Laboratory evolution of robust and enantioselective Baeyer-Villiger monooxygenases for asymmetric catalysis

Article abstract of DOI:10.1021/ja906212k

The Baeyer-Villiger Monooxygenase, Phenylacetone Monooxygenase (PAMO), recently discovered by Fraaije, Janssen, and co-workers, is unusually thermostable, which makes it a promising candidate for catalyzing enantioselective Baeyer-Villiger reactions in organic chemistry. Unfortunately, however, its substrate scope is very limited, reasonable reaction rates being observed essentially only with phenylacetone and similar linear phenyl-substituted analogs. Previous protein engineering attempts to broaden the range of substrate acceptance and to control enantioselectivity have been met with limited success, including rational design and directed evolution based on saturation mutagenesis with formation of focused mutant libraries, which may have to do with complex domain movements. In the present study, a new approach to laboratory evolution is described which has led to mutants showing unusually high activity and enantioselectivity in the oxidative kinetic resolution of a variety of 2-aryl and 2-alkylcyclohexanones which are not accepted by the wild-type (WT) PAMO and of a structurally very different bicyclic ketone. The new strategy exploits bioinformatics data derived from sequence alignment of eight different Baeyer-Villiger Monooxygenases, which in conjunction with the known X-ray structure of PAMO and induced fit docking suggests potential randomization sites, different from all previous approaches to focused library generation. Sites harboring highly conserved proline in a loop of the WT are targeted. The most active and enantioselective mutants retain the high thermostability of the parent WT PAMO. The success of the "proline" hypothesis in the present system calls for further testing in future laboratory evolution studies.

Full text of DOI:10.1021/ja906212k

Published on Web 10/06/2009
Laboratory Evolution of Robust and Enantioselective
Baeyer-Villiger Monooxygenases for Asymmetric Catalysis
Manfred T. Reetz* and Sheng Wu
Max-Planck-Institut fu¨r Kohlenforschung, Kaiser-Wilhelm-Platz 1, 45470 Mu¨lheim an der Ruhr, Germany
Received July 24, 2009; E-mail: reetz@mpi-muelheim.mpg.de
Abstract: The Baeyer-Villiger Monooxygenase, Phenylacetone Monooxygenase (PAMO), recently dis-
covered by Fraaije, Janssen, and co-workers, is unusually thermostable, which makes it a promising
candidate for catalyzing enantioselective Baeyer-Villiger reactions in organic chemistry. Unfortunately,
however, its substrate scope is very limited, reasonable reaction rates being observed essentially only
with phenylacetone and similar linear phenyl-substituted analogs. Previous protein engineering attempts
to broaden the range of substrate acceptance and to control enantioselectivity have been met with limited
success, including rational design and directed evolution based on saturation mutagenesis with formation
of focused mutant libraries, which may have to do with complex domain movements. In the present study,
a new approach to laboratory evolution is described which has led to mutants showing unusually high
activity and enantioselectivity in the oxidative kinetic resolution of a variety of 2-aryl and 2-alkylcyclohex-
anones which are not accepted by the wild-type (WT) PAMO and of a structurally very different bicyclic
ketone. The new strategy exploits bioinformatics data derived from sequence alignment of eight different
Baeyer-Villiger Monooxygenases, which in conjunction with the known X-ray structure of PAMO and induced
fit docking suggests potential randomization sites, different from all previous approaches to focused library
generation. Sites harboring highly conserved proline in a loop of the WT are targeted. The most active and
enantioselective mutants retain the high thermostability of the parent WT PAMO. The success of the “proline”
hypothesis in the present system calls for further testing in future laboratory evolution studies.
Introduction
appropriate monoxygenases, specifically Baeyer-Villiger Mo-
nooxygenases (BVMOs) in the presence of oxygen (air) as the
Catalytic asymmetric Baeyer-Villiger (BV) reactions of
ketones with formation of chiral esters or lactones constitute
synthetically attractive transformations in organic chemistry.1
This type of enantioselective C-C bond activation can be
catalyzed by chiral transition metal complexes1,2 or organo-
catalysts1-3 using stoichiometric amounts of H2O2, per-acids,
or alkylhydroperoxides. However, high levels of enantioselec-
tivity and activity are only possible with strained ketones such
as cyclobutanone derivatives. Enzymatic processes employing
oxidant, are an attractive alternative,1c,4 as originally shown by
Taschner using a cyclohexanone monooxygenase (CHMO).5
The catalytic profile of these enzymes in terms of activity,
substrate scope, and enantioselectivity is remarkable, yet notable
limitations persist to this day.1c,4 These monooxygenases are
flavin-dependent (Scheme 1),4,6 which means that cofactor
regeneration of expensive NADPH is necessary when using
isolated enzymes.7 Fusion proteins comprising BVMOs and
phosphite dehydrogenase7i (as an NADPH regeneration enzyme)
are likewise known.8 Thus far, biocatalytic BV reactions have
generally been performed with whole cells, biotechnological
processes that have been up-scaled in some cases,9 but seldom
used in normal organic chemistry laboratories.
(1) Reviews of Baeyer-Villiger reactions:(a) Krow, G. R. Org. React.
(N. Y.) 1993, 43, 251–798. (b) Renz, M.; Meunier, B. Eur. J. Org.
Chem. 1999, 737–750. (c) ten Brink, G.-J.; Arends, I. W. C. E.;
Sheldon, R. A. Chem. ReV. 2004, 104, 4105–4123.
(2) (a) Bolm, C.; Schlingloff, G.; Weickhardt, K. Angew. Chem. 1994,
106, 1944–1946. Angew. Chem., Int. Ed. Engl. 1994, 33, 1848-1849.
(b) Gusso, A.; Baccin, C.; Pinna, F.; Strukul, G. Organometallics 1994,
13, 3442–3451. (c) Uchida, T.; Katsuki, T. Tetrahedron Lett. 2001,
42, 6911–6914. (d) Ito, K.; Ishii, A.; Kuroda, T.; Katsuki, T. Synlett
2003, 643–646. (e) Strukul, G. Angew. Chem. 1998, 110, 1256–1267.
Angew. Chem., Int. Ed. 1998, 37, 1198-1209. (f) Frison, J.-C.; Palazzi,
C.; Bolm, C. Tetrahedron 2006, 62, 6700–6706. (g) Colladon, M.;
Scarso, A.; Strukul, G. Synlett 2006, 3515–3520.
(4) Reviews of Baeyer-Villiger Monooxygenases as catalysts in organic
chemistry6:(a) Kayser, M. M. Tetrahedron 2009, 65, 947–974. (b)
Stewart, J. D. Curr. Org. Chem. 1998, 2, 195–216. (c) Flitsch, S.;
Grogan, G. In Enzyme Catalysis in Organic Synthesis; Drauz, K.,
Waldmann, H., Eds.; Wiley-VCH: Weinheim, 2002; Vol. 2, pp 1202-
1245. (d) Alphand, V.; Carrea, G.; Wohlgemuth, R.; Furstoss, R.;
Woodley, J. M. Trends Biotechnol. 2003, 21, 318–323. (e) Brzos-
towicz, P. C.; Walters, D. M.; Thomas, S. M.; Nagarajan, V.; Rouvie`re,
P. E. Appl. EnViron. Microbiol. 2003, 69, 334–342. (f) Mihovilovic,
M. D. Curr. Org. Chem 2006, 10, 1265–1287. (g) Wohlgemuth, R.
Eng. Life Sci. 2006, 6, 577–583. (h) Kamerbeek, N. M.; Janssen, D. B.;
Van Berkel, W. J. H.; Fraaije, M. W. AdV. Synth. Catal. 2003, 345,
667–678.
