Protein engineering is an important methodology to inves-
tigate and understand the basic function of proteins, but
especially to change their properties as enzymes are fre-
quently used in biocatalysis.[1] Rational protein design (in
which distinct amino acid substitutions are introduced guided
by computer modeling based on the 3D structure of the
protein) and directed evolution (the creation of random
mutant libraries followed by screening or selection to identify
desired variants) are the two major concepts used for protein
engineering. Whereas rational design is limited by the
available information (such as, structure, knowledge about
mechanism, substrate binding mode), directed evolution
approaches are often hampered by the huge sequence space
making high-throughput screening or selection methods a
necessity.[2] Screening is usually performed in 96-well micro-
titer plates and hence most researchers usually analyze only a
few thousand clones per directed evolution round. Selection
methods allow for a much higher throughput (105–108 clones),
but are mostly restricted to problems, where complementa-
tion of a key step in the metabolism needs to take place. One
exception is the use of in vitro compartmentalization (IVC) or
single-cell compartmentalization in combination with fluo-
rescence activated cell sorting (FACS),[3] but this requires
in vitro protein biosynthesis of the enzyme of interest (in case
of IVC), stable substrates over the entire procedure, and the
generation of a fluorescent product. Moreover, only enzymes
with novel activities can be discovered, but the identification
of variants with improved properties is difficult to establish
with this system. Another recently published alternative is the
use of cell surface display in combination with FACS.[4]
Although this method could successfully be used to identify
more enantioselective variants of an esterase or lipase, the
required biotin tyramides are laborious to synthesize and the
protein must be processed and displayed in an active form on
the surface of the cell.[4]
growth (due to release of carbon source; glycerol) or by
monitoring of the pH shift (due to release of acid).[5]
Possibly the most useful property of enzymes in biocatal-
ysis is their enantioselectivity (E value). However, enzymes
often do not display the desired enantioselectivity towards
industrially interesting chiral compounds. Consequently
improvement of enantioselectivity[6] or even inversion of
enantiopreference[7] have been extensively studied by
directed evolution, but solely by screening in microtiter
plates. Alternatively, inhibition of enzymes by enantiomers of
sulfoxides was used for enantioselective screening, as one
enantiomer was a better inhibitor than the other.[8] Selective
binding of antibodies was also described as principle to
determine the enantiomeric excess.[9]
One concept to establish an in vivo selection system to
discover mutants with altered enantioselectivity is based on
linking the survival of the microbial host with one enantiomer
and to cause cell death by the opposite enantiomer. Thus, with
this “carrot and stick” approach desired variants should, in
principle, be accessible because only surviving cells need to be
further analyzed. Reetz and Rꢀggeberg could show that for
yeast strains in the presence of either an acetic acid ester of
pantolactone (supporting growth) or a fluoroacetic acid ester
of pantolactone (causing cell death) differential growth
occurs, but no improvement in ee value was reported.[10] In a
follow-up study, they could further refine the principle and
identified variants of lipase CAL-B with slightly inverted but
not high enantioselectivity towards 1,2-O-isopropylidene
glycerol (IPG, WT (wild type): E = 1.9, favoring the (R)-
enantiomer to mutants with E = 3–8 favoring the (S)-enan-
tiomer).[11] Similarly, Quax and co-workers[12] investigated a
lipase from Bacillus subtilis using IPG, but linked either to
aspartate or a phosphonate. Release of the amino acid
supported growth of an aspartate auxotroph E. coli strain
whereas the phosphonate lead to inhibition of the lipase. They
also reported a mutant with inverted enantioselectivity (up to
73% ee); improved E-values were not reported.
Another limitation in typical directed evolution experi-
ments is that all variants generated in a library need to be
investigated. In some cases only the active mutants are
studied, but still the majority of them do not possess the
desired property and hence time and consumables are spent
on them. A genetic selection system would enable this
limitation to be overcome and allow the investigation of a
much larger sequence space. In previous work, we could
successfully establish an agar plate based selection method to
identify variants of an esterase from Pseudomonas fluorescens
(PFE) capable of hydrolyzing a sterically hindered 3-hydrox-
yester. Active mutants can be identified by either enhanced
In both examples, the throughput was very low (80 or
2500 colonies, respectively) as selection was still based on
agar plate screening. In addition, the success was only partial,
as the mutants identified displayed only moderate enantiose-
lectivity (E < 10).
Herein, we report an in vivo selection method coupled
with flow cytometric[13] analysis to allow ultra-high through-
put identification of esterase variants with altered enantiose-
lectivity in the kinetic resolution of 3-phenyl butyric acid by
coupling one enantiomer of the carboxylic acid to either
glycerol—serving as carbon source—or 2,3-dibromopropa-
nol[14]—serving as toxic compound (Scheme 1). The use of this
pair has the additional advantage, that both alcohols are
highly similar with respect to steric demands and hence the
important requirement “you get what you screen for” for
successful directed evolution experiments is fulfilled too.
Further advantages of our selection system are that no
auxothroph E. coli strains are needed and the substrates are
cleaved by intracellularly expressed enzymes; hence a surface
display technique is not required.
[*] Dr. E. Fernꢀndez-ꢁlvaro, Dr. R. Snajdrova, Dr. H. Jochens,
Dipl.-Biol. T. Davids, Dr. D. Bçttcher, Prof. Dr. U. T. Bornscheuer
Institute of Biochemistry, Dept. of Biotechnology &
Enzyme Catalysis, Greifswald University
Felix-Hausdorff-Strasse 4, 17487 Greifswald (Germany)
E-mail: dominique.boettcher@uni-greifswald.de
[**] We thank the Deutsche Bundesstiftung Umwelt (grant AZ 13198)
and the DFG (grant Bo1862/4-1) for financial support.
Incubation of this pseudo-racemic mixture with E. coli
liquid cultures containing the esterase mutant libraries should
Supporting information for this article is available on the WWW
Angew. Chem. Int. Ed. 2011, 50, 8584 –8587
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
8585