Natural diversity to guide focused directed evolution

Article abstract of DOI:10.1002/cbic.201000284

Simultaneous multiple site-saturation mutagenesis was performed at four active-site positions of an esterase from Pseudomonas fluorescens to improve its ability to convert 3-phenylbutyric acid esters (3-PBA) in an enantioselective manner. Based on an appropriate codon choice derived from a structural alignment of 1751 sequences of α/β-hydrolase fold enzymes, only those amino acids were considered for library creation that appeared frequently in structurally equivalent positions. Thus, the number of mutants to be screened could be substantially reduced while the number of functionally intact variants was increased. Whereas the wild-type esterase showed only marginal activity and poor enantioselectivity (E_true=3.2) towards 3-PBA-ethyl ester, a significant number of hits with improved rates (up to 240-fold) and enantioselectivities (up to E_true=80) were identified in these "smart" libraries.

Full text of DOI:10.1002/cbic.201000284

DOI: 10.1002/cbic.201000284
Natural Diversity to Guide Focused Directed Evolution
Helge Jochens and Uwe T. Bornscheuer*[a]
Simultaneous multiple site-saturation mutagenesis was per-
formed at four active-site positions of an esterase from Pseudo-
monas fluorescens to improve its ability to convert 3-phenylbu-
tyric acid esters (3-PBA) in an enantioselective manner. Based
on an appropriate codon choice derived from a structural
alignment of 1751 sequences of a/b-hydrolase fold enzymes,
only those amino acids were considered for library creation
that appeared frequently in structurally equivalent positions.
Thus, the number of mutants to be screened could be sub-
stantially reduced while the number of functionally intact var-
iants was increased. Whereas the wild-type esterase showed
only marginal activity and poor enantioselectivity (Etrue =3.2)
towards 3-PBA-ethyl ester, a significant number of hits with
improved rates (up to 240-fold) and enantioselectivities (up to
E
true =80) were identified in these “smart” libraries.
Introduction
During the past two decades a broad range of novel protein
engineering methodologies have emerged.[1] In particular, di-
rected evolution has developed to a state-of-the-art tool to
modify and tailor proteins towards a desired property. Ran-
domization of genes followed by screening or selection
toward mutants with beneficial properties enabled the im-
provement of nearly every protein feature: thermostability,[2]
pH stability,[3] solvent stability,[4] catalytic efficiency,[5] enantiose-
lectivity,[6] substrate specificity,[7] and even the introduction of
new activities into protein scaffolds;[8] this explains the impor-
tance of this tool for biocatalysis.
assumption that mutations often occurring within the set of
sequences suggest a properly folded and active protein. This
concept goes beyond back-to-consensus mutation concepts,
for which it could be already shown that this can have a posi-
tive effect on the properties of proteins[11] as the structural
alignment within a large superfamily allows coverage of a
broad set of similar-to-consensus mutations.
Results and Discussion
Creation of a “smart” library
The intelligent utilization of rapidly increasing protein infor-
mation from X-ray structures, (genome) sequences, and bio-
chemical data combined with the ongoing progress in the
computer technology, tends to replace the classical directed
evolution—random mutagenesis followed by screening or se-
lection of huge libraries—by more sophisticated and targeted
methods. Hence a more “rational random” approach is consid-
ered to cover protein diversity in “smart” mutant libraries with
a high probability of finding desired proteins. Examples are the
iterative saturation mutagenesis (ISM) and CASTing methods,[9]
which have been successfully used to efficiently evolve en-
zymes, although only libraries of modest size (5000 to 20000
variants) needed to be screened. Another concept is the ex-
ploitation of neutral, genetic drift,[10] which is based on a prese-
lection of structurally intact proteins within a randomized li-
brary by using the natural substrate. This assumes that the
active variants in the resulting pool of mutants have sufficient
plasticity to be challenged towards a new property such as an
alternative substrate.
