Iterative saturation mutagenesis accelerates laboratory evolution of enzyme stereoselectivity: Rigorous comparison with traditional methods

Article abstract of DOI:10.1021/ja1030479

Efficacy in laboratory evolution of enzymes is currently a pressing issue, making comparative studies of different methods and strategies mandatory. Recent reports indicate that iterative saturation mutagenesis (ISM) provides a means to accelerate directed evolution of stereoselectivity and thermostability, but statistically meaningful comparisons with other methods have not been documented to date. In the present study, the efficacy of ISM has been rigorously tested by applying it to the previously most systematically studied enzyme in directed evolution, the lipase from Pseudomonas aeruginosa as a catalyst in the stereoselective hydrolytic kinetic resolution of a chiral ester. Upon screening only 10 000 transformants, unprecedented enantioselectivity was achieved (E = 594). ISM proves to be considerably more efficient than all previous systematic efforts utilizing error-prone polymerase chain reaction at different mutation rates, saturation mutagenesis at hot spots, and/or DNA shuffling, pronounced positive epistatic effects being the underlying reason.

Full text of DOI:10.1021/ja1030479

Published on Web 06/10/2010
Iterative Saturation Mutagenesis Accelerates Laboratory
Evolution of Enzyme Stereoselectivity: Rigorous Comparison
with Traditional Methods
Manfred T. Reetz,*,† Shreenath Prasad,† Jose´ D. Carballeira,‡
Yosephine Gumulya,† and Marco Bocola§
Max-Planck-Institut fu¨r Kohlenforschung, Kaiser-Wilhelm-Platz 1, 45470 Mu¨lheim an der Ruhr,
Germany, Industrial Biotransformations SerVice, Scientific Park of Madrid, C/Santiago Grisol´ıa
2, Tres Cantos, 28760 Madrid, Spain, and Institut fu¨r Biophysik und Physikalische Biochemie,
UniVersita¨t Regensburg, UniVersita¨tsstrasse 31, 93053 Regensburg, Germany
Received April 12, 2010; E-mail: reetz@mpi-muelheim.mpg.de
Abstract: Efficacy in laboratory evolution of enzymes is currently a pressing issue, making comparative
studies of different methods and strategies mandatory. Recent reports indicate that iterative saturation
mutagenesis (ISM) provides a means to accelerate directed evolution of stereoselectivity and thermostability,
but statistically meaningful comparisons with other methods have not been documented to date. In the
present study, the efficacy of ISM has been rigorously tested by applying it to the previously most
systematically studied enzyme in directed evolution, the lipase from Pseudomonas aeruginosa as a catalyst
in the stereoselective hydrolytic kinetic resolution of a chiral ester. Upon screening only 10 000 transformants,
unprecedented enantioselectivity was achieved (E ) 594). ISM proves to be considerably more efficient
than all previous systematic efforts utilizing error-prone polymerase chain reaction at different mutation
rates, saturation mutagenesis at hot spots, and/or DNA shuffling, pronounced positive epistatic effects
being the underlying reason.
Introduction
meaningful comparisons assessing a broad range of mutagenesis
methods and strategies are essentially unknown.4 Methodology
Directed evolution1 constitutes a method for engineering
essentially any property of enzymes as catalysts in synthetic
organic chemistry and biotechnology, including stability2 and
stereoselectivity.1i,3 The most often used gene mutagenesis
methods are error-prone polymerase chain reaction (epPCR),
saturation mutagenesis, and DNA shuffling. However, it is not
at all clear what the best strategy actually is because statistically
development in the quest to make laboratory evolution more
efficient and therefore faster than in the past is currently an
important focus of research.1,5 The challenge is to maximize
the quality of mutant libraries, defined in terms of the frequency
of superior mutants (hits) in a given library and the actual degree
(3) Reviews of directed evolution of stereoselective enzymes:1g,12 (a)
Reetz, M. T. J. Org. Chem. 2009, 74, 5767–5778. (b) Reetz, M. T. In
Manual of Industrial Microbiology and Biotechnology, 3rd ed.; Baltz,
R. H., Demain, A. L., Davies, J. E., Bull, A. T., Junker, B., Katz, L.,
Lynd, L. R., Masureka, P., Reeves, C. D., Zhao, H., Eds.; ASM Press:
Washington, DC, 2010; pp 466-479.
† Max-Planck-Institut fu¨r Kohlenforschung.
‡ Scientific Park of Madrid.
§ Universita¨t Regensburg.
(1) Recent reviews of directed evolution:2,3 (a) Johannes, T. W.; Zhao,
H. Curr. Opin. Microbiol. 2006, 9, 261–267. (b) Lutz, S.; Bornscheuer,
U. T. Protein Engineering Handbook; Wiley-VCH: Weinheim, 2009;
Vols. 1-2. (c) Turner, N. J. Nat. Chem. Biol. 2009, 5, 567–573. (d)
Ja¨ckel, C.; Kast, P.; Hilvert, D. Annu. ReV. Biophys. Biomol. Struct.
2008, 37, 153–173. (e) Bershtein, S.; Tawfik, D. S. Curr. Opin. Chem.
Biol. 2008, 12, 151–158. (f) Romero, P. A.; Arnold, F. H. Nat. ReV.
Mol. Cell Biol. 2009, 10, 866–876. (g) Reetz, M. T. In Asymmetric
Organic Synthesis with Enzymes; Gotor, V., Alfonso, I., Garc´ıa-
Urdiales, E., Eds.; Wiley-VCH: Weinheim, 2008; pp 21-63. (h) Kaur,
J.; Sharma, R. Crit. ReV. Biotechnol. 2006, 26, 165–199. (i) Hida, K.;
Hanes, J.; Ostermeier, M. AdV. Drug DeliVery ReV. 2007, 59, 1562–
1578. (j) Otten, L. G.; Hollmann, F.; Arends, I. W. C. E. Trends
Biotechnol. 2009, 28, 46–54. (k) Shivange, A. V.; Marienhagen, J.;
Mundhada, H.; Schenk, A.; Schwaneberg, U. Curr. Opin. Chem. Biol.
2009, 13, 19–25.
(4) Examples of directed evolution studies in which two mutagenesis
methods were compared: (a) Parikh, M. R.; Matsumura, I. J. Mol.
Biol. 2005, 352, 621–628. (b) Pavlova, M.; Klvana, M.; Prokop, Z.;
Chaloupkova, R.; Banas, P.; Otyepka, M.; Wade, R. C.; Tsuda, M.;
Nagata, Y.; Damborsky, J. Nat. Chem. Biol. 2009, 5, 727–733. (c)
Paramesvaran, J.; Hibbert, E. G.; Russell, A. J.; Dalby, P. A. Protein
Eng., Des. Sel. 2009, 22, 401–411. (d) Tang, S.-Y.; Fazelinia, H.;
Cirino, P. C. J. Am. Chem. Soc. 2008, 130, 5267–5271. (e) Joern,
J. M.; Meinhold, P.; Arnold, F. H. J. Mol. Biol. 2002, 316, 643–656.
(f) Wang, T.-W.; Zhu, H.; Ma, X.-Y.; Zhang, T.; Ma, Y.-S.; Wei,
D.-Z. Mol. Biotechnol. 2006, 34, 55–68. (g) Zaccolo, M.; Williams,
D. M.; Brown, D. M.; Gherardi, E. J. Mol. Biol. 1996, 255, 589–603.
(h) Reetz, M. T.; Wilensek, S.; Zha, D.; Jaeger, K.-E. Angew. Chem.,
Int. Ed. 2001, 40, 3589–3591.
