D. B. Janssen et al.
2ꢁ103 and 8ꢁ103 clones were obtained on a square LB+Amp
plate after overnight incubation at 378C. From the pooled colonies,
plasmid DNA was isolated and stored as library.
phylogenetic analysis of homologous haloalkane dehalogenas-
es. The resulting site-restricted mutagenesis (SRM) libraries
were constructed with restricted ambiguous codon sets, which
were conveniently selected with a codon search algorithm (Co-
Finder) that will be described in detail elsewhere. To optimize
coverage of any desired set of amino acids at a specific posi-
tion in a library, CoFinder will suggest mixtures of primers,
each with a specific partly undefined (ambiguous codon), that
are used in the same mutagenic PCR reactions.
Library prescreening: Chemically competent E. coli TOP-10 cells
were transformed with library plasmid DNA and plated on pH indi-
cator plates for prescreening. Indicator plates contained eosin
(40 mgLꢀ1), methylene blue (6.5 mgLꢀ1), arabinose (200 mgLꢀ1),
and ampicillin (50 mgLꢀ1).[40] Hydrobromic acid production by indi-
vidual bacterial colonies could be detected after overnight growth
at 378C. For this, plates were incubated for 15 min at room tem-
perature after spotting of bromoethane (600 mL) on a filter paper
that was mounted in the lid of the plates. The plates were sealed
with parafilm and incubated in a fume hood. Bacterial colonies
expressing active haloalkane dehalogenase developed an intense
purple color, and cell material was immediately transferred to 96-
well plates containing LB medium (150 mL) plus ampicillin. After
overnight growth at 378C with shaking, glycerol (30 mL, 50%, v/v)
was added to each well and the plates were frozen at ꢀ808C and
stored at ꢀ208C until further use.
With these strategies we obtained enantiocomplementary
haloalkane dehalogenase variants that catalyze the asymmetric
conversion of TCP, a highly toxic industrial waste compound.
The structure of TCP means that it is a very challenging target,
because any enantioselectivity is caused by different binding
modes of this prochiral substrate in the transition state even
though TCP offers few interactions that restrict its active site
dynamics. The highly diverged variants that were obtained
after five rounds of evolution differ at 25 positions, yet were
discovered by chiral GC screening of just 5500 clones. The best
R- and S-enantioselective mutants carry 13 and 17 mutations,
respectively, relative to DhaA31, and produce the chiral build-
ing blocks (R)-2,3-dichloropropan-1-ol with 90% ee and (S)-2,3-
dichloropropan-1-ol with 97% ee. The approach followed here
thus appears to be an attractive strategy for directed evolution
of enantiodivergent enzymes for the conversion of prochiral
TCP. The approach might be also be useful for other enzymes
in which catalytic properties are governed by subtle differen-
ces in substrate positioning, especially if high-throughput ex-
pression or enzyme performance assays are unavailable.
Screening for enantioselective TCP conversion: Deep-well 96
square well plates (polypropylene, Waters 186002482) containing
LB+amp liquid medium (200 mL) were inoculated from frozen
stocks by use of a pinned stamp (Enzyscreen). After 6 h of growth
at 378C under shaking conditions the log-phase cultures were
induced by addition of LB+Amp (1 mL) containing arabinose
(0.02%) and growth was allowed to continue under the same con-
ditions for 18 h. Pure TCP (3.5 mL) was then added directly to each
culture and the plates were capped and incubated for 5 h under
the same conditions. Both unreacted TCP and produced DCP were
extracted from the cultures with chloroform (220 mL) containing
dodecane (0.05%, v/v) as an internal standard. The organic layer
was carefully transferred to phase separation plates (Whatman 96-
well Unifilter 7720–7229–1) and the filtrate was collected in collec-
tion plates (Corning 3370) containing molecular sieves powder
(4 ꢂ, ꢂ50 mg). Plates were sealed and incubated for 10 min under
shaking conditions and the drying agent was removed by centrifu-
gation at 2500g for 10 min at 168C. The chloroform extract was
transferred to polypropylene 96-deep-well plates (Greiner 780215)
containing glass inserts (300 mL, Waters 150820) and plates were
closed with aluminum seals.
Experimental Section
Molecular modeling: To predict the structure of mutated proteins
accurately,
a computational protocol within YASARA (http://
rounds of optimization. In each round, the mutated side-chains are
energy-optimized in implicit water by sampling discrete side-chain
conformations (rotamers). This is followed by a gradient energy
minimization of the mutated side-chains and the nearby protein in
explicit water. Prior to prediction of the mutant DhaA structures,
TCP was docked in pro-R and pro-S orientations suitable for cataly-
sis into the template PDB file 1BN6[33,42] with the aid of Auto-
dock 4.0 as described earlier.[43,32b] Figures were prepared by use of
Pymol software (http://www.pymol.org).
For chiral GC we used a Shimadzu 17A dual-line GC with a PAL in-
jection system (CTC Analytics), an FID detector, and GC-solutions
software. Separation was performed with a Hydrodex-b-TBDAc
column, isothermic at 1388C. Retention times were 1.83 min (TCP),
3.35 min ((R)-DCP) and 3.50 min ((S)-DCP). Samples from one plate
were injected with time intervals of 4.5 min and run in a single iso-
thermal run, allowing measurement of 384 samples in 24 h.
Cultivation and expression: The dhaA31 gene[27c] was kindly pro-
vided by Prof. Jiri Damborsky (Masaryk University, Brno, Czech Re-
public). The gene was cloned with an N-terminal His6 tag in the
pBAD/Myc-His expression vector with use of NdeI and XhoI restric-
tion sites.[31b] The construct was transformed to and expressed in
E. coli XL10-Gold or TOP10. Cells were grown at 378C in liquid cul-
ture or on plates of Luria–Bertani (LB) medium supplemented with
ampicillin (50 mgmLꢀ1).
Thermostability: Thermal unfolding of mutant enzymes was moni-
tored by the fluorescence-based thermal stability method de-
scribed by Ericsson et al.[45] This method is based on a fluorescence
increase upon binding of Sypro Orange (Life Technologies, Carls-
bad, CA, USA) to hydrophobic protein surfaces that become ex-
posed during unfolding. Increases in fluorescence emission at
575 nm were monitored by excitation at 490 nm with a MyiQ real-
time PCR machine (Biorad) with increasing block temperatures
from 20 to 908C at 18Cminꢀ1. Protein solutions containing 300-
fold diluted Sypro Orange (7.5 mL) in MilliQ water and purified pro-
tein (0.4 mgmLꢀ1, 17.5 mL) in TEMG buffer were prepared in iQ 96-
well real-time PCR plates. After sealing with iQ 96-Well PCR plate
seals (Biorad) the temperature gradient was started. The first deriv-
ative of the measured fluorescence change with temperature was
Library construction: All libraries were constructed by use of Quik-
Change Lightning single-site- or multi-site-directed mutagenic PCR
amplifications (Stratagene). Some modifications of the general pro-
tocol provided with the kit were made. DpnI digestion of parental
DNA was carried out at 378C over 40 min instead of 5 min and ul-
tracompetent cells of the strain XL10-Gold were directly trans-
formed with DpnI-digested PCR product (4 mL). Generally between
146
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2012, 13, 137 – 148