To verify our concept, the amino acid distributions at four posi-
tions (W28, V121, F198, and V225) near the active site of PFE
were determined because they have been reported to influ-
ence the enantioselectivity of this enzyme.[12] The distribution
was determined by using the software 3DM (http://www.bio-
prodict.nl) that builds protein superfamily-specific databases.[13]
Because PFE is an esterase, we choose the family of a/b-hydro-
lase fold enzymes[14] as a representative example for a large
protein family and created a 3DM-database that included
1,751 structurally related enzymes (mainly carboxyl ester hy-
drolases such as esterases and lipases, Figure 1, top; http://fun-
gen.wur.nl/ABHDB). Due to the assembly of the database it is
ensured that for every amino acid position included in the
alignment an amino acid at an equivalent structural position
exists in any sequence covered in the 3DM-database (within in
a certain cutoff). Thus, the program allows the highly reliable
determination of amino acid distributions within a huge
number of proteins (Figure 1, bottom right). Frequently occur-
ring residues were considered to be allowed (A) because they
We developed an alternative approach to create a “smart” li-
brary and demonstrated its applicability to improve the enan-
tioselectivity of an esterase from Pseudomonas fluorescens
(PFE) to convert esters of the chiral compound 3-phenyl buty-
ric acid. For this concept, we used a huge set of structurally
similar enzymes from the a/b-hydrolase fold superfamily and
determined the frequency of certain amino acids with the
[a] Dr. H. Jochens, Prof. Dr. U. T. Bornscheuer
Department of Biotechnology and Enzyme Catalysis
Institute of Biochemistry, Greifswald University
Felix-Hausdorff-Strasse 4, 17487 Greifswald (Germany)
Fax: (+49)3834-86-80066
E-mail: uwe.bornscheuer@uni-greifswald.de
ChemBioChem 2010, 11, 1861 – 1866
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1861

U. T. Bornscheuer et al.
Figure 1. 3DM compares structurally related sequences as shown here for a/b-hydrolase fold enzymes (Top: 3DM alignment showing a representative portion
of all 1751 structurally related proteins in the family) and allows the determination of amino acid distributions within all 1751 sequences for each position to
be extracted (bottom right: amino acid distribution at a representative single position). To predict “smart” libraries, amino acids that appear very rarely were
defined as not allowed (NA) as they presumably lead to unfolded enzymes and frequently occurring residues were termed allowed (A) because they should
lead to correctly folded and active proteins (bottom left).
presumably lead to correctly folded and active protein. Rarely
(or never) occurring mutations were defined as not allowed
(NA) because their rareness suggests that they are deleterious
to the protein function and hence were discriminated against
during the evolutionary process.
tigate if indeed the preselection of allowed amino acids led to
correctly folded and active variants within the library.
Table 1. Classification of amino acids at four active site positions in PFE
into allowed and not allowed according to their frequency in an align-
ment of 1751 esterases. In addition, the codons used to create the libra-
ries are shown together with the corresponding encoded amino acids.
In order to create a smart library at these four positions with
optimum coverage of the allowed (library A) amino acids,
appropriate codons were deduced (Table 1), and the resulting
library was created by site-saturation mutagenesis.
Position Allowed amino
acids
Unallowed amino Codon Encoded
acids amino acids
28
121
198
225
A, V, I, L, F, W, Y, S, C, M, P, H, Q, D, E,
T, G, N, K
A, V, I, L, S, T, G, P, F, W, Y, C, M, H, Q,
D, E, K, R
A, V, I, L, F, W, Y, M, C, S, H, Q, D, E, K,
T, G, P, N
A, V, I, L, C, M, T, P, F, W, Y, S, G, H, N,
Q, E, K, R
KBS A, V, L, F, W, S,
G, C
RBC A, V, I, S, T, G
R
The preselection of amino acid substitutions in the “smart
library” determines the number of correctly folded and
active esterases
N
KKK V, L, F, W, G, C
DYA A, V, I, L, S, T
R
Library A was first screened for activity towards the widely
used esterase substrate p-nitrophenyl acetate (pNPA) to inves-
D
1862
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1861 – 1866

Enzyme Catalysis
For a proper validation of this approach, two additional li-
braries were created and screened that contain either all pro-
teinogenic amino acids (NNK, 32 codons for 20 amino acids) or
only those that appear rarely on the targeted positions in the
alignment (NA). Hence, the NA library served as a negative
control, assuming that the majority of its members cannot be
expressed as correctly folded and/or active proteins. The NNK
library represents full coverage of all possible combinations of
amino acids at the four positions and hence must include
members of libraries A and NA. However, the number of indi-
viduals within this library is 204 =160000 and about three mil-
lion colonies need to be screened to cover 95% of all possible
combinations according to the Poisson statistic.[15]
ity (Eappꢀ5) than the wild-type (Eapp =2). Furthermore, the abso-
lute number of more-selective variants was significantly higher
than in library NNK. We were pleased to observe that the
increase in enantioselectivity did not occur at the expense of
specific activity. Rather, variants from library A showed much
higher activity and enantioselectivity compared to wild-type
PFE. In contrast, the proteins in library NA showed no signifi-
cant improvement at all, and only a few enzymes in library
NNK gave moderately increased activity and selectivity
(Figure 2).