(3) (a) Murahashi, S.-I.; Ono, S.; Imada, Y. Angew. Chem. 2002, 114,
2472–2474. Angew. Chem., Int. Ed. 2002, 41, 2366-2368. (b) Imada,
Y.; Iida, T.; Murahashi, S.-I.; Naota, T. Angew. Chem. 2005, 117,
1732–1734. Angew. Chem., Int. Ed. 2005, 44, 1704-1706. (c) Wang,
B.; Shen, Y.-M.; Shi, Y. J. Org. Chem. 2006, 71, 9519–9521. (d)
Peris, G.; Miller, S. J. Org. Lett. 2008, 10, 3049–3052. (e) Malkov,
A. V.; Friscourt, F.; Bell, M.; Swarbrick, M. E.; Koe`ovsky´, P. J. Org.
Chem. 2008, 73, 3996–4003. (f) Xu, S.; Wang, Z.; Zhang, X.; Zhang,
X.; Ding, K. Angew. Chem. 2008, 120, 2882–2885. Angew. Chem.,
Int. Ed. 2008, 47, 2840-2843.
(5) (a) Taschner, M. J.; Black, D. J. J. Am. Chem. Soc. 1988, 110, 6892–
6893. (b) Taschner, M. J.; Black, D. J.; Chen, Q.-Z. Tetrahedron:
Asymmetry 1993, 4, 1387–1390. (c) Taschner, M. J.; Peddada, L.
J. Chem. Soc., Chem. Commun. 1992, 1384–1385.
9
15424 J. AM. CHEM. SOC. 2009, 131, 15424–15432
10.1021/ja906212k CCC: $40.75  2009 American Chemical Society

Baeyer-Villiger Monooxygenases
A R T I C L E S
Scheme 1. Simplified Mechanism of BVMOs4,6
Inspite of the past successes,1c,4 the instability of BVMOs
constitutes a point of concern.10 When using the isolated
enzymes in conjunction with an enzymatic NADPH regeneration
system (or isolated fusion proteins) in place of whole cells, this
factor needs to be considered. Greater robustness would also
be advantageous when employing whole cells, either as such
or in immobilized form, because this would impart greater
robustness to the system under operating conditions. For these
reasons, the discovery by Fraaije, Janssen, et al. of a thermo-
stable BVMO, namely phenylacetone monooxygenase (PAMO),
was a major step forward.11 Moreover, the X-ray structure
analysis of PAMO (in the absence of NADP+), reported in 2004
by Malito, Mattevi, and co-workers,12 provided for the first time
a means to consider the mechanism of BVMOs in greater detail
than in the past.4 In particular, it was suggested that Arg337
stabilizes the Criegee intermediate by H-bonding,12 which helps
considerably when unravelling the details of the mechanism.
Indeed, we utilized this information in discussing the stereo-
chemical outcome of previous BVMO catalyzed reactions such
as those reported for CHMO and mutants thereof generated by
directed evolution,13 yet a detailed interpretation was not
possible. Most recently, Lau, Berghuis, and co-workers reported
the X-ray structure of a CHMO from an environmental
Rhodococcus strain not only bound with FAD but also with
NADP+,14 showing an open and a closed form in the absence
of a substrate or inhibitor. This is highly revealing, but a precise
interpretation of the stereochemical outcome of CHMO cata-
lyzed reactions remains a challenge, which may have to do with
the complex domain movements and the “sliding” cofactor.
Unfortunately, the substrate scope of the robust wild-type
(WT) PAMO is very narrow, reasonable activity being observed
essentially only with phenylacetone or similar linear phenyl-
substituted ketones.11 In previous protein engineering studies
of PAMO, we utilized rationally designed amino acid deletions15
and saturation mutagenesis using the Combinatorial Active-Site
Saturation Test (CAST)16 at loop positions 441-444 next to
the putative binding pocket (Figure 1), drastically reduced amino
acid alphabets being used on the basis of sequence alignment
of eight BVMOs.17 These positions, being part of a longer loop,
constitute a “bulge” in PAMO, absent in CHMO, and were
therefore considered to be promising sites for protein engineer-
ing. However, strategies utilizing deletions or randomization at
positions 441-444 expanded the substrate scope only minimally.
The basic result was the evolution of mutants that catalyze the
oxidative kinetic resolution of 2-phenylcyclohexanone with
(6) Isolation, characterization and/or mechanistic studies of CHMOs and
other BVMOs:(a) Turfitt, G. E. Biochem. J. 1948, 42, 376–383. (b)
Donoghue, N. A.; Norris, D. B.; Trudgill, P. W. Eur. J. Biochem.
1976, 63, 175–192. (c) Schwab, J. M.; Li, W. B.; Thomas, L. P. J. Am.
Chem. Soc. 1983, 105, 4800–4808. (d) Walsh, C. T.; Chen, Y.-C. J.
Angew. Chem. 1988, 100, 342–352. Angew. Chem., Int. Ed. Engl. 1988,
27, 333-343. (e) Brzostowicz, P. C.; Walters, D. M.; Thomas, S. M.;
Nagarajan, V.; Rouvie`re, P. E. Appl. EnViron. Microbiol. 2003, 69,
334–342. (f) Sheng, D.; Ballou, D. P.; Massey, V. Biochemistry 2001,
40, 11156–11167.
(7) (a) Zhao, H.; van der Donk, W. A. Curr. Opin. Biotechnol. 2003, 14,
583–589. (b) van der Donk, W. A.; Zhao, H. Curr. Opin. Biotechnol.
2003, 14, 421–426. (c) de Gonzalo, G.; Ottolina, G.; Carrea, G.;
Fraaije, M. W. Chem. Commun. (Cambridge, U. K.) 2005, 3724–3726.
(d) Hollmann, F.; Taglieber, A.; Schulz, F.; Reetz, M. T. Angew. Chem.
2007, 119, 2961–2964. Angew. Chem., Int. Ed. 2007, 46, 2903-2906.
(e) Hollmann, F.; Hofstetter, K.; Schmid, A. Trends Biotechnol. 2006,
24, 163–171. (f) Urlacher, V. B.; Schmid, R. D. Curr. Opin. Chem.
Biol. 2006, 10, 156–161. (g) Zambianchi, F.; Pasta, P.; Carrea, G.;
Colonna, S.; Gaggero, N.; Woodley, J. M. Biotechnol. Bioeng. 2002,
78, 489–496. (h) Lee, W.-H.; Park, J.-B.; Park, K.; Kim, M.-D.; Seo,
J.-H. Appl. Microbiol. Biotechnol. 2007, 76, 329–338. (i) Woodyer,
R.; van der Donk, W. A.; Zhao, H. Biochemistry 2003, 42, 11604–
11614.
(13) (a) Reetz, M. T.; Brunner, B.; Schneider, T.; Schulz, F.; Clouthier,
C. M.; Kayser, M. M. Angew. Chem. 2004, 116, 4167–4170. Angew.
Chem., Int. Ed. 2004, 43, 4075-4078. See also directed evolution for
increasing the enantioselectivity of the BVMO from Pseudomonas
fluorescens: (b) Kirschner, A.; Bornscheuer, U. T. Appl. Microbiol.
Biotechnol. 2008, 81, 465–472.