(5) Papers addressing the question of efficiency in directed evolution:4
(a) Lutz, S.; Patrick, W. M. Curr. Opin. Biotechnol. 2004, 15, 291–
297. (b) Kazlauskas, R. J.; Bornscheuer, U. T. Nat. Chem. Biol. 2009,
5, 526–527. (c) Fox, R. J.; Huisman, G. W. Trends Biotechnol. 2008,
26, 132–138. (d) Reetz, M. T.; Kahakeaw, D.; Lohmer, R. ChemBio-
Chem 2008, 9, 1797–1804. (e) Park, S.; Morley, K. L.; Horsman, G. P.;
Holmquist, M.; Hult, K.; Kazlauskas, R. J. Chem.Biol. 2005, 12, 45–
54.
(2) Reviews of directed evolution of thermostability: (a) Eijsink, V. G. H.;
Gåseidnes, S.; Borchert, T. V.; van den Burg, B. Biomol. Eng. 2005,
22, 21–30. (b) Bommarius, A. S.; Broering, J. M. Biocatal. Biotrans-
form. 2005, 23, 125–139. (c) Wintrode, P. L.; Arnold, F. H. AdV.
Protein Chem. 2001, 55, 161–225. (d) Amin, N.; Liu, A. D.; Ramer,
S.; Aehle, W.; Meijer, D.; Metin, M.; Wong, S.; Gualfetti, P.;
Schellenberger, V. Protein Eng., Des. Sel. 2004, 17, 787–793.
9
9144 J. AM. CHEM. SOC. 2010, 132, 9144–9152
10.1021/ja1030479  2010 American Chemical Society

Accelerating Laboratory Evolution of Enzyme Stereoselectivity
A R T I C L E S
Chart 1
of catalyst improvement.5d High-quality libraries require less
screening effort,5d which to this day is the bottleneck of
laboratory evolution.1,6 Motivated by the credo “quality, not
quantity”,5a we have recently developed iterative saturation
mutagenesis (ISM), according to which appropriate sites in the
protein, comprising one or more amino acid positions, are first
randomized with formation of focused libraries.5d,7 The gene
of a given hit then serves as a template for performing saturation
mutagenesis at the respective other sites, and the process is
repeated until the desired catalyst quality has been achieved.
Stereoselectivity,7a-c substrate acceptance (rate),7b and
thermostability7c can be handled, the criteria for choosing the
proper randomization sites being different according to the
catalytic property under study. When addressing stereoselectivity
and/or substrate scope, sites aligning the complete binding
pocket are considered in a process termed combinatorial active-
site saturation test (CAST).7a,b,8 Rather than arbitrarily targeting
a certain site for saturation mutagenesis and then possibly
turning to another residue, as we4h,9 and others have done
previously for various purposes,1,10 iterative CASTing consti-
tutes methodical systematization and is therefore a useful
acronym to distinguish it from saturation mutagenesis at
alternative (remote) sites. Since only small mutant libraries in
the range of 100-3000 transformants are generally required,
the screening effort is minimized. However, meaningful assess-
ment relative to other methods is possible only if direct
comparisons are made. Here we present the first comprehensive
undertaking of this kind, which throws light not only on ISM
itself but also on conventional approaches used thus far in
directed evolution.1
To date, the most thoroughly investigated case of directed
evolution concerns the lipase A from Pseudomonas aeruginosa
(PAL) as a catalyst in the hydrolytic kinetic resolution of the
chiral ester rac-1 (Chart 1), an enzyme that we have focused
on since the mid-1990s.4h,9,11,12 Using PAL as the model protein
and epPCR as the gene mutagenesis method, we provided proof-
of-principle of laboratory evolution of enantioselective enzymes
as a potentially prolific source of catalysts for asymmetric
transformations.11 Subsequently we tested a variety of other
mutagenesis techniques and strategies in order to evolve
enhanced or reversed stereoselectivity, thereby setting the stage
for comparing mutagenesis methods. As part of several com-
prehensive studies involving PAL, various approaches were
explored, including four successive rounds of epPCR at low
mutation rate,11 a fifth round,12 epPCR at high mutation rate,4h
saturation mutagenesis at hot spots identified by epPCR,9
saturation mutagenesis at a four-residue site aligning the binding
pocket,4h and DNA shuffling of various mutants obtained at low
and high error-rate epPCR4h as well as DNA shuffling with
simultaneous saturation mutagenesis at two hot spots.4h,12 Today
these strategies are conventional. Turning from one hot spot to
another using saturation mutagenesis or epPCR was part of these
exercises, but systematization as in ISM was not considered.
Following the screening of about 50 000 transformants, a variant
1H8 was identified showing a selectivity factor of E ) 51 in
favor of (S)-1, which is a notable increase relative to the
performance of wild-type (WT) PAL (E ) 1.1). It is character-
ized by six point mutations, Asp20Asn/Ser53Pro/Ser155Met/
Leu162Gly/Thr180Ile/Thr234Ser, only Leu162Gly being next
to the binding pocket.4h,12
(6) Reviews covering screening and selection systems in directed
evolution: (a) Reetz, M. T. In Enzyme AssayssHigh-throughput
Screening, Genetic Selection and Fingerprinting; Reymond, J.-L., Ed.;
Wiley-VCH: Weinheim, 2006; pp 41-76. (b) Reymond, J.-L. Enzyme
AssayssHigh-throughput Screening, Genetic Selection and Finger-
printing; Wiley-VCH: Weinheim, 2006. (c) Lin, H.; Cornish, V. W.
Angew. Chem., Int. Ed. 2002, 41, 4402–4425. (d) Boersma, Y. L.;
Dro¨ge, M. J.; Quax, W. J. FEBS J. 2007, 274, 2181–2195. (e) Taylor,
S. V.; Kast, P.; Hilvert, D. Angew. Chem., Int. Ed. 2001, 40, 3310–
3335. (f) Reymond, J.-L.; Fluxa`, V. S.; Maillard, N. Chem. Commun.
(Cambridge, U.K.) 2009, 34–46.
(7) (a) Reetz, M. T.; Wang, L.-W.; Bocola, M. Angew. Chem., Int. Ed.
2006, 45, 1236–1241 (erratum p 2556). (b) Bougioukou, D. J.; Kille,
S.; Taglieber, A.; Reetz, M. T. AdV. Synth. Catal. 2009, 351, 3287–
3305. (c) Reetz, M. T.; Carballeira, J. D. Nat. Protoc. 2007, 2, 891–
903.
(8) (a) Reetz, M. T.; Bocola, M.; Carballeira, J. D.; Zha, D.; Vogel, A.
Angew. Chem., Int. Ed. 2005, 44, 4192–4196. (b) Reetz, M. T.;
Carballeira, J. D.; Peyralans, J.; Ho¨benreich, H.; Maichele, A.; Vogel,
A. Chem.sEur. J. 2006, 12, 6031–6038.
(9) Liebeton, K.; Zonta, A.; Schimossek, K.; Nardini, M.; Lang, D.;
Dijkstra, B. W.; Reetz, M. T.; Jaeger, K.-E. Chem. Biol. 2000, 7, 709–
718.
(10) Early examples of saturation mutagenesis with formation of focused
libraries:4h,9 (a) Dube, D. K.; Loeb, L. A. Biochemistry 1989, 28, 5703–
5707. (b) Climie, S.; Ruiz-Perez, L.; Gonzalez-Pacanowska, D.;
Prapunwattana, P.; Cho, S.-W.; Stroud, R.; Santi, D. V. J. Biol. Chem.