To investigate the extent to which amino acid exchanges in-
fluenced substrate specificity, activity against pNPA was plotted
against activity for 3-PB-pNP (Figure 3). This clearly shows, that
many variants with higher activity towards the sterically more
demanding 3-phenyl butyric acid ester were found in library A
compared to library NNK. In addition, library A contained es-
terase variants that had reduced activity towards pNPA and at
the same time a better ability to hydrolyze 3-PB-pNP, although
the more activated acetate is still the better substrate.
Analysis of the number of desired hits relative to the total li-
brary size revealed that the diversity chosen for library A is by
far less deleterious to the proteins’ function compared to the
completely randomized NNK library. More than 50% of the var-
iants in library A showed at least 10% of wild-type activity to-
wards pNPA whereas only 2% of the members in library NNK
fulfilled this criteria. As expected, rarely occurring amino acids
(library NA) lead to loss of protein function and only one
enzyme (corresponding to 0.2%) showed more than 10% of
wild-type activity. Furthermore it is worth mentioning that
almost all variants in library A showed significantly higher
product formation than the uncatalyzed reaction; this indicates
that the vast majority of the proteins are structurally intact.
Two variants M1 (V121I, F198G, V225A) and M2 (V121I,
F198C) that were derived from library A and were chosen due
to their high enantioselectivities and activities towards 3-PB-
pNP in the screening were purified and characterized in more
detail and kinetic constants were determined by using the
spectrophotometric assay. However, to determine the “true” E
value—which represents the ratio of reaction velocities to-
wards both substrate enantiomers under competitive condi-
tions—racemic 3-phenyl butyric acid ethyl ester was used. For
M2, the enantioselectivity in the hydrolysis of this ester was
substantially higher with Etrue =80, compared to the almost
nonselective wild type (Etrue =3.2). Furthermore, the specific
activity was enhanced 240-fold. Surprisingly, the mutant M1
Frequently occurring amino acids positively influence
substrate specificity and enantioselectivity
Subsequently, representative numbers of variants (A: 520,
NNK: 515, NA: 542) in the three libraries were screened for in-
creased activity and enantiose-
lectivity by using the sterically
demanding 3-phenyl butyric
acid p-nitrophenyl ester (3-PB-
pNP) as a chiral model substrate.
For this, variants were produced
in microtiter plates and then dis-
tributed into two plates for
measuring the activity towards
either enantiomer of (R)- or (S)-3-
PB-pNP by following the release
of p-nitrophenol at 410 nm. The
apparent enantioselectivity (Eapp
)
of each variant was then calcu-
lated from the ratio of the initial
rates determined for each enan-
tiomer (Figure 2).
Although it can not be ex-
pected that frequently occurring
amino acids must be linked to
altered substrate range or enan-
tioselectivity, we found that 36
members (out of 520 variants
screened) derived from library A
Figure 2. Apparent E value vs. activity towards the faster-reacting enantiomer of 3-PB-pNP; black: collection of
exhibited higher enantioselectiv- wild-type enzymes; red: library NA; green: library NNK; blue: library A.
ChemBioChem 2010, 11, 1861 – 1866
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1863

U. T. Bornscheuer et al.
the properties of cysteine might
have caused the beneficial be-
havior observed for M2. Hence,
our approach still represents a
useful compromise between li-
brary size to be screened and
the chance to find a variant with
improved properties. The size of
the smart library is only ~10000
compared to three million indi-
viduals to be screened in the
NNK-library (for coverage of
95% of all possible combina-
tions). Although one might miss
certain beneficial mutations, the
>300-fold reduction in screen-
ing effort clearly makes the
design of small but smart libra-
ries advantageous as consuma-
bles, and time and labor can be
substantially reduced.
