL-W120R showed some activity to-
wards both enantiomers (S enantiomer: 0.25 UmgÀ1; R enan-
tiomer: 0.13 UmgÀ1).
Activity of the variants in the synthesis of (R)-mandelonitrile
First experiments for the synthesis of (R)-mandelonitrile with
cleared lysate of E. coli showed that several variants demon-
strated improved conversion and/or enantioselectivity in com-
parison to GtHNL-WT (Table 2).
Table 2. Conversion of benzaldehyde and ee of the (R)-cyanohydrin enan-
tiomer catalysed by GtHNL variants obtained from random and site-di-
rected mutagenesis.[a]
GtHNL variant
Conversion [%]
ee (R) [%]
I3L/T20A
I25M/L36Q
I109L/A117V
A40H
96.5
92.4
98.4
99
97.6
38.4
97.9
95.4
98.8
84.6
94.1
93.6
96.3
87.9
94.5
73.7
n.d.[b]
95.4
96.3
88.7
In the synthesis of 2-chloromandelonitrile with cell-free
lysate, GtHNL-F19V and GtHNL-F19V/F88V showed higher con-
versions (97.8 and 94.7%, respectively) than GtHNL-WT (89%
conversion with 36.1% ee (R)) but exhibited inverted enantio-
selectivities (26 and 33.9% ee (S), respectively). Conversely, the
variant W120R converted 99.8% of the substrate to (R)-2-chlor-
omandelonitrile with a significantly improved ee of 71.9% in
comparison to the wild-type enzyme. Both variants were signif-
icantly less active and enantioselective than GtHNL-WT in the
synthesis of mandelonitrile (Figure S5). The results suggest that
the choice of the substrate in the screening affects the result:
because racemic 2-chloromandelonitrile was used as the
screening substrate, variants with higher activity were identi-
fied independent of their enantioselectivity. W120 is located in
the active site close to the metal-binding site (Figure 1). As-
suming that the carbonyl group of 2-chlorobenzaldehyde
points towards the metal, the chlorine at the ortho position
could come in close contact with the bulky tryptophan side
chain and thus steric hindrance could cause a slightly different
binding position, which resulted in lower conversion and enan-
tioselectivity of the wild-type enzyme (Figure S6). A mutation
at this position creates more space for the chlorine, and the
positively charged arginine could interact with the electroneg-
A40R
H106N
H106Y
Q110H
V42T
WT
[a] Reaction conditions: two-phase system consisting of 500 mL of cell-
free lysate (22.5 mg of the total protein with ꢀ50% cupin in 100 mm
sodium acetate buffer, pH 4.0) as the aqueous phase and 1 mL of MTBE
containing 0.5m benzaldehyde and 2m HCN as the organic phase,
1000 rpm, T=58C, t=24 h. [b] Not detected.
Most variants that resulted from the random libraries
showed similar conversions and enantioselectivities as did
GtHNL-WT (Figure S5). However, significantly increased conver-
sion and ee values were obtained with GtHNL-I3L/T20A,
GtHNL-I25M/L36Q, and GtHNL-I109L/A117V (Table 2). Improved
conversions and enantioselectivities were again observed with
several variants from site-saturation mutagenesis libraries, that
is, GtHNL-A40H, GtHNL-A40R, GtHNL-V42T, and GtHNL-Q110H.
Surprisingly, GtHNL-H106N showed substantially reduced con-
ChemCatChem 2015, 7, 325 – 332
www.chemcatchem.org
327
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
ative chlorine. F19 is located in the already open entrance
region of the active site of GtHNL. The exchange of F19 with
the smaller amino acids valine or leucine could enable the
entry of larger substrates; however, the larger space can favour
non-productive binding or binding in a different orientation,
which results in reduced or inverted enantioselectivity.
sodium oxalate buffer, pH 5.5, which revealed a Michaelis con-
stant (Km) of 12.8Æ0.4 mm and a turnover number (kcat) of
51Æ1.2 sÀ1 for GtHNL-A40H/V42T/Q110H compared with a Km
of 6.1Æ0.2 mm and a kcat of 0.075Æ0.003 sÀ1 for GtHNL-WT.
Substrate scope of GtHNL-A40H/V42T/Q110H
Purified GtHNL-A40H/V42T/Q110H was subjected to substrate
screening in the synthesis of cyanohydrins including several al-
dehydes (Table 3). Hydroxypivaldehyde was tested at pH 2.4
and 4.0, but was not converted by the variant, which was not
Combination of beneficial mutations
The beneficial mutations identified in the site-saturation muta-
genesis libraries (A40H, A40R, V42T, and Q110H) were com-
bined. Whereas most of the double variants
showed only a slight increase in conversion and/
Table 3. Conversion and enantioselectivity of purified GtHNL-A40H/V42T/Q110H towards
or enantioselectivity in comparison to the single
variants, the double mutant GtHNL-A40H/V42T
was significantly better (data not shown). Most
promising, however, was the triple mutant
GtHNL-A40H/V42T/Q110H, which had already ach-
ieved nearly full conversion of benzaldehyde after
2 h (98%), with an excellent ee (R) of >99.9%
(Figure 2).