(8) Torres Pazmin˜o, D. E.; Snajdrova, R.; Baas, B.-J.; Ghobrial, M.;
Mihovilovic, M. D.; Fraaije, M. W. Angew. Chem. 2009, 120, 2307–
2310. Angew. Chem., Int. Ed. 2008, 47, 2275-2278.
(9) (a) Hilker, I.; Wohlgemuth, R.; Alphand, V.; Furstoss, R. Biotechnol.,
Bioeng. 2005, 92, 702–710. (b) Baldwin, C. V. F.; Wohlgemuth, R.;
Woodley, J. M. Org. Process Res. DeV. 2008, 12, 660–665.
(10) Walton, A. Z.; Stewart, J. D. Biotechnol. Prog. 2004, 20, 403–411.
(11) (a) Fraaije, M. W.; Wu, J.; Heuts, D. P. H. M.; van Hellemond, E. W.;
Spelberg, J. H. L.; Janssen, D. B. Appl. Microbiol. Biotechnol. 2005,
66, 393–400. (b) Zambianchi, F.; Fraaije, M. W.; Carrea, G.; de
Gonzalo, G.; Rodr´ıguez, C.; Gotor, V.; Ottolina, G. AdV. Synth. Catal.
2007, 349, 1327–1331. (c) Rodr´ıgues, C.; de Gonzalo, G.; Torres
Pazmin˜o, D. E.; Fraaije, M. W.; Gotor, V. Tetrahedron: Asymmetry
2008, 19, 197–203.
(14) Mirza, I. A.; Yachnin, B. J.; Wang, S.; Grosse, S.; Bergeron, H.; Imura,
A.; Iwaki, H.; Hasegawa, Y.; Lau, P. C. K.; Berghuis, A. M. J. Am.
Chem. Soc. 2009, 131, 8848–8854.
(15) (a) Bocola, M.; Schulz, F.; Leca, F.; Vogel, A.; Fraaije, M. W.; Reetz,
M. T. AdV. Synth. Catal. 2005, 347, 979–986. (b) Schulz, F.; Leca,
F.; Hollmann, F.; Reetz, M. T. Beilstein J. Org. Chem. 2005, 1, 10.
(16) (a) Reetz, M. T.; Bocola, M.; Carballeira, J. D.; Zha, D.; Vogel, A.
Angew. Chem. 2005, 117, 4264–4268. Angew. Chem., Int. Ed. 2005,
44, 4192-4196. (b) Reetz, M. T.; Kahakeaw, D.; Sanchis, J. Mol.
BioSyst. 2009, 5, 115–122.
(12) Malito, E.; Alfieri, A.; Fraaije, M. W.; Mattevi, A. Proc. Natl. Acad.
Sci. U.S.A. 2004, 101, 13157–13162.
(17) Reetz, M. T.; Wu, S. Chem. Commun. (Cambridge, U. K.) 2008, 5499–
5501.
9
J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009 15425

A R T I C L E S
Reetz and Wu
away (second sphere residue). Since we succeeded in evolving
pronounced activity and high enantioselectivity in BV reactions
using very small mutant libraries (and thus minimal experimental
effort), while maintaining high thermostability inherent in WT
PAMO, our study constitutes a significant step forward in this
field of catalytic oxidation. Moreover, due to their high
thermostability, it is possible to use the BVMO mutants in
isolated form in combination with a secondary alcohol dehy-
drogenase (2°ADH) as an NADPH regeneration system with
isopropanol serving as the stoichiometric reductant, if so desired.
Results and Discussion
Exploratory Experiments Based on the Conventional
Approach to Saturation Mutagenesis. In directed evolution of
enantioselective enzymes based on saturation mutagenesis with
formation of focused libraries,20,21 a strategy first reported in
2001,22 it is necessary to choose randomization sites next to
the binding pocket, preferably in a systematic way as in
CASTing.16,20,21 If the X-ray structure (or homology model) is
available, preferably with a bound substrate or inhibitor, the
choice is generally straightforward. However, if no structural
data of substrate or inhibitor bound enzyme is accessible, the
choice may be difficult to make, certainly when studying
fluxional enzymes known to undergo substantial internal move-
ments (in the extreme case allosteric motion). BVMOs appear
to belong to this category.1c,4 As noted above, loop optimization
by randomizing those positions in PAMO which appear to align
the putative binding pocket of the ligand-free enzyme, positions
441-444 (Figure 1), led only to partial success.17 In hope of
obtaining a more realistic “picture” of the putative binding
pocket, we performed induced fit docking experiments of WT
PAMO utilizing the X-ray data10 and phenylacetone as the
substrate. The result indicates that the substrate and catalytically
important Arg337 are both located on the re-side of the flavin
(FAD), substantiating expectations (Figure 2). The distance
between the oxygen atom of phenylacetone and the C4a atom
of FAD is about 6.6 Å, which means that the intermediate
deprotonated flavin-hydroperoxide (Scheme 1) appears to be
spatially close enough to the carbonyl function of the substrate
for smooth BV reaction to occur.
Close inspection of the docking result reveals the presence
of 17 amino acids within the range of about 5 Å around the
substrate, namely Gln152, Leu153, Ser196, Lys336, Arg337,
Ile339, Gly388, Phe389, Asp390, Ala391, Leu392, Ala442,
Met446, Ser500, Trp501, Met515 and Leu516 (Figure S1/
Supporting Information), which are all potential sites for
CASTing. At this stage several strategic decisions appeared
logical, e.g., the generation of 17 individual saturation mu-
tagenesis libraries, or libraries resulting from grouping several
amino acid positions together. Based on Figure 1 (and Figure
S1/Supporting Information) and previous experience regarding
loop optimization,17 we proceeded by choosing five randomiza-
tion sites (Figure S2/Supporting Information), specifically 441/
442/443/444 (site A), 445/446 (B), 152/153 (C), 338/339 (D)
Figure 1. Excerpt of WT PAMO structure based on the X-ray data in the
absence of a substrate or inhibitor, featuring the oxidized flavin (FAD),
Arg337 presumably stabilizing the Criegee intermediate, and the 441-444
loop segment (“bulge”) directly aligning the putative binding pocket (chosen
in previous studies for protein engineering).15,17
reasonable reaction rate and high enantioselectivity (E ) 30-70
in favor of the respective R-lactone), compared to the WT
PAMO which shows a minimal rate and thus low conversion
(<10% after extended reaction times) as well as stereorandom
behavior (E ≈ 1).17 Later work showed that 2-alkyl substituted
cyclohexanone derivatives such as 2-methyl- or 2-ethylcyclo-
hexanone are not at all accepted by these mutants nor by WT
PAMO.18 In another study, rationally designed PAMO mutant
Met466Gly proved to be a useful catalyst for enantioselective
sulfoxidation of some prochiral thio-ethers, but broadening the
substrate scope of BV reactions proved to be difficult.11b Thus,
the need to perform further research was obvious.
Directed evolution of enzymes has emerged as a powerful
method to engineer such catalytic properties as thermostability,
robustness toward hostile solvents as well as substrate scope
and enantioselectivity, error-prone PCR, DNA shuffling and
saturation mutagenesis being the most popular gene mutagenesis
techniques.19,20 When targeting substrate scope and/or enanti-
oselectivity using saturation mutagenesis, amino acid positions
aligning the binding pocket are generally chosen as randomiza-
tion sites with formation of so-called focused libraries.15,16,19-23
Here we describe an unconventional strategy for focused library
production which involves randomization at sites quite different
from those based on the conventional procedure, specifically
by focusing on a residue which appears to be spatially further
(18) Wu, S.; Reetz, M. T. Unpublished results, 2008.