Figure 3. Activity towards pNPA vs. activity toward the faster reacting enantiomer of 3-PB-pNP; Black: collection
of wild-type enzymes; red: library NA; green: library NNK; blue: library A.
Conclusions
Focused directed evolution in an
enzyme’s active site represents a
showed only very residual activity towards the ethyl ester and
no detectable enantiopreference. This result is remarkable be-
cause it converted the corresponding p-nitrophenyl ester with
53-fold increased catalytic efficiency. At the same time the ac-
tivity towards pNPA was 80-fold decreased; this corresponds to
an overall 4250-fold change in substrate specificity for wild-
type enzyme versus M1 (Table 2).
very powerful approach to tailor the substrate specificity and
enantioselectivity of different proteins.[9] Most often site-satura-
tion mutagenesis is performed in an iterative manner. This
means that only one site (which might contain more than one
amino acid position) is saturated at any one time. The best var-
iant of each round of evolution serves as a template for the
next round.
Interestingly, the best mutant M2 carries a cysteine at posi-
tion 198, a residue that was defined as not allowed in the 3DM
alignment, but was encoded due to the set of codons chosen
for library creation. However, sequence analysis of additional
mutants that showed improved enantioselectivity and activity
during ester hydrolysis indicated that the small size rather than
However, in our approach, we aimed to saturate all four tar-
geted positions simultaneously to allow possible cooperative
effects that might get lost in an iterative approach. That self-
imposed requirement and the limited availability of suitable se-
lection assays or ultra-high-throughput screening methods to
engineer enantioselectivity forced us to develop an approach
that is able to decrease the
mutant library to a reasonable
size. Reetz et al. suggested using
the codon NDT instead of
NNK,[15] which, in our case,
would have decreased the theo-
retical number of mutants to be
screened from three million to
62000 but still encodes for 12
amino acids including all four
properties (acidic, basic, hydro-
phobic, aromatic). Our approach
suggests covering only those
amino acids that appear fre-
quently in natural enzymes at
identical positions because these
have proven during natural evo-
lution that they are not destruc-
Table 2. Kinetic constants of the two variants M1 (V121I, F198G, V225A) and M2 (V121I, F198C) towards 3-PB-
pNP or pNPA, and enantioselectivity towards 3-PB-ethyl ester (3-PB-Et) compared to wild-type PFE.
[a]
Enzyme
Substrate
KM [m]
kcat [sÀ1
]
kcat/KM [sÀ1 mÀ1
]
Eapp
Etrue
wild-type
pNPA
2.3ꢁ10À4
8.7ꢁ10À4
9.7ꢁ10À4
n.d.[c]
4.2ꢁ10À4
9.9ꢁ10À4
5.3ꢁ10À4
n.d.
2.5ꢁ10À4
2.0ꢁ10À4
2.0ꢁ10À3
n.d.
74
3.2ꢁ105
3.3
2.9
n.d.
4.0ꢁ103
175.4
10.2
n.a.[b]
1.1
n.a.
n.a.
n.a.
3.2[d]
n.a.
n.a.
n.a.
n.a.
n.a.
n.a.
n.a.
80[d]
(R)-3-PB-pNP
(S)-3-PB-pNP
(R,S)-3-PB-Et
pNPA
(R)-3-PB-pNP
(S)-3-PB-pNP
(R,S)-3-PB-Et
pNPA
2.9ꢁ10À3
2.8ꢁ10À3
n.d.
n.a.
n.a.
17.2
M1
M2
1.7
0.1736
5.4ꢁ10À3
n.d.
n.d.
n.a.
n.a.
10.0
0.9
3.6ꢁ103
41
(R)-3-PB-pNP
(S)-3-PB-pNP
(R,S)-3-PB-Et
8.2ꢁ10À3
8.2ꢁ10À3
n.d.
4.1
n.d.
n.a.
[a] Calculated from [(kcat/KM)(R)-3-PB-pNP/(kcat/KM)(S)-3-PB-pNP]. [b] Not applicable. [c] Not determined. [d] Specific activity
[mmolminÀ1 mgÀ1]: wild-type: 8ꢁ10À3, M2: 1.9.