1990, 265, 18776–18779. (c) Teplyakov, A. V.; van der Laan, J. M.;
Lammers, A. A.; Kelders, H.; Kalk, K. H.; Misset, O.; Mulleners,
L. J. S. M.; Dijkstra, B. W. Protein Eng. 1992, 5, 413–420. (d)
Graham, L. D.; Haggett, K. D.; Jennings, P. A.; Le Brocque, D. S.;
Whittaker, R. G.; Schober, P. A. Biochemistry 1993, 32, 6250–6258.
(e) Warren, M. S.; Benkovic, S. J. Protein Eng. 1997, 10, 63–68. (f)
Antikainen, N. M.; Hergenrother, P. J.; Harris, M. M.; Corbett, W.;
Martin, S. F. Biochemistry 2003, 42, 1603–1610. (g) Gabor, E. M.;
Janssen, D. B. Protein Eng., Des. Sel. 2004, 17, 571–579. (h) Rui, L.;
Cao, L.; Chen, W.; Reardon, K. F.; Wood, T. K. J. Biol. Chem. 2004,
279, 46810–46817. (i) Schmitzer, A. R.; Le´pine, F.; Pelletier, J. N.
Protein Eng., Des. Sel. 2004, 17, 809–819. (j) Chockalingam, K.; Chen,
Z.; Katzenellenbogen, J. A.; Zhao, H. Proc. Natl. Acad. Sci. U.S.A.
2005, 102, 5691–5696.
Mechanistically, activated serine at position 82 as part of the
catalytic triad Asp229/His251/Ser82 adds nucleophilically to the
ester carbonyl function with formation of the usual oxyanion
in the rate- and stereochemistry-determining step.13 A theoretical
study subsequently revealed enhanced enantioselectivity as being
due to a relay mechanism, and also predicted that only two of
the six accumulated point mutations are actually necessary for
achieving high enantioselectivity, namely Ser53Pro/Leu162Gly,
(11) Reetz, M. T.; Zonta, A.; Schimossek, K.; Liebeton, K.; Jaeger, K.-E.
Angew. Chem., Int. Ed. Engl. 1997, 36, 2830–2832.
(12) Reetz, M. T. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 5716–5722.
(13) Nardini, M.; Lang, D. A.; Liebeton, K.; Jaeger, K.-E.; Dijkstra, B. W.
J. Biol. Chem. 2000, 275, 31219–31225.
9
J. AM. CHEM. SOC. VOL. 132, NO. 26, 2010 9145

A R T I C L E S
Reetz et al.
which was substantiated experimentally by generating the
respective double mutant (E ) 63).14 This was a triumph of
theory, but it also suggested that the strategies that we were
applying, although successful, were far from optimal. Superflu-
ous point mutations accompanied by excessive screening labor
are likely to be common in laboratory evolution, although this
point is rarely illuminated.1 In view of the extensive available
dataregardingtheevolutionofenantioselectivePALvariants,4h,9,11,12
we surmised that this experimental platform is ideal for testing
ISM and consequently for comparing directed evolution methods
in a systematic and rigorous manner.
Results and Discussion
Strategies for Choosing ISM Sites. It is well known that
lipases harboring esters as substrates have “two” binding
pockets, one for the acid and the other for the alcohol part of
the substrate, catalytically active serine as part of the triad Asp/
His/Ser lying between them.15 In the present case, the activated
hydroxy function of Ser82 in the catalytic triad Asp229/His251/
Ser82 adds nucleophilically to the carbonyl function of the ester
rac-1 with formation of the respective short-lived oxyanion.
Since the mutagenesis/screening experiments were performed
with PAL using the hydrolytic kinetic resolution of the ester
rac-1 having the stereogenic center in the acid part, we focused
solely on the acid-binding pocket in order to select appropriate
residues for saturation mutagenesis. Guided by the analysis of
the crystal structure of PAL (PDB code 1ex9),13 we selected
six amino acid residues, Met16, Leu17, Leu159, Leu162,
Leu231, and Val232, for saturation mutagenesis (Figure 1). The
distance values of the R-C-atom of all these six residues to the
R-C-atom of the acid moiety (C4) from the phosphonate
inhibitor bound in the crystal structure of PAL, corresponding
to the stereocenter C2 of substrate 1 bound as the oxyanion,
are as follows: Met16, 5.3 Å; Leu17, 8.4 Å; Leu159, 8.5 Å;
Leu162, 6.5 Å; Leu231, 7.0 Å; and Val232, 7.3 Å.
When deciding on how to group these six single residues
into CAST sites for saturation mutagenesis, the problem of
oversampling needs to be considered.5d,7c,16 The six amino acid
positions could be classified, for example, as six single-residue
randomization sites. Each saturation mutagenesis library would
then require, in the case of NNK codon degeneracy encoding
all 20 canonical amino acids, the screening of about 100
transformants for 95% library coverage (N, adenine/cytosine/
guanine/thymine; K, guanine/thymine). Alternatively, it is
possible to consider, for example, three, two-residue CAST sites,
each requiring an oversampling of about 3000 transformants
for similar library coverage. These values were calculated using
the CASTER computer aid7c which is based on previous
algorithms assuming the absence of amino acid bias.16b Group-
ing into three, two-residue sites involves a greater screening
effort in the case of the initial libraries, but the situation reverses
Figure 1. Schematic representation of amino acid residues considered for
saturation mutagenesis, based on the X-ray structure of WT-PAL:13 sites
A (Met16/Leu17, green), B (Leu159/Leu162, blue), and C (Leu231/Val232,
yellow) around the active site Ser82 (stick representation in gray and red)
in the acid-binding pocket (white circle). The red circle marks the alcohol-
binding pocket, in the case at hand harboring the p-nitrophenyl moiety of
rac-1. At the top of picture, helix and loop in pink (right, Asp113-Leu156)
and purple (left, Pro203-Asn228) represent lid 1 and lid 2, respectively.18
in the subsequent iterative steps if the complete ISM scheme is
fully investigated (Table 1). In the case of a reduced amino
acid alphabet using, for example, DNT codon degeneracy (D,
adenine/guanine/thymine; T, thymine) encoding 11 amino acids
(Ala, Asn, Asp, Cys, Gly, Ile, Phe, Ser, Thr, Tyr, and Val) or
NDT encoding 12 amino acids (Phe, Leu, Ile, Val, Tyr, His,
Asn, Asp, Cys, Arg, Ser, and Gly), the screening effort is sharply
reduced in both cases,7c but again double-residue sites appear
to have a statistical advantage over single-residue analogues in
an ISM scheme. Moreover, we speculated that in such a
grouping cooperative (not just additive) effects could operate
between the two point mutations within each set and also
between sets of mutations in the subsequent iterative steps.17
In contrast, cooperative effects are not possible in the initial
libraries when choosing single-residue sites. In the present study
we therefore opted for three sites, each comprising two amino
acid positions, namely A (Met16/Leu17), B (Leu159/Leu162),
and C (Leu231/Val232), shown in Figure 1, which also features
catalytically active Ser82 as well as lid 1 and lid 2.18
Generation of Mutants with High Stereoselectivity. When
considering an ISM system consisting of three sites, each site
being visited once in a given upward pathway, a total of 15
saturation mutagenesis libraries defining six different pathways
are theoretically possible (Table 1 and Figure 2a). A systematic
search provides a means to identify the best pathway. In the
present case we used NNK codon degeneracy and screened
about 3000 transformants in each of the three initial libraries.