different aldehydes after 2 and 24 h.[a]
Substrate[b]
t=2 h
t=24 h
Mutant
WT
Mutant
WT
Conv. [%] ee [%]
Conv. [%] ee [%]
Conv. [%] ee [%]
Conv. [%] ee [%]
Moreover, purified GtHNL-A40H/V42T/Q110H
showed a specific activity of 137Æ9.95 UmgÀ1 in
the cyanogenesis of 18 mm (R)-mandelonitrile in
citrate/phosphate buffer, pH 5.5, which corre-
sponds to an approximately 490-fold increase in
activity compared with GtHNL-WT under the
same reaction conditions. In contrast to GtHNL-
WT, purified GtHNL-A40H/V42T/Q110H showed
significant residual activity even at pH 4.0 and 4.5
(2.3Æ0.04 and 10.3Æ0.09 UmgÀ1, respectively).
Kinetic measurements of purified GtHNL-A40H/
V42T/Q110H and GtHNL-WT were performed in
1a
2a
3a
4a
5a
6a
87.9
62.1
85.4
70.7
98.2
n.d.[e]
>99.9 n.x.[c]
99.1 24.5
97.9 26.4
>99.9 17.4
99.5[d] 77.9
n.x.
14.9 99.9
À13.9 94.7
>99.9 97.0
74.8[d] 98.3
98.6
99.8 50.0
89.3
23.8
99.4 60.5
98.5 78.9
>99.9 21.8
98.1[d] 95.8
À13.5
77.8
72.2[d]
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
[a] Reaction conditions: two-phase system consisting of 500 mL of purified enzyme [4.7 mg
of GtHNL-A40H/V42T/Q110H (mutant) and 3.7 mg of GtHNL-WT in 100 mm sodium acetate,
pH 4.0] as the aqueous phase and 1 mL of MTBE containing 0.5m aldehyde and 2m HCN
as the organic phase, 1000 rpm, T=58C. [b] 1a=benzaldehyde; 2a=2-chlorobenzalde-
hyde; 3a=3-phenylpropanal; 4a=3-phenylprop-2-enal; 5a=2-furylaldehyde; 6a=hy-
droxypivaldehyde.[c] Not determined. [d] Change of product configuration as a conse-
quence of the Cahn–Ingold–Prelog rule. [e] Not detected.
surprising because at pH 4.0 it is mainly found as a dimer and
the enzyme was not active at pH 2.4. For all other substrates,
excellent enantioselectivities and good to excellent conver-
sions were achieved (Table 3). Reducing the applied amount of
the enzyme to a tenth still resulted in excellent enantioselec-
tivities but with reduced conversion after 2 h and full conver-
sion after 24 h (Table S2).
Compared to literature data, the excellent enantioselectivity
in particular makes GtHNL-A40H/V42T/Q110H a highly compet-
itive alternative for the synthesis of chiral cyanohydrins. The R-
selective AtHNL from A. thaliana resulted in 99% conversion of
3-phenylpropanal but only 68% ee after 22 h (however with
only 50 mm aldehyde and 0.25m HCN).[20] While PaHNL-WT
showed moderate to excellent conversions and enantioselec-
Figure 2. Conversion of benzaldehyde and ee (R) obtained with GtHNL-WT
(grey) and GtHNL-A40H/V42T/Q110H (light grey) at different time points.
Conversions are shown as solid lines, and ee values are shown as dashed
lines. Reaction conditions: two-phase system consisting of 500 mL of cell-free
lysate (22.5 mg total protein with ꢀ50% cupin in 100 mm sodium acetate
buffer, pH 4.0) as the aqueous phase and 1 mL of MTBE containing 0.5m
benzaldehyde and 2m HCN as the organic phase, 1000 rpm, T=58C. The
product (black line) of the negative control (empty pET26b vector) is race-
mic.
tivities depending on the substrate, several variants of PaHNL
were created in recent years that achieved excellent conver-
sions and enantioselectivities (e.g., 2-chlorobenzaldehyde:
98.5%;
3-phenylpropanal:
96.7%;
3-phenylprop-2-enal:
97.6%).[15,16,18,21,22]
ChemCatChem 2015, 7, 325 – 332
www.chemcatchem.org
328
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Structural analysis of GtHNL-A40H/V42T/Q110H
troduced residues are positioned in such a way that a network
of hydrogen bonds is formed (Figure 3e), which thus defines
their own conformations as well as the positions of water mol-
ecules in the active sites.
The GtHNL-A40H/V42T/Q110H variant was crystallised in a dif-
ferent space group as compared with the wild-type enzyme,[7]
and its structure was determined at 2.1 ꢁ resolution (Table S3).