(19) Reviews of directed evolution:(a) Lutz, S.; Bornscheuer, U. T. Protein
Engineering Handbook, Vol. 1-2; Wiley-VCH: Weinheim, Germany,
2009. (b) Arndt, K. M.; Mu¨ller, K. M. Protein Engineering Protocols
(Methods in Molecular Biology), Vol. 352; Humana Press: Totowa,
NJ, 2007. (c) Arnold, F. H.; Georgiou, G. Methods and Molecular
Biology, Vol. 230; Humana Press: Totowa, NJ, 2003. (d) Ja¨ckel, C.;
Kast, P.; Hilvert, D. Annu. ReV. Biophys. Biomol. Struct. 2008, 37,
153–173. (e) Brakmann, S.; Schwienhorst, A. EVolutionary Methods
in Biotechnology - CleVer Tricks for Directed EVolution, Wiley-VCH:
Weinheim, Germany, 2004. (f) Hibbert, E. G.; Baganz, F.; Hailes,
H. C.; Ward, J. M.; Lye, G. J.; Woodley, J. M.; Dalby, P. A. Biomol.
Eng. 2005, 22, 11–19. (g) Rubin-Pitel, S. B.; Zhao, H. Comb. Chem.
High Throughput Screening 2006, 9, 247–257. (h) Kaur, J.; Sharma,
R. Crit. ReV. Biotechnol. 2006, 26, 165–199. (i) Bershtein, S.; Tawfik,
D. S. Curr. Opin. Chem. Biol. 2008, 12, 151–158. (j) Reetz, M. T. In
Asymmetric Organic Synthesis with Enzymes; Gotor, V., Alfonso, I.,
Garc´ıa-Urdiales, E., Eds.; Wiley-VCH: Weinheim, Germany, 2008;
pp 21-63.
(21) (a) Reetz, M. T.; Carballeira, J. D. Nat. Protoc 2007, 2, 891–903. See
also discussion regarding close and remote mutations:19j (b) Park, S.;
Morley, K. L.; Horsman, G. P.; Holmquist, M.; Hult, K.; Kazlauskas,
R. J. Chem. Biol. 2005, 12, 45–54. (c) Horsman, G. P.; Liu, A. M. F.;
Henke, E.; Bornscheuer, U. T.; Kazlauskas, R. J. Chem.sEur. J. 2003,
9, 1933–1939.
(20) Reviews of directed evolution of enantioselective enzymes:19j (a) Reetz,
M. T. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 5716–5722. (b) Reetz,
M. T. In Protein Engineering Handbook; Lutz, S., Bornscheuer, U. T.,
Eds.; Wiley-VCH: Weinheim, Germany, 2009; Vol. 2.
(22) Reetz, M. T.; Wilensek, S.; Zha, D.; Ja¨ger, K.-E. Angew. Chem. 2001,
113, 3701–3703. Angew. Chem., Int. Ed. 2001, 40, 3589-3591.
(23) Reetz, M. T.; Kahakeaw, D.; Lohmer, R. ChemBioChem 2008, 9,
1797–1804.
9
15426 J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009

Baeyer-Villiger Monooxygenases
A R T I C L E S
Scheme 2. Oxidative Kinetic Resolution of rac-1 Catalyzed by
PAMO Mutantsa
a Note that the R/S-designation may change in accord with the CIP
priority of the R-substitutent of the ketone and lactone.
Figure 2. Induced fit docking of phenylacetone in the X-ray structure of
WT PAMO10 featuring this substrate (green sticks) in the putative binding
pocket and FAD (magenta sticks).
and 334/335 (E). In order to reduce the screening effort, which
is the bottleneck in any directed evolution study,19,20 we chose
for the different positions reduced amino acid alphabets17,23 by
applying the respective codon degeneracy (Table S1/Supporting
Information). The degree of oversampling corresponding to 95%
coverage, calculated by the computer aid CASTER21a,24 (based
on known algorithms assuming the absence of amino acid bias)25
was found to be within a reasonable range which can be handled
by the screening system used in the present study.
We envisioned the laboratory evolution of PAMO mutants
which would not just accept 2-phenylcyclohexanone (1a) in a
fast reaction, but essentially any 2-substituted cyclohexanone
derivative, for example, ketones 1b-1l, which would broaden
the substrate scope of robust PAMO enzymes considerably. As
the model reaction for mutagenesis/screening, the BV transfor-
mation of ketone 1a was chosen. We reasoned that if mutants
displaying substantial activity were to be found in any of the
libraries, then they could be tested for the other more challenging
ketones as well. The hope was that active and stereoselective
mutants can be identified in this way without performing
additional mutagenesis/screening experiments.
Unfortunately, no hits were found in libraries B, C, D, or E,
while library A contained, not unexpectedly, several improved
mutants that were already known to us from our original
(441-444)-loop optimization, which however do not accept
2-alkylcyclohexanone derivatives.17,18 Currently, it is difficult
to explain these negative results. Rather than continuing the
search for sensitive CAST sites or using larger amino acid
alphabets for saturation mutagenesis (which could well be
successful), we turned to a different strategy.
Figure 3. Illustration of the putative binding pocket of PAMO based on
the induced fit docking model. Phenylacetone and the loop segment
441-444 are shown in cyan and red, respectively, whereas Pro437 and
Pro440 are pictured in blue.
Bioinformatics Approach to Saturation Mutagenesis. In our
previous study17 regarding partial loop optimization at positions
441-444, which according to the ligand-free X-ray structure
in combination with the induced fit docking model all have
direct contact with the binding pocket (Figure 3), sequence
alignment of WT PAMO and seven other BVMOs was used as
a guide for choosing reduced amino alphabets in saturation
mutagenesis experiments (Figure 4). In contrast to this approach,
in the present study we focus on position 440, which according
to Figure 3 seems to be a “second sphere” residue,19a that is,
the first residue not in apparent direct contact with the binding
pocket. Due to the dynamics of PAMO, which may be expected
from this enzyme, and the lack of structural data of the ligand-
bound form, we realized that this is an uncertain issue. For this
very reason we were curious to test position 440, since situations
of this kind may arise in diverse future studies of other enzymes
as well. It also struck us that proline is highly conserved at this
position in BVMOs (Figure 4), an amino acid that is known to
impart a certain degree of rigidity in proteins.26 Increasing
flexibility by substituting proline for another amino acid could
(24) The CASTER computer aid for designing saturation mutagenesis
libraries is available free of charge from the authors homepage: http://
www.mpi-muelheim.mpg.de/mpikofo_home.html.
(25) (a) Rui, L.; Kwon, Y. M.; Fishman, A.; Reardon, K. F.; Wood, T. K.
Appl. EnViron. Microbiol. 2004, 70, 3246–3252. (b) Bosley, A. D.;
Ostermeier, M. Biomol. Eng. 2005, 22, 57–61. (c) Mena, M. A.;
Daugherty, P. S. Protein Eng., Des. Sel. 2005, 18, 559–561. (d)
Denault, M. ; Pelletier, J. N. In Protein Engineering Protocols; Vol.
352; Arndt, K. M., Mu¨ller, K. M., Eds; Humana Press: Totowa, NJ,
2007: pp 127-154.
(26) Fersht, A. Structure and Mechanism in Protein Science; W. H. Freeman
and Co.: New York, 2000.