1864
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1861 – 1866

Enzyme Catalysis
12.5 nmolmLÀ1), reverse primer (1 mL, 12.5 nmolmLÀ1), Pfu+ DNA
polymerase (0.3 mL). The mixture was separated in two PCR tubes
and the reaction was performed by using the following conditions:
1) 1 cycle 958C, 180 s; 2) 16 cycles 958C, 45 s; 588C, 45 s; 728C,
360 s; 1 cycle 728C, 600 s. Afterwards DpnI (0.5 mL) was added and
the samples were incubated for 1 h at 378C followed by enzyme
denaturation for 10 min at 808C. Chemo-competent E. coli cells
(DH5a) were transformed with the amplified plasmid (5 mL) and
plated out on LBAMP-plates. First, position 28 was randomized by
using the protocol described above. For the next rounds of satura-
tions, plasmid isolation was performed for a certain number of
clones and used as the starting point for subsequent QuikChange
reactions. The number of clones picked depended on the calculat-
ed number for mutants within each library for each round of satu-
ration but did not exceed a total of 600 clones.
tive to the integrity of the protein and thus are likely to deliver
a mutant library of high quality with respect to activity. Indeed,
almost all mutants in library A showed activity that was signifi-
cantly higher than the background and more than 50%
showed more than 10% of the wild-type activity towards
pNPA.
In our experiment the codons were chosen in such a way
that as many as possible amino acids that were classified as al-
lowed were encoded whereas all disallowed residues could be
left out. Due to the nature of the degenerate codons this
might not be clearly possible in every case, and thus at posi-
tions 21 and 198 a not allowed cystein was incorporated in li-
brary A. However, computer tools like CASTER[16] can be used
to assist the elucidation of suitable codons to achieve the best
possible coverage of only allowed residues.
Primers: W28X-A-FW: 5’-GTT CAG CCA CGG TKB SCT ACT GGA TGC
CG-3’, W28X-A-RV: 5’-CGG CAT CCA GTA GSV MAC CGT GGC TGA
AC-3’; V121X-A-FV: 5’-GCT GCT GGG CGC CRB CAC CCC GCT GTT
CG-3’, V121X-A-RV: 5’-CGA ACA GCG GGG TGV YGG CGC CCA GCA
GC-3’; F198X-A-FW: 5’-GAT TGC GTC ACC GCG KKK GCC GAA ACC
GAC TTC-3’, F198X-A-RV: 5’-GAA GTC GGT TTC GGC MNN CGC GGT
GAC GCA ATC-3’; V225X-A-FW: 5’-GGC GAC CAG ATC DYA CCG TTC
GAG ACC-3’, V225X-A-RV: 5’-GGT CTC GAA CGG TRH GAT CTG GTC
GCC-3’; W28X-NA-FW: 5’-GTT CAG CCA CGG TSV MCT ACT GGA
TGC CG-3’, W28X-NA-RV: 5’-CGG CAT CCA GTA GKB SAC CGT GGC
TGA AC-3’; V121X-NA-FW: 5’-GCT GCT GGG CGC CBR KAC CCC GCT
GTT CG-3’, V121X-NA-RV: 5’-CGA ACA GCG GGG TMY VGG CGC
CCA GCA GC-3’; F198X-NA-FW: 5’-GAT TGC GTC ACC GCG VRM
GCC GAA ACC GAC TTC-3’, F198X-NA-RV: 5’-GGC GAC CAG ATC
BRK CCG TTC GAG ACC-3’; V225X-NA-FW: 5’-GAA GTC GGT TTC
GGC KYB CGC GGT GAC GCA ATC-3’, V225X-NA-RV: 5’-GGT CTC
GAA CGG MYV GAT CTG GTC GCC-3’; W28X-NNK-FW: 5’-GTT CAG
CCA CGG TNN KCT ACT GGA TGC CG-3’, W28X-NNK-RV: 5’-CGG CAT
CCA GTA GMN NAC CGT GGC TGA AC-3’; V121X-NNK-FW: 5’-GCT
GCT GGG CGC CNN KAC CCC GCT GTT CG-3’, V121X-NNK-RV: 5’-
CGA ACA GCG GGG TMN NGG CGC CCA GCA GC-3’; F198X-NNK-
FW: 5’-GAT TGC GTC ACC GCG NNK GCC GAA ACC GAC TTC-3’,
F198X-NNK-RV: 5’-GAA GTC GGT TTC GGC MNN CGC GGT GAC
GCA ATC-3’; V225X-NNK-FW: 5’-GGC GAC CAG ATC NNK CCG TTC
GAG ACC-3’, V225X-NNK-RV: 5’-GGT CTC GAA CGG MNN GAT CTG
GTC GCC-3’.