(14) (a) Bocola, M.; Otte, N.; Jaeger, K.-E.; Reetz, M. T.; Thiel, W.
ChemBioChem 2004, 5, 214–223. (b) Reetz, M. T.; Puls, M.;
Carballeira, J. D.; Vogel, A.; Jaeger, K.-E.; Eggert, T.; Thiel, W.;
Bocola, M.; Otte, N. ChemBioChem 2007, 8, 106–112.
¨
(15) (a) Orrenius, C.; Haeffner, F.; Rotticci, D.; Ohrner, N.; Norin, T.; Hult,
K. Biocatal. Biotransform. 1998, 16, 1–15. (b) Ema, T. Curr. Org.
Chem. 2004, 8, 1009–1025.
(16) (a) Rui, L.; Kwon, Y. M.; Fishman, A.; Reardon, K. F.; Wood, T. K.
Appl. EnViron. Microbiol. 2004, 70, 3246–3252. (b) Patrick, W. M.;
Firth, A. E. Biomol. Eng. 2005, 22, 105–112. (c) Bosley, A. D.;
Ostermeier, M. Biomol. Eng. 2005, 22, 57–61. (d) Denault, M.;
Pelletier, J. N. In Protein Engineering Protocols; Arndt, K. M., Mu¨ller,
K. M., Eds.; Humana Press: Totowa, NJ, 2007; Vol. 352, pp 127-
154.
(17) Reetz, M. T.; Sanchis, J. ChemBioChem 2008, 9, 2260–2267.
(18) Cherukuvada, S. L.; Seshasayee, A. S. N.; Raghunathan, K.; Anishetty,
S.; Pennathur, G. PLoS Comput. Biol. 2005, 1, e28, 182-189.
9
9146 J. AM. CHEM. SOC. VOL. 132, NO. 26, 2010

Accelerating Laboratory Evolution of Enzyme Stereoselectivity
A R T I C L E S
Table 1. Statistical Consequences as a Function of Grouping Single CAST Residues into Randomization Sites
total number of transformants screened in complete
ISM scheme for 95% coverage
no. of pathways in
complete ISM scheme
no. of libraries in
complete ISM scheme
type of grouping
using NNK codon
using NDT or DNT codon
6 single-residue sites
3 double-residue sites
2 triple-residue sites
1 six-residue site
720
6
2
1956
15
4
183 864 (94 per lib.)
45 990 (3066 per lib.)
392 652 (98 163 per lib.)
3.21 × 109
66 504 (34 per lib.)
6450 (430 per lib.)
20 700 (5175 per lib.)
8.95 × 106
1
1
Medium-throughput screening (700 transformants/day) was
performed using our original ee-assay in which a UV/vis plate
reader monitors the initial reaction rates of (R)- and (S)-1
separately, the relative slopes serving as an indicator for the
occurrence of a hit having enhanced enantioselectivity11 (Figure
S1, Supporting Information). Whereas the libraries generated
at sites A and C failed to contain significantly improved variants,
the library at site B was found to harbor several hits, the best
one being variant 1F8 with a single mutation Leu162Asn,
showing a selectivity factor of E ) 8 in favor of (S)-1. The
results of a kinetic study of mutant 1F8 and of some other
prominent mutants arising from randomization at sites B and C
are summarized in Table 2.
compound, or both. Kinetic data regarding the hydrolytic
reactions of (S)-1 and (R)-1 were obtained separately using
purified WT-PAL, the previously obtained best mutant 1H8 with
six mutations,4h and the two newly evolved mutants of the
present study, 1F8 (Leu162Asn) and 1B2 (Met16Ala/Leu17Phe/
Leu162Asn), respectively. The results summarized in Table 2
point to a remarkable effect, namely that in the case of the best
mutant of the present study (1B2), the rate of reaction kcat of
the (S)-1 has been increased dramatically (37-fold), a slight
decrease (5.6-fold) in reaction rate of the (R)-enantiomer
contributing additionally to the high overall enantioselectivity
(E ) 594). Together these effects lead to a perfect kinetic
resolution.
We then utilized the gene of variant 1F8 as a template for
performing two separate second-round randomization experi-
ments at the other two sites, A (Met16/Leu17) and C (Leu231/
Val232), respectively, despite the fact that the initial libraries
generated thereat had not provided any significantly improved
single or double mutants in terms of enhanced enantioselectivity.
In order to decrease the screening effort, we considered a
reduced amino acid alphabet on the basis of NDT or DNT codon
degeneracy (Table 1), a strategy that we had previously
developed in order to increase the efficacy of directed evolution
based on saturation mutagenesis.5d Since both sites harbor
leucine, we opted for DNT codon degeneracy for a simple
reason: DNT does not encode leucine, thereby reducing the
amount of “starting” 1F8 variants in the respective libraries and
increasing the number of new mutants in the screened library.
It is known that in any saturation mutagenesis library, WT or
“starting” mutants always appear, which reduces quality.5d,7a
In the present case of a two-residue site, about 430 transformants
need to be screened for 95% library coverage5d,7 (Table 1), and
we settled for 500. The best hit (1B2) originated from the
second-round library at site A, containing two new mutations,
Met16Ala/Leu17Phe. Mutant 1B2 is thus characterized by three-
point mutations Met16Ala/Leu17Phe/Leu162Asn. It proved to
be so enantioselective that application of the Sih equation19 for
determining the E-value is no longer meaningful. Kinetic
characterization revealed a selectivity factor of E ) 594 (S) in
the model transformation involving (S)- and (R)-1 (Table 2).
Upon targeting site C using the gene of 1F8 as the template, no
significantly improved mutants were found. At this point,
screening amounted to the evaluation of a total of only 10 000
transformants, yet unprecedented (S)-enantioselectivity had
evolved. The pathways considered thus far are shown schemati-
cally in Figure 2. From a practical viewpoint, further mutagen-
esis, e.g., using 1B2 as a template to randomize at site C, was
not necessary.
Unveiling the Reason for the Efficacy of ISM. It is clear that
the sum of the single mutation Leu162Asn (site B) obtained in
the initial randomization experiment and the two mutations
Figure 2. (a) Scheme of an ISM system composed of three sites, A, B,
and C. Solid green connections define the best pathway in the present PAL
study, dotted red lines indicating saturation mutagenesis which failed to
provide significantly improved PAL mutants and dotted black lines denoting
unexplored saturation mutagenesis experiments. (b) Real-scale scheme of
the best pathway, WTfBfA, leading to the best mutant 1B2 characterized
by mutations Met16Ala/Leu17Phe/Leu162Asn and showing a selectivity
factor of E ) 594 (S) in the hydrolytic kinetic resolution of rac-1.
Kinetic Studies. Enhancing the enantioselectivity of a kinetic
resolution can come about by increasing the reaction rate of
one enantiomer, or by decreasing the rate of the mirror-image
(19) Chen, C.-S.; Fujimoto, Y.; Girdaukas, G.; Sih, C. J. J. Am. Chem.
Soc. 1982, 104, 7294–7299.
9
J. AM. CHEM. SOC. VOL. 132, NO. 26, 2010 9147

A R T I C L E S
Reetz et al.