The asymmetric unit contains 20 molecules of the variant ar-
ranged in five tetramers. The tetramers as well as the overall
fold of the individual chains are conserved in the variant
enzyme, as indicated by the overall root mean square devia-
tion (for Ca atoms) between the two structures, which does
not exceed 0.35 ꢁ when individual chains (or tetramers) are su-
perimposed (Figure 1).
The minimal requirements for HNL activity are the presence
of a catalytic base and a positive charge or general positive
electrostatic potential in the binding pocket.[23] By assuming
that the hydroxyl group of the cyanohydrin (or the carbonyl
oxygen of the aldehyde or ketone) is bound to the manganese
ion, we propose that the negative charge at the cyano group
or cyanide is better stabilised by the additional positive charg-
es provided by the two histidine residues in comparison to the
wild-type enzyme (Figure S6).
The active site structure, however, demonstrates profound
changes that correlate with the measured activity increase in
the variant. With three amino acid substitutions, a much nar-
rower cavity is created in the vicinity of the manganese-bind-
ing site (Figure 3a,b). This cavity is more hydrophilic than in
the wild-type enzyme and possibly provides additional positive
charges through His40 and His110 (Figure 3c,d). The newly in-
Transfer of beneficial mutations to the cupin HNL from
Acidobacterium capsulatum
To investigate the general importance of these amino acids in
the active site of HNLs with a cupin fold, the beneficial muta-
tions A40H, V42T, and Q110H were transferred as
single and triple mutations to a recently discovered
highly similar cupin HNL from Acidobacterium capsu-
latum ATCC 51196 (AcHNL),[9] and tested in the cyano-
genesis and synthesis of (R)-mandelonitrile (Tables S4
and S5). GtHNL and AcHNL share 77% sequence
identity, and most active site amino acids such as
A40, V42, and Q110 are identical. The purified AcHNL
triple mutant AcHNL-A40H/V42T/Q110H showed
a specific activity of 139Æ11.2 UmgÀ1 in the cyano-
genesis of 18 mm (R)-mandelonitrile in citrate/phos-
phate buffer, pH 5.5, which is identical to that of
GtHNL-A40H/V42T/Q110H. Moreover, purified AcHNL-
A40H/V42T/Q110H was investigated in the bioconver-
sion of benzaldehyde and showed 99.4% conversion
and 99.7% ee (R) after 24 h.
Conclusions
Random and site-saturation mutagenesis of the man-
ganese-dependent hydroxynitrile lyase (HNL) from
Granulicella tundricola (GtHNL) originated a remark-
able number of variants with improved HNL activity
in the cyanogenesis of (R)-mandelonitrile and, more
importantly, in the synthesis of cyanohydrins. GtHNL-
A40H/V42T/Q110H, a combinatorial variant of three
beneficial amino acid exchanges in the active site,
was superior to all other variants. Moreover, the gen-
eral importance of these amino acids in the active
site of HNLs with a cupin fold was confirmed by the
successful transfer of these amino acid exchanges to
another cupin HNL, the cupin HNL from Acidobacteri-
Figure 3. Structural analysis of the active site of GtHNL-A40H/V42T/Q110H (b,d: c: ex-
changed amino acids) compared with GtHNL-WT (a,c: c). The manganese ion is de-
picted as a purple non-bonded sphere. a,b) Comparison of the size and shape of the
cavities that harbour the active site of the WT and the variant enzymes, respectively, de-
picted as a meshed cloud. c,d) Comparison of the distribution of water molecules in the
active sites (red non-bonded spheres). e) Zoom-in of the network of hydrogen bonds in
the active site of GtHNL-A40H/V42T/Q110H. The figures were prepared with the program
PyMOL.
um capsulatum.
The excellent enantioselectivity combined with its
high expression level in Escherichia coli makes
GtHNL-A40H/V42T/Q110H a highly competitive alter-
native for the synthesis of chiral cyanohydrins, such
as 2-chlorobenzaldehyde cyanohydrin, (R)-2-hydroxy-
ChemCatChem 2015, 7, 325 – 332
www.chemcatchem.org
329
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
4-phenylbutyronitrile, and (R)-2-hydroxy-4-phenyl-3-butene ni-
trile, which serve as intermediates for the synthesis of pharma-
ceuticals, for example antiplatelet agents or a class of angio-
tensin-converting enzyme inhibitors.