The tremendous reduction in library size coupled with an
enrichment of active variants within the library that possess a
sufficient variability of shape and chemical properties in the
active-site region make our approach a reasonable concept.
Moreover, screening of only a relatively small fraction (5%) of
the total library has already given mutants that show remarka-
bly improved enantioselectivity while simultaneously increas-
ing the specific activity. Thus our approach enables researchers
to perform multiple-site-saturation mutagenesis at the same
time without generating libraries that are unmanageably large
by using conventional screening methods. For ultra-high-
throughput systems and certain in vivo and in vitro selection
systems, our concept could even be more useful because one
can easily target up to ten positions simultaneously and thus
generate enormous diversity without destroying the overall
structure of the targeted proteins.
The success of this approach is certainly dependent on the
quality of the determination of amino acid distribution on the
positions to be mutated. In our case, we used a structure-
guided multiple-sequence alignment (3DM) that collected
mainly esterases that were very well aligned. However, this ap-
proach should not be limited by the availability of such com-
putational tools. A carefully chosen collection of protein se-
quences or even better structures of a set of functional and
structural related enzymes (in our case esterases) could enable
a similarly accurate amino acid distribution even if the number
of sequences included is much smaller than in our experiment.
When collecting the data set, the value should be set on a fre-
quent occurrence of the targeted activity within the alignment
because an under-representation could compromise the deter-
mination of the most promising amino acid distribution. How-
ever, the increasing sequence and structure data as well as on-
going progress in the development of computational methods
will further facilitate the application of this concept.
Cultivation and cell harvest: The transformants were grown on
agar plates at 378C until an appropriate size was achieved and
then picked by a colony picker (Geneworx, Holzkirchen, Germany),
placed in microtiter plates containing LBAMP medium (180 mL), and
incubated overnight at 378C. The cell culture (50 mL) was added to
fresh LBAMP medium (150 mL). After 4 h incubation at 378C at
200 rpm, protein production was induced as described below and
the cells were incubated for an additional 16 h at 308C and
200 rpm. In contrast to the protocol given below, the cells were
disrupted in only a small amount of lysis buffer (100 mL). All other
steps remained the same.
Synthesis of (R)- and (S)-3-PB-pNP: Either (R)- or (S)-3-phenyl bu-
tyric acid (2.6 mL) were dissolved in CH2Cl2 (20 mL). After adding
two drops of DMF and SOCl2 (2 mL), the reaction mixture was
stirred at 08C for 2 h. The residual SOCl2 was removed, the reaction
product was dissolved in CH2Cl2 (75 mL), and ZnCl2 (2 g) and p-ni-
trophenol (5.68 g) were added. The mixture was refluxed for 3 h.
Afterwards the mixture was cooled with ice and extracted twice by
using Et2O. To remove residual p-nitrophenol, Na2CO3 was added,
and the organic phase was washed a few times with H2O. Finally,
the solvent was removed, and the product was purified by using
Experimental Section
Construction of the library: The libraries were constructed by four
subsequent QuikChange reactions. The following reaction mixture
was prepared: Distilled H2O (40.7 mL), reaction buffer (10ꢁ, 5 mL),
dNTP (1 mL, 10 mm each), plasmid pJOE2792.1 (1 mL, 50 nmolmLÀ1
)
containing the gene encoding PFE,[17] forward primer (1 mL,
ChemBioChem 2010, 11, 1861 – 1866
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1865

U. T. Bornscheuer et al.
silica gel chromatography (solvent: n-hexane/EtOAc 1:1). 1H NMR
(CDCl3): d=1.42 (d, 3H), 2.9 (m, 2H), 3.4 (m, 1H), 7.1 (m, 2H), 7.3
(m, 5H), 8.2 ppm (m, 2H); 13C NMR (CDCl3): 22 (q), 37 (d), 43 (t),
122 125 127 129 (4d), 145 (s), 146 (s), 170 ppm (s); yields: (R)-3-PB-
pNP: 50%; (S)-3-PB-pNP: 55%
butyric acid methylester, 21.1 min; (R)-phenyl butyric acid methyl-
ester, 22.2 min; (S)-phenyl butyric acid ethylester, 31.3 min; (R)-
phenyl butyric acid ethylester, 32.7 min.