Table 2. Kinetic Study of Various Mutants of Wild-Type P. aeruginosa Lipase (WT-PAL) Obtained from Iterative Saturation Mutagenesis,
Deconvolution Study of Best Mutant (1B2), and Conventional Methods, (S)-1 and (R)-1 Serving Separately as Substrates
(S)-1
(R)-1
rounds of
saturation
mutagenesis
site of ISM
mutation
Km
kcat
k
cat/Km
Km
kcat
kcat/Km
(s-1 M-1
entry
mutant
(×10-4 M) (×10-3 s-1
)
(s-1 M-1
)
(×10-4 M) (×10-3 s-1
)
)
E
1
2
3
4
5
6
7
8
9
WT-PAL
Leu162Gly (2C11)
Leu162Ile (2H9)
Leu162Thr (19F12)
8.6
3.4
9.4
13.2
5.6
2.9
3.4
22.0
3.9
6.9
10.1
37.4
3.2
13.6
167
42.1
0.4
1374
16.1
1.4
1.0
38.6
43.5
9.4
14.5
126
75.2
1.4
4041
7.3
3.6
1.4
38.2
7.9
1.9
1.2
9.7
7.9
5.0
7.9
20.6
9.5
7.5
29.9
0.3
0.3
30.1
7.4
0.8
5.4
5.7
1.5
1.0
5.5
37.8
1.6
2.5
31.0
9.4
1.6
6.8
2.8
1.6
1.3
0.9
1.1(S)
5.9(S)
5.8(S)
4.1(S)
8.0(S)
1.1(R)
594(S)
2.6(S)
2.2(S)
1.1(S)
42.4(S)
B
B
B
B
C
B/A
1st round
1st round
1st round
1st round
1st round
2nd round
Leu162Asn (1F8)
Leu231Ile/Val232Cys (1C7)
Met16Ala/Leu17Phe/Leu162Asn (1B2)
Met16Ala/Leu17Phe
mutant of deconvolution study
mutant of deconvolution study
mutant of deconvolution study
best mutant from conventional
method of enzyme evolution12
Met16Ala/Leu162Asn
10 Leu17Phe/Leu162Asn
11 Asp20Asn/Ser53Pro/Ser155Met/
62.6
Leu162Gly/Thr180Ile/Thr234Ser (1H8)
Met16Ala/Leu17Phe (site A) accumulated in the iterative step
induces a dramatic increase in stereoselectivity. Since a double
mutant corresponding to Met16Ala/Leu17Phe was not found
in the initial saturation mutagenesis library at site A, it is likely
that strong cooperative effects operating between the mutations
are involved. In order to validate this, deconvolution experiments
were performed by generating appropriate mutants using site-
directed mutagenesis. For this purpose variants Met16Ala/
Leu17Phe, Met16Ala/Leu162Asn, and Leu17Phe/Leu162Asn
were tested in kinetic experiments using (S)-1 and (R)-1 (Table
2). It can be seen that the epistatic interaction of mutation
Leu162Asn with the set of two mutations Met16Ala/Leu17Phe
results in a dramatic cooperative effect, the two respective
variants having selectivity factors E of only 8.0(S) and 2.6(S),
respectively. Together they orchestrate unprecedented enanti-
oselectivity amounting to E ) 594 (S). Moreover, letting the
observed point mutation at position 162 interact individually
with those at positions 16 and 17 in the form of variants
Met16Ala/Leu162Asn (E ) 2.2) and Leu17Phe/Leu162Asn (E
) 1.1), respectively, actually reduces enantioselectivity relative
to the single mutant Leu162Asn (E ) 8.0). We conclude that
the decision to group the originally identified six single amino
acid positions into a set of three, two-residue sites is indeed
optimal, and that a grouping defined by six single-residue sites
would not have been successful in the present system. However,
this does not mean that the process of grouping into single-
residue sites always ends up in failure.7b
Substrate Scope of Best Mutant Met16Ala/Leu17Phe/
Leu162Asn. From the viewpoint of organic chemists or bio-
technologists, an enzyme variant evolved for a specific substrate
should also be a viable catalyst for a set of structurally different
compounds, first and foremost in terms of activity.3 We therefore
tested the best mutant, 1B2, as a catalyst in the hydrolytic kinetic
resolution of racemic compounds 4-11 (Chart 2). Since the
actual directed evolution experiments were performed using
rac-1 in the attempt to increase enantioselectivity, there was
no stringent reason to expect wide substrate acceptance.
Remarkably, enhanced activity relative to WT-PAL was ob-
served in all cases, demonstrating that substrate scope has been
widened (Table 3). For example, in the case of substrate 4, kcat/
Km characterizing the reaction of the preferred (S)-enantiomer
increases by a factor of 5 × 103 upon going from WT-PAL to
Chart 2
9
9148 J. AM. CHEM. SOC. VOL. 132, NO. 26, 2010

Accelerating Laboratory Evolution of Enzyme Stereoselectivity
A R T I C L E S
Table 3. Altered Substrate Scope (Rate) and Enhanced Enantioselectivity upon Going from WT-PAL to ISM-Evolved Mutant 1B2 (Met16Ala/
Leu17Phe/Leu162Asn), Measured by Kinetics
mutant
Km (×10-4 M)
kcat (×10-3 s-1
)
kcat/Km (s-1 M-1
)
Km (×10-4 M)
kcat (×10-3 s-1
(R)-4
)
kcat/Km (s-1 M-1
)
E
(S)-4
0.2
161
WT
1B2
8.1
17.2
0.25
93.6
28.2
8.5
3.0
31(S)
(S)-5
0.1
239
(R)-5
0.1
0.9
WT
1B2
2.3
25.0
0.4
95.6
6.6
10.1
0.2
0.9
2.0(R)
106(S)
(S)-6
0.2
2949
(R)-6
0.4
134
WT
1B2
1.4
23.8
1.4
1239
4.4
59.1
0.9
22.7
1.6(S)
55(S)
mutant 1B2. Thus, when addressing substrate acceptance (rate),
ISM appears to be far more effective than our previous strategy
based on combining point mutations obtained by saturation
mutagenesis at two different sites.8b We prefer not to speculate
on the reason for this finding. Moreover, the kinetic studies of
rac-4, rac-5, and rac-6 using variant 1B2 also showed that good
to excellent stereoselectivity was achieved in favor of the (S)-
substrates, the E-values amounting to 31, 106, and 55, respec-
tively (Table 3). In the case of the other substrates, moderate
but still respectable enantioselectivity resulted, the selectivity
factors being derived by application of the Sih equation:19 rac-
7, E ) 12; rac-8, E ) 17; rac-9, E ) 23; rac-10, E ) 20; and
rac-11, E ) 18, all in favor of the respective (S)-enantiomer.
Unveiling the Source of Enhanced Enantioselectivity. In order
to gain some insight regarding the source of enhanced reaction
rate and enantioselectivity of the superior mutant 1B2 (Met16Ala/
Leu17Phe/Leu162Asn), molecular modeling and molecular
dynamics (MD) simulations were performed, similar to those
reported previously for the originally evolved mutant 1H8
having six point mutations.14 As before, the crystal structure of
WT-PAL13 was utilized as a starting point, and amino acid
mutations were introduced using the modeling suite YASARA.20
The substrates (R)- and (S)-1 were introduced as tetrahedral
intermediates (oxyanions) covalently bound to catalytically
active Ser82.
can then be considered to be a much smaller residue than
Leu162, which is characterized by a fairly space-filling hydro-
phobic isobutyl side chain pointing into the binding pocket of
WT-PAL. In the case of the best earlier mutant, 1H8 (E ) 51),4h
having six point mutations, a space-providing effect was also
induced by a mutational change at the same amino acid position,
although the molecular basis of the particular point mutation
observed, Leu162Gly, is obviously different.14 Point mutation
Met16Ala in the best variant, 1B2, also contributes to the
enlargement of the binding pocket, but it is the cooperativity
acting between the three point mutations Met16Ala/Leu17Phe/
Leu162Asn which is responsible for the overall effect regarding
increased activity and enhanced enantioselectivity. Interestingly,
the third point mutation in variant 1B2, namely Leu17Phe, leads
to an energetically favorable interaction of the phenyl group
with the p-nitrophenyl moiety of the substrate.