Screening of libraries
The transformants from each GtHNL random library were grown
on LB agar (supplemented with 100 mgLÀ1 of ampicillin). The colo-
nies were transferred to 384-well plates (Nunc, Thermo Scientific)
filled with 2xTY medium (55 mL per well) containing ampicillin
(final 100 mgLÀ1 per well) with a QPix picking robot (Genetix, now
Molecular Devices LLC). The transformants from site-saturation mu-
tagenesis libraries were picked into the wells of 96-well plates
(filled with 100 mL of LB medium containing 100 mgLÀ1 of ampicil-
lin) with sterile toothpicks. E. coli TOP10F’ cells harbouring
pMS470-GtHNL or empty pMS470D8 plasmid were added to each
plate as positive and negative controls, respectively. All plates were
incubated at 378C overnight and were then used as templates for
transferring colonies to nylon membranes (Biodyne A, 0.2 mm; Pall
Life Sciences) with a pin stamp. A colony-based colorimetric filter
assay, as described previously[19] (for details, see the Supporting In-
formation), using (R)-mandelonitrile or racemic 2-chloromandeloni-
trile as a substrate was used.
Experimental Section
General
All chemicals were purchased from Sigma–Aldrich or Carl Roth
GmbH, if not stated otherwise. Materials for molecular biology
were obtained from Thermo Fisher Scientific, if not specifically
mentioned. E. coli TOP10F’ was used for basic cloning work and as
a host for mutant libraries. E. coli BL21-Gold(DE3) was used for pro-
tein expression. HPLC analyses were performed on an Agilent 1100
instrument (Agilent Technologies) equipped with an autosampler,
a multi-wavelength detector, and a Daicel Chiralcel OD-H column
(250 mmꢂ4.6 mmꢂ5 mm; Chiral Technologies). For GC analysis, an
HP 6890 GC system (Hewlett-Packard) equipped with a flame ioni-
zation detector and a CP-Chirasil-DEX CB column (25 mꢂ0.32 mm,
0.25 mm film; Agilent Technologies) was used.
Protein expression and purification
Precultures of E. coli BL21-Gold(DE3) cells harbouring pET26b(+)
constructs were grown in LB medium (100 mL) supplemented with
kanamycin (40 mgLÀ1) at 378C overnight. They were used for the
inoculation of LB medium (330 mL) containing kanamycin
(40 mgLÀ1) to an optical density at l=600 nm (OD600) of approxi-
mately 0.03. Cultivation was performed at 378C until an OD600 of
0.6–0.8 was reached. Protein expression was induced by adding
isopropyl-b-d-thiogalactoside (0.1 mm), and cultivation was contin-
ued at 258C for 20 h. Moreover, MnCl2 (0.1 mm) was added to the
cultures at the induction time. Cells were harvested at 3600g,
20 min, and 48C; resuspended in sodium phosphate buffer (30 mL,
10 mm, pH 7.0); and disrupted by sonication (Sonifier S-250, Bran-
son Ultrasonics Corporation; 80% duty cycle, output control 8) for
6 min under continuous cooling. Cell-free lysates were obtained by
centrifugation (48000g, 1 h, 48C). For cyanohydrin synthesis,
cleared lysates were concentrated to 50 mgmLÀ1 with Vivaspin 20
centrifugal concentrators (10 kDa molecular mass cut-off; Sartor-
ius). The lysates were stored at À208C.
Cloning and mutagenesis
For details of the used methods, see the Supporting Information.
Random mutagenesis was performed by an error-prone poly-
merase chain reaction (PCR). The error rate of the polymerase was
increased by the addition of MnCl2, an increased amount of MgCl2,
and unbalanced nucleotide concentrations. The plasmid pMS470-
GtHNL (codon optimised for E. coli[7]) was used as the template for
the amplification of the 396 bp target sequence. The PCR product
was purified and recloned into pMS470D8[24] by using NdeI and
HindIII restriction sites. The desalted ligation mixture was trans-
formed into E. coli TOP10F’ cells. The average mutation rate was
estimated by sequencing randomly selected variants (LGC Genom-
ics). In addition, to exclude the metal-binding amino acids from
mutagenesis, random mutagenesis libraries of three regions of the
gene were constructed (Figure S1) through a standard error-prone
PCR with DreamTaq DNA polymerase or through amplification with
Mutazyme II (Agilent Technologies). The resulting PCR products
were then used as megaprimers for the site-directed mutagenesis
of pMS470-GtHNL. Site-directed and site-saturation mutagenesis of
pMS470-GtHNL were performed with a single-primer reactions in
parallel mutagenesis protocol.[25] Degenerated oligonucleotides
(Table S6) used for site-saturation mutagenesis contained NNK trip-
lets at the targeted positions. Thus, a mixture of 32 mutants con-
taining all 20 proteinogenic amino acids at respective positions
was created. DpnI-digested PCR products were used for the trans-
formation of electrocompetent E. coli TOP10F’ cells.