Acknowledgements
Activity assay
Alteration of substrate specificity: After performing the cell disrup-
tion as described above, the crude cell extract (15 mL) was added
to phosphate buffer (255 mL, 50 mm, pH 7.5), mixed, and aliquots
(100 mL) of these solutions were transferred to a new microtiter
plate. Activity was measured by following the absorbance at
410 nm for 5 min after adding pNPA solution (10 mL, 10 mm in
DMSO).
We thank the German Research Foundation (DFG, Grant Bo1862/
4-1) for financial support and Prof. Romas Kazlauskas (UofM, St.
Paul, USA) for useful discussions.
Keywords: directed evolution · enantioselectivity · esterases ·
site-saturation mutagenesis · substrate specificity
Alteration of enantioselectivity: After cell harvest, supernatant (2ꢁ
30 mL) was transferred into two microtiter plates each containing
phosphate buffer (100 mL, 50 mm, pH 7.5). The activity of the mu-
tants in both plates were measured by following the absorbance
at 410 nm for 10 min after adding substrate solution (30 mL, (R)-or
(S)-3-PB-pNP, 1 mm in MeCN). Afterwards the plates were incubat-
ed for 1 h at 378C and the absorbance was measured again as an
one-point measurement. The activity of mutants was determined
from the first 10 min for variants showing high activity and from
initial absorbance and absorbance after 60 min for those showing
low activities.
[1] a) R. J. Kazlauskas, U. T. Bornscheuer, Nat. Chem. Biol. 2009, 5, 526–529;
b) Protein Engineering Hanbook, (Eds.: S. Lutz, U. T. Bornscheuer), Wiley-
VCH, Weinheim, 2009.
[2] L. Giver, A. Gershenson, P. O. Freskgard, F. H. Arnold, Proc. Natl. Acad.
Sci. USA 1998, 95, 12809–12813.
[3] Q. Wang, T. Xia, Biotechnol. Lett. 2008, 30, 937–944.
[4] J. J. Hao, A. Berry, Protein Eng. 2004, 17, 689–697.
[5] D. Wilkinson, N. Akumanyi, R. Hurtado-Guerrero, H. Dawkes, P. F.
Knowles, S. E. V. Phillips, M. J. McPherson, Protein Eng. 2004, 17, 141–
148.
[6] M. Schmidt, D. Hasenpusch, M. Kꢃhler, U. Kirchner, K. Wiggenhorn, W.
Langel, U. T. Bornscheuer, ChemBioChem 2006, 7, 805–809.
[7] A. Hidalgo, A. Schliessmann, R. Molina, J. Hermoso, U. T. Bornscheuer,
Protein Eng. 2008, 21, 567–576.
[8] B. Seelig, J. W. Szostak, Nature 2007, 448, 828–831.
[9] a) M. T. Reetz, L. W. Wang, M. Bocola, Angew. Chem. 2006, 118, 1258–
1263; Angew. Chem. Int. Ed. 2006, 45, 1236–1241; b) M. T. Reetz, J. D.
Carballeira, A. Vogel, Angew. Chem. 2006, 118, 7909–7915; Angew.
Chem. Int. Ed. 2006, 45, 7745–7751; Angew. Chem. 2006, 118, 7909–
7915; c) S. Bartsch, R. Kourist, U. T. Bornscheuer, Angew. Chem. 2008,
120, 1531–1534; Angew. Chem. Int. Ed. 2008, 47, 1508–1511; Angew.
Chem. 2008, 120, 1531–1534; d) R. Caglio, F. Valetti, P. Caposio, G. Gri-
baudo, E. Pessione, C. Giunta, ChemBioChem 2009, 10, 1015–1024.
[10] a) J. D. Bloom, S. T. Labthavikul, C. R. Otey, F. H. Arnold, Proc. Natl. Acad.
Sci. USA 2006, 103, 5869–5874; b) J. D. Bloom, P. A. Romero, Z. Lu, F. H.
Arnold, Biol. Direct 2007, 2, 17; c) R. D. Gupta, D. S. Tawfik, Nat. Methods
2008, 5, 939–942.