The creation of more space does not necessarily need to
correlate with higher reaction rate. Indeed, the reshaping of the
binding pocket induces a second effect which is crucial, namely
additional stabilization of the oxyanion beyond the normal
stabilizing interactions in traditional oxyanion formation. This
provides additional stabilization of the transition state of the
rate- and stereochemistry-determining step, a phenomenon which
is possible with the favored (S)-1 but not with the disfavored
(R)-enantiomer. The effect comes about by additional H-bonding
between His83 and the oxyanion originating from (S)-1 (Figure
4). In the case of the disfavored substrate (R)-1, severe steric
clash between the R-methyl group at the stereogenic center and
the imidazole moiety of His83 occurs, preventing any gain in
stabilization. Moreover, the simulation in the case of (R)-1 points
to a twisted binding mode which results in a less productive
conformer, because now the usual oxyanion stabilization arising
from His251 is largely disrupted. This model is in full accord
One important result is the realization that additional space
at the binding pocket in mutant 1B2 is created by the point
mutation Leu162Asn, thereby inducing significant rate enhance-
ment of the reaction of the model compound and the additional
bulky substrates (Figure 3). Accordingly, the side chain of
Asn162 was found in a rotamer in which it reaches away from
the active site by undergoing an H-bond relationship with Ser158
and His83 near the surface of the enzyme. In effect, Asn162
Figure 3. Comparison of the acid-binding pocket of WT-PAL (left) and model of the best mutant, 1B2 (right), with bound (S)-1 to active site Ser82. The
strong cooperative effect between mutations Met16Ala/Leu17Phe/Leu162Asn leads to the creation of a larger binding pocket and thereby to the accommodation
of the long aliphatic arm of (S)-1 (right) better than WT-PAL (left). In mutant 1B2 (right), a new extended H-bond network indicated by green lines occurs
between Asn162, Ser158, His83, carbonyl oxygen of (S)-1, and Ala16, the side chain of Asn162 pointing away from the acid-binding pocket.
9
J. AM. CHEM. SOC. VOL. 132, NO. 26, 2010 9149

A R T I C L E S
Reetz et al.
Figure 4. Comparison of oxyanion-stabilizing H-bonds (indicated by green lines) in mutant 1B2 in a productive binding mode for (S)-1 (left) and in a
nonproductive orientation as found for the less reactive (R)-1 (right). In the nonproductive orientation, the catalytic H-bond toward His251 from the catalytic
triad is lost, and the negatively charged oxyanion is stabilized by only two H-bonds. Due to disruption of the H-bond network in nonproductive binding
mode (right), Asn162 is found in different conformers, floating toward the substrate binding space (red double-headed arrows) and occupying space thereat.
with the results of the kinetic study (Table 2). Indeed, the
calculated difference in stabilization energy of the two enanti-
omers amounts to about 3 kcal/mol. Finally, we speculated that
water could also bind in the newly created pocket, thereby
increasing the difference in energy. Although a stable minimum
structure was identified, it was found that, during the simulation,
water moves around and is less frequently observed near the
oxyanion than His83.
As already delineated, previous efforts based on standard
approaches using epPCR at different mutation rates, various
forms of saturation mutagenesis, and DNA shuffling resulted
in a selectivity factor of E ) 51, the overall effort requiring the
screening of about 50 000 transformants.12 In sharp contrast,
ISM as applied in the present study has provided a much
better mutant having dramatically improved enantioselectivity
(E ) 594), while requiring a highly reduced screening effort
of only 10 000 transformants. Deconvolution experiments
showed that superfluous mutations do not occur when
applying ISM, in contrast to the original method based on a
combination of epPCR, saturation mutagenesis, and DNA
shuffling approaches,14b and that the method is characterized
by the occurrence of strong cooperative effects acting between
point mutations and sets of mutations. We have observed this
effect in a previous ISM study concerning the enhancement of
enantioselectivity of an epoxide hydrolase,17 and we conclude
that positive epistasis of this kind is the best prerequisite for
high-quality libraries in directed evolution. In the present case,
additivity alone would not have resulted in a significant increase
in stereoselectivity. Finally, the best mutant proved to be a viable
catalyst in the hydrolysis of a set of bulky chiral esters not
accepted by WT-PAL, enantioselectivity being respectable to
excellent (E ) 12-106). Thus, without performing any ad-
ditional mutagenesis experiments, it is possible to “get more
than what you screen for”.7b Since we did not focus on substrate
scope in the actual mutagenesis experiments, these findings
should be considered as being fortuitous. However, ISM can,
in fact, be applied in order to address the issue of substrate
acceptance (rate) directly.7b Recently, this was demonstrated
once more by Ba¨ckvall and co-workers, who applied ISM in
the quest to engineer enantioselectivity and substrate scope of
lipase A from Candida antarctica (CALA).21b
Conclusions
More than 15 years of research in directed evolution of
enzymes have shown that some degree of catalyst improvement
can always be achieved, irrespective of the applied gene
mutagenesis method. This record reflects the power of laboratory
evolution as a protein engineering method, yet the pressing issue
currently concerns efficacy.1-5 It is not only industrial biotech-
nology that needs methods for rapid directed evolution enabling
reliable timeframes for the discovery and production of effective
catalysts for new tasks. The challenge is of a fundamental nature,
requiring basic research. Our contribution in this mission is
iterative saturation mutagenesis (ISM), which we previously
illustrated in the evolution of stereoselectivity of an epoxide
hydrolase7a and of an enoate reductase7b and in thermostabili-
zation of a lipase,7c but it was not clear how general this
approach is and especially how it compares with other directed
evolution methods.
In order to subject ISM to rigorous comparison with other
directed evolution approaches, and at the same time to extend
its application, we revisited in the present study the previously
most intensively investigated enzyme in this respect, namely
the lipase from Pseudomonas aeruginosa (PAL) as a catalyst
in the hydrolytic kinetic resolution of the chiral ester rac-1.4h,9,11,12
(20) Krieger, E.; Darden, T.; Nabuurs, S. B.; Finkelstein, A.; Vriend, G.
Proteins: Struct., Funct. Bioinf. 2004, 57, 678–683.