9
J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009 15427

A R T I C L E S
Reetz and Wu
Table 1. Kinetic Resolution of rac-1e using PAMO Mutants
Derived from a Saturation Mutagenesis Library (Entries 1-5) or
Site-Specific Mutagenesis (Entries 6-7) Both at Position 440
entry
mutant
E-value
Km (mM)
kcat (s-1
)
kcat/Km (M-1 s-1
)
1
2
3
4
5
6
7
Pro440Phe
Pro440Leu
Pro440Ile
Pro440Asn
Pro440His
Pro440Trp
Pro440Tyr
26
>200
>200
>200
34
0.89
1.6
2.7
2.2
1.0
1.3
1.9
1.2
1300
450
240
680
830
1000
580
0.72
0.66
1.5
0.83
1.3
>200
95
1.1
Table 2. E-values in the oxidative kinetic resolution of substrates
rac-1a-1d and rac-1f-1la
Figure 4. Alignment of eight BVMO protein sequences highlighting loop
segment 441-444 (red box) and positions 437 and 440 (arrows) indicating
the new sites for saturation mutagenesis. Code: Phenylacetone monooxy-
genase (PAMO) from Thermobifida fusca (1w4x), steroid monooxygenase
(STMO) from Rhodococcus rhodochrous (BAA24454), cyclohexanone
monooxygenase (CHMO) from Acinetobacter sp. NCIMB 9871 (BAA86293),
cyclohexanone monooxygenase 1 (CHMO1) from BreVibacterium sp. HCU
(AAG01289), cyclohexanone monooxygenase 2 (CHMO2) from BreVibac-
terium sp. HCU (AAG01290), cyclohexanone monooxygenase 3 (CHMO3)
from Acinetobacter sp. (P12015), cyclopentanone monooxygenase (CPMO)
from Comamonas testosterone (CAD10798), and cyclododecanone mo-
nooxygenase (CDMO) from Rhodococcus ruber (AAL14233).
entry
mutant
rac-1a
rac-1b
rac-1c
rac-1d
rac-1f
rac-1g
1
2
3
4
5
6
Pro440Leu
Pro440Ile
Pro440Asn
Pro440His
Pro440Tyr
Pro440Trp
42
25
18
12
11
3.4
63
ND
18
11
ND
ND
-
-
-
-
4
>200
>200
>200
>200
ND
>200
>200
>200
>200
ND
>200
>200
>200
33
ND
ND
30
ND
ND
entry
mutant
rac-1h
rac-1i
rac-1j
rac-1k
rac-1l
7
8
9
10
11
12
Pro440Leu
Pro440Ile
Pro440Asn
Pro440His
Pro440Tyr
Pro440Trp
>200
34
>200
37
ND
ND
-
-
-
>200
69
22
84
119
45
-
-
>200
20
49
146
>200
40
well have an impact on the spatial position of the neighboring
residues and perhaps of the entire loop by inducing some kind
of rearrangement. Along this line of reasoning, a second nearby
position also harboring proline, namely Pro437, was likewise
considered for saturation mutagenesis.
-
>200
>200
>200
9
ND
19
a ND: not determined, -: low conversion.
Two focused libraries were produced by saturation mutagen-
esis at positions 437 and 440, respectively, using NNK
degeneracyencodingall20proteinogenicaminoacids(QuikChange/
Stratagene).27 This time we considered for screening purposes
a particularly “difficult” substrate, rac-2-ethylcyclohexanone
(rac-1e), as the model compound in oxidative kinetic resolution.
As already mentioned, this compound is not at all transformed
by WT PAMO, even after an extended reaction duration of
several days.18 Each library was screened by evaluating about
200 transformants. This was achieved by an activity test,
followed by enzyme isolation and kinetic resolution of rac-1e
in conjunction with the known robust 2°ADH from Thermoa-
naerobacter ethanolicus as the NADPH regeneration enzyme28
using isopropanol as the reductant. Assuming the absence of
amino acid bias, this degree of oversampling should be more
than sufficient for at least 95% coverage.21a,24
Whereas the focused library corresponding to position 437
failed to contain any active clones, saturation mutagenesis at
position 440 resulted in the identification of a number of active
and enantioselective mutants. Ten hits were identified and
sequenced, five of them showing particularly high activity (the
other five were no longer considered for further study, namely
Pro440Cys, Pro440Thr, Pro440Lys, Pro440Arg, and Pro440Gly).
The selectivity factors E of the five active mutants as well as
the results of kinetic studies are summarized in Table 1 (entries
1-5). It can be seen that unusually high enzyme activity and
enantioselectivity were achieved by this simple and rapid
experiment.
studies,19-21 although very simple to apply, occurs with some
degree of amino acid bias. This may mean that some of the
expected amino acid substitutions leading to potential hits are
not observed in the screening process even though oversampling
has been performed.15b,21a,23 In directed evolution studies using
saturation mutagenesis this is tolerable, since improved mutants
are generally found.19 Nevertheless, we wanted to check this,
which in the present system is an easy endeavor because it
involves a single amino acid position. All of the “missing”
mutants were prepared by conventional site-specific mutagenesis
and tested in the model reaction. Of the nine relevant cases,
namelyPro440Ala,Pro440Val,Pro440Ser,Pro440Asp,Pro440Glu,
Pro440Gln, Pro440Tyr, Pro440Trp and Pro440Met, two new
mutants proved to be active, namely Pro440Tyr and Pro440Trp
(Table 1, entries 6-7). Thus, in the present system conventional
saturation mutagenesis coupled with theoretically sufficient
oversampling assuming the absence of amino acid bias results
in 80% hit-identification. This is the first time that such
information has been generated, which may be of interest in
future directed evolution studies based on saturation mutagen-
esis. Interestingly, it can be seen that very different types of
amino acids replacing Pro440 lead to active and enantioselective
catalysts, which means that the factors contributing to catalyst
improvement, on a molecular level, must be quite different.
Although no (bio)catalyst can ever be “universal”, organic
chemists seek to develop catalysts which function well for a
range of different substrates. Therefore, without performing any
more mutagenesis experiments, we were keen to test the best
mutants, engineered for a single substrate rac-1e, as catalysts
in the kinetic resolution of a variety of structurally different
racemic ketones: rac-1a-1d, rac-1f-1l and the bicyclic
substrate rac-3. In these experiments, isolated enzyme mutants
were again employed in conjunction with the in vitro use of
2°ADH for NADPH regeneration. As in the case of the model
We and others have previously pointed out that conventional
saturation mutagenesis27 of the type used here and in most other
(27) Hogrefe, H. H.; Cline, J.; Youngblood, G. L.; Allen, R. M. Biotech-
niques 2002, 33, 1158–1165.
(28) (a) Zheng, C.; Pham, V. T.; Phillips, R. S. Bioorg. Med. Chem. Lett.
1992, 2, 619–622. (b) Burdette, D. S.; Tchernajencko, V.; Zeikus, J. G.
Enzyme Microb. Technol. 2000, 27, 11–18.
9
15428 J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009

Baeyer-Villiger Monooxygenases
A R T I C L E S
Table 4. Regio- and Enantioselectivity of PAMO Mutants As
Table 3. Kinetic Studies of Mutant Pro440Phe as a Catalyst in the
Kinetic Resolution of Substrates rac-1a-1l
Catalysts in the BV Reaction of rac-3
entry
substrate
E-value
Km (mM)
kcat (s-1
)
kcat/Km (M-1 s-1
)
“normal” lactone 4
“abnormal” lactone 5
1
2
3
4
5
6
7
8
9
rac-1a
rac-1b
rac-1c
rac-1d
rac-1e
rac-1f
rac-1g
rac-1h
rac-1i
rac-1j
rac-1k
rac-1l
12
7
117
95
26
145
39
102
>200
145
48
0.81
0.26
0.13
3.1
0.89
0.70
0.34
1.1
0.73
0.26
0.16
3.8
1.3
1600
3000
1500
520
1300
2400
3200
1500
360
mutant
conv. (%)a ee (%) abs. conf.
ee (%)
abs. conf.
ratio(4/5)
0.77
0.19
1.6
1.2
1.7
1.1
1.7
0.26
0.41
0.33
1.5
WT PAMO
Pro440Leu
Pro440Ile
Pro440Phe
Pro440Tyr
Pro440Trp
Pro440Asn
Pro440His
31
85
89
99
89
86
85
82
86
92
92
94
92
87
92
92
(1S,5R)
(1S,5R)
(1S,5R)
(1S,5R)
(1S,5R)
(1S,5R)
(1S,5R)
(1S,5R)
63
99
93
98
97
97
97
95
(1S,5R)
(1R,5S)
(1R,5S)
(1R,5S)
(1R,5S)
(1R,5S)
(1R,5S)
(1R,5S)
72:28
52:48
49:51
48:52
46:54
48:52
49:51
47:53
10
11
12
1600
2100
390
a Reaction time: 10 h.