For purification of GtHNL-WT and variants thereof, and of AcHNL-
WT and variants thereof, the enzymes were expressed as described
above and the cell pellets were resuspended in buffer A (50 mm
BIS-TRIS buffer, 30 mm NaCl) and lysed as described above. The pH
of the buffer was set according to the theoretical isoelectric point
of the wild-type enzyme and variants (Table S7), which was calcu-
lated by using the program ProtParam.[26] Cleared lysates were pu-
rified by using anion-exchange chromatography with Q Sepharose
Fast Flow columns (HiTrap Q FF, 3ꢂ5 mL; GE Healthcare, Uppsala,
Sweden) applying a step of 10% (or 20% for GtHNL-W120R,
GtHNL-K93E, and GtHNL-I3T variants) buffer B (50 mm BIS-TRIS
buffer, 1m NaCl, the same pH as the respective buffer A). The frac-
tions were analysed by using sodium dodecyl sulfate polyacryl-
amide gel electrophoresis, pooled, and then concentrated. GtHNL-
WT and GtHNL-A40H/V42T/Q110H were further purified by using
size-exclusion chromatography with a HiLoad 16/600 Superdex
75 pg column (125 mL; GE Healthcare). Proteins were eluted with
sodium phosphate buffer (50 mm, pH 7.0) containing NaCl
(100 mm). GtHNL-A40H/V42T/Q110H used for crystallography was
eluted in TRIS-HCl (20 mm; pH 7.0) containing NaCl (200 mm). The
fractions were analysed by using sodium dodecyl sulfate polyacryl-
amide gel electrophoresis, and those containing the desired pro-
Improved GtHNL variants were subcloned into plasmid pET26b(+)
by using NdeI and HindIII restriction sites. The beneficial mutations
of GtHNL were combined and transferred to AcHNL from Acidobac-
terium capsulatum ATCC 51196[9] as single and triple mutations by
using a single-primer reactions in parallel mutagenesis protocol (as
described above) with pET26b(+)-GtHNL-Q110H and pET26b(+)-
AcHNL and then pET26b(+)-AcHNL-Q110H as the template. The re-
sulting constructs were confirmed by sequencing (LGC Genomics)
and transformed into electrocompetent E. coli BL21-Gold(DE3)
cells.
tein were pooled and concentrated to 10–20 mgmLÀ1
.
ChemCatChem 2015, 7, 325 – 332
www.chemcatchem.org
330
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Alternatively, for the GC analysis of the conversion of benzaldehyde
(1a), 2-chlorobenzaldehyde (2a), 3-phenylpropanal (3a), 3-phenyl-
prop-2-enal (4a), 2-furaldehyde (5a), and hydroxypivaldehyde (6a),
samples of the organic phase (50 mL) were withdrawn after 2 and
24 h and used for derivatisation with dichloromethane (850 mL),
acetic acid anhydride (100 mL), and pyridine (50 mL) at RT for 1 h.
The acetylated cyanohydrins were analysed with a 6890N gas chro-
matograph (Agilent Technologies) equipped with a PAL autosam-
pler (CTC Analytics AG), a Varian CP7503, a CP-Chirasil-DEX CB
column (25 mꢂ320 mmꢂ0.25 mm), and a flame ionization detector.
GC parameters were as follows: carrier gas: constant pressure
mode at 100 kPa H2; injector: 2508C. 1a: Temperature programme:
608C, 108CminÀ1 to 1408C, 2 min at 1408C; retention times: alde-
hyde 4.5 min, (R)-cyanohydrin acetate 9.6 min, (S)-cyanohydrin ace-
tate 10.3 min; 2a: temperature programme: 1008C, 108CminÀ1 to
1258C, 18CminÀ1 to 1308C, 108CminÀ1 to 1708C, 1 min at 1708C;
retention times: aldehyde 2.0 min, (R)-cyanohydrin acetate 6.8 min,
(S)-cyanohydrin acetate 7.1 min; 3a: temperature programme:
1008C, 108CminÀ1 to 1708C, 2 min at 1708C; retention times: alde-
hyde 2.6 min, (R)-cyanohydrin acetate 6.2 min, (S)-cyanohydrin ace-
tate 6.5 min; 4a: temperature programme: 1408C, 18CminÀ1 to
1508C, 108CminÀ1 to 1708C; retention times: aldehyde 1.9 min,
(R)-cyanohydrin acetate 7.1 min, (S)-cyanohydrin acetate 9.3 min;
5a: temperature programme: 1058C, isotherm; retention times: al-
dehyde 0.8 min, (S)-cyanohydrin acetate 3.3 min, (R)-cyanohydrin
acetate 4.6 min; 6a: temperature programme: 1108C, 0 min,
108CminÀ1 to 1308C, 208CminÀ1 to 1708C, 0.5 min at 1708C; re-
tention times: aldehyde 0.9 min, (R)-cyanohydrin acetate 2.5 min,
(S)-cyanohydrin acetate 2.6 min.