9
9150 J. AM. CHEM. SOC. VOL. 132, NO. 26, 2010

Accelerating Laboratory Evolution of Enzyme Stereoselectivity
A R T I C L E S
Table 4. List of Forward Primers
name of primer
sequence of primer
name of library/mutant generated
WT_LibA_NNK_F
WT_LibB_NNK_F
WT_LibC_NNK_F
1F8_LibA_DNT_F
1F8_LibC_DNT_F
Met16Ala/Leu17Phe_F
CTGGCCCACGGCNNKNNKGGCTTCGACAACATC
TCACTGGGCTCGNNKGAGTCGNNKAACAGCGAGGGTGCC
GCCAACGACGGCNNKNNKGGCACCTGCAGTTCGC
CTGGCCCACGGCDNTDNTGGCTTCGACAAC
CACCGCCAACGACGGCDNTDNTGGCACCTGCAGTTCG
TCACTGGGCTCGCTGGAGTCGCTGAACAGCGAGGGTGCC
library A of WT-PAL
library B of WT-PAL
library C of WT-PAL
library A of mutant 1F8
library C of mutant 1F8
mutant Met16Ala/Leu17Phe using plasmid
of mutant 1B2
Met16Ala/Leu162Asn_F
Leu17Phe/Leu162Asn_F
CTGGCCCACGGCGCGCTCGGCTTCGACAACATC
CTGGCCCACGGCATGTTTGGCTTCGACAACATC
mutant Met16Ala/Leu162Asn using plasmid
of mutant 1B2
mutant Leu17Phe/Leu162Asn using plasmid
of mutant 1B2
spin miniprep kit (Qiagen) was used for plasmid preparation from
cells of P. aeruginosa PABST7.1 and E. coli DH5 R.
We conclude that ISM in the form of iterative CASTing is a
viable method to address the challenging problem of narrow
substrate scope and/or poor stereoselectivity of enzymes,
traditional limitations that are often encountered in white
biotechnology. When faced with the problem of how to group
individual CAST residues into ISM sites, we recommend two-
or three-residue sites in combination with a reduced amino acid
alphabet, because this maximizes the probability of cooperative
epistatic interactions along evolutionary pathways. Recently,
other groups likewise confronted with the numbers problem in
laboratory evolution5d have applied ISM in the form of CASTing
successfully for a variety of purposes, including enhancing and
inverting the enantioselectivity of the lipase from Candida
antarctica A,21 expanding substrate scope and enhancing
enantioselectivity of an arylmalonate decarboxylase,22 introduc-
ing promiscuous reactivity in an amidohydrolase with the
evolution of phosphotriesterase activity,23 inverting the stereo-
selectivity of the esterase from Burkholderia gladioli,24 turning
a cellobiose- into a lactose-phosphorylase,25 altering coenzyme
binding specificity of a xylose reductase (NADH versus
NADPH),26 and modulating lignin biosynthesis for better
utilization of plants in paper-making, biofuel production, and
agriculture.27 Unfortunately, in those studies, comparisons with
other methods of laboratory evolution were not made. Neverthe-
less, in all cases, only small libraries were needed. ISM is a
knowledge-driven approach which requires structural data, be
it for addressing stereoselectivity/substrate scope or thermosta-
bility. When such information is lacking, the limitation of ISM
becomes apparent.
Culture Conditions of Bacterial Strains. LB agar and 2x
Luria-Bertani (LB) broth supplemented with carbenicillin (100 µg/
mL) and tetracycline (50 µg/mL) were used to grow the
pUCPCL6AN containing P. aeruginosa PABST7.1, while LB agar
and LB broth supplemented with only carbenicillin (100 µg/mL)
were used to culture the pUCPCL6AN containing E. coli DH5R.
P. aeruginosa PABST7.1 can grow in the presence of tetracycline-
containing LB. The incubation temperature of P. aeruginosa
PABST7.1 and E. coli DH5R on LB agar plate was 37 °C, while
in broth it was 30 and 37 °C, respectively, with shaking at 250
rpm (for culture in 100-500 mL Erlenmeyer flask) or 800 rpm
(for deep-well plates).
Library Generation, Storage, Screening, and Sequencing.
Saturation mutagenesis libraries were created at sites A (Met16/
Leu17), B (Leu159/Leu162), and C (Leu231/Val232) using forward
primers (Table 4, see below) and a silent reverse primer (GGAGC-
CGGTGAGCGTGGGTCTCGCGG, 3750-3725 bp of pUCPCL6AN)
according to an improved PCR-based method for creating saturation
mutagenesis libraries.33 PCR reactions were performed with KOD
Hot Start polymerase (Novagen, USA) using these primers and WT-
PAL plasmid (pUCPCL6AN isolated from E. coli DH5R) as the
template DNA. PCR conditions used were 98 °C, 10 min; five
cycles of (96 °C, 2 min; 60 °C, 1 min; 72 °C, 4 min) for the
generation of megaprimer; 20 cycles of (98 °C, 1 min; 68 °C, 10
min); and final extension at 72 °C, 14 min. The PCR products were
treated with DpnI to digest the parent plasmid (2 h at 37 °C with
10 units of DpnI × two times) and purified with PCR purification
kit (QIAquick, Qiagen, Germany). The purified PCR product was
electroporated into freshly prepared P. aeruginosa electrocompetent
Finally, whereas CASTing5d,7,8 constitutes systematization of
saturation mutagenesis at sites aligning the binding pocket and
is therefore based on previous work regarding focused
libraries,4h,10 it is also possible to choose alternative randomiza-
tion sites rationally, e.g., at second-sphere residues28 or at remote
residues which can be expected to induce allosteric effects in
the absence of effectors.29 The utility of ISM in such cases has
yet to be explored.
(21) (a) Sandstro¨m, A. G.; Engstro¨m, K.; Nyhle´n, J.; Kasrayan, A.;
Ba¨ckvall, J.-E. Protein Eng., Des. Sel. 2009, 22, 413–420. (b)
Engstro¨m, K.; Nylen, J.; Sandstro¨m, A. G.; Ba¨ckvall, J.-E. J. Am.
Chem. Soc. 2010, 132, 7038–7042.
(22) Okrasa, K.; Levy, C.; Wilding, M.; Goodall, M.; Baudendistel, N.;
Hauer, B.; Leys, D.; Micklefield, J. Angew. Chem., Int. Ed. 2009, 48,
7691–7694.
(23) Hawwa, R.; Larsen, S. D.; Ratia, K.; Mesecar, A. D. J. Mol. Biol.
2009, 393, 36–57.
(24) Ivancic, M.; Valinger, G.; Gruber, K.; Schwab, H. J. Biotechnol. 2007,
129, 109–122.
Experimental Section
(25) De Groeve, M. R. M.; De Baere, M.; Hoflack, L.; Desmet, T.;
Vandamme, E. J.; Soetaert, W. Protein Eng., Des. Sel. 2009, 22, 393–
399.
General. All chiral p-nitrophenyl esters (1-11) were synthesized
in our laboratory by the standard procedure.8b
(26) Liang, L.; Zhang, J.; Lin, Z. Microb. Cell Fact. 2007, 6, 36.
(27) Bhuiya, M.-W.; Liu, C.-J. J. Biol. Chem. 2010, 285, 277–285.
(28) Reetz, M. T.; Wu, S. J. Am. Chem. Soc. 2009, 131, 15424–15432.
(29) Wu, S.; Acevedo, J. P.; Reetz, M. T. Proc. Nat. Acad. Sci. U.S.A 2010,
107, 2775–2780.
Plasmid Preparation. Designed pUCPCL6AN plasmid, contain-
ing wild-type LipA gene of Pseudomonas aeruginosa PAO1,13 is
a source of expression of lipase in expression host lipase-negative
P. aeruginosa PABST7.1. Plasmids isolated from P. aeruginosa
PABST7.1 are unmethylated30,31 and are transformed into chemical
competent cells of Escherichia coli DH5R32 for methylation.
Plasmid isolated from E. coli is used as template for PCR, which
can easily be digested by restriction endonuclease DpnI (New
England Biolabs) after PCR to remove the template DNA. QIAprep
(30) Stover, C. K.; et al. Nature (London, U.K.) 2000, 406, 959–964.