91
note that WT PAMO favors the formation of 4 as the (1S,5R)-
enantiomer and also produces some of 5 as the (1S,5R)-
enantiomer, whereas the mutants lead to a reversal of enanti-
oselectivity in the case of regioisomer 5 (Table 4).
compound rac-1e, the other substrates are likewise not accepted
by WT PAMO, with the exception of the 2-phenyl derivatives
rac-1a-1b which show minimal conversion under extended
reaction times but poor enantioselectivity (E ) 1-2). Gratify-
ingly, many of the mutants are indeed excellent catalysts in the
oxidative kinetic resolution of all of the tested 2-substituted
cyclohexanone derivatives (Table 2). In the majority of cases
the standard reaction conditions ensure the desired conversion
of about 50% within 10 h. Table S5 in the Supporting In-
formation shows the ee-values of all substrates and products at
defined conversion.
Variant Pro440Phe, found to be most active for rac-1e, was
then used as a catalyst in further oxidative kinetic resolutions
involving all substrates. The results summarized in Table 3 point
to some remarkable effects. Mutant Pro440Phe shows very high
activity for almost all substrates. In the case of 2-alkyl substrates
(Table 3, entries 4-7), the catalytic efficiency kcat/Km correlates
well with the chain length. The longer the side-chain, the higher
the catalytic efficiency. We explain this by noting that the
affinity (Km) increases as the length of the alkyl chain increases,
while the activity (kcat) remains in the same range. This is an
interesting phenomenon, especially when viewing the respective
enantioselectivity, which shows an irregular pattern. It is
particularly striking and of synthetic significance that even
compounds having fairly bulky substituents such as cyclohexyl
or benzyl react rapidly (Table 3, entries 10 and 11). In the case
of 2-phenylcyclohexanone (rac-1a), a notably high value of kcat/
Km is observed for mutant Pro440Phe, a result that needs to be
compared to the published results of a kinetic study using WT
PAMO as the catalyst in the kinetic resolution of the same
substrate (kcat ) 0.06 s-1; Km ) 5; kcat/Km ) 10).9 This means
a rate acceleration by a factor of 1.6 × 102. In the case of the
majority of substrates not at all accepted by WT PAMO, it was
not possible to quantify the rate acceleration, but it must be
substantially higher.
Thermostability of Evolved PAMO Mutants. While perform-
ing the above experiments, we noticed qualitatively that the
mutants appear to retain high thermostability characteristic of
WT PAMO. In order to check quantitatively whether the
thermostability of the PAMO variants has suffered to any
notable degree as a consequence of mutagenesis, the T5060 values
were measured for the best mutants. This is the temperature at
which a heat treatment for one hour reduces enzyme activity
by 50%.29 The measured T50 value of WT PAMO (58.2 °C)
60
and those of the best mutants Pro440Phe (56.2 °C), Pro440Tyr
(56.7 °C), Pro440Trp (56.3 °C) Pro440His (55.8 °C), Pro440Leu
(56.0 °C), Pro440Ile (56.0 °C) and Pro440Asn (56.3 °C) show
that the mutational changes do not impair the unique high
thermostability of the enzyme to any significant degree (Figure
5). This was also checked by monitoring the residual activity
as a function of time, the BV-reaction of phenylacetone serving
as the transformation. No loss of activity was observed when
the mutant enzymes were kept at 50 °C for 40 h (data not
shown). Again, it can be seen that thermostability is not
significantly impaired by the mutational changes.
Conclusions
In further work, we tested some of the mutants in the regio-
and stereoselective kinetic resolution of a structurally very
different substrate, namely rac-3 with formation of the two
regioisomers 4 and 5, the so-called “normal” and “abnormal”
lactones, respectively.4 Previous studies regarding the use of
other BVMOs in whole cells have in some cases provided high
enantioselectivities (>95% ee) and varying degrees of regiose-
lectivity.4 The WT PAMO shows a relatively low rate with this
substrate, the “normal” lactone 4 being preferred at moderate
enantioselectivity.15 Here again it was of interest to test some
of the mutants evolved for substrate rac-1e without resorting
to any new mutagenesis experiments (Table 4). Importantly,
enzyme activity has increased in all cases. It is interesting to
We have devised a new strategy in directed evolution in order
to construct a robust experimental platform for asymmetric
Baeyer-Villiger reactions based on the thermostable monooxy-
genase PAMO. Since WT PAMO accepts only phenylacetone
and structurally similar linear phenyl-substituted ketones with
reasonable rates, the goal was to expand the substrate scope,
i.e., to increase the reaction rate significantly, and to reach high
(29) Reviews of enzyme thermostabilization by protein engineering:(a)
Eijsink, V. G. H.; Gåseidnes, S.; Borchert, T. V.; van den Burg, B.
´
Biomol. Eng 2005, 22, 21–30. (b) O’Fa´ga´in, C. Enzyme Microb.
Technol. 2003, 33, 137–149. (c) Bommarius, A. S.; Broering, J. M.
Biocatal. Biotransform. 2005, 23, 125–139.
9
J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009 15429

A R T I C L E S
Reetz and Wu
the shape, positioning and dynamics of a loop and therefore
the environment at the nearby binding pocket and/ influence
complex domain movements. However, presently we do not for
certain that such a mutation does in fact increase flexibility.
The actual laboratory evolution was performed with a single
substrate, rac-2-ethylcyclohexanone (1e), by applying saturation
mutagenesis at two sites, which required the screening of only
200 transformants in each case. The mini-library at one of the
single residue sites resulted in half a dozen hits showing high
activity and enantioselectivity. Screening a total of only 400
transformants in the present project is an extremely small
number in laboratory evolution, the usual libraries employing
other techniques generally comprising 103-106 mutants.19 These
biocatalysts, differing by only one point mutation with respect
to the WT PAMO, were subsequently tested with 12 other
structurally different ketones without performing any further
mutagenesis/screening experiments, which led to the discovery
that the mutants not only retain their thermostability, but also
have a respectrably broad substrate acceptance and excellent
stereoselectivity. Generalizing the “proline” hypothesis,30 es-
pecially in loops of enzymes that undergo (unidentified) internal
motions, as well as identifying the reasons for enhanced activity
and enantioselectivity in the system described herein are the
subjects of future studies.
Figure 5. Thermostability of WT-PAMO and its mutants as measured by
the relative residual activity as a function of temperature, the BV-
transformation of phenylacetone serving as the model reaction.
enantioselectivity, without compromising thermostability. This
was indeed achieved. Mutants were evolved which show
unusually high activity and enantioselectivity in the oxidative
kinetic resolution of a variety of structurally different 2-sub-
stituted aryl- and alkylcyclohexanone derivatives and of a
structurally unrelated bicyclic ketone. High thermostability is
a notable advantage when using BVMOs in whole-cell setups,
but also when choosing to apply them in isolated enzyme form.