HNL activity assay
The activity of the enzymes in the cyanogenesis of (R)-mandeloni-
trile or (R)- and (S)-chloromandelonitrile was determined spectro-
photometrically by following the increase in benzaldehyde and 2-
chlorobenzaldehyde absorption at 280 and 300 nm, respectively, in
a plate reader at 258C. The reaction buffer (130 mL; either 100 mm
citrate/phosphate buffer, pH 4.0, 4.5, 5.0, and 5.5; 100 mm sodium
oxalate buffer, pH 4.0, 4.5, and 5.0; or 100 mm MES oxalate buffer,
pH 5.5) was mixed with the enzyme sample (20 mL; 1–200 mg of
the total purified protein per well), and the reaction was started by
adding a cyanohydrin [50 mL; (R)-mandelonitrile dissolved in 3 mm
citrate/phosphate buffer, pH 3.5, or 3 mm oxalic acid, pH 2.6, or 2-
chloromandelonitrile dissolved in 40% DMSO in 3 mm citrate/phos-
phate buffer, pH 3.2, final concentration 18 mm]. Blank reactions
contained a buffer (20 mL) instead of the enzyme sample. The
blank reaction corresponded to the chemical cyanogenesis of the
cyanohydrin under the respective reaction conditions. The result-
ing slopes (DA/minblank
enzyme-catalysed reactions (DA/minenzyme). The activity was calcu-
) were subtracted from those of the
lated by using the Beer–Lambert law and molar extinction
coefficients
e280 nm =1.376 LmmolÀ1 cmÀ1
and
e300 nm =
1.521 LmmolÀ1 cmÀ1 for benzaldehyde and 2-chlorobenzaldehyde,
respectively. All measurements were performed as triplicates. The
apparent kinetic data were obtained from the initial rate measure-
ments at pH 5.5 (100 mm sodium oxalate buffer) and 258C by
using (R)-mandelonitrile concentrations from 1.5 to 48 mm and
with purified wild-type GtHNL (150 mg per well) or purified GtHNL-
A40H/V42T/Q110H (0.6 mg per well), as described above.
One mandelonitrile cyanogenesis unit is defined as the amount of
enzyme that catalyses the formation of 1 mmol of benzaldehyde
from mandelonitrile dissolved in aqueous buffer in 1 min at 258C.
Crystallization and structure determination of GtHNL-A40H/
V42T/Q110H
Initial crystallization trials of GtHNL-A40H/V42T/Q110H were per-
formed with an Oryx 8 robot (Douglas Instruments) by using the
sitting drop method. Protein concentrations of 9 and 18 mgmLÀ1
(in 20 mm TRIS-HCl, pH 7.0, 200 mm NaCl) were used. Drops of the
enzyme solution (0.8 mL) were mixed with the reservoir solution
(0.8 mL) by using JCSG-plus and Morpheus crystallization screens
(Molecular Dimensions). All plates were incubated at 168C. Initial
crystals were observed after 3 days under several conditions. Crys-
tallization conditions were optimised, and a complete data set to
2.1 ꢁ resolution was collected from the beamline ID23-2 at the Eu-
ropean Synchrotron Radiation Facility (Grenoble)[28] from a crystal
obtained under condition 1.15 of the JCSG-plus screen (0.2m
sodium thiocyanate, 20% w/v polyethylene glycol 3350).
Cyanohydrin synthesis
All reactions involving cyanides were performed in a well-ventilat-
ed hood equipped with an HCN detector (Drager Pac III). The reac-
tions were performed in two-phase systems, with the organic
phase (1 mL) containing the substrate (0.5m aldehyde), the internal
standard (2% v/v triisopropylbenzene), and HCN (2m) in MTBE).
The aqueous phase (500 mL) comprised the enzyme (cleared lysate
with a total protein concentration of 45 mgmLÀ1 containing
ꢀ50% cupin, or purified enzyme with a concentration of 1–
10 mgmLÀ1) in sodium acetate buffer (100 mm, pH 4.0) or sodium
citrate buffer (100 mm, pH 2.4). The pH of the acidified samples
was controlled before use. HCN was produced in situ and extracted
in MTBE as described earlier by Okrob et al.[27] Benzaldehyde, 2-
chlorobenzaldehyde, and 2-furaldehyde were freshly distilled; hy-
droxypivaldehyde was monomerised by gentle heating before use.
The reactions were performed in 2 mL reaction tubes on a thermo-
mixer shaking at 1000 rpm and 58C. Samples were withdrawn
from the organic stock solution before adding the aqueous phase
as a reference, which represented the initial amount of the sub-
strate and internal standard. For HPLC analysis of the conversion of
benzaldehyde, samples of the organic phase (10 mL) were with-
drawn at appropriate time points and diluted 1:100 with HPLC
eluent (n-heptane/iPr/trifluoroacetic acid=96:4:0.1). Chiral HPLC
analysis was performed with a Daicel CHIRALCEL OD-H column)
with a flow rate of 0.9 mLminÀ1, temperature 258C, and UV detec-
tion at 210 and 254 nm. 1a: retention times: triisopropylbenzene
3.8 min, benzaldehyde 7.5 min, (S)-mandelonitrile 29.1 min, (R)-
mandelonitrile 30.6 min.