[11] a) M. Lehmann, C. Loch, A. Middendorf, D. Studer, S. F. Lassen, L. Pasa-
montes, A. P. G. M. van Loon, M. Wyss, Protein Eng. 2002, 15, 403–411;
b) K. M. Polizzi, J. F. Chaparro-Riggers, E. Vazquez-Figueroa, A. S. Bom-
marius, Biotechnol. J. 2006, 1, 531–536.
[12] S. Park, K. L. Morley, G. P. Horsman, M. Holmquist, K. Hult, R. J. Kazlaus-
kas, Chem. Biol. 2005, 12, 45–54.
Cultivation, purification and characterization of PFE wild-type
and mutants M1 and M2: LBAmp media in shake flasks (250 mL)
was inoculated with an overnight culture of wild-type PFE or the
mutants (2.5 mL). When an OD600 of 0.5 had been reached, protein
expression was induced by adding l-rhamnose solution (2.5 mL,
20% w/v), and cultures were incubated for 5 h. Afterwards the cells
were centrifuged (15 min at 4355g), washed twice with Tris–HCl
(20 mm, pH 7.5, 500 mm NaCl) and disrupted by using a FastPrep-
24 bead mill. After centrifugation (15 min at 8173g) the enzyme
was purified by an ꢂktaPurifier by using a 5 mL HisTrap crude FF
column. The protein was eluted from the column with Tris–HCl
(20 mm, pH 7.5, 500 mm NaCl) containing imidazol (300 mm). To
remove imidazol from the solution, the protein was desalted by
using a 60 mL Q-Sepharose (G-25) column. Protein content was de-
termined by the Bradford method.
Determination of kinetic constants of PFE (wild-type) M1, M2:
Kinetic constants were determined by measuring the activities of
purified enzymes at different substrate concentrations (pNPA: 10–
0.05 mm in 10% DMSO; 3-PB-pNP: 1.6–0.05 mm in 25% acetoni-
trile). Calculation was based on calibration curves.
[13] a) R. K. Kuipers, H. J. Joosten, E. Verwiel, S. Paans, J. Akerboom, J. van
der Oost, N. G. Leferink, W. J. van Berkel, G. Vriend, P. J. Schaap, Proteins
2009, 76, 608–616; b) N. G. H. Leferink, M. W. Fraaije, H.-J. Joosten, P. J.
Schaap, A. Mattevi, W. J. van Berkel, J. Biol. Chem. 2009, 284, 4392–
4397.
[14] D. L. Ollis, E. Cheah, M. Cygler, B. Dijkstra, F. Frolow, S. M. Franken, M.
Harel, S. J. Remington, I. Silman, J. Schrag, Protein Eng. 1992, 5, 197–
211.
[15] M. T. Reetz, D. Kahakeaw, R. Lohmer, ChemBioChem 2008, 9, 1797–
1804.
[16] http://www.mpi-muelheim.mpg.de/kofo/institut/arbeitsbereiche/reetz/
reetz_e.html.
Biocatalysis: To determine the true E-values of M1, M2 and wild-
type, (R,S)-3-phenyl butyric acid ethyl ester (20 mm) was incubated
in phosphate buffer (800 mL, 50 mm, pH 7.5) and enzyme solution
(200 mL) at 378 and 550 rpm. Samples (100 mL) were taken at differ-
ent times. After HCl (1n, 40 mL) had been added, the reaction mix-
ture was extracted with CH2Cl2 (3ꢁ400 mL). The combined organic
phases were dried with sodium sulfate and transferred into a new
reaction vial. Afterwards the 3-phenyl butyric acid was derivatized
using trimethylsilyl diazomethane (20 mL) for 30 min. After stop-
ping the reaction by adding HOAc (40 mL) and removing all sol-
vents by vacuum centrifugation (Speedvac), the compounds were
dissolved in CH2Cl2 (200 mL). All enantiomers were separated by
GC–MS (QP2010, Shimadzu Corp.). Column temperature: 1008C, In-
jector: 2208C, Pressure: 116.1 kPa, Retention times: (S)-3-phenyl
[17] N. Krebsfꢃnger, F. Zocher, J. Altenbuchner, U. T. Bornscheuer, Enzyme
Microb. Technol. 1998, 22, 641–646.
Received: May 14, 2010
Published online on August 2, 2010
1866
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1861 – 1866