(31) Jacquel´ın, D. K.; Filiberti, A.; Argaran˜a, C. E.; Barra, J. L. Biochem.
J. 2005, 388, 879–887.
(32) Sambrook, J.; Russell, D. W. Molecular Cloning: A Laboratory
Manual; Cold Spring Habor Laboratory Press: New York, 2001.
(33) Sanchis, J.; et al. Appl. Microbiol. Biotechnol. 2008, 81, 387–397.
9
J. AM. CHEM. SOC. VOL. 132, NO. 26, 2010 9151

A R T I C L E S
Reetz et al.
cells, made by the method of Enderle and Farwell,34 and selected
for antibiotics resistance on LB agar carbenicillin (100 µg/mL) and
tetracycline (50 µg/mL) plates. Colonies appeared in 24-36 h and
were picked by toothpicks, transferred to preculture media, 900
mL of 2x LB broth medium containing carbenicillin (100 µg/mL)
and tetracycline (50 µg/mL) in a 96-square-well storage plate
(capacity, 2.2 mL; Thermo Scientific, UK), and incubated at 30 °C
under shaking of 800 rpm. Preculture and 70% glycerol were mixed
in 1:1 volume to make 200 µL in a 96-round-well glycerol stock
storage plate (NUNC; Thermo-Fisher Scientific, Denmark) and
stored at -80 °C. A fresh 900 µL of 2x LB broth was inoculated
from 25 µL preculture by replication program of liquid handling
robotics (TECAN, Switzerland), allowed to grow at 30 °C for 5 h,
induced by 0.1 mM isopropyl ꢀ-thiogalactopyranoside (IPTG), and
allowed to express lipase for 5 h more. The plates were centrifuged
at 4000 rpm for 60 min, and then the supernatant was used for
lipase assay by UV/vis screening method21 using (S)-1 and (R)-1
separately. Positive hits were sequenced by Eurofins Medigenomix
GmbH, Germany.
Site-Specific Mutagenesis. The above-described PCR protocol,
DpnI digestion, transformation, and screening methods were used
to generate mutants (Met16Ala/Leu17Phe, Met16Ala/Leu162Asn,
and Leu17Phe/Leu162Asn) of the deconvolution study using the
plasmid of mutant 1B2 (isolated from E. coli DH5alfa) as template
DNA and three different forward primers (Table 4) along with
reverse primer.
Kinetic Measurements. WT-PAL and mutants were purified
using the reported method,35 with minor modifications. The purity
of the enzyme was tested by SDS-PAGE, and the concentration
of the purified enzyme was estimated by the Bradford method (Bio-
Rad protein assay kit). To access the enantioselectivity of WT-
PAL and mutants, both enantiomers of p-nitrophenyl esters of 1,
4, 5, and 6 were used separately at various concentration ranges,
their hydrolysis was monitored at 25 °C in potassium phosphate
buffer (50 mM, pH 7.5, 1.0% Triton X-100), and the values of Km
and kcat were derived from the corresponding MichaelissMenten
plots. E-factors of WT-PAL and mutants were calculated as (kcat/
Km)S/(kcat/Km)R or (kcat/Km)R/(kcat/Km)S (Tables 2 and 3).
substrates 7-11 were performed by using a Chiralpak AD-3 chiral
column, 150 mm × 4.6 mm i.d. Conditions: mobile phase,
n-heptane/2-propanole/TFA ) 98:2:0.1; flow, 1 mL/min; temper-
ature profile, 298 K; detection, UV at 254 nm. GC analysis was
performed as described previously.8b
Molecular Modeling and MD Simulation. The molecular
modeling suite YASARA-structure version 9.5.420 was employed,
utilizing the AMBER0336 force field for the protein and the general
amber force field37 (GAFF) for the ligands throughout this study.
The partial charges for the ligands and the covalently bound
tetrahedral intermediate were computed using the AM1/BCC
procedure38 as implemented in YASARA-structure.20 The starting
point for molecular modeling was the crystal structure of WT-PAL13
(PDB 1ex9). The substrates (R)- and (S)-1 were introduced as
tetrahedral intermediates (oxyanions) covalently bound to catalyti-
cally active Ser82 γ-O.
All structures were solvated in a water box and neutralized.39
The structures were initially minimized using steepest descent and
the simulating annealing procedures. All mutations were introduced
sequentially using YASARA-structure. MD simulations were
carried out at 300 K over 2.5 ns in an NPT ensemble using PME.40
The first 500 ps was treated as the equilibration phase. All
simulations were performed individually for both enantiomers using
various starting geometries and rotamers for the mutated residues.
The 2.5 ns MD trajectories were sampled every 2.5 ps, resulting in
100 simulation frames per run, which were evaluated and addition-
ally minimized to derive statistical averages and properties of the
corresponding local minima. Finally, the energy of the bound
tetrahedral intermediate was calculated for every simulation frame,
and the distances for all hydrogen bonds stabilizing the oxyanion
were also computed.
Acknowledgment. We thank Marcus Hermes for the synthesis
of all chiral p-nitrophenyl esters (1, 4-11), J. Rosentreter for GC
analyses, and A. Deege for HPLC analyses. This work was financed
by the DFG (Schwerpunkt 1170) and the FCI.
Supporting Information Available: Medium-throughput as-
say11 for evaluating mutants as catalysts in the hydrolytic kinetic
resolution of rac-1; complete refs 30 and 33. This material is
available free of charge via the Internet at http://pubs.acs.org.
Reaction and Analysis Conditions for Determining the
Conversion and Enantioselectivity of Substrates rac-7-11. The
reactions were performed as follows: Potassium phosphate buffer
(750 µL, 50 mM, pH 7.5, 1.0% Triton X100) and substrates rac-
7-11 in acetonitrile (100 µL, 10 mg/mL) were added to the purified
enzymes (150 µL equiv to 11-15 µg of protein).The reaction
mixture was shaken at 800 rpm for 90 min at 30 °C and was then
acidified by adding 10% HCl (20 µL). The acidified solution was
extracted twice with 400 µL of methyl tert-butyl ether (MTBE).
The extracted organic phase was collected in a vial and analyzed
by HPLC. The analyses of the enantioselective conversion of
JA1030479
(36) Duan, Y.; Wu, C.; Chowdhury, S.; Lee, M. C.; Xiong, G.; Zhang,
W.; Yang, R.; Cieplak, P.; Luo, R.; Lee, T.; Caldwell, J.; Wang, J.;
Kollman, P. J. Comput. Chem. 2003, 24, 1999–2012.
(37) Wang, J.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D. A.
J. Comput. Chem. 2004, 25, 1157–1174.
(38) Jakalian, A.; Jack, D. B.; Bayly, C. I. J. Comput. Biol. 2002, 23, 1623–
1641.
(39) Krieger, E.; Nielsen, J. E.; Spronk, C. A. E. M.; Vriend, G. J. Mol.
Graph. Model. 2006, 25, 481–486.
(34) Enderle, P. J.; Farwell, M. A. BioTechniques 1998, 25, 954–958.
(35) Singh, S.; Banerjee, U. C. Process Biochem. (Oxford, U.K.) 2007,
42, 1063–1068.
(40) Essmann, U.; Perera, L.; Berkowitz, M. L. J. Chem. Phys. 1995, 103,
8577–8593.
9
9152 J. AM. CHEM. SOC. VOL. 132, NO. 26, 2010