Data were processed by using the XDS program suite[29] and soft-
ware from the CCP4 suite.[30] The structure was solved by molecular
replacement using the program Phaser[31] and the structure of
GtHNL-WT [Protein Data Bank (PDB) entry 4bif] as the search tem-
plate.[7] Structure rebuilding and refinement were performed with
Coot[32] and Refmac5.[33] The final structure was validated by using
the program MolProbity.[34] Atomic coordinates and structure fac-
tors were deposited in the PDB under the accession number 4UXA.
Thermal shift assay
The thermal shift assay was performed as described in the litera-
ture[35] with an Applied Biosystems 7500 Fast Real-Time PCR
system (Life Technologies) and 96-Well Optical Reaction plates.
Each well contained purified protein (5 mg; GtHNL-WT and variants)
and 1X SYPRO Orange dye (Sigma–Aldrich) in various buffers
ChemCatChem 2015, 7, 325 – 332
www.chemcatchem.org
331
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[10] A. Łyskowski, K. Steiner, I. Hajnal, G. Steinkellner, H. Schwab, K. Gruber,
Acta Crystallogr. Sect. F 2012, 68, 451–454.
[11] Y. Asano, M. Dadashipour, M. Yamazaki, N. Doi, H. Komeda, Prot. Eng.
Des. Sel. 2011, 24, 607–616.
[12] M. Avi, R. M. Wiedner, H. Griengl, H. Schwab, Chem. Eur. J. 2008, 14,
11415–11422.
[13] J. von Langermann, D. M. Nedrud, R. J. Kazlauskas, ChemBioChem 2014,
15, 1931–1938.
(100 mm citrate/phosphate buffer, pH 7.5–3.0, 100 mm sodium oxa-
late buffer, pH 7.5–3.0, 100 mm sodium acetate buffer, pH 5.5–4.0,
added as 0.5m stock, the protein storage buffer, 20 mm TRIS-HCl,
200 mm NaCl, pH 7.5 was used as a reference) and a total volume
of 25 mL. The temperature was increased from 20 to 958C (step
size: 18C). The increase of the fluorescence of the SYPRO Orange
probe during the unfolding of the protein was measured with exci-
tation at 490 nm and emission at 575 nm. Measurements were per-
formed in duplicate or triplicate.
[14] D. Okrob, J. Metzner, W. Wiechert, K. Gruber, M. Pohl, ChemBioChem
2012, 13, 797–802.
[15] A. Glieder, R. Weis, W. Skranc, P. Pçchlauer, I. Dreveny, S. Majer, M. Wub-
bolts, H. Schwab, K. Gruber, Angew. Chem. Int. Ed. 2003, 42, 4815–4818;
Angew. Chem. 2003, 115, 4963–4966.
Acknowledgements
[16] Z. Liu, B. Pscheidt, M. Avi, R. Gaisberger, F. S. Hartner, C. Schuster, W.
Skranc, K. Gruber, A. Glieder, ChemBioChem 2008, 9, 58–61.
[17] B. Pscheidt, Z. B. Liu, R. Gaisberger, M. Avi, W. Skranc, K. Gruber, H.
Griengl, A. Glieder, Adv. Synth. Catal. 2008, 350, 1943–1948.
[18] R. Weis, R. Gaisberger, W. Skranc, K. Gruber, A. Glieder, Angew. Chem. Int.
Ed. 2005, 44, 4700–4704; Angew. Chem. 2005, 117, 4778–4782.
[19] B. Krammer, K. Rumbold, M. Tschemmernegg, P. Pçchlauer, H. Schwab,
J. Biotechnol. 2007, 129, 151–161.
[20] J. Andexer, J. von Langermann, A. Mell, M. Bocola, U. Kragl, T. Eggert, M.
Pohl, Angew. Chem. Int. Ed. 2007, 46, 8679–8681; Angew. Chem. 2007,
119, 8833–8835.
[21] W. Skranc, A. Glieder, K. Gruber, R. Weis, O. Maurer, R. Gaisberger,
WO2006076965.
[22] W. Skranc, A. Glieder, K. Gruber, R. Weis, R. Luiten, WO2004083424 B2.
[23] W. Becker, E. Pfeil, Biochem. Z. 1965, 346, 301–321.
[24] D. Balzer, G. Ziegelin, W. Pansegrau, V. Kruft, E. Lanka, Nucl. Acids Res.
1992, 20, 1851–1858.
[25] O. Edelheit, A. Hanukoglu, I. Hanukoglu, BMC Biotechnol. 2009, 9, 61.
[26] E. Gasteiger, C. Hoogland, A. Gattiker, S. Duvaud, M. R. Wilkins, R. D.
Appel, A. Bairoch in The Proteomics Protocols Handbook (Ed.: J. M.
Walker), Humana Press, 2005, pp. 571–607.