Although the former is generally preferred in industrial applica-
tions,4 we have demonstrated that the latter in vitro process poses
no problems, a secondary alcohol dehydrogenase in combination
with the reductant isopropanol serving as the NADPH regenera-
tion system.15b Thus, the present study constitutes a further step
to more economically and ecologically viable BVMO-mediated
stereoselective Baeyer-Villiger reactions in organic chemistry.9
Just as importantly, a new strategy in laboratory evolution
was proposed and implemented experimentally. It goes beyond
the traditional generation of focused libraries at sites aligning
the binding pocket,16,17,19-23 and may therefore be useful in
other future directed evolution studies. Whenever a given
enzyme under study is likely to undergo notable conformational
changes due to inherent flexibility, including domain move-
ments, and only its ligand-free X-ray structure is known, then
the traditional approach based on saturation mutagenesis at sites
seemingly aligning the binding pocket may prove to be
problematic. This was the case in the present study. Therefore,
a bioinformatics approach to defining randomization sites likely
to be hot spots was devised in which the first step involves
sequence alignment of eight BVMOs. Guided by the identifica-
tion of high homology sites and the results of induced fit docking
of phenylacetone in the ligand-free X-ray structure, we pro-
ceeded to focus on what appeared to be second sphere residues
(although they may not actually be so!). Since in any enzyme
the number of second sphere residues is much higher than those
directly next to the binding pocket,21 systematic saturation
mutagenesis as in CASTing16 at all of them would entail a labor-
intensive screening effort. In order to circumvent this, an
appropriate choice regarding a select number of such random-
ization sites (hot spots) had to be made. The present study shows
that the focus on proline sites in a loop can be rewarding.30
This choice was made in the expectation that the introduction
of a more “flexible” amino acid in place of proline can change
Experimental Section
Induced Fit Molecular Docking. Commercial software was used
in all cases.31 The ligand phenylacetone was treated using the
program LigPrep of Maestro 8.0 (Schro¨dinger LLC). OPLS_2005
force field was used to optimize the geometry and minimize the
energy. Wide type PAMO crystal structure (PDB code: 1w4x) was
optimized by the Protein Preparation Wizard. The missing hydrogen
atoms were added, bond orders and formal charges were adjusted,
and all the water molecules were removed. After hydrogen bonds
were optimized, the obtained structure was further subjected to
minimize the energy using the default 0.3 Å rmsd tolerance.
The induced fit docking studies were performed using the
Induced Fit Docking protocol of Maestro 8.0. The center of the
Glide enclosing box was the centroid of residues Arg 337, Gln
152, Leu 153, Ala442 and Trp 501 based on the catalytic
mechanism of PAMO. In the first step of Glide ligand docking,
Arg 337 was chosen to temporarily remove the side chains and
replaced with Ala. The Coulomb-vdW scaling factors were changed
to the default values of 0.7 and 0.5 for the protein and ligand,
respectively. A maximum of 20 poses were retained. In the second
phase of docking, the Prime application reversed the temporary
replacement of Arg 337 with Ala, refined nearby residues, and
optimized side chains. In the final docking phase, the ligand was
redocked into all induced fit protein structures that were within 30
kcal/mol of the lowest energy structure using the Glide SP scoring
function. A maximum of 20 poses were needed. Each was ranked
using the composite induced fit score. Each pose was further
analyzed manually to determine the substrate binding mode, the
distance between the flavin C4a atom and the oxygen atom of the
ligand. The best induced fit docking parameters were as follows:
docking score ) -5.05 kcal/mol, IDFScore ) -1355.9 kcal/mol,
prime energy ) -27018.2 kcal/mol. The distance between the
oxygen atom of phenylacetone and C4a atom of FAD was 6.64 Å.
Mutant Library Preparation. The plasmid pPAMO contains
the WT PAMO gene (pamo) under the control of the PBAD
promoter.11a The libraries A, B, C, D, E, F, G and H at the
corresponding sites were created by the QuikChange PCR method27
with the template pPAMO and the primers (Table S2/Supporting
(31) Schro¨dinger Suite 2007 Induced Fit Docking protocol; Glide version
4.5, Schro¨dinger LLC, New York, NY, 2005; Primer version 1.6, LLC,
New York, NY, 2005.
(30) Conversely, proline scanning throughout a loop may likewise constitute
a useful strategy.
9
15430 J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009

Baeyer-Villiger Monooxygenases
A R T I C L E S
Information). The reaction (50 µL final volume) contained: 10 ×
KOD buffer (5 µL), MgCl2 (2 µL, 25 mM), dNTP (5 µL, 2 mM
each), primers (5 µL, 2.5 µM each), template plasmid (2 µL, 10
ng µL-1) and 1 unit of KOD DNA polymerase. The PCR cycle
consisted of an initial denaturation step at 94 °C for 3 min followed
by cycling at 94 °C for 1 min and 72 °C for 14 min for 20 cycles,
then a final elongation for 35 min at 72 °C. The template plasmid
in 26 µL PCR amplification reaction was removed by digestion
with 1 unit of Dpn I (New England Biolabs) in 3 µL of NEB buffer
4 for 2-3 h at 37 °C. The resulting PCR product was used to
transform into electrocompetent Escherichia coli TOP10 cells. The
cells were spread on LB agar plates containing 100 µg mL-1
carbenicilline. For the primers used in site-specific mutagenesis,
see Table S3/Supporting Information.
Library Screening. Individual colonies were placed into 2.2-
mL 96-deep-well plates containing 800 µL of LB media with 100
µg mL-1 carbenicilline by a colony picker QPIX (Genetix, New
Milton, UK). After cell growth at 37 °C overnight with shaking at
800 rpm, 10 µL of each preculture was transferred into a new plate
containing 800 µL of TB media supplemented with 0.1% L-
arabinose as inducer and 100 µg mL-1 carbenicilline. The duplicate
plates were grown for additional 24 h to induce PAMO expression.
The cultures were centrifuged at 4000 rpm and 4 °C for 6 min and
the supernatants were discarded. The original plates were stored.
Each cell pellet was resuspended in 600 µL of 50 mM Tris-HCl
(pH 8.0) containing 1 mg mL-1 lysozyme and 4 units of Dnase I.
Lysis were performed at 37 °C and 800 rpm for 3 h. Cell debris
was precipitated by centrifugation at 4000 rpm and 4 °C for 30
min and 50 µL of each cleared supernatant transferred to a 1.1-mL
96-deep-well plate. Then in each well, 50 µL of the secondary
alcohol dehydrogenase (2°ADH) crude extract (about 10 U),15b 10
µL of 1 mM NADP+, 10 µL of 100 mM rac-1a (or rac-1e) in
acetonitrile and 380 µL of 50 mM Tris-HCl (pH 8.0) containing 5
mM isopropanol were added. The reaction plates were incubated
at 37 °C and 800 rpm for 24 h. Four-hundred microliters of ethyl
acetate was then added to each well, and a plastic cover was used
to cover the plate tightly. The plate was vibrated vigorously to
extract the substrate and product from the solution. After centrifu-
gation, 200 µL of organic layer in each well was transferred into
a new glass-made 96-deep-well plate, and subjected to GC analysis
for screening. Active clones were collected and the results
reproduced. The entire gene of the indentified hits was sequenced
to confirm there were not any other mutations.
Procedure for the Enzymatic Oxidation of Substrates. In a
typical protocol, 1 mL reaction system consists of 2.5 mM substrate,
5 mM isopropanol, 100 µM NADP+, 0.1-0.5 µM purified PAMO
mutant enzyme, secondary alcohol dehydrogenase (2 U), 2.5%
acetonitrile and 50 mM Tris-HCl buffer (pH 8.0). For rac-1b, 1
equivalent of additional ꢀ-cyclodextrin was also added to increase
the solubility. The mixture in 8 mL glass tube with a sealed cap
(to avoid evaporation of 2-alkyl substituted substrates) was shaken
at 200 rpm and 30 °C for times established to control the conversion
rate less than 50% for kinetic resolution. The reaction was stopped,
worked up by extraction with ethyl acetate (containing 200 mg L-1
internal standard as nonane, dodecane, hexadecane and octadecane).
The sample was analyzed by achiral and chiral GC in order to
determine the conversion of and the enantiomeric excess of the
residual ketones and produced lactones. Control experiments were
performed for all the tested substrates with purified WT PAMO
enzymes (about 5 µM). It should be noted that a minimal amount
(<3%) of 2-methylcyclohexanol was produced in the case of
substrate rac-1d. For all the other substrates, no reduction products
were observed. As for oxidation of rac-3, the parallel experiments
were performed with the same concentration of enzymes (WT-
PAMO and mutants, 0.3 µM) and rac-3 (2.5 mM). The other
components in the reaction system were the same as above (without
ꢀ-cyclodextrin) and the reaction time was 10 h.
Starting Substrates (Ketones) and Products (Lactones).
Ketones rac-1g, 1i, and 1j were prepared by alkylation as
described.32a Ketone rac-1b and 1c were prepared by the reported
method,32b and the others purchased from Sigma-Aldrich, Acros,
TGI, and Fluka (used without further purification). The racemic
lactones were prepared by conventional BV-oxidation of the
corresponding ketones using m-chloroperoxybenzoic acid (mCP-
BA). All of the synthesized compounds were isolated by silica gel
1
chromatography and identified by H and 13C NMR.
Identification of Products from Biotransformation. After
extraction as described above, the ketones and lactones were
identified by comparison of their retention times on GC (HP-5 and
the corresponding chiral separation columns) with authentic racemic
compounds.
Determination of the Enantiomeric Excess of Residual
Ketones and Produced Lactones. The ee values of residual ketones
1 and lactones 2 were determined by chiral GC directly except for
ketone 1b and lactone 2b, whose ee values were determined
separately. Table S4 (Supporting Information) lists the details
regarding the chiral separation conditions for each compound tested.
The absolute configurations were made by comparison with
authentic samples and referring to the literature,4b the respective
lactones being prepared by using the recombinant CHMO from
Acinetobacter sp. NCIMB 9871 as a catalyst in the reaction of
ketones 1d-1i and 1l, and P3-PAMO mutant for preparing 2a and
2k.33,15a The (tentative) assignment of the absolute configuration
of lactone 2b-2c and 2j was made on the basis of analogy. All
E-values were calculated using the equation of Sih.34
Steady-State Kinetics. The activities of the purified enzymes
were determined spectrophotometrically by monitoring the decrease
in the level of NADPH over time at 25 °C and 340 nm (ε340 )
6.22 mM-1 cm-1). The reaction mixture (1.0 mL) contained 50
mM Tris-HCl (pH 8.0), 100 µM NADPH, 0.2-4 mM substrate,
2% acetonitrile, 0.1-0.6 µM purified enzyme. The kinetics
measurements were performed on a Molecular Devices Spectramax
Enzyme Expression and Purification. The plasmid pPAMO
(WT and the corresponding mutants) was transformed into Es-
cherichia coli TOP 10 cells and grown at 37 °C in LB media with
100 µg mL-1 carbenicilline. Two mL of overnight cultures of WT
PAMO and mutants in LB media containing 100 µg mL-1
carbenicilline were transferred into 200 mL TB media with 0.1%
L-arabinose as inducer and 100 µg mL-1 carbenicilline, and
incubated at 37 °C with shaking at 250 rpm for 24 h. Cells were
harvested by centrifugation and washed once with 0.9% NaCl
solution. The cell pellets were resuspended in 10 mL 50 mM Tris-
HCl buffer (pH 8.0) and lysed by sonification. The cell debris was
removed by centrifugation at 10 000 rpm for 30 min at 4 °C. The
supernatant was filtered and loaded on a GE Healthcore HisTrap
FF Crude columm (5 mL) pre-equilibrated with 50 mM Tris-HCl
buffer containing 0.5 M NaCl and 5 mM imidazol. Impurity was
removed by imidazol at the concentration of about 25 mM and the
enzyme was eluted by 50 mM Tris-HCl buffer with 0.5 M NaCl
and 200 mM imidazol. The enzyme fraction was desalted and
concentrated by ultrafiltration using Millipore’s Amicon Ultra-15
centrifugal filter devices and then dissolved in 50 mM Tris-HCl
buffer (pH 8.0) and stored at -80 °C. The purity of the enzyme
was measured by SDS-PAGE (Figure S3/Supporting Information).
The concentration of the purified enzyme was determined by
measuring the absorbance at 441 nm using an extinction coefficient
(32) (a) Malosh, C. F.; Ready, J. M. J. Am. Chem. Soc. 2004, 126, 10240–
10241. (b) Xie, J.-H.; Liu, S.; Huo, X.-H.; Cheng, X.; Duan, H.-F.;
Fan, B.-M.; Wang, L.-X.; Zhou, Q.-L. J. Org. Chem. 2005, 70, 2967–
2973.
(33) (a) Stewart, J. D.; Reed, K. W.; Martinez, C. A.; Zhu, J.; Chen, G.;
Kayser, M. M. J. Am. Chem. Soc. 1998, 120, 3541–3548. (b) Berezina,
N.; Kozma, E.; Furstoss, R.; Alphand, V. AdV. Synth. Catal. 2007,
349, 2049–2053.
(34) Chen, C. S.; Fujimoto, Y.; Grirdaukas, G.; Shi, C. J. J. Am. Chem.
Soc. 1982, 104, 7294–7299.
of 12.4 mM-1 cm-1
.
9
J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009 15431

A R T I C L E S
Reetz and Wu
(Molecular Devices GmbH, Germany). The obtained data were
fitted to the Michaelis-Menten equation by nonlinear regression
analysis. For measuring the thermostability of the WT PAMO and
mutants, solutions of the purified enzymes (about 30 µM) were
incubated at 50, 55, 57, and 60 °C for 1 h. The residual activity
was determined as described above in the presence of 2 mM
phenylacetone. The thermostability of the mutant enzymes was also
measured as a function of time at 50 °C, the residual activity being
determined as above.
Supporting Information Available: Induced fit docking model
(Figure S1), picture of sites 437 and 440 in PAMO at which
saturation mutagenesis was applied (Figure S2), choice of codon
degeneracies (Table S1), primers used in generating libraries
A-G (Table S2), primers used in saturation mutagenesis at
position 440 (Table S3), purification of PAMO and mutants
(Figure S3), chiral GC separation conditions (Table S4) and
ee-values of substrates 1a-l and products 2a-l in kinetic
resolution at defined conversions (Table S5). This material is
available free of charge via the Internet at http://pubs.acs.org.
Acknowledgment. Financial aid by the Fonds der Chemischen
Industrie is gratefully acknowledged. We also thank Petra Wede-
mann for synthesizing some of the compounds, Jutta Rosentreter
for the carefully performed chiral GC analyses, and Dr. Despina
Bougiokou for helpful discussions.
JA906212K
9
15432 J. AM. CHEM. SOC. VOL. 131, NO. 42, 2009