This work has been supported by the Federal Ministry of Science,
Research and Economy (BMWFW), the Federal Ministry of Traffic,
Innovation and Technology (bmvit), the Styrian Business Promo-
tion Agency SFG, the Standortagentur Tirol and ZIT—Technology
Agency of the City of Vienna through the competence centers for
excellent technologies (COMET) funding program managed by
the Austrian Research Promotion Agency FFG. This project has re-
ceived funding from the European Union’s Seventh Framework
Programme for research, technological development, and dem-
onstration under grant agreement no. 289646 (KyroBio). We ac-
knowledge the ESRF for the provision of synchrotron-radiation fa-
cilities and thank the ESRF scientific and technical personnel for
assistance in using the beamline ID23-2. We thank Ing. Karin
Reicher, MSc, for helping with cyanohydrin synthesis and protein
purification and Sarah Trunk, MSc, for performing the thermal
shift assay.
[27] D. Okrob, M. Paravidino, R. V. A. Orru, W. Wiechert, U. Hanefeld, M. Pohl,
Adv. Synth. Catal. 2011, 353, 2399–2408.
[28] D. Flot, T. Mairs, T. Giraud, M. Guijarro, M. Lesourd, V. Rey, D. van Brussel,
C. Morawe, C. Borel, O. Hignette, J. Chavanne, D. Nurizzo, S. McSwee-
ney, E. Mitchell, J. Synchr. Rad. 2010, 17, 107–118.
Keywords: biocatalysis · cupins · cyanohydrin syntheses ·
protein engineering · lyases
[29] W. Kabsch, Acta Crystallogr. Sect. D 2010, 66, 125–132.
[30] M. D. Winn, C. C. Ballard, K. D. Cowtan, E. J. Dodson, P. Emsley, P. R.
Evans, R. M. Keegan, E. B. Krissinel, A. G. W. Leslie, A. Mccoy, S. J. McNi-
cholas, G. N. Murshudov, N. S. Pannu, E. A. Potterton, H. R. Powell, R. J.
Read, A. Vagin, K. S. Wilson, Acta Crystallogr. Sect. D 2011, 67, 235–242.
[31] A. J. Mccoy, R. W. Grosse-Kunstleve, P. D. Adams, M. D. Winn, L. C. Storo-
ni, R. J. Read, J. Appl. Crystallogr. 2007, 40, 658–674.
[32] P. Emsley, K. Cowtan, Acta Crystallogr. Sect. D 2004, 60, 2126–2132.
[33] G. N. Murshudov, A. A. Vagin, E. J. Dodson, Acta Crystallogr. Sect. D
1997, 53, 240–255.
[34] V. B. Chen, W. B. Arendall, J. J. Headd, D. A. Keedy, R. M. Immormino,
G. J. Kapral, L. W. Murray, J. S. Richardson, D. C. Richardson, Acta Crystal-
logr. Sect. D 2010, 66, 12–21.
[35] U. B. Ericsson, B. M. Hallberg, G. T. DeTitta, N. Dekker, P. Nordlund, Anal.
Biochem. 2006, 357, 289–298.
[1] M. Gruber-Khadjawi, M. H. Fechter, H. Griengl in Enzyme Catalysis in Or-
ganic Synthesis (Eds.: K. Drauz, H. Grçger, O. May), Wiley-VCH, Wein-
heim, 2012, pp. 947–990.
[2] E. Lanfranchi, K. Steiner, A. Glieder, I. Hajnal, R. A. Sheldon, S. van Pelt,
M. Winkler, Recent Pat. Biotechnol. 2013, 7, 197–206.
[3] T. Purkarthofer, W. Skranc, C. Schuster, H. Griengl, Appl. Microbiol. Bio-
technol. 2007, 76, 309–320.
[4] M. Winkler, A. Glieder, K. Steiner, in Comprehensive Chirality, Synthetic
Methods VI—Enzymatic and Semi-Enzymatic, Vol. 7 (Eds.: E. M. Carreira,
H. Yamamoto), Elsevier, Amsterdam, 2012, pp. 350–371.
[5] J. N. Andexer, J. V. Langermann, U. Kragl, M. Pohl, Trends Biotechnol.
2009, 27, 599–607.
[6] M. Dadashipour, Y. Asano, ACS Catal. 2011, 1, 1121–1149.
[7] I. Hajnal, A. Łyskowski, U. Hanefeld, K. Gruber, H. Schwab, K. Steiner,
FEBS J. 2013, 280, 5815–5828.
[8] Z. Hussain, R. Wiedner, K. Steiner, T. Hajek, M. Avi, B. Hecher, A. Ses-
sitsch, H. Schwab, Appl. Environ. Biotechnol. 2012, 78, 2053–2055.
[9] R. Wiedner, M. Gruber-Khadjawi, H. Schwab, K. Steiner, Comp. Struct.
Biotechnol. J. 2014, 10, 58–62.
Received: September 15, 2014
Published online on December 3, 2014
ChemCatChem 2015, 7, 325 – 332
www.chemcatchem.org
